technical sessions - ipu.ru

TECHNICAL SESSIONS

Monday, 8:30-10:00

� MA-01Monday, 8:30-10:00RB-Alfa

Keynote Lecture: Professor IgnacioGrossmannStream: Keynote, Tutorial and Plenary LecturesKeynote sessionChair: Vincent Mousseau, LGI, Ecole Centrale Paris, Grande voiedes vignes, 92 295, Chatenay Malabry, France,[email protected]

1 - Challenges in the Application of Mathematical Pro-gramming in the Enterprise-wide Optimization ofProcess IndustriesIgnacio Grossmann, Chemical Engineering, Carnegie MellonUniversity, 5000 Forbes Ave, 15217, Pittsburgh, PA, UnitedStates, [email protected]

Enterprise-wide optimization (EWO) is a new emerging area that liesat the interface of chemical engineering and operations research, andhas become a major goal in the process industries due to the increasingpressures for remaining competitive in the global marketplace. EWOinvolves optimizing the operations of supply, production and distribu-tion activities of a company to reduce costs and inventories. A majorfocus in EWO is the optimization of manufacturing plants as part of theoverall optimization of the supply chain. Major operational items in-clude production planning, scheduling, and control. This talk providesan overview of major modeling and computational challenges in thedevelopment of deterministic and stochastic linear/nonlinear mixed-integer optimization models for planning and scheduling for the opti-mization of plants and entire supply chains that are involved in EWOproblems. We illustrate the application of these ideas in four majorproblems: a) integration of planning and scheduling in batch processesthat lead to large-scale mixed-integer linear programs, b) optimizationof responsive process supply chains that lead to large-scale bicriterionmixed-integer optimization problems, c) optimization of distribution-inventory planning of industrial gases that lead to integration of plan-ning and vehicle-routing models, d) optimization of oilfield infrastruc-tures under uncertainty that lead to multistage stochastic programmingproblems with endogenous uncertain parameters. We outline the so-lution methods that have been developed. Also, these problems havebeen addressed in collaboration with industry, and have led to substan-tial economic savings.

� MA-02Monday, 8:30-10:00RB-Beta

Scheduling Applications

Stream: SchedulingInvited sessionChair: Malgorzata Sterna, Institute of Computing Science, PoznanUniversity of Technology, Piotrowo 2, 60-965, Poznan, Poland,[email protected]

1 - Scheduling chains of divisible computationsMaciej Drozdowski, Dept. of Computer Science, Institute ofComputing Science, Poznan Univeristy of Technology,Piotrowo 2, 60-965, Poznan, Poland,[email protected], Joanna Berlinska

We analyze scheduling chains of divisible computations. In the chain,one application produces input load for the next one. MapReduce ap-plications are example of 2-stage chains. We propose a divisible load

model and give an algorithm for scheduling multilayer divisible com-putations. The algorithm is tested experimentally. Structures of theobtained schedules and performance of such applications are studied.

2 - Scheduling malleable tasks with arbitrary processingspeed functionsMaciej Machowiak, Poznan University of Technology,Poland, [email protected], Mikhail Y.Kovalyov, Jan Weglarz, Maksim Barketau

The problem of scheduling n tasks in a system with m processors tominimize the makespan is considered. Task are malleable when can beexecuted by several processors at a time, its processing speed dependson the number of assigned processors, and a set of processors assignedto the same task can change over time. The processing speed of a taskis a strictly increasing function of the number of processors allocatedto this task. We present results for arbitrary n and m: polynomial timealgorithms for convex and concave functions and an enumerative algo-rithm for arbitrary functions.

3 - Fast Truck-Packing of 3D boxesGrzegorz Pawlak, Institute of Computing Science, PoznanUniversity of Technology, ul. Piotrowo 2, 60-965, Poznan,Poland, [email protected], JoannaJozefowska, Erwin Pesch, Dawid Kowalski, Michal Morze

The practical 3D container packing problem was considered. Theproblem was formulated for the real packing constrains appearing inthe sales and logistics departments of household equipment factory.The formal MIP formulation has been presented and fast heuristic al-gorithm was proposed. The appropriate computational experiment hasbeen performed. The solution can be implemented in the real containerpacking problem for the rectangular boxed products.

4 - Genetic Algorithm Supporting Supply Process inCharitable OrganizationsMalgorzata Sterna, Institute of Computing Science, PoznanUniversity of Technology, Piotrowo 2, 60-965, Poznan,Poland, [email protected], MateuszCichenski, Mateusz Jarus, Michal Miskiewicz, JaroslawSzymczak

We study the optimization problem arising in charitable organizationsduring supply process. Such institutions are especially interested inminimizing the total cost of purchase which consists of the prices atwhich particular products are bought and the cost of their transporta-tion. We present the formal mathematical model of the problem, whichis NP-hard, and propose list heuristics as well as the genetic algorithmsolving it. The efficiency of implemented methods was checked inextensive computational experiments.

� MA-03Monday, 8:30-10:00RB-L1

Business Excellence in Logistics:Intelligent and Interactive Routing

Stream: Business Excellence in LogisticsInvited sessionChair: Martin Josef Geiger, Logistics Management Department,Helmut-Schmidt-University, Holstenhofweg 85, 22041, Hamburg,Germany, [email protected]: Murat Koksalan, [email protected]

1 - An Interactive Approach for Multi-objective Route Se-lection ProblemDiclehan Tezcaner, Industrial Engineering, Middle EastTechnical University, Endustri Muhendisligi Bolumu, OrtaDogu Teknik Universitesi, 06800, Ankara, Turkey,[email protected], Murat Koksalan

1

MA-04 EURO 25 - Vilnius 2012

Multi-objective (MO) route selection problem is a combination of twocombinatorial problems: MO shortest path problem (finding the effi-cient paths between the nodes to be visited) and MO traveling salesper-son problem (finding the efficient tours made up of the efficient paths).We develop an interactive approach that helps in finding the best solu-tion of a decision maker who has a quasiconcave utility function. Wedemonstrate our approach on the routing problem for unmanned airvehicles with minimization of distance traveled and minimization ofradar detection threat objectives.

2 - Delivery Scheduling Using IntelligenceGulgun Kayakutlu, Industrial Engineering, Istanbul TechnicalUniversity, Macka, 34367, Istanbul, Turkey,[email protected], Gokcen TurkelDynamic scheduling is an indispensable need in today’s businessworld.To satisfy the customer,on-time delivery is crucial for thebusiness to continue in spite of continuous change.This study aimsto propose a model to predict the optimum time of van deliver-ies.Factors influencing the delivery process are depicted from vari-ous researches.Prediction is modeled as an artificial neural networkusing Back Propagation.The model is applied in a case study of vandeliveries.Achievements of this study will not only contribute to theresearch,but will also open a new angle in the logistics vision.

3 - A holistic optimization problem for aviation commu-nity noiseDaniel Zachary, CRTE, CRP Henri Tudor, 66, rue deLuxembourg, L-4421, Esch sur Alzette, Luxembourg,[email protected], Ulrich LeopoldIn our presentation, we give an overview of the Luxembourg Avia-tion community noise situation and discuss a holistic approach to noiseabatement and cost sharing. The problem is framed as a multi-indicatorassessment and is solved using a minimal cost objective. Recent pastresults are shown (Luxembourg Aviation Community Noise - LANproject, 2007-2009), a cooperative project with the air freight com-pany Cargolux Airlines International S.A.. New ideas are discussed,including the tuning of the aircraft scheduling, routes, aircraft type,operational procedures and a 3D-trajectory problem.

� MA-04Monday, 8:30-10:00RB-L2

Optimization of shared urbantransportation systems

Stream: Optimization in Public TransportInvited sessionChair: Gonçalo Correia, Department of Civil Engineering,University of Coimbra, Faculdade de Ciências e Tecnologia. Dep.Eng. Civil, sala SA 2.8 Rua Luis dos Reis Santos, 3030, Coimbra,Portugal, [email protected]

1 - Strategic design of public bicycle sharing systemswith service level constraintsJenn-rong Lin, Department of Transportation Science,National Taiwan Ocean University, 2, Pei-Ning Road, 20224,Keelung, Taiwan, [email protected], Ta-hui YangWe formulate and analyze a strategic design model for public bicyclesharing systems with service level considerations. In considering theinterests of users and investors, the proposed model attempts to deter-mine the number and locations of bike stations, the network structureof bike paths connected between the stations, and the travel paths forusers between each pair of origins and destinations. A numerical ex-ample is created to illustrate the proposed model. Sensitivity analysisis performed to gain better insights into knowing how important pa-rameters affect the design of the system.

2 - Testing the validity of the MIP approach for locatingcarsharing stations in one-way systemsDiana Rita Ramos Jorge, Department of Civil Engineering,University of Coimbra, Rua Luis dos Reis Santos, 3030-788,Coimbra, Portugal, [email protected], GonçaloCorreia, Cynthia Barnhart

Previous research has proposed using a mixed integer programming(MIP) model for choosing the location of stations in one-way carshar-ing systems as an approach to solve vehicle stock imbalance. However,MIP models have difficulties including reality-based factors that canhave a substantial impact on carsharing. Thus we test the validity ofthose solutions developing a simulation model including demand vari-ability and relocation operations. The first tentative results show thatthese factors influence the company profit, affecting significantly theperformance of the deterministic solutions.

3 - Assessing the profitability of carsharing fleet distri-butions based on usage patternsMathias Goeppel, GR/PAP, Daimler AG, Building 10, Room2.3.036, Wilhelm-Runge-Str. 11, 89081, Ulm, Germany,[email protected]

One-way carsharing as seen with car2go, where vehicle pick-up anddrop-off locations are not predefined, challenges the operator. The pre-dominant difficulty lies in matching vehicle locations with customerdemand, as well as the development of solutions for fleet maintenance.This article focuses on the formulation of a reliable measure of prof-itability for the involved vehicles. The measure is based on a time-spatial decomposition of customer data. It enables the operator to de-cide whether specific areas can be run profitably and allows for ad-vanced scheduling of the service agents.

4 - A shared-taxi agent based model. Testing severalmarket configurations for the Lisbon municipalityGonçalo Correia, Department of Civil Engineering, Universityof Coimbra, Faculdade de Ciências e Tecnologia. Dep. Eng.Civil, sala SA 2.8 Rua Luis dos Reis Santos, 3030, Coimbra,Portugal, [email protected], Luis Martinez, José Viegas

This paper presents the conceptual design of a new agent-based modelto simulate the operation and performance of shared taxi systems. Theproposed model incorporates a matching algorithm between the sup-ply and demand, finding the most adequate option, considering multi-ple objectives, as minimum travel time for passengers and balance innumber of services for taxis. The paper will also evaluate the severalpossible market configurations, measuring the performance of the sys-tem from an operator, client and society perspective, finding the mostbalanced solution in terms of fleet size.

� MA-05Monday, 8:30-10:00RB-L3

Maritime fleet size and mixStream: Maritime TransportationInvited sessionChair: Trond A. V. Johnsen, MARINTEK, SINTEF, Otto Nielsenveg 10, 7052, Trondheim, Norway,[email protected]

1 - Containership speed and fleet size optimisation withsemi-elastic demand: an application to northerneurope-south america tradeAli Cheaitou, College of Engineering - Industrial Engineeringand Management Department, University of Sharjah, SharjahUniversity City, 27272, Sharjah, United Arab Emirates,[email protected], Pierre Cariou

Increasing the number of vessels in a container liner service while re-ducing speed, known as slow steaming strategy, is modelled in thispaper. We focus on the case of optimal speed under semi-elastic de-mand, for which containerized perishable product transport is sensitiveto time. This paper investigates if slow steaming is still optimal whenworking to maximise the total profit on the cycle. Furthermore, a nu-merical application is carried out for a direct Northern Europe to EastCoast of South America container service, a route selected due to thehigh volume of perishable products.

2

EURO 25 - Vilnius 2012 MA-07

2 - On solving the feeder containership network designproblem via metaheuristic approachesOlcay Polat, Production Management, TU Berlin, TU berlin(H95-H9157), Strasse des 19. Juni 135, 10623, Berlin, Berlin,Germany, [email protected], Hans-Otto Guenther, OsmanKulak

Feeder ports serve as regional hubs from where containers are dis-tributed in the respective region by use of feeder ships. The feeder net-work design problem (FNDP) determines the smallest ship fleet sizewith routes to minimize total travelled distance with time, capacityand service frequency restrictions. In this study, a mathematical modelis proposed and metaheuristic approaches are developed to solve theFNDP. The developed approaches have been tested using case studiesfrom Turkey. The numerical results of the study show that Turkishports have great potential as hub ports in its region.

3 - Fleet size and mix in liner shippingTrond A. V. Johnsen, MARINTEK, SINTEF, Otto Nielsenveg 10, 7052, Trondheim, Norway,[email protected], Lars Magne Nonås, KjetilFagerholt, Bjørn Egil Asbjørnslett

Fleet size and mix is a core decision for all shipping companies. In anygiven market situation, the number of vessels and the variety of vesseltypes will determine the profit potential for the company. We presentan optimization model for improved support for strategic fleet size andmix decision-making. The model has been implemented and tested incooperation with a world-leading liner shipping company.

� MA-06Monday, 8:30-10:00RB-Gamma

Matheuristics in Transportation

Stream: MatheuristicsInvited sessionChair: Fabien Tricoire, Department of Business Administration,University of Vienna, Chair for Production and OperationsManagement, Brünner Straße 72, 1210, Vienna, Austria,[email protected]

1 - Fresh bovine skin transportation from slaughters totannery with special time windows and capacity con-straintsGiorgio Romanin-Jacur, Management and Engineering,University of Padova, Stradella San Nicola, 3, 36100,Vicenza, Italy, [email protected], Carlo Filippi

On a road network a tannery, slaughters and truck depots are placed.Different capacity trucks start from depots, load fresh bovine skins sup-plied by slaughters and unload them at the tannery. Time windows areimposed to truck starting, skins’ loading and skins’ unloading; skinpreservation times and capacity constraints shall be respected. Totalcost,due to truck utilization and to paths and times shall be minimized.Quick and easy low cost solution algorithms are required. A mixed in-teger programming model has been built and both exact and heuristictechniques are suggested and tested.

2 - Freight train composition with minimum shuntingoperationsUwe T. Zimmermann, Institute of Mathematical Optimization,TU Braunschweig, Pockelsstrasse 14, 38106, Braunschweig,Germany, [email protected], Ronny Hansmann

Planning freight train schedules in dense rail networks provides anenormous challenge. Resulting optimization models include a tremen-dous number of eligible train routes and departure times subject totight infrastructure requirements. From a BMBF-supported cooper-ation with Deutsche Bahn, we outline first results focusing on traincomposition. OD-routed rail cars are assigned to scheduled trains.Optimal locations of rail cars within trains minimize the number ofthe expensive shunting operations in rail yards. We present MIP-models,heuristics,and some computational experience on DB-data.

3 - Hybrid heuristics for solving a maritime short sea in-ventory routing problemAlexandrino Delgado, Matemática, Universidade de Aveiro,Campus Universitário de Santiago, 3810-193, Aveiro,Portugal, Portugal, [email protected], Agostinho Agra,Marielle Christiansen, Luidi Simonetti

We consider a fuel oil distribution problem where an oil company isresponsible for the routing and scheduling of ships between ports suchthat the demand for various fuel oil products is satisfied during theplanning horizon. The objective is to determine distribution policiesthat minimize the total cost (routing and operations costs), while in-ventory levels are maintained within their limits. We propose an arc-load flow formulation of the problem. In order to obtain good fea-sible solutions for several months we propose and compare differenthybridization strategies.

� MA-07Monday, 8:30-10:00RB-Eta

Novel and emerging VRPs

Stream: New Trends in Vehicle RoutingInvited sessionChair: Jorge E. Mendoza, Institute of Applied Mathematics,Université Catholique de l’Ouest, 3 Place André Leroy, 49008,Angers, France, [email protected]

1 - A Multiple Plan Approach for the Dynamic TechnicianRouting and Scheduling ProblemVictor Pillac, Automatique - Productique, Ecole des Mines deNantes, 4 Rue Alfred Kastler, La Chantrerie, 44300, Nantes,France, [email protected], Christelle Gueret, AndresMedaglia

The Dynamic Technician Routing and Scheduling Problem (DTRSP)deals with a crew of technicians that serves dynamically appearing re-quests. In the DTRSP, each technician has a set of skills, tools, andspare parts, while requests require a subset of each. The problem isthen to design a set of tours of minimal total duration such that eachrequest is visited exactly once, within its time window, by a compat-ible technician, and to dynamically insert new requests into existingtours. We propose a Multiple Plan Approach to solve the DTRSP andillustrate its performance on benchmark instances.

2 - Capacitated Vehicle Routing Problem with LoadingConstraints: A Case in the UKNasrin Asgari, Mathematics, University of Portsmouth,United Kingdom, [email protected], Xiang Song,Dylan Jones, Tim Pigden

Recently it has been raised to solve Capacitated vehicle routingproblem (CVRP) considering loading and unloading conditions likesequential loading (Last-In First-Out), vehicle capacity and itemsfragility. We have dealt with a real case in the UK that distributesconsumer items like dairy, food and fruit. Each depot covers 100-150customers. The case is solved with respect to existing limitations in-cluding loading constraints; drivers rule, shift and pattern; and cus-tomer time-window. Order splitting is allowed in the case, whilst nospilt on deliveries has been allowed in the literature.

3 - The Minimum Duration Truck Driver SchedulingProblemAsvin Goel, Telematique.eu, Germany, [email protected]

Truck driver scheduling problems are important subproblems of real-life vehicle routing and scheduling problems because rest periods re-quired by government regulations have a significant impact on traveland arrival times. This paper identifies common constraints imposedby hours of service regulations world wide and presents a mixed inte-ger programming formulation for the truck driver scheduling problem.A dynamic programming approach is presented and its effectiveness isdemonstrated for working hour regulations in the United States and inthe European Union.

3


4 - A GRASP with heuristic concentration for the vehiclerouting problem with stochastic demandsJorge E. Mendoza, Institute of Applied Mathematics,Université Catholique de l’Ouest, 3 Place André Leroy,49008, Angers, France, [email protected], Juan G.Villegas

The vehicle routing problem with stochastic demands consists in de-signing transportation routes of minimal expected cost to satisfy aset of customers with random demands of known probability distri-butions. This paper proposes a GRASP that uses randomized route-first, cluster-second heuristics, a variable neighborhood descent, anda post-optimization procedure specially tailored for the problem. Ex-periments on standard instances show that our approach is competitivewith the state-of-the-art method in terms of both accuracy and effi-ciency

� MA-08Monday, 8:30-10:00RB-Epsilon

Network Optimization 1

Stream: Network OptimizationInvited sessionChair: Ana Bautzer, ISCAL / CIO, Av. Miguel Bombarda, 20,1069-035, Lisboa, Portugal, [email protected]

1 - Recent Developments in the Ring Spur AssignmentProblemPaula Carroll, Management Information Systems, UCD,Quinn Business School, Belfield, 4, Dublin,[email protected], Bernard Fortz, Martine Labbé, SeánMcGarraghy

We consider the Ring Spur Assignment Problem (RSAP), an NP-hardsurvivable telecommunications network design problem. It arises inthe context of exploiting existing resources efficiently in strategic net-work development and complements the Sonet Ring Assignment Prob-lem (SRAP) of determining the set of link capacities to install. We pro-pose a complete IP formulation and present recent developments andthe results of a cutting plane implementation on benchmark test data.

2 - ILP Models for the PON Access Network DesignProblemMaria João Lopes, Departamento de Métodos Quantitativos,University Institute of Lisbon (ISCTE - IUL) and CIO, Av.das Forcas Armadas, 1649-026, Lisboa, Portugal,[email protected], Luis Gouveia, Amaro de Sousa

A PON is an optical access network connecting a Central Office to aset of terminals using splitters installed on intermediate nodes. In thePON access network design problem, terminals are clustered in a min-imum number of PONs and we have to define for each PON where toinstall splitters and how to connect all elements through fibres. Thereare costs associated with intermediate nodes, splitter types and fibreconnections. We define the minimum cost design problem in the con-text of densely populated urban areas, proposing different formulationsand valid inequalities for this problem.

3 - Robust optimization of optical fiber access networksdeploymentsCédric Hervet, Orange Labs/CNAM, France,[email protected], Marie-Christine Costa, FayeAlain, Matthieu Chardy, Stanislas Francfort

Due to the recent increase in bandwidth requirements, telecommuni-cation operators have to plan the deployment of optical fiber networksthrough Gigabit Passive Optical Network technology (FTTH GPON).One great challenge is to design this network while not knowing whoand where the future subscribers will be. We focus on the problem ofthe robust optical fiber network deployment under demand uncertainty.A two-stage robust optimization model is proposed for this problem,as well as two robust solution methods extending classical results fromBen-Tal et al. and Babonneau et al.

4 - Models for a Steiner multi-ring network design prob-lem with revenuesAna Bautzer, ISCAL / CIO, Av. Miguel Bombarda, 20,1069-035, Lisboa, Portugal, [email protected], LuisGouveia, Ana Paias, José PiresThe Steiner multi-ring network design problem with revenues is theproblem of designing a set of rings that pass through a central depotand through all the customers in a set R and some of the customers ina set S. Besides the usual arcs link costs, revenue profits between eachpair of customers in the same ring are also considered. The objectiveis to minimize the difference between the total routing cost and totalprofit. We introduced some integer programming formulations for theproblem and compared their linear relaxations values. Some valid in-equalities are also presented.

� MA-09Monday, 8:30-10:00RB-Zeta

Cutting and Packing 1

Stream: Cutting and PackingInvited sessionChair: A. Miguel Gomes, INESC TEC, Faculdade de Engenharia,Universidade do Porto, Rua Dr. Roberto Frias s/n, 4200-465,Porto, Portugal, [email protected]

1 - Heuristics for the three-dimensional loading capaci-tated vehicle routing problemLeonardo Junqueira, Department of Production Engineering,Federal University of São Carlos, Rua Dona Alexandrina,Num. 2471, Apto. 8, 13566-290, São Carlos, São Paulo,Brazil, [email protected], Reinaldo MorabitoIn this work we present heuristics for the capacitated vehicle routingproblem with three-dimensional loading constraints. The problem con-sists in finding minimum cost delivery routes for a set of identical vehi-cles that, departing from and returning to the depot, visit all customersonly once, while considering constraints related to cargo stability, loadbearing and multi-dropping. The routing and loading modules of themethod are built on classical heuristics of the vehicle routing and con-tainer loading literatures. The results of some computational experi-ments are presented and discussed.

2 - A heuristic for the vehicle routing problem with back-hauls and three-dimensional loading constraintsAndreas Bortfeldt, Dept. of Information Systems, Universityof Hagen, Profilstrasse 8, 58084 Hagen, Germany, 58084,Hagen, FR Germany, Germany,[email protected], Lars MoenchIn the vehicle routing problem with backhauls (VRPB) there are de-livery and pick up customers. We consider the 3L-VRPB, i.e. a com-bination of the VRPB and 3D loading. The given vehicles have a 3Dloading space and the cargo of a customer consists of 3D boxes. Aheuristic is proposed including a variable neighborhood search algo-rithm for routing and a tree search algorithm for packing boxes. Re-sults are presented for 95 test instances.

3 - Inter-depot transportation planning involving palleti-sation and truck loading.Maria Teresa Alonso Martínez, Department of mathematics,University of Castilla-La Mancha, Edificio Infante Don JuanManuel, Avda.de España s/n, 02071, Albacete, Spain,[email protected], Ramon Alvarez-Valdes,Joaquim Gromicho, Francisco Parreno, Gerhard F. Post, JoseTamaritIn a large logistic company the inter-depot planning problem has tobe solved every day. The products are placed on pallets and then thepallets loaded into trucks. The main objective is to send the fewestnumber of trucks, covering the demand of each depot and satisfyinga number of constraints concerning the way in which pallets are builtand placed into the truck. Our proposal is to solve the problem usinga metaheuristic in two phases, one for building a feasible solution andthe other for improving the solution. In both phases palletisation andtruck loading are simultaneously considered.

4


4 - A study on multiobjective rectilinear packing prob-lemsA. Miguel Gomes, INESC TEC, Faculdade de Engenharia,Universidade do Porto, Rua Dr. Roberto Frias s/n, 4200-465,Porto, Portugal, [email protected], Marisa Oliveira, Ma

Eduarda Pinto Ferreira

In many application areas, like in Facility Layout and in VLSI de-sign, one wishes to pack, without overlap, a set of rectilinear polygonswhile minimizing simultaneously more than one objective, leading tothe Multi-objective Rectilinear Packing Problem. The most commonobjectives are the minimization of the enclosing rectangle area andthe minimization of the sum of the distances between a subset of thepieces. The purpose of this work is to study and analyze the rela-tionship between these two objectives, i.e. to study if they really areconflicting objectives.

� MA-10Monday, 8:30-10:00RB-Theta

Applications

Stream: Production and the Link with Supply ChainsInvited sessionChair: Hicham Chehade, ICD- LOSI, University of Technology ofTroyes, 12 Rue Marie Curie, 10000, Troyes, France,[email protected]: Lionel Amodeo, Charles Delaunay Institute, University ofTechnology of Troyes, 12 Rue Marie Curie BP2060, 10000,Troyes, France, [email protected]

1 - Determining means of serial processors and param-eters of lot rectifying inspection planAshraf El-Ga’aly, Systems Engineering, King FahdUniversity of Petroleum & Minerals, KFUPM P.O. Box 5008,31261, Dhahran, Eastern, Saudi Arabia,[email protected], Shokri Selim

We consider determining the optimal means of a set of machines inseries. Production is in lots. A lot is processed on each machine se-quentially. After each machine, a sample of the lot is examined. If thenumber of defectives does not exceed a threshold the lot is forwardedto the next machine; otherwise, the lot is rectified and sold. Currentsolution methods are slow and may not get an optimal solution. Weexploit a special structure of the problem that is used to design a fastsolution method. We also determine the parameters of the rectificationinspection plan for each machine.

2 - High speed methods for volume computation andapplications to robust optimization in Supply ChainManagementRahul Paul, Computer Science, International Institute OfInformation Technology, Bangalore, 26/C, Electronic City,Hosur Road, 560100, Bangalore, Karnataka, India,[email protected], G. N. Srinivasa Prasanna, KundanKumar, Subhrajit Debnath

Randomized polynomial time polyhedral volume algorithms (Lovasz-Vempala, 2003) are complex and difficult to implement. We extend ex-act algorithms (Vinci, B. Büeler and K. Fukuda), to introduce a cheby-ball variant of Direct Monte Carlo with affine transformations, extend-ing their applicability from 9 to 20 dimensions. The volume is used inan information theoretic framework to estimate the information con-tent under uncertainty of a polytope/scenario in a robust optimizationframework for Supply Chain Management.

3 - Optimal replenishment policy for deteriorating itemswith time dependent demandShivraj Singh, Mathematics, D.N. College, Shivraj Singh,Dept. of Mathematics„ D.N.(P.G.) College, Meerut, U.P.,250001, Meerut, Uttar Pradesh, India,[email protected]

This paper considers the impact of learning on set up cost on the Eco-nomic Order Quantity model for deteriorating items subject to trape-zoidal type consumption rate and shortages. Deterioration models con-sider gradual reduction in quality and quantity of products over time.In this study the deterioration rate is a Weibull distribution functionof time. A model where each cycle begins with shortages and endswithout shortages is proposed. The model is developed over an infinitetime horizon to determine optimal replenishment policy by minimizingtotal system running cost.

4 - Optimization of the processing times of cyclically re-peated intersecting operation setsBoris Rozin, Operations research laboratory, United Instituteof Informatics Problems of NAS of Belarus, Surganov Str.,6,220012, Minsk, Belarus, [email protected], GenrikhLevin

A cyclically repeated sequence of intersecting sets of operations is ex-ecuted in a system and the operations of each set are activated simulta-neously. The expenses for each operation decrease with the increase ofits processing time, the expenses for the whole cycle are proportionalto its time. The problem is to determine the processing times of theoperations to minimize the total expenses under the constraint on thecycle time. The proposed method uses decomposition techniques anddynamic programming. Examples of such systems are presented.

� MA-11Monday, 8:30-10:00RB-Iota

Inculsion of foreign transportationressourcesStream: Transportation PlanningInvited sessionChair: Jörn Schönberger, University of Bremen, 28359, Bremen,Germany, [email protected]

1 - Heuristic Strategies for Bundle Bidding in TransportAuctionsTobias Buer, Department of Business Studies & Economics,Chair of Logistics, University of Bremen, Wilhelm-Herbst-Str.5, 28359, Bremen, – Please Select (only U.S. / Can / Aus),Germany, [email protected], Herbert Kopfer

The procurement of transport services via a combinatorial auction isconsidered from a freight carrier’s point of view. The freight carrierhas to decide on which subsets of the tendered freight requests to placea bid. This problem is denoted as the bid generation problem (BGP).To solve the BGP, we propose heuristic bidding strategies based onsynergies between requests. By means of a computational study, thestrategies are compared to a brute-force strategy which generates allpossible bundle bids. Based on this comparison, trade-offs are dis-cussed.

2 - Benefits of a heterogeneous fleet — the ecologicalperspectiveHerbert Kopfer, Department of Business Studies &Economics, Chair of Logistics, University of Bremen,Wilhelm-Herbst-Strasse 5, 28359, Bremen, Germany,[email protected], Heiko Kopfer, Jörn Schönberger

Road haulers are more and more enforced to integrate emission-relatedissues in their daily operations planning. We present an innovative ap-proach that aims at exploiting the ecologic benefits of a heterogeneousfleet. Vehicles are classified by their fuel consumption with respect totheir dead weight and actual payload. Here, the linearity between trav-elled distances and cargo weight becomes void. We demonstrate thepredominance of a heterogeneous fleet over a homogeneous fleet withrespect to the reduction of green house gas emissions.

5


3 - Collaborative Transportation Operations Planning ina Real-world Scenario - Handling of BackhaulsAndrea Nagel, Information Systems Research Group,FernUniversität - University of Hagen, 58084, Hagen,Germany, [email protected], Giselher Pankratz,Hermann Gehring

In this contribution we examine a cooperation of four producers in thefood and beverages industry, which decided to coordinate their distri-bution activities by inter-organisational transportation planning. Wemodeled this scenario as a Rich Vehicle Routing Problem. One of thereal-world restrictions is the occurrence of backhauls. Different needsfor the transportation of the goods induce different backhaul policiesthat have to be covered by the model. We identify and characterizethese policies and propose a strategy for a improved handling of back-hauls in a real-world-scenario.

4 - Enrichment of Mobile Freight Auctions with LocationSensitivity: Systemic Design ConsiderationsDimitrios Emiris, Industrial Management & Technology,University of Piraeus, 80 Karaoli & Dimitriou Street, 18534,Piraeus, Greece, [email protected], Charis Marentakis

The use of the Auction Classification Ecosystem (ACE) recently pre-sented enabled the systemic design of many auction types, includingB2B ones that are widely used in trading of freight services. Thepresent proposal secures the systemic design of advanced auctions andcapitalizes on the ACE for shaping a progressive and evolutionary ap-proach to implement advanced auctions with mobility and locationsensitivity and to incorporate additional sources of information suchas participants’ behavior. The model parameters are fine tuned usingobservations from actual or experimental findings.

� MA-12Monday, 8:30-10:00RB-Omicron

Fuzzy Goal Programming andOptimization Methods

Stream: Fuzzy Systems, Neural Networks & ArtificialIntelligenceInvited sessionChair: Mariano Jimenez-Lopez, Economía Aplicada I, Universityof the Basque Country, Plaza de Oñati 1, 20018, San Sebastian,Spain, [email protected]

1 - Soft computing applied to portfolio selection with un-certain dataCarlos Cruz Corona, DECSAI, Universidad de Granada,Spain, [email protected], Ricardo Coelho Silva, JosèL. Verdegay

Nowadays, Soft Computing is employed with great success in the de-sign, construction and utilization of a wide range of products and sys-tems whose functioning is directly based on the human beings reasonways. This is specifically patent in the case of the real-life optimiza-tion problems. In this work a novel approach that solves quadratic pro-gramming problems under different kind of uncertainties in its data isproposed. As an illustration, a portfolio selection problem with uncer-tain data is formulated and solved as a fuzzy quadratic programmingproblem.

2 - Fuzzy goal programming for material requirementplanning under uncertainty and integrity conditionsJosefa Mula, Research Centre on Production Management andEngineering, Universitat Politècnica de València, EscuelaPolitécnica Superior de Alcoy, Plaza Ferrándiz y Carbonell, 2,03801, Alcoy, Alicante, Spain, [email protected], ManuelDíaz-Madroñero, Mariano Jimenez-Lopez

To solve a material requirement planning problem of a first tier sup-plier in an automobile supply chain a fuzzy goal programming modelis proposed. The goals are to minimize production and inventory costs,

idle time and backorder quantities. The epistemic uncertainty is con-sidered in the demand, available and required capacity. The integrityconditions for the main decision variables are considered. To model therelative importance between goals, instead of a crisp weight structure,fuzzy binary relations are used. An example based on modifications ofreal-world industrial problems is shown.

3 - A Novel Approach for Multiobjective ProgrammingProblems with fuzzy objective functionsMonga K Luhandjula, Decision Sciences, University of SouthAfrica, Muckleneuk Campus, 0003, Pretoria, Gauteng, SouthAfrica, [email protected]

In this paper we propose a novel approach for Multiobjective Program-ming Problems with fuzzy number coefficients. The main idea behindour approach is to approximate involved fuzzy numbers by their re-spective nearest interval counterparts. An oracle that associates witha given fuzzy number its nearest interval plays a pivotal role in thisregard. Our method contrasts markedly with techniques based on def-fuzification operators which replace a fuzzy set by a single real numberleading to a lost of too many important information. A numerical ex-ample is also provided for the sake of illustration.

4 - Sequential Goal Programming with Fuzzy Hierarchy:Application to Sustainable InvestmentsMariano Jimenez-Lopez, Economía Aplicada I, University ofthe Basque Country, Plaza de Oñati 1, 20018, San Sebastian,Spain, [email protected], Amelia Bilbao-Terol, MarArenas-Parra, Verónica Cañal

A portfolio selection model satisfying Socially Responsible Investment(SRI) criteria is shown. SRI bears in mind financial goals and personalvalues. To evaluate SRI criteria, social, environmental and ethical in-dicators are introduced by a fuzzy-MCDM approach. The evaluationof the financial criteria is carried out by the Expected Value at the Endand by the Conditional Value at Risk; both are incorporated, along withthe SRI criteria into a two-stage GP problem with imprecise sequentialhierarchies. Our model optimizes investment and social performance.

� MA-13Monday, 8:30-10:00RB-Tau

Lot-sizing and Related Topics 1

Stream: Lot-sizing and Related TopicsInvited sessionChair: Bernardo Almada-Lobo, Industrial Engineering andManagement, Faculty of Engineering of Porto University, Rua Dr.Roberto Frias s/n DEIG, 4200-465, Porto, Portugal,[email protected]

1 - Integrated scheduling of continuous casters and hotstrip mills in the steel industry: a block planning ap-proachPedro Amorim, Industrial Engineering and Management,Faculty of Engineering of University of Porto, Rua DoutorRoberto Frias, 4200-465, Porto, Portugal,[email protected], Imke Mattik, Hans-Otto Guenther

This work addresses the joint scheduling of continuous caster and hotstrip mill processes in the steel industry. Traditionally, slab yards areused to decouple these two stages. However, the rising importanceof energy costs gives motivation for an integrated scheduling. Foreach of the processes a mixed-integer optimization model based onthe block planning principle is presented. This approach develops pro-duction schedules that take technological sequences of steel grades andmilling programs into account. Instances from a real-world steel plantare solved and the results are discussed.

6


2 - Lot sizing and scheduling of chemical commodityproducts: An evolutionary block planning applica-tionAndreas Schöpperl, Production Management, TU Berlin,Strasse des 17. Juni 135, Sekr. H95, 0623, Berlin, Berlin,Germany, [email protected], Hans-OttoGuenther

Nowadays companies in many industries are faced with an increasedproduct variety and the need to quickly respond to short-term customerorders. Hence a specific challenge is to continuously adapt produc-tion schedules in a cost-efficient way when new customer orders ar-rive. Evolutionary planning is proposed as a general concept support-ing the continuous development of production schedules. Specifically,we adopt the block planning principle as a practical tool for schedul-ing product variants in a predetermined sequence. As a case study theproduction of chemical commodities is considered.

3 - Combining column generation and metaheuristicsfor solving a parallel machine scheduling problemwith job splittingCarina Pimentel, DEGEI, University of Aveiro/AlgoritmiResearch Center, Campus Universitário de Santiago, Portugal,3810-193, Aveiro, Aveiro, Portugal, [email protected],Filipe Alvelos, Diogo Alves

In this talk we present the problem of scheduling n independent jobson m identical parallel machines with the objective of minimizing totaltardiness, considering jobs release dates, sequence independent setupsand lot splitting. A time indexed mixed integer programming com-pact model and a decomposition model will be presented. The de-composition model is solved by a SearchCol algorithm. SearchColis a recently proposed general framework for decomposable integerprogramming/combinatorial optimization problems based on the hy-bridization of column generation and metaheuristics.

4 - Neighbourhood-based hybrid algorithm for pulp andpaper production planningGonçalo Figueira, Industrial Engineering and Management,Faculty of Engineering of Porto University, 4200-465, Porto,Portugal, [email protected], Maristela Santos,Bernardo Almada-Lobo

Motivated by a real-world case study in the pulp and paper industry,we tackle the short-term production planning and scheduling in an in-tegrated mill. Our approach combines a Variable Neighborhood Searchprocedure (which manages the setup variables), a specific heuristic (todetermine the discrete production speeds constrained to a maximumvariation) and an exact solver for the optimization of production andflow movement decisions. We explore different strategies to speed-up the solution procedure and test alternative variants with instancesbased on real data from the case study.

� MA-14Monday, 8:30-10:00RB-Omega

Scheduling

Stream: Scheduling under Resource ConstraintsInvited sessionChair: Wieslaw Kubiak, Faculty of Business Administration,Memorial University, Prince Philip Drive, A1B 3X5, St. John’s,NL, Canada, [email protected]

1 - Product Rate Variation problem on parallel machinesand divisor methods of apportionmentJoanna Jozefowska, Institute of Computing Science, PoznañUniversity of Technology, Piotrowo 2, 60-965, Poznañ,Wielkopolska, Poland, [email protected]

House monotone methods of apportionment may be successfully ap-plied to solving the PRV problem defined on a single machine. PRVminimization on multiple parallel machines may be defined in twoways: either the product rate is related to the total production volume

or it is related to schedule length. We show that the divisor methods ofapportionment may fail in constructing an optimal solution in case ofvolume oriented product rate.

2 - An Efficient Algorithm for Finding Ideal SchedulesDariusz Dereniowski, Department of Algorithms and SystemModeling, Gdansk University of Technology, ul. GabrielaNarutowicza 11/12, 80-233, Gdansk, Poland,[email protected], Ed Coffman, Wieslaw Kubiak

In this talk we address the problem of scheduling UET jobs with re-lease dates and precedence constraints on two identical processors. Wesay that a schedule is ideal if it minimizes both maximum and totalcompletion time simultaneously. We give an instance of the problemshowing that ideal schedules do not exist in general when preemptionsare allowed. If preemptions are not allowed, then ideal schedules doexist for general precedence constraints, and we describe an algorithmfor finding ideal schedules in O(n3) time, where n is the number ofjobs.

3 - Coordinating Subcontractor Scheduling with Divisi-ble Jobs and Private InformationBehzad Hezarkhani, Faculty of Business Administration,Memorial University, General Office, Faculty of BusinessAdministration, Memorial University of Newfoundland, A1B3X5, St. John’s, Newfoundland and Labrador, Canada,[email protected], Wieslaw Kubiak

We consider a decentralized setting with several manufacturer agentsand a subcontractor where subcontracting is done to reduce themakespans. We show that the VCG mechanism with Clarke pivot rulegives rise to a simple closed-from formula for payments. Next, we in-troduce pricing schemes that support the efficient solution, i.e. agentsindividually choose their allocations in the efficient solution. We showthat subcontractor’s revenue with such supporting prices are at least asmuch as that in the VCG mechanism with Clarke pivot rule and can bearbitrarily close to the maximum possible.

4 - Time auction as a tool for solving multiagentscheduling problemsPiotr Modlinski, Institute of Control and ComputationEngineneering, Warsaw University of Technology, ul.Nowowiejska 15/19, 00-665, Warsaw, Poland,[email protected]

The problem of scheduling disjoint, indivisible tasks takes on a newmeaning in a situation when a number of agents have contradictoryrequirements. Attempts to centralized solving the problem do not pro-duce satisfying results because agents are utilitarian. Bilateral negotia-tions are not sufficient mechanism in the general case, and multilateralones are too complex. In the paper we consider usability of the timeauction. It bases on multi-commodity auction mechanism. Issues ofeffectiveness and incentive compatibility are considered.

� MA-16Monday, 8:30-10:00RB-2103

Advances in Nonsmooth Optimization

Stream: Nonsmooth OptimizationInvited sessionChair: Adilson Elias Xavier, Graduate School of SystemsEngineering and Computer Sciences, Federal University of Rio deJaneiro, P.O. Box 68511, Ilha do Fundão - Centro Tecnologia -H319, 21941-972, Rio de Janeiro, RJ, Brazil, [email protected]

1 - A Simple Resilient Backpropagation Algorithm forNon-Smooth Optimisation ProblemsApostolos Kotsialos, School of Engineering and ComputingSciences, Durham University, South Road, DH1 3LE,Durham, United Kingdom, [email protected]

7


This paper reports on a non-smooth optimisation algorithm based onnon-continuous gradient functions. The resilient backpropagation al-gorithm is employed. Results from test problems in their uncon-strained, bound constrained and non-smooth constrained versions, arepresented. The suggested algorithm’s performance is compared withthose of the more sophisticated bundle methods, that have appearedin the literature; better performance is shown in some cases. Furtherpositive features include the algorithm’s scalability and simple imple-mentation requirements.

2 - Solving the Continuous Multiple Allocation p-HubMedian Problem by the Hyperbolic Smoothing Ap-proach: Computational ResultsAdilson Elias Xavier, Graduate School of SystemsEngineering and Computer Sciences, Federal University ofRio de Janeiro, P.O. Box 68511, Ilha do Fundão - CentroTecnologia - H319, 21941-972, Rio de Janeiro, RJ, Brazil,[email protected], Claudio GesteiraThis paper considers the problem of finding the least expensive hub lo-cations, given demands between pairs of cities. The problem leads to amin-sum-min formulation that is strongly non-differentiable. The pro-posed method overcomes this difficulty with the hyperbolic smoothingstrategy, which has been proven able to solve different non-smoothproblems. The most important feature of the methodology is the lowdimension of the subproblems. The robustness and the efficiency ofthe method are shown through a set of computational experiments withlarge problems with up to 1000 cities.

3 - Solving the Minimum Sum of L1 Distances Cluster-ing Problem by Hyperbolic Smoothing and Partitioninto Boundary and Gravitational RegionsSergio B. Villas-Boas, PESC, UFRJ / COPPE, Department ofSystems Engineering and Computer Science (PESC),Graduate School of Engineering (COPPE), Federal Universityof Rio de Janeiro, 21.941-972, Rio de Janeiro, Rio de Janeiro,Brazil, [email protected], Vinicius Layter XavierIt is considered the minimum sum of distances clustering problem, fordistances measured using the L1 or Manhattan metric (MSDC-L1). Itis a min-sum-min problem, strongly non differentiable.It is proposed the AHSC-L1 method, by combining 2 techniquesthat drastically simplify the computational tasks. (1) Use HyperbolicSmoothing Clustering (HSC), that adopts a smoothing strategy usinga special C_infinity completely differentiable function. (2) Partitionof the set of observations into two non overlapping groups: "data infrontier" and "data in gravitational regions".

4 - Solving the Fermat-Weber Problem with the Hyper-bolic Smoothing MethodVinicius Layter Xavier, Systems Engineering and ComputerSciences Depart., Federal University of Rio de janeiro, CT -Bloco H - sala H319, Ilha do Fundao, 21941-972, Rio deJaneiro, RJ, Brazil, [email protected], Felipe França,Adilson Elias Xavier, Priscila LimaThe minimization of the sum of the distances of observations to theircluster centroids, a non-differentiable and a non-convex problem witha large number of local minima, is tackled. The hyperbolic smooth-ing strategy solves a sequence of low dimension differentiable uncon-strained optimization sub-problems, which gradually approaches theoriginal problem. The reliability and the efficiency of the method areillustrated via a set of computational experiments. It must be empha-sized that the proposed methodology can be analogously applied to thesolving of the Fermat-Weber location problem.

� MA-17Monday, 8:30-10:00RB-2105

Location and Supply Chain Management

Stream: Applications of Location AnalysisInvited sessionChair: Teresa Melo, Business School, Saarland University ofApplied Sciences, Waldhausweg 14, D-66123, Saarbrücken,Germany, [email protected]

1 - On the Feasibility of Establishing a Northern WesternAustralian Beef Abattoir as a Facility Location Prob-lemRodolfo Garcia-Flores, Mathematics, Informatics andStatistics (CMIS), Commonwealth Scientific and IndustrialResearch Organisation (CSIRO), 71 Normanby Rd., Clayton,3168, Melbourne, VIC, Australia,[email protected], Andrew Higgins, AndreasErnst

The trade of livestock from north Western Australia (WA) has recentlybecome more vulnerable to policy changes, whereas the feasibility ofbuilding and operating local abattoirs has become highly sensitive toconstruction costs and prices of slaughtered meat. We formulate a fa-cility location problem that selects segments of the road network toupgrade, and abattoirs from a set of potential sites in the Pilbara regionof WA. The model provides insight on the trade off between resourceallocation to facilities and links. We present initial results and proposean algorithm based on decomposition.

2 - Design of a pooled distribution network: a casestudyOlivier Péton, Automatic Control and Industrial Engineering,LUNAM Université, Ecole des Mines de Nantes, IRCCyNUMR CNRS 6597, 4 rue alfred kastler, 44307, NANTES,France, [email protected], Fabien Lehuédé,Christian Leroux, Xavier Perraudin

We consider a cluster of neighboring competing companies which wishto pool their distribution network. These companies frequently deliversmall quantities of goods to hundreds of common retailers spread in alarge territory. Their main goal is to reduce transportation cost andenvironmental footprint thanks to full truckload transportation. Wemodel this problem as a sequence of MILPs with the following steps:location of one consolidation warehouse and several regional cross-docks, design of truckload routes, assignment of retailers to regionalwarehouses and loading of trucks.

3 - Comparing classical performance measures for amulti-period logistics network design problemTeresa Melo, Business School, Saarland University ofApplied Sciences, Waldhausweg 14, D-66123, Saarbrücken,Germany, [email protected], Isabel Correia,Francisco Saldanha-da-Gama

A two-echelon network design problem is studied. Location and trans-portation decisions are considered together with the sizing of storageareas for product families. A finite set of capacity levels for each fam-ily is available at each potential location. Two mixed-integer linearmodels are proposed that differ in their objective functions: cost min-imization v. profit maximization. Valid inequalities are developed tostrengthen the formulations. The impact of these performance mea-sures on network design are investigated through a numerical study.An extension to a bi-objective case is studied.

� MA-18Monday, 8:30-10:00RB-2107

Financial Crisis Modelling

Stream: Financial Modelling & Risk Management (con-tributed)Contributed sessionChair: Blanka Škrabic Peric, Department of Quantitative Methods,Faculty of Economics, Matice hrvatske 31, Split, Croatia,[email protected]

8


1 - Spanish savings banks in the credit crunch: coulddistress have been predicted before the crisis? Amultivariate statistical analysisMartí Sagarra, Department of Business Economics,Universitat Autònoma de Barcelona, Departamentd’Economia de l’Empresa, Facultat d’Economia i Empresa -Edifici B, 08193, Bellaterra (Cerdanyola del Vallès), Spain,Spain, [email protected], Cecilio Mar-molineroSpanish savings Banks (Cajas) have had a long and distinguished his-tory. However, they have been heavily affected by the banking crisis of2007. Many of them had to merge with other institutions or were res-cued. We show that, before the crisis, there were structural differencesbetween successful Cajas and those that had to be rescued. The tech-nical approach is based on Multidimensional Scaling Analysis (MDS).MDS has the advantage that the main characteristics of the study canbe presented in a visual form, and thus facilitate communication of theresults.

2 - On the transition of Japanese corporate rating struc-ture under the recent credit crisesMotohiro Hagiwara, School of Commerce, Meiji University,1-1 Kandasurugadai Chiyoda-ku, 101-0062, Tokyo, Japan,[email protected], Yasuhiro Matsushita, KatsuakiTanakaThe aim of this study is to check the transition of recent bond ratingstructure under the credit crises based on accounting informations giv-ing Altman Z-score. Japanese corporate bond ratings are transformedto normally distributed variables using published 5 -Year actual defaultprobability, and are modeled as functions by key ratios giving Z-scoreusing ANN. This study found some facts concerning the transition ofrating structure in recent several years and time consistency of ratinginformations under the crises are discussed based on models beforecrises.

3 - Using assignment problem in financial products’ de-signMaria Mavri, Business Administration, University of theAegean, 8 Michalon Street, 82100, Chios, Greece,[email protected], Michail BekiarisToday, the world economic crises are forcing banks to change their op-erational internal processes. We argue that a service system can effec-tively be represented as a manufacturing system, and more specificallythat any product or service offered by a bank to its clients could bedesigned following a procedure similar to those used in "production’.In this study we use the concept of assignment problem in order to dis-tribute tasks and jobs to bank units, that develop new products, suitablefor satisfying customers’ diverse and different needs.

4 - Recent crisis, credit boom and nonperforming loansin CEE countries-panel data evidenceBlanka Škrabic Peric, Department of Quantitative Methods,Faculty of Economics, Matice hrvatske 31, Split, Croatia,[email protected], Josip ArnericThe paper investigates impact of credit boom during 2000th and re-cently crisis on nonperforming loans in CEE counties. Namely, thebeginning of 2000th was characterized by many features as the for-eign capital inflow, entry of foreign banks and economic growth. Thishas resulted in increased competition in the banking system and rapidcredit growth. Dynamic panel data is used to analyze the impact ofrapid credit growth on nonperforming loans in 16 CEE countries dur-ing period 1994-2010. Empirical analysis is performed on aggregateddata for each country.

� MA-19Monday, 8:30-10:00RB-2111

Risk management in commoditiesmarketsStream: Commodities and Financial ModelingInvited sessionChair: Massimo Panella, University of Rome "La Sapienza", Dpt.of Information Engineering, Electronics and Telecommunications

(DIET), Via Eudossiana, 18, 00184, Rome, Italy,[email protected]: Rita Decclesia, Economic and Social Analysis, SapienzaUniversity of Rome, Piazza Aldo Morto 5, 00185, Rome, Italy,[email protected]: Bogdan Iftimie, Applied Mathematics, Academy ofEconomic Studies of Bucharest, 010374, Bucharest, Romania,[email protected]

1 - Energy commodities:price relationships the key is-sueRita Decclesia, Economic and Social Analysis, SapienzaUniversity of Rome, Piazza Aldo Morto 5, 00185, Rome,Italy, [email protected]

This paper examines the time series econometric relationship betweenthe fossil fuels and electricity in Europe and in US.When data have unitroots as in this case a cointegrating relationship may provide effectiveresults. An optimization framework is also set to get the correct rollingcorrelation measure which provides hint on short ter,m relationships.Long runa and short term relationship represent the key element to setup effective risk management strategies.

2 - A Mixed Integer Linear Programming Approach toMarkov Chain BootstrappingRoy Cerqueti, Department of Economics and FinancialInstitutions, University of Macerata, Via Crescimbeni, 20,62100, Macerata, Italy, [email protected], Paolo Falbo,Cristian Pelizzari, Federica Ricca, Andrea Scozzari

The Markov Chain bootstrap method is one of the most acknowledgedtools to make forecasts and study the statistical properties of an evolu-tive phenomenon. The underlying idea is to replicate the phenomenonon the basis of an available sample, which is assumed to evolve accord-ing to a Markov Chain. This paper deals with the relevant problem ofreducing the dimension of the transition probability matrix by main-taining the original features of the sample. To this aim, we advancetwo mixed integer linear programs. We present also some numericalexamples based on the electricity market.

3 - Financial Optimization Modeling in RRonald Hochreiter, Finance, Accounting and Statistics, WUVienna University of Economics and Business, Augasse 2-6,1090, Vienna, Austria, [email protected]

Simplifying the task of modeling financial optimization problems isimportant. Many commercial products have been created to supportthe financial modeling process, but none of these products has beenadopted by a significantly large number of users. Simplifications arecrucial to implement optimization models into business processes suc-cessfully. We outline an approach on how to simplify financial opti-mization modeling using R and external optimization modeling lan-guages as well as by building model generators for specific applicationproblems.

� MA-20Monday, 8:30-10:00RB-2113

Financial Time Series Analysis andForecasting

Stream: Data Mining in the Financial SectorInvited sessionChair: Marcus Hildmann, Information Technology and ElectricalEngineering, ETH Zurich, ETL G 24.2, Physikstrasse 3, 8092,Zürich, Switzerland, [email protected]

9


1 - Carbon Price Sensitivity and Sectoral AnalyisisOmer Kayhan Seyhun, Banking and Financial InstitutionsDepartment, Risk Centralization Division, Central Bank ofTurkey, Istiklal Street No 10, Ulus, 06100, Ankara, Turkey,[email protected] examine the fundamental drivers of carbon prices and their sen-sitivity to various sectors; we investigate various related interactions.Multivariate statistical techniques are used to attest the link betweencarbon prices and sectoral level variables. It is not clear how to pricethe future damages of the climate change and the increasing naturaldisaster occurrence. It is expected that the correlation will be a signif-icant tool to observe the sensitivity of carbon prices with energy andagricultural sectors.

2 - Online Financial Data Streams Value at Risk Segmen-tation MethodsDima Alberg, Industrial Engineering and Management, SCESami Shamoon College of Engineering, Bazel/Bialik Sts,84100, Beer-Sheva, Israel, [email protected] novel algorithms FISW and FISWAB for an efficient segmenta-tion of financial data streams are presented. The proposed algorithmsimplement adaptive sliding window approach whose size recomputedonline according to the rate of change observed from the data itself.The main advantages of the algorithms that they are time and memoryefficient, use the Value at Risk as a tuning parameter and do not requirechoosing lot of input parameters. The accuracy and computation timeof the proposed algorithms will be compared to real financial streamdata.

3 - Modeling the market decision problems for electricalenergy balancing systemsMariusz Kaleta, Inistitute of Control & ComputationEngineering, Warsaw University of Technology, Nowowiejska15/19, 00-665, Warsaw, Poland, [email protected] work is a step for standardization of balancing problems descrip-tion (similarly to Kendall’s notation in theory of queues) for electri-cal energy markets, what in effect should facilitate understanding andcomparing research works, exchanging data cases and mechanism im-plementation. We formulate an abstract model of balancing problem,which consists in so called market decision problems and balancingmechanism. We provide a parametrization of the market decision prob-lems space. To prove our concept we apply it to build an expressivemarket simulation tool.

� MA-21Monday, 8:30-10:00RB-2115

Financial Mathematics and OR 1Stream: Financial Mathematics and ORInvited sessionChair: Tansel Avkar, Department of Mathematics, Middle EastTechnical University, 06531, Ankara, Turkey,[email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - Different approaches for evaluating a Portfolio ofR&D ProjectsAnabela Costa, Quantitative Methods, CIO/ ISCTE - InstitutoUniversitário de Lisboa, Av. das Forças Armadas, 1649-026,Lisboa, Portugal, [email protected], José PaixãoWe present three approaches to valuing a portfolio of R&D projectswith a budget constraint. One approach is based on a dynamic pro-gramming model where stages are associated to decisions concerningthe progress of the projects. The second one is based on a Monte Carlosimulation approach determining the parallel valuation for the projects.Finally, the third approach consists of a 0-1 integer linear program forfinding the optimal decisions in terms of the overall value of the port-folio. We report computational experience concerning the differentapproaches.

2 - Time series forecasting and optimal stopping prob-lem in market index portfolioLukas Pichl, Department of Information Science,International Christian University, Osawa 3-10-2, Mitaka,181-8585, Tokyo, [email protected]

Time series forecasting of a single stock title, within an entire mar-ket index, and across different markets is an active area of opera-tions research ranging from artificial neural networks across economet-ric models to information compression methods in symbolic analysismodels. We survey the recent developments and present our findingson the correlations in the dynamics of stock index and its individualconstituents in Japan, using high frequency tick data aggregated at var-ious time scales, and demonstrating the presence of stylized facts.

3 - Portfolio Optimization with Hybrid UncertaintyAlexander Yazenin, Applied Mathematics Department, TverState University, Zhelyabova, 33, 170100, Tver,[email protected]

In the paper the mathematical model of the minimal risk portfoliowith hybrid uncertainty (possibility/probability) is presented and in-vestigated. In the model the profitabilities are explicated with fuzzyrandom variables. Their second order moments those form the portfo-lio risk function are non-fuzzy. In the model the expected value level ispresented with possibilistic/probabilistic restrictions. The behavior ofthe investment possibilities set is investigated depending on the levelof possibility/necessity and probability. The results are demonstratedon the model example.

4 - Multicriteria decision making in comparison of objec-tive and subjective criteria of companies’ successBranka Marasovic, Faculty of Economics Split, 21000, Split,Croatia, [email protected], Ivana Tadic

Although, financial indicators usually are taken as the most importantindicators of success, lately authors argue about the same relevance inusage of subjective indicators. The aim of paper is to research whetherthe most successful companies regarding financial ratios (objectivegroup of criteria) show the similar results by other groups of criteria, ashuman resource management evaluation and evaluation of entire com-panies’ success (subjective groups of criteria). Selected companies willbe ranked by each group of criteria using multicriteria decision makingmethod PROMETHEE II.

� MA-22Monday, 8:30-10:00RB-2117

Soft Computing and SCM Applications

Stream: Fuzzy Sets and Soft ComputingInvited sessionChair: Kaj-Mikael Bjork, IAMSR, Abo Akademi University andArcada, University of Applied Sciences, 20520, Turku, Finland,[email protected]

1 - A tabu search vehicle routing problemKaj-Mikael Bjork, IAMSR, Abo Akademi University andArcada, University of Applied Sciences, 20520, Turku,Finland, [email protected]

Heuristic optimization methods provide a solid platform to solve bigsized (realistic) problems even in the application domain of NP-hardproblems. Many transportation optimization problems are such thatthey can be solved with rigorous methods (such as MILP-methods, forinstance) for only smaller sized problems. Even if heuristic methodswill not provide global convergence, they will be applicable to realisticbig-sized problems. In this paper, a Tabu search method is applied to atransportation problem and more specifically a vehicle routing problemrelevant for the Finnish forest industry.

10


2 - A Multi-item Fuzzy Economic Production QuantityProblem with Backorders and Limited Storage SpaceMagnus Westerlund, Economics, IT and Media, Arcada,Jan-magnus jansonsplats 1, 00550, Helsinki, Finland,[email protected], Kaj-Mikael Bjork, JózsefMezei

We extend a fruitful fuzzy EOQ (Economic Order Quantity) track ofresearch with a multi-item model that takes storage limitation andbackorders into consideration. In the paper, finite production rate isconsidered (i.e. the products are produced to stock and not purchasedwhich is a typical situation in process industry). In the model, the ordercycle time, the demand and the backorders are represented by triangu-lar fuzzy numbers which allows for taking expert opinion into accountwhen modeling uncertainties. Signomial global optimization (SGO)algorithm is employed to find the optimal solution.

3 - Systems of (max, min)-linear equations and inequal-itiesMartin Gavalec, Department of Information TechnologiesFIM, University of Hradec Kralove, Rokitanského 62, 50003,Hradec Kralove, Czech Republic, [email protected],Karel Zimmermann

By (max, min)-linear equation (inequality) we understand an equation(inequality) between two (max, min)-linear functions of n variables.Properties of the solution sets of finite systems of (max, min)-linearequations and inequalities are investigated in the contribution. Theobtained results are then used for solving optimization problems withmax-separable objective function and constraints described by a sys-tem of (max, min)-linear equations and inequalities. Applications tosome multiple fuzzy goal decision problems as well as some general-izations are discussed.

4 - Measuring transitivity of fuzzy pairwise comparisonmatrix in DM problemsJaroslav Ramik, Dept. of Math. Methods in Economics,Silesian University, School of Business, University Sq.1934/3, 73340, Karvina, Czech Republic, [email protected],Petr Korviny

A pair-wise comparison matrix is the result of pair-wise comparison apowerful method in multi-criteria optimization. When comparing twoelements, DM assigns the value representing the element of the pair-wise comparison matrix. The consistency ratio is defined in AHP. Insome situations another interpretation is convenient. The DMs pref-erences can be represented by a fuzzy preference relation. We inves-tigate relations between several types of transitivity of fuzzy relationsand multiplicative preference relations. We also define the grade oftransitivity.

� MA-23Monday, 8:30-10:00RB-Delta

Generalized DifferentiationStream: Generalized Differentation and OptimizationInvited sessionChair: Alexander Kruger, School of Science, InformationTechnology & Engineering, University of Ballarat, UniversityDrive, Mount Helen, P.O. Box 663, 3353, Ballarat, Victoria,Australia, [email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - On Weak Subdifferentials and ExhaustersMustafa Soyertem, Mathematics, Anadolu University,Anadolu University Yunus Emre Campus Department ofMathematics, 26470, Eskisehir, Turkey,[email protected], Mahide Kucuk, Ryszard Urbanski,Jerzy Grzybowski, Yalcin Kucuk, Ilknur Atasever, DidemTozkan

Exhausters were defined by Demyanov (1999). Then, relationships be-tween exhausters and generalized subdifferentials were given by De-myanov and Roshchina (2007, 2008). In this work, a special classof exhausters is obtained by using weak subdifferentials defined byAzimov and Gasimov (1999). Some examples of positively homoge-neous functions whose upper and lower exhausters calculated by weaksubdifferential are given. Moreover, some properties of this class ofexhausters are examined.

2 - Exhausters, Weak Subdifferentials and OptimalityConditionsDidem Tozkan, Department of Mathematics, AnadoluUniversity, Anadolu Universitesi Yunus Emre Kampusu, FenFakültesi Matematik Bolumu, 26470, Eskisehir, Turkey,[email protected], Mahide Kucuk, Ryszard Urbanski,Jerzy Grzybowski, Yalcin Kucuk, Ilknur Atasever, MustafaSoyertem

Demyanov defined the concept of exhausters (1999) and then optimal-ity conditions in terms of proper and adjoint exhausters were givenby Demyanov and Roshchina (2006, 2008). In this study, we givesome examples of evaluating exhausters of positively homogeneousfunctions via weak subdifferentials defined by Azimov and Gasimov(1999). Furthermore, some optimality conditions are examined by us-ing relationships between exhausters and weak subdifferentials.

3 - Relationships between gw-subdifferentials and ra-dial epiderivatives for nonconvex vector functionsIlknur Atasever, Department of Mathematics, AnadoluUniversity, Anadolu Universitesi Fen Fakultesi MatematikBolumu, 26470, Eskisehir, Turkey, [email protected],Yalcin Kucuk, Mahide Kucuk

The notion gw-subdifferential for vector valued functions was definedand necessary and sufficient conditions for gw-subdifferentiability ofnonconvex vector valued functions were given by Küçük et. al. (2011).In this work, we examine relations between radial epiderivative definedby Kasimbeyli (2009) and gw-subdifferential of vector valued func-tions. Furthermore, under some assumptions, we proved that the radialepiderivative of a vector valued function is an element of the supre-mum of the set which is defined by using gw-subgradients of the givenfunction.

4 - Minkowski-Radstrom-Hormander cone and reducinginfinite exhaustersJerzy Grzybowski, Faculty of Mathematics and ComputerScience, Adam Mickiewicz University, Wieniawskiego 1,61-712, Poznan, Poland, [email protected], Ryszard Urbanski,Mahide Kucuk, Yalcin Kucuk, Ilknur Atasever, MustafaSoyertem, Didem Tozkan

We introduce the notion of generalized Minkowski-Radstrom-Hormander cone. We also embed the convex cone C(X) of allnonempty closed convex subsets of real Hausdorff topological vec-tor space X and the convex cone W(Rn) of all nonempty convex sub-sets of Rn into respective Minkowski-Radstrom-Hormander cones. Wealso present equivalent conditions for reducing infinite exhausters ofclosed convex sets to a pair of convex sets (an element of Minkowski-Radstrom-Hormander cone).

� MA-24Monday, 8:30-10:00CC-A11

Cooperation and Logistics

Stream: Game Theory and LogisticsInvited sessionChair: Ignacio García-Jurado, Department of Mathematics,Coruna University, Faculty of Computer Science, Campus deElvina, 15071, Coruna, Spain, [email protected]

11


1 - A polynomial expression for the Owen value in themaintenance cost gameJulian Costa, Faculty of Computer Science, Universidade daCoruña, Campus de Elviña, 15071, A Coruña, Spain,[email protected]

The maintenance cost games are a class of cooperative games withtransferable utility where the main issue is how to share the mainte-nance costs of a facility among the agents. We propose the use of theOwen value for the cost allocation problem associated with a game inthis class when there is a system of a priori unions. In the general case,the Owen value has exponential complexity. In this paper, we providea cubic polynomial expression for the Owen value of a maintenancecost game with a priori unions.

2 - Minimum cost Steiner tree problemsSilvia Lorenzo-Freire, Department of Mathematics,Universidade da Coruña, Campus de Elviña, 15071, ACoruña, Spain, [email protected], Gustavo Bergantinos,Leticia Lorenzo, Juan Vidal-Puga

Consider a group of agents located at different geographical placesthat are interested in some resource provided by a common supplier.Agents can be served directly from the supplier or indirectly throughother agents or public switches. The first problem that we need to solveis how to provide all the agents with the resource with a minimal cost.This problem is known as the Steiner tree problem and it is NP-hard.So, in order to find a solution, heuristics need to be applied. In this pa-per, we allocate the cost of constructing this approximated tree amongthe agents in a stable way.

3 - Scheduling jobs with a common due date via combi-natorial gamesIrinel Dragan, Mathematics, University of Texas, 411S.Nedderman Dr., Pickard Hall, 76019-0408, Arlington,Texas, United States, [email protected]

Scheduling jobs with a common due date as a sequence that has a min-imal total deviation from the due date can be done in the simplest case,by means of J.J.Kanet’s algorithm (1981). Assuming that it is knownhow to do it for a given set of jobs and its subsets, we generate a co-operative cost game with transferable utilities. As recommended byH.P. Young in his paper included in Handbook of Game Theory, vol.2,some efficient values are fair solutions and we discuss the fairness bymeans of the cost excesses. yjr nucleolus is the most fair solution.

4 - On graphs which can or cannot induce Chinese Post-man games with a non-empty coreDaniel Granot, University of British Columbia, Vancouver,Canada, [email protected]

We study the Chinese postman (CP) cooperative game induced by aconnected, weighted, undirected graph G, wherein players reside atedges of G and a postman, starting from a post-office location (i.e.,vertex of G), needs to traverse all edges before returning to the post-office. We provide a complete characterization of all connected graphsfor which there exists a positive edge-cost function such that the in-duced CP game has a non-empty core, and, consequently, we derive acomplete characterization of all connected graphs for which there doesnot exist a positive edge-cost function which induces a CP game with anon-empty core. Membership in these classes of graphs can be verifiedin strongly polynomial time.

� MA-25Monday, 8:30-10:00CC-A23

Energy Distribution: Planning andOptimization

Stream: Energy, Environment and Climate 1Contributed sessionChair: Carmen Anido, Economic Analysis: Quantitatve Economy,Autonoma University of Madrid, Faculty of Economics,Cantoblanco., 28049, Madrid, Spain, [email protected]

1 - Unit Commitment optimisation through matheuris-ticsAna Viana, INESC TEC/ISEP, Campus da FEUP, Rua Dr.Roberto Frias, 4220-465, Porto, Portugal,[email protected], Dewan Fayzur Rahman, Joao PedroPedrosoUnit Commitment is a classical problem in power production planning,consisting of deciding which electric generators must be committed toproduction in each period of a given planning horizon, and definingthe production level at which each generator should operate. In thiswork two approaches are explored: one based on "Local Branching’and another where a Particle Swarm Optimisation (PSO) algorithmcooperates with the MIP solver. Thorough computational tests in awell-known set of benchmark instances show the effectiveness of theapproaches.

2 - Optimising Distributed Energy Operations in Build-ingsAfzal Siddiqui, Statistical Science, University CollegeLondon, Gower Street, WC1E 6BT, London, UnitedKingdom, [email protected], Markus Groissböck,Somayeh Heydari, Ana Mera Vazquez, Eugenio PereaOlabarria, Michael StadlerDeregulation of the energy sector has created new markets for produc-ers as well as opportunities for consumers to meet their needs in a morecustomised way. However, to realise economic and energy efficiencygains, small-scale consumers at the building level require more tailoreddecision support. We model the effect of active equipment control onthe zone temperature taking into account the external temperature, so-lar gains, the building’s shell and internal loads. These serve as con-straints in a cost-minimisation problem, providing an optimal dispatchfor installed equipment.

3 - Compatibility of optimized distributed storage pro-files for grid operators and energy tradersStefan Nykamp, University Twente, NL / RWE Deutschland,Bad Bentheim, Germany, [email protected] implementation of storage capacities in power grids is seen as anessential element for a successful integration of PV and wind gen-eration. Also on distribution grid level storage capacities are evalu-ated and installed in pilot projects to overcome massive reinforcementneeds. In this talk, the optimal storage profile of a distribution gridoperator with real measured energy profiles is presented. This peakshaving scenario is compared to an optimal storage profile of an en-ergy trader with arbitrage purposes. The correlation analysis revealscomplementary or supplementary operating modes.

4 - A hybrid genetic algorithm for the optimal allocationof remote controlled switches in radial distributionsystemLaura Assis, DENSIS, Universidade Estadual de Campinas,Avenida Albert Einstein n 400 - Cidade Universitária ZeferinoVaz - Barão Geraldo - Campinas - São Paulo - Brasil,13081-970, Campinas, SP, Brazil, [email protected],Fábio Usberti, Christiano Lyra, Fernando J. Von Zuben, JoséFederico VizcainoRegulatory agencies define target reliability indices to increase thequality of electrical power supply. These indices can be improved withthe solution of specific combinatorial optimization problem. This worktackles the problem of finding the best number, location and type ofswitches to be installed in a radial power distribution network. The ob-jective is to improve system reliability, while minimizing the total costof the solution. An effective hybrid genetic algorithm that evaluatesand optimizes the allocation of remote controlled sectionalizers and tieswitches is proposed.

� MA-26Monday, 8:30-10:00CC-A24

OR Applications in Oil and Gas

Stream: OR in the Oil and Gas SectorsInvited sessionChair: Krystsina Bakhrankova, Applied economics, SINTEF -

12


Technology and society, Box 4760 Sluppen, S. P. Andersens veg 5,7465, Trondheim, Norway, [email protected]

1 - Stochastic MIP modeling of a natural gas-poweredindustrial parkGerardo Perez Valdes, IOT, NTNU, Trondheim, Norway,[email protected], Kjetil Midthun, Michal Kaut,Vibeke NørstebøWe present an investment decision model for a natural gas-based indus-trial park. The model is a stochastic MIP model with random prices ofraw materials. The potential plants in the park are modeled from a casestudy from Norway, whose motivation is to consider the potential forincreasing the domestic use of natural gas. Natural gas is abundant inthe country, but its domestic use is limited due to both availability ofother energy sources and concerns over carbon emissions. Our resultshave an emphasis on showing model properties in terms of model sizeand solution times.

2 - Strategic behaviour in the crude oil market - one-stage vs. two-stage oligopoly modelsDaniel Huppmann, DIW Berlin, Mohrenstraße 58, 10117,Berlin, Germany, [email protected], Clemens HaftendornOligopoly-fringe models are used to investigate market power in re-source markets, combining Cournot and competitive suppliers in amixed complementarity setting. While computationally convenient,this may lead to results that are not Nash equilibria. We investigatewhich situations are particularly problematic, and under which circum-stances these results are similar to two-stage games. We also revisittax models of resource extraction as an alternative approach to modeloligopoly-fringe markets. Several model setups are then compared us-ing data on the crude oil market over the past decade.

3 - Stochastic mixed integer programming for integratedportfolio planning in the LNG supply chainKristin Tolstad Uggen, Applied economics and operationsresearch, SINTEF Technology and society, S.P. Andersens v5, 7465, Trondheim, Norway, [email protected], Adrian Werner,Marte Fodstad, Arnt-Gunnar LiumWe present a model to support strategic planning by global playersin the liquefied natural gas market. The model takes an integratedportfolio perspective and addresses uncertainty in future price devel-opments. Decision variables include investments and disinvestmentsin infrastructure and vessels, chartering of vessels, the timing of LNGand natural gas contracts, and spot market trades. Industry-motivatednumerical cases are discussed to provide benchmarks for the potentialincreases in profits that can be obtained.

4 - Modeling gas infrastructure investments under un-certaintyOzge Ozdemir, ECN, Netherlands, [email protected], Jeroende JoodeA continuous challenge for gas infrastructure companies across Europeis to optimally accommodate the future need for gas infrastructure in-vestments given the uncertainties in gas markets. These uncertaintiesmay originate from the role for gas in the future energy mix, globalgas market developments (e.g., shale gas), and the possible cut-offsby a major supplier which threats security of supply. This paper ex-plores how these uncertainties can be tackled and the outcomes canbe translated into robust strategies that prevent undesirable gas marketoutcomes in a range of possible scenarios.

� MA-27Monday, 8:30-10:00CC-A25

Empirical Research on DecisionProcesses - Methods & Findings

Stream: Decision ProcessesInvited sessionChair: Gilberto Montibeller, Dept. of Management, LondonSchool of Economics, Houghton Street, WC2A 2AE, London,United Kingdom, [email protected]

1 - Exploring IT project managers’ decision making pro-cesses in-situ, in-actu, in-toto: an empirical study ofday-to-day decision making in medium size softwaredevelopmentAna Barcus, Management, LSE, 65 Melbourne Ave, M6K1K6, Toronto, Ontario, Canada, [email protected] research explores real-time decision-making during the deliveryphase of small to medium size software development projects. The em-pirical research was conducted in two software companies, recordingregular day-to-day project meetings. Twelve software projects weretracked, totaling 144 recorded project meetings, resulting in over 94hours of data. Seventeen decision episodes, spanning over a numberof meetings, were identified in-actu. The study examines in more de-tail how the perception of contextual factors impacts on project team’sdecision making processes.

2 - The impact of need for closure on model-supportedgroup conflict managementEtienne Rouwette, Nijmegen School of Management,Radboud University Nijmegen, 6500 HK, Nijmegen,Netherlands, [email protected], L. Alberto FrancoNeed for closure, the desire for definite knowledge on some issue, hasimportant impacts on decision making. Decision makers those are highin need for closure seize on information and then freeze on early cues.We focus on the impact of need for closure on conflicts in decisionmaking groups. The groups in our study use Value Focused Think-ing to jointly develop a model. We compare groups high and low inneed for closure with regard to a) conflicts and how these are resolved;b) how features of the model are used and c) group outcomes such asconsensus and satisfaction.

3 - Post-catastrophe decision makingIan Durbach, Statistical Sciences, University of Cape Town,Department of Statistical Sciences, University of Cape TownRondebosch, 7701, Cape Town, Western Cape, South Africa,[email protected], Gilberto MontibellerUnexpected high-impact events present decision makers with dramaticnew information but also highlight their previous state of ignoranceabout the nature of possible outcomes, termed ’sample state igno-rance’. We examine post-catastrophe decision making using an ex-periment based on a simplified decision context. In the experiment,decision makers need to predict share prices and experience, for someseries, catastrophic falls in prices. We assess the impact of confrontingsuch events on two dimensions of judgement and decision making: therange of future estimates and attitudes towards risk.

� MA-28Monday, 8:30-10:00CC-A27

OR in Quality Management 1

Stream: OR in Quality ManagementInvited sessionChair: Ipek Deveci Kocakoç, Econometrics, Dokuz EylulUniversity Faculty of Economics and Administrative Sciences,Dokuz Eylul Universitesi Ikt.Id.Bil.Fak., Buca, 35160, Izmir,Turkey, [email protected]

1 - Dealing with multiple quality characteristics as re-sponses in prediction modeling: An aggregation ap-proachLeman Esra Dolgun, Industrial Engineering, AnadoluUniversity, 26555, Eskisehir, Turkey,[email protected], Gulser Koksal, Nimetullah BurnakThis study deals with aggregation of multiple quality characteristics(QCs) into a single measure reflecting the decision maker’s expertise.Types of interactions among QCs are identified through structured sur-veys with quality control experts, designed in this study. Appropriateaggregation models are proposed for the single measure representingthe multiple QCs, depending on the identified interactions. Any (singleresponse) prediction model can be used to relate product and processconditions to the resulting measure. Such a model provides valuablefeedback for quality improvement.

13


2 - A supervised learning procedure for monitoringmean and covariance simultaneouslyIpek Deveci Kocakoç, Econometrics, Dokuz Eylul UniversityFaculty of Economics and Administrative Sciences, DokuzEylul Universitesi Ikt.Id.Bil.Fak., Buca, 35160, Izmir, Turkey,[email protected], Eralp DoguThe signal issued by the simultaneously used mean vector control chartand/or covariance matrix control chart triggers the process profession-als to investigate the special cause. This signal does not always in-dicate that the root cause actually occurred at that particular point intime. Change point methods have been widely used to bridge the gapbetween monitoring and special cause identification. In this study,a supervised learning approach based on artificial neural networks isproposed to monitor location and dispersion simultaneously in multi-variate domain.

3 - Optimization on selection problem of solder paste in-spection machinesGokce Baysal, Econometrics, Dokuz Eylul University,Iktisadi ve Idari Bilimler Fakültesi, EkonometriBl,Dokuzcesmeler Kampusu, Buca, 35160, Izmir, Turkey,[email protected], Ipek Deveci KocakoçOne of the most important factors to give an inspection automationdecision is the rate of false decisions made by the inspection machine.This paper aims to provide help in selection of a solder paste inspection(SPI) machine for a printed circuit board (PCB) production line. Forthis purpose, two SPI machines have been tested for the same PCBs.After inspection, false "fail" decisions are reviewed by a qualified hu-man inspector and false rejection rate is determined. By utilizing aminimization model, the machine that gives the lowest false decisionsis recommended for buying.

4 - An adaptive Bayesian scheme for joint monitoring ofprocess mean and varianceGeorge Nenes, Mechanical Engineering, University ofWestern Macedonia, Bakola & Sialvera, 50100, Kozani,Greece, [email protected], Sofia PanagiotidouWe present a model for the economic optimization of a process opera-tion where two assignable causes may occur, affecting both the processmean and the variance. Thus, the process may operate in control, un-der the effect of either one of the assignable causes, or under the effectof both of them. The Bayes theorem is used to determine the proba-bilities of operating under the effect of each assignable cause, at eachsampling instance. Based on these probabilities, the parameters of thecontrol scheme are optimized economically. The proposed model issignificantly superior to simpler approaches.

� MA-29Monday, 8:30-10:00CC-A29

Mixed-Integer Quadratic Problems

Stream: Mixed-Integer Non-Linear ProgrammingInvited sessionChair: Adam Letchford, Department of Management Science,Lancaster University, Management School, LA1 4YX, Lancaster,United Kingdom, [email protected]

1 - A New Separation Algorithm for the Boolean Quadricand Cut PolytopesMichael Sørensen, Dept. of Economics and Business, AarhusUniversity, Fuglesangs Alle 4, 8210, Aarhus V, Denmark,[email protected], Adam LetchfordWe present separation algorithms for the polytopes associated withzero-one quadratic programming and the max-cut problem. Our ap-proach is based on separation of 0,1/2-cuts. We present a system ofweakened linear inequalities for the Boolean quadric polytope that isclosed under switching, and which allows the 0,1/2-separator to runin polynomial time. Using the so-called covariance mapping, we ob-tain an efficient separation algorithm for the cut polytope as well. Thelatter algorithm separates all odd bicycle wheel and all (p,2)-circulantinequalities.

2 - Cutting Planes from a Convex Quadratic Relaxationof the Stable Set ProblemFabrizio Rossi, Ingegneria e scienze dell’Informazione eMatematica, University of L’Aquila, Via Vetoio, I-67010,L’Aquila, Italy, [email protected], MoniaGiandomenico, Adam Letchford, Stefano SmriglioA great deal of research has been focused on finding both linear andsemidefinite relaxations for the Stable Set problem, but the resultingbranch-and-bound algorithms have not been completely successful inpractice. We propose an approach based on the construction of anellipsoid that (i) contains the Stable Set Polytope, and (ii) whose asso-ciated upper bound equals the Lovász theta number. This ellipsoid isthen exploited to derive strong (linear) cutting planes. Extensive com-putational results demonstrate that embedding these cutting planes ina branch-and-cut framework can be profitable.

3 - Nonstandard Semidefinite Bounds For Solving Ex-actly 0-1 Quadratic ProblemsFrederic Roupin, LIPN, CNRS-UMR 7030, Université Paris13, 99 avenue, Jean-Baptiste Clément, 93430, Villetaneuse,France, [email protected], NathanKrislock, Jerome MalickWe present an improved algorithm for finding exact solutions to Max-Cut and other binary quadratic programming problems. The algorithmuses a branch-(and-cut-)and-bound paradigm, using standard valid in-equalities and nonstandard semidefinite bounds. More specifically, weadd a quadratic regularization term to the strengthened semidefiniterelaxation in order to use a quasi-Newton method. The ratio of thetightness of the bounds to the time required to compute them can becontrolled by two real parameters.

4 - On the Gap Inequalites for the Max-Cut ProblemAdam Letchford, Department of Management Science,Lancaster University, Management School, LA1 4YX,Lancaster, United Kingdom, [email protected],Laura Galli, Konstantinos KaparisThe gap inequalities, introduced by Laurent and Poljak, form an ex-tremely general class of cutting planes for the max-cut problem. Re-cently, we have made several advances in the study of these inequal-ities. First, we have adapted them to general mixed-integer quadraticprograms. Second, we have proved various results concerned with thecomplexity of the inequalities. Third, we have designed and coded thefirst ever cutting-plane algorithm for the max-cut problem based ongap inequalities. Preliminary computational results are promising. Inparticular, the upper bounds obtained are very strong.

� MA-30Monday, 8:30-10:00CC-A31

Fat Tail Models in FinanceStream: Emerging Applications of Fat Tail Models inFinancial Modeling and EngineeringInvited sessionChair: Audrius Kabasinskas, Dept. of Mathematical Research inSystems, Kaunas University of Technology, Studentu 50-220, LT -51368, Kaunas, Lithuania, [email protected]

1 - Application of estimates of alpha-stable distributionto distress forecastAudrius Kabasinskas, Dept. of Mathematical Research inSystems, Kaunas University of Technology, Studentu 50-220,LT - 51368, Kaunas, Lithuania, [email protected],Zivile KalsyteApplication of alpha-stable distribution parameters to train a specificneural network is presented. Trained neural network was used to fore-cast behavior of distress in the US Healthcare sector. Distress valueof a company is calculated from financial ratios Gross Margin (GM),Current Ratio (CR) and Total Asset Turnover (TAT). Indicators for pre-diction of distress value were selected by genetic algorithm. Our re-sults shows that neural network trained with alpha-stable parameters ofgiven indicators has smaller forecast error comparing to trained with 31financial ratios.

14


2 - Hurst coefficient and alpha-stable parameter foranalysis of financial seriesRoman Rodriguez Aguilar, Economy, Instituto PolitecnicoNacional, kopoma 523, toltecas 166, C-1013, 14100, Mexico,Distrito Federal, Mexico, [email protected] paper analyzes the relationship between the parameter alpha ofalpha-stable distribution and the Hurst coefficient. It seeks to explorethe violation of two major assumptions in the modeling of financialseries (assume that the series are normally distributed and that the suc-cessive returns are independent) by estimating the Hurst coefficient andthe parameter alpha. One of the main results was the presence of frac-tal features and heavy tails in the serie of exchange rate peso-dollar forsome periods in different magnitudes, such differences are accentuatedduring periods of crisis.

3 - Xpress-Mosel: Modelling support for distributed, re-mote, and cloud computing with applications in Fi-nanceSusanne Heipcke, Xpress team, FICO, 54 rue Balthazar deMontron, 13004, Marseille, France,[email protected], Oliver Bastert, Yves ColombaniXpress-Mosel is a modelling and programming language that offersadvanced functionality for the implementation of optimization algo-rithms and also for the design of complete applications. Among theprominent features of Mosel figures its support for distributed mod-eling and problem solving on heterogeneous computing environments(including virtual environments). We shall discuss examples of opti-mization applications, particularly from the finance sector, that havebeen deployed using these advanced functionalities of the Mosel envi-ronment.

4 - High-Frequency Financial Data: A Mixed-Stable ap-proachAudrius Kabasinskas, Dept. of Mathematical Research inSystems, Kaunas University of Technology, Studentu 50-220,LT - 51368, Kaunas, Lithuania, [email protected],Igoris Belovas, Leonidas SakalauskasWith the introduction of electronic trading an enormous quantity oftrading data became available. High-frequency financial data can re-veal events and laws that are impossible to identify with monthly,weekly or daily data. We propose a mixed-stable approach to modelintra-daily data from German DAX component stocks returns. In prac-tice we often observe a large number of zero returns in high-frequencyreturn data due to the underlying asset price does not change at agiven very short time intervals. Our model is designed to capture theseunique features observed in high-frequency return data.

� MA-31Monday, 8:30-10:00CC-A33

Game Theory and Social Networks 1

Stream: Game Theory and Social NetworksInvited sessionChair: Juan Tejada, Estadística e Investigación Operativa I,Complutense Unversity of Madrid, Universidad Complutense deMadrid, Plaza de Ciencias, 3, 28040, Madrid, Spain,[email protected]

1 - Analysis of Network Formation with Learning Dy-namicsMichiharu Masui, Business Administration, IshinomakiSenshu University, 1 Shinmito Minamizakai, 9868580,Ishinomaki, Japan, [email protected] a game theoretic context, it is assumed that the decision maker hasperfect rationality. However, recently, many studies have restricted theperfect rationality or the structure of the game. These restrictions arecalled bounded rationality. In this study, we denote bounded rationalityby learning dynamics and complex networks. Learning dynamics andcomplex networks are widely used to describe the real world. The pur-pose of this study is to examine the behavior of the decision maker withlearning dynamics and the formation of networks by the interaction ofdecision makers.

2 - Centrality in Weighted Social Networks. A Game The-oretic Approach.Conrado M. Manuel, Statistics and Operation Research III,Complutense University of Madrid, Avenida Puerta de Hierros/n, 28040, Madrid, Spain, Spain, [email protected],Enrique González-Arangüena

We consider the case of networks in which ties are not just eitherpresent or absent, but have some weight attached to them, that we willinterpret as a function of duration, intimacy or intensity of the rela-tions. Moreover, we will suppose that actors in the network are alsoplayers in a cooperative TU game representing the interests that moti-vate their interactions. We propose as centrality measure for each nodeits probabilistic Myerson value (Calvo et al., 1999 and Gómez et al.,2008) assuming that the game is a symmetric one and thus, no a prioridifferences among players exist.

3 - Clustering networks by a new edge betweennessmeasure: A cooperative game theoretical approachDaniel Gomez Gonzalez, Estadistica e InvestigacionOperativa III, Complutense University, Av Puerta de Hierrro,28040, Madrid, Spain, [email protected], Javier Castro

The Girvan-Newman algorithm is one of the most used algorithms inclustering networks. This algorithm is based on the edge betweennessmeasure. In this work we present a new centrality measure based ongame theory that fixes the bad performance (in terms of modularity)that presents the classical edge betweenness measure when it is ap-plied to the Girvan-Newman algorithm. We test this new measure insome well-known examples in clustering network as: the Karate Clubnetwork, the author network, the dolphin network and Les miserablesnetwork.

4 - Computing centralities and the Myerson value inlarge social networksJuan Tejada, Estadística e Investigación Operativa I,Complutense Unversity of Madrid, Universidad Complutensede Madrid, Plaza de Ciencias, 3, 28040, Madrid, Spain,[email protected], Javier Castro, Daniel Gomez Gonzalez,Elisenda Molina

We present an approximate, but surprisingly accurate, method for cal-culating the Myerson value. This value is computable in polynomialtime for most of the games used to measure the centrality or impor-tance of a node in a network. The proposed method will be used inknown problems in order to measure the error and in real large prob-lems in order to prove its usefulness. The methods presented here cancalculate, with the same accuracy, the Myerson value of any type ofgame. The complexity is polynomial if the calculation of the charac-teristic function is computable in polynomial time.

� MA-32Monday, 8:30-10:00CC-A34

Combining Methods & Multimethodology

Stream: Soft OR and Problem Structuring MethodsInvited sessionChair: Alberto Paucar-Caceres, Business School, ManchesterMetropolitan University, Aytoun Building, Aytoun Street, M13GH, Manchester, United Kingdom, [email protected]

1 - Combining hard-soft OR for generative design think-ingPolyxeni Vassilakopoulou, Mechanical Engineering, NationalTechnical University of Athens, Heroon Polytechniou 9„Zografou, 15780, Athens, Greece, [email protected],Vassilis Tsagkas, Nicolas Marmaras

We aim to contribute to design research combining hard-soft OR tosupport circles of convergent-divergent thinking. We build upon sys-tem approaches (Churchman 1971; Ackoff 1974; Checkland 1981) anduse established OR techniques to nurture explorative, generative de-sign (Martin 2009). We illustrate with a case from banking. Unlike

15


Mingers et al 2009, we start with DEA modeling to probe questioningof the initial briefing (using different input-output arrangements) bring-ing to surface different worldviews and continue with SSM to developpurposeful activity models that guide the design process.

2 - Contrasting the ’expected’ versus ’realized’ benefitsof multi-methodologyFelipe Henao, Faculty of Management Science and Economy,Universidad Icesi, Universidad Icesi, Calle 18 # 122 - 135,Pance, 760031, Cali, Valle del cauca, Colombia,[email protected], L. Alberto Franco

Multi-Methodology has been extensively employed by OR/MS schol-ars to cope with the complexity of real-world situations. The literaturehas focused on how to design interventions and the challenges facedby OR researchers in deploying MM in practice. However, little at-tention has been paid to assess MM’s impacts from the participants’perspective and there is little understanding of the link between MM’sexpected and realized benefits. This paper addresses this gap by com-paring the intervention’s expected versus realized impacts, and relatethose to relevant group decision theory.

3 - Applying Soft Systems Methodology and SystemsDynamics to Environmental Management: A bound-ary critique frameworkAlberto Paucar-Caceres, Business School, ManchesterMetropolitan University, Aytoun Building, Aytoun Street, M13GH, Manchester, United Kingdom, [email protected],Diane Hart, Ricardo Rodriguez-Ulloa

We introduce the use of Ulrich’s Critical Systems Heuristics to orga-nize second order critical reflexion in systems practice and action re-search. The paper draws on critical systems thinking and illustratesits application in the context of an intervention using Soft SystemsMethodology (SSM) and Systems Dynamics (SD) to evaluate the en-vironmental impact of mining operations in southern Peru. We sketchthe context in which the systemic intervention was carried out report-ing on the initial SSM and SD stages to outline a 2nd order boundarycritique tool for systemic interventions.

4 - Structuring Community Issues - working with com-munity members to identify and (re)structure localproblematic issuesRebecca Herron, Community Operational Research Unit,University of Lincoln, LN6 7TS, Lincoln, United Kingdom,[email protected]

Problem Structuring Methods often take as their starting point a stake-holder group and an issue requiring exploration. Community ORspecifically works with groups of community members — taking asone of its meta-goals the establishment of community-based activitiesthat explore and develop problematic notions such as ’the commongood’. We give an example, highlight how workshops can be devel-oped flexibly in community settings, and the issue of how we can mea-sure the impact of our interventions.

� MA-33Monday, 8:30-10:00CC-A37

OR for Business and IndustrialDevelopment

Stream: OR for Development and Developing Coun-triesInvited sessionChair: Honora Smith, School of Mathematics, University ofSouthampton, Highfield, SO17 1BJ, Southampton, Hampshire,United Kingdom, [email protected]: Tatjana Staube, Institute of Civil Construction and RealEstate Economics, Riga Technical university, 1/7 Meza Street;Room 210, LV1048, Riga, Latvia, [email protected]

1 - Mining Industrial Transformations and Resource Re-Allocation Using a New MCDM Model – Case of Tai-wan and PolandMei-Chen Lo, National United University, 360, Miaoli,Taiwan, [email protected], Gwo-Hshiung Tzeng,Tian-Jong Hwu, Jerzy Michnik, Tadeusz Trzaskalik, MaciejNowakTechnology change dynamically and rapidly, therefore the industrialtransformation become an important issues for appropriate resourceallocations as well as rearrangement of its usefulness. In this study,the related concerns are discussed, and a hybrid MCDM model basedon DANP is used to assess the mining industrial transformation rela-tive performance viewed by different groups of experts from Taiwanand Poland. The procedures find the improvement directions and effi-cient allocations of resources for the mining industrial transformationsto achieve the aspiration levels.

2 - Using DEA to Inform an Integrated Response to Pol-icy Change: The Case of the South African PrivateHospital IndustryShivani Ramjee, Actuarial Science, University of Cape Town,Room 5.40, Robert Leslie Commerce Building, 7700, CapeTown, Western Cape, South Africa, [email protected] private hospital sector in South Africa has been criticized for theirperceived role in rising private medical expenditure and the associatedhealth-system inefficiencies. We use DEA to aid this sector in for-mulating an integrated response to National Health Insurance policyproposals. Operational, financial and clinical data for 43 hospitals for2007-2009 was used. It is clear that case-mix and quality of care dataare critical to engaging with policymakers. We identify the need forprivate hospitals to be permitted to employ doctors and highlight theconstraint posed by nurse attrition.

3 - Determining the Position of E7 Countries among De-veloping Countries by Kohonen NetworksHüseyin Tatlidil, Statistics, Hacettepe University, HacettepeUniversitesi Beytepe Kampusu Istatistik Bolumu, Beytepe,06532, Ankara, Turkey, [email protected], NurbanuBursaIn this study, developing 100 countries, grouped as low-middle andupper-middle income categories on the database of the World Bank,are clustered to their infrastructure investment sizes between the dataof years 1990 and 2011 by using Kohonen Networks and it is foundin which clusters Brazil, China, India, Indonesia, Mexico, Russia andTurkey known as E7 countries are. As Kohonen Networks cannot forma set of rules, C&RT Decision Tree is applied. Also, it is aimed to findthe geometric locations of E7 countries and the countries in the sameclusters as them through Multidimensional Scaling.

4 - Does Latvia use its advantageous location?Tatjana Staube, Institute of Civil Construction and Real EstateEconomics, Riga Technical university, 1/7 Meza Street; Room210, LV1048, Riga, Latvia, [email protected], InetaGeipeleAs small scale country with 0.5% from EU population and 70 timeslarger than Liechtenstein Latvia has economically advantageous geo-graphical location linking large and powerful investors countries likeRussia and Scandinavian states block. Latvia could cumulate all itsforce on servicing these big neighbors, but it does not. For this pur-pose it must have strong professional specialists and well developedinfrastructure. The rating of local specialists has been remarkably ris-ing for the latest decade due to integration into Europe and Latvianinhabitants’ natural ambitious leadership.

� MA-34Monday, 8:30-10:00CC-A39

Game SolutionsStream: Cooperative Game TheoryInvited sessionChair: Encarnación Algaba, Applied Mathematics II, SevilleUniversity, C/Camino de los Descubrimientos s/n, Isla de laCartuja., 41092, Sevilla, Spain, [email protected]

16


1 - Equilibria in Load Balancing GamesBo Chen, Warwick Business School, University of Warwick,Gibbet Hill Road, CV4 7AL, Coventry, United Kingdom,[email protected]

A Nash equilibrium (NE) in a multi-agent game is a strategy profilethat is resilient to unilateral deviations. A strong Nash equilibrium(SE) is one that is stable against coordinated deviations of any coali-tion. We show that, in the load balancing games, NEs approximate SEswell in the sense that the benefit of each member of any coalition fromcoordinated deviations is quantifiably quite limited. Furthermore, aneasily recognizable special subset of NEs exhibit even better approxi-mation of SEs. To establish the above results, we introduce and applya powerful graph-theoretic tool.

2 - Dealing with time issues in testing revealed prefer-ence axiomsFabrice Talla Nobibon, Decision Sciences and InformationManagement, KU Leuven, Naamsestraat 69, 3000, Leuven,Belgium, [email protected], YvesCrama, Frits Spieksma

This paper proposes a graph-coloring approach for dealing with theproblem of testing revealed preference axioms while taking into ac-count the timing of the observations. We translate well-known axiomsof the revealed preference theory into graph properties leading to ageneralized graph coloring problem with an additional vertex orderingconstraint. We derive a greedy algorithm for solving the problem whenthe obtained graph property is hereditary, and a dynamic programmingalgorithm when the property is not hereditary. In both cases if the prop-erty can be verified in polynomial time then these algorithms also runin polynomial time.

3 - Equilibrium, coalitive and computational aspects ofPareto-Nash-Stackelberg GamesValeriu Ungureanu, Department of Applied Mathematics,State University of Moldova, Gh. Asachi str., 58, ap.38,MD-2009, Chisinau, Moldova, Republic Of,[email protected]

Equilibrium Principles for Hierarchical and Mixture of Hierarchicaland Simultaneous Games are defined and examined by the means re-action sets, efficient reaction sets and its intersection. The set of equi-libria normally consists by more than one element and may have thecardinality of the continuum. If all players form a coalition, they ap-ply usually the criterion of Pareto optimality. Otherwise, there are alot of other possibilities. So, equilibrium, coalitive and computationalaspects of Pareto-Nash-Stackelberg Games are treated.

4 - Harsanyi power solutions for union stable systemsEncarnación Algaba, Applied Mathematics II, SevilleUniversity, C/Camino de los Descubrimientos s/n, Isla de laCartuja., 41092, Sevilla, Spain, [email protected], Jesus-MarioBilbao, Rene van den Brink

This paper analyzes solutions for games in which partial cooperationis based on union stable systems. In this context, we provide axiomaticcharacterizations of the Harsanyi power solutions which distribute theHarsanyi dividends proportional to weights determined by a powermeasure for union stable systems. Moreover, the Myerson value is theHarsanyi solution for the equal power measure and on a special sub-class the position value coincides with the Harsanyi solution obtainedfor the influence power measure.

� MA-35Monday, 8:30-10:00CC-A41

OR Industry; OR Software

Stream: OR in Industry and Software for OR (con-tributed)Contributed sessionChair: Belarmino Adenso-Diaz, Engineering School at Gijon,Universidad de Oviedo, Campus de Viesques, 33204, Gijon,Spain, [email protected]

1 - High-performance local search for TV media plan-ning on TF1Frédéric Gardi, e-lab, Bouygues SA, 40 rue Washington,75008, PARIS, France, [email protected]

In 2012, the leading French TV Channel TF1 launched a new offerconsisting in selling some packages of commercials, and not only com-mercials individually. In this offer, TF1 is responsible for planning theadvertising campaigns of its clients while satisfying some audience andbudget constraints, in others. Having outlined the resulting combina-torial optimization problem, we present the high-performance local-search heuristic developed by the Bouygues e-lab to solve it. Thisalgorithm is daily used by TF1 to automate and optimize the sale ofnearly 100,000 commercials per year.

2 - Cost optimization in planning the dubbing of a filmMa Luisa Carpente, Matemáticas, Universidade da Coruña,Facultade de Informática. Campus de Elviña, 15071, ACoruña, Spain, [email protected], Ana Cerdeira-Pena, SilviaLorenzo-Freire

In this paper we propose a binary programming model for optimizingcosts in a dubbing studio. This cost is directly related to the planningof the dubbing of a film. The model, programmed using the packageAIMMS, can always obtain optimal schemes for a wide range of films.Despite this, the time for obtaining many of the solutions is not entirelysatisfactory for practical handling in the production. That is why wehave designed an alternative algorithm that obtains optimal schemes infew seconds.

3 - Multistage Multiproduct Advertising BudgetingCesar Beltran-Royo, Estadística e Investigación Operativa,Universidad Rey Juan Carlos, Calle Tulipán, s/n, 28933,Móstoles, Madrid, Spain, [email protected], HuizhenZhang, Luis Blanco, Jose Almagro

In this paper we present a market optimization model to allocate the an-nual advertising budget in a multisage setting. Market response modelsallow to estimate the effect that marketing mix variables have on per-formance measures, such as sales or market share. Market optimiza-tion models combine utility functions with market response modelsin order to find an optimal marketing mix. We analyze the concavityproperties of this (non-linear optimization) model and present somenumerical examples that show its effectiveness in order to be used inthe advertising industry.

� MA-36Monday, 8:30-10:00CC-A43

DEA and Performance Measurement:Metodology 1

Stream: DEA and Performance MeasurementInvited sessionChair: Timo Kuosmanen, Aalto University School of Economics,00100, Helsinki, Finland, [email protected]

1 - Aggregation in cross-efficiency evaluations with in-duced ordered weighted averaging (IOWA) operatorsInmaculada Sirvent, Centro de Investigación Operativa,Universidad Miguel Hernández, Avd. de la Universidad, s/n,03202, Elche, Alicante, Spain, [email protected], Teresa Leon,Nuria Ramón, José L. Ruiz

We propose an enhancement of the cross-efficiency evaluationthrough the aggregation of cross-efficiencies by using induced orderedweighted averaging (IOWA) operators. To be specific, an ordering is tobe induced in the rows of the matrix of cross-efficiencies, so the IOWAoperator weights can be attached accordingly. Both the choice of thecriterion that induces such ordering and the specification of the ornessdegree allow the DM to incorporate his/her preferences on the relativeimportance that should be attached to the cross-efficiencies providedby the different DMUs.

17


2 - Performance Measurement Tools for HR Manage-ment related Processes based on System IdentityAnalysisJitka Civinova, Department of Informatics andTelecommunications, Czech Technical University, CzechRepublic, [email protected] identity method is a tool enabling the identification of opera-tion critical subjects and parameters. The aim of this paper is to designa set of measurement techniques to control these potential weak pointsin the process related to the knowledge management and to providethe best solution to stabilize them. The term stable system defines asystem that can be for its reliability and objectivity applied repeatedlywithout scattering resulting values. By monitoring individual elementsof the process, the optimization of complex process is achieved.

3 - One-stage and two-stage DEA estimation of the ef-fects of contextual variablesTimo Kuosmanen, Aalto University School of Economics,00100, Helsinki, Finland, [email protected], AndrewJohnsonIn two-stage data envelopment analysis (2-DEA) efficiency scores areregressed on z-variables representing operational conditions. We showconsistency of 2-DEA under more general conditions than earlier stud-ies. However, the finite sample bias of DEA carries over to the secondstage, causing bias in the regression coefficients. Using the fact thatDEA is a special case of convex nonparametric least squares, a newone-stage estimator (1-DEA) is developed to directly incorporate z-variables to standard DEA. Monte Carlo simulations show that 1-DEAperforms systematically better than 2-DEA.

4 - Estimation Efficiency of Teachers through Data En-velopment AnalysisVenkata Hrd, Business Management,C/O. Prof. Ramana, S.V.University, TIRUPATI, School of Commerce, Managementand Computers, 2H/3B, Mayura Apartments, V.V. Plaza,Reddy & Reddy Colony, TIRUPATI, 517501, TIRUPATI,Andhra Pradesh, India, [email protected] Envelopment Analysis to evaluate the efficiency of faculty basedon various faculty attributes and outputs at the student and institutelevel.The inputs are research grants, feed-back forms,IT, infrastructure,staff assistance, incentives for case,book writings, number of hoursspent on seminars or conferences.Outputs are consistency of rankingof institute or teacher, number of books, research papers, case studiespublished,number of presentations in conferences and seminars andfund development.DMUs are faculty members.

� MA-37Monday, 8:30-10:00CC-Act

Bioinformatics IStream: Computational Biology, Bioinformatics andMedicineInvited sessionChair: Jacek Blazewicz, Institute of Computing Science, PoznanUniversity of Technology, ul.Piotrowo 2, 60-965, Poznan, Poland,[email protected]: Metin Turkay, Department of Industrial Engineering, KocUniversity, Rumelifeneri Yolu, Sariyer, 34450, Istanbul, Turkey,[email protected]

1 - A data mining case study on brain cancer patientsFadime Uney-Yuksektepe, Industrial Engineering, IstanbulKultur University, E5 Karayolu Londra Asfalti Uzeri, AtakoyKampusu, 34156, Istanbul, Turkey, [email protected] data mining study is carried out in order to analyze the brain cancerpatients. Previous brain cancer patients’ data is taken from an oncolo-gist who works in a neurosurgery department of a private hospital. Theaim of the study is to find out useful information that will help the doc-tors to predict the survival status and survival length of patients. Thedata set on hand is studied by different data classification algorithms.The most suitable and efficient method is determined and useful rulesand models that will help the doctors are obtained.

2 - Markov models of voltage gating of gap junctionchannelsHenrikas Pranevicius, Business Informatics, KaunasUniversity of Technology, Studentu 50, Kaunas, Lithuania,Lithuania, [email protected], Feliksas Bukauskas,Mindaugas Pranevicius, Osvaldas Pranevicius, SauliusVaiceliunasThe major goal of this study was to create a continuous time Markovmodels of voltage gating of gap junction (GJ) channels formed of con-nexin protein. This goal was achieved using the piece linear aggregate(PLA) formalism to describe function of GJs and transforming PLAinto Markov process. Proposed method was implemented to creatingmodels of voltage gating of GJ channels containing in each hemichan-nel of the GJ channel fast and slow gating mechanisms, operating be-tween open and closed states.

3 - Inferring two-generation kinship (or genealogy) frommicrosatellite samplesTanya Berger-Wolf, University of Illinois at Chicago, 60607,Chicago, United States, [email protected], Daehan Won,Chun An Chou, W. Art Chaovalitwongse, Bhaskar Dasgupta,Ashfaq Khokhar, Mary V. AshleyWe propose a combinatorial optimization framework for parental re-construction of a single-generation population using microsatellitedata. Without prior information about the population, our optimizationframework uses the combinatorial concepts of the Mendelian laws toreconstruct sibling groups and in turn identify the associated parentalgenotypes. The proposed framework was tested on real and simulateddata sets with different mating patterns: monogamy, semi-polygamy,and polygamy. The results show that our framework can efficiently andaccurately infer the real parental genotypes.

� MA-38Monday, 8:30-10:00HH-Colombus

Integer Linear and Linear MultiobjectiveOptimization

Stream: Multiobjective OptimizationInvited sessionChair: Walter Habenicht, Business Administration, University ofHohenheim, Lst. fuer IBL(580 B), 70593, Stuttgart, Germany,[email protected]

1 - Enumerative Cuts in Integer Linear Vector Optimiza-tionWalter Habenicht, Business Administration, University ofHohenheim, Lst. fuer IBL(580 B), 70593, Stuttgart, Germany,[email protected] paper deals with a conceptual framework of a hybrid method forsolving linear integer vector optimization problems. It combines cut-ting planes with an enumeration approach. The concept is based onenumerative cuts, which is well established approach in (single cri-terion) integer linear optimization. In this paper we show how enu-merative cuts can be integrated in an interactive procedure for solvinginteger linear vector optimization problems.

2 - Graphical exploration of the weight space in multiob-jective integer linear programmingMaria João Alves, Faculty of Economics, University ofCoimbra / INESC-Coimbra, Av. Dias da Silva, 165, 3004-512,Coimbra, Portugal, [email protected], João Paulo CostaWe present an approach that determines, at each time, a subset of theindifference region in the weight space associated with an efficient so-lution of a multiobjective integer linear programming (MOILP) prob-lem. Based on the graphical exploration of these regions, we proposean algorithm to compute all supported efficient solutions "around’ analready known solution, or even all the supported efficient solutions,of a tri-objective problem. We also show a computer implementationof the algorithm.

18


3 - A multicriteria linear programming model for measur-ing efficiency in productive processesDavid Alcaide Lopez de Pablo, Estadistica, InvestigacionOperativa y C., Universidad de La Laguna, Avda. AstrofisicoFrancisco Sanchez, 38206, La Laguna, Tenerife, Spain,[email protected], Rafaela Dios-Palomares, Angel M. Prieto

The efficiency in production has been usually analysed with the pro-duction frontier function. Efficiency scores are based on distances tothe frontier. Improvements regard to the total consumption of eachinput. However, in many cases, each input is detached in its stagesand trade—off among them is possible. Such analysis provides thetotal optimal consumption of each input, as data envelopment analy-sis does, and the most efficient assignment in each stage. This paperstudies technical efficiency with this perspective. A non—parametricalmethodology and a multicriteria model are proposed.

4 - Factoraggregation and its application for solvingbilevel linear programming problems.Pavels Orlovs, Department of Mathematics, University ofLatvia, Zellu street 8, LV-1002, Riga, Latvia,[email protected], Olga Montvida

We introduce the notion of a factoraggregation, which is a special con-struction of a general aggregation operator, based on an ordinary ag-gregation operator. We apply the factoraggregation for analysis of op-timal solution of bilevel linear programming problems. The aggrega-tion observes lower level objective functions considering the classes ofequivalence generated by an objective function at the upper level. Theproposed method is illustrated with numerical and graphical examples.

� MA-39Monday, 8:30-10:00HH-Cousteau

MADM Applications I

Stream: MCDA: New Approaches and ApplicationsInvited sessionChair: Chin-Tsai Lin, Department of Business Administration,Ming Chuan University, 5F, No.130, Jihe Rd., Shihlin District„11111, Taipei, Taiwan, [email protected]: Chie-bein Chen, International Business, National DongHwa University, 1, Sec. 2, Da-hsueh Rd. Shou-feng, 974, Hualien,Taiwan, [email protected]

1 - A storage location assignment problem for a pick-and-pass warehousing system with multiple pickersand congestion considerationsJason Chao-Hsien Pan, Department of BusinessAdministration, Takming University of Science andTechnology, No.56, Sec.1, Huanshan Rd., NeihuDistrict,Taipei City, Taiwan 11451, R.O.C., Taiwan,[email protected], MingHung Wu

This paper develops a performance evaluation method for a pick-and-pass system by describing the operation of a picker as a GI/G/1openqueueing network to estimate the expected response time of the pickerin a picking zone. Based on the proposed method, this study providestorage assignment algorithm to balance the workload of each zoneso that the performance of the system can be improved. A simulationmodel is used to implement the proposed algorithm and to compare thethroughput for different polices as well. The results indicate that theproposed algorithm outperforms existing policies.

2 - The Effects of Individualism/Collectivism Orienta-tionsKuochung Chang, International Business, National DongHwaUniversity, Taiwan, [email protected]

The purpose of study was to examine the relationship between indi-vidualism/collectivism orientations, self-construals, and quality moti-vation. Participants were chosen from three Industrial Parks in Taiwan.The findings indicated that collectivism orientation not only had signif-icant effect on the technological dimension of quality motivation, but

also had indirect effect through the individuals’ interdependent self-construal. The study also found that independent self-construal me-diated the relationship between individualism orientation and the hu-manistic dimension of quality motivation.

3 - An EOQ Model with Defective Items and Imperfect In-spection ProcessChia-Huei Ho, Business Administration, Ming ChuanUniversity, 250 Zhong Shan N. Rd.,Sec. 5„ 111, Taipei,Taiwan, [email protected], Liang-yuh Ouyang,Chia-hsien Su

This paper considers an EOQ model with defective goods in inwardpurchases and imperfect inspection process. We assume that screeningerrors (including type I and type II) may occur due to imperfect inspec-tion quality, and retailer invests capital to improve inspection quality.The objective is to find the optimum order quantity and the power ofthe test such that total cost per unit time is minimized. Mathematicalanalysis is used to find the optimal solution. Numerical examples andsensitivity analysis will be presented as well to demonstrate the resultsof the proposed model.

4 - Market Effectiveness and Reinstatement of Top Man-agersChin-Tsai Lin, Department of Business Administration, MingChuan University, 5F, No.130, Jihe Rd., Shihlin District„11111, Taipei, Taiwan, [email protected], Yi-HsienWang, Jung-Ho Lu, Hwa-Rong Shen, Yang-Sheng Wang

This study investigated the impact of the reinstatement of former CEOon the stock market reaction of listed corporations in Taiwan. There-fore, corporate decision often to reinstate former CEO, because ex-pected former CEO with experience in managerial practices can im-prove corporate performance. The aim of this study is to provide prac-tical guidelines to companies for decision making, and helpful infor-mation for adjust expectations of investors to alter portfolio allocation.

� MA-40Monday, 8:30-10:00HH-Livingstone

MADM Applications V


1 - The study on the festival event experience benefitstructure relationship model — a case of KaohsiungZuoying Wannian Folklore FestivalTang-Chung Kan, Department of Travel Management,National Kaohsiung University of Hospitality and Tourism, 1,Sung-ho Rd. Hsia-Kang, 81267, Kaohsiung, Taiwan,[email protected], Hung-ju Chien, Vivien Y.C. Chen

Event’s place attachment and tourist’s experience quality, are the im-portant key factors of festival event success and event benefits. Thepurpose of this research is to investigate the relationships among theplace attachment, experience quality and events benefits of the Kaohsi-ung Zuoying Wannian Folklore Festival. Use LISREL analysis systemto investigate linear structure relation model. The result can be developthe relationship model among place attachment, experience quality andevent benefits’s aspect.

2 - Dealer selection for an automotive company usingELECTRE I and TOPSIS methodologiesElif Maç, Industrial Engineering, Kocaeli University / Naturaland Applied Sciences, Turkey, [email protected],Zerrin Aladag

19


Sales & Aftersales network is of vital importance for customer satisfac-tion with automotive companies. New technologies, competitive priceand charming design increase new customer potential and of coursecustomer satisfaction; but these are not enough to sustain long termsatisfaction and loyalty. Companies have to select accurate contacts fordealer role. Multicriteria analysis is meaningful to integrate various as-pects into a uniform evaluation process. We propose to use ELECTREI and TOPSIS to choose one dealer among 6 dealership applicants foran automotive company in Turkey.

3 - An empirical analysis between Japanese Govern-ment and NPOs in the New PeriodKuo Tzu-hsuan, Department of Applied Japanese„ NationalKaohsiung University Of Hospitality And Tourism, Taiwan,NO,311 sonchung street Kaohsuing, NO, 807, Kaohsiung,Taiwan, Taiwan, [email protected], Vivien Y.C. Chen,Tang-Chung Kan

After the 1990s, Japan has been deeply influenced by internationaltrends as well as its own political and economic development. This dis-sertation constructs a society-centered theory to explore and to analyzethe relationship between the Japanese government and NPOs duringthe new period. I examine the change and the development by divid-ing the progress into "traditional period’ and "new period’ with afterthe cold war in the 1990s. This study is emphasized on showing howimportant Japanese government and NPOs cooperative partnership.

4 - A multi-objective strategy for Pareto set refinementAdriano Lisboa, ENACOM - Handcrafted Technologies,31255530, belo horizonte, minas gerais, Brazil,[email protected], Fellipe Santos, DouglasVieira, Rodney Saldanha, Marcus Lobato

This paper introduces a strategy for Pareto set refinement in order toallow where there will be more samples of the Pareto set. The funda-mental idea is to find the Pareto set vertices and then use refinementcriteria in order to generate new seeds for extra Pareto set samples.The strategy is especially efficient when coupled to intrinsic multi-objective optimization algorithms. The strategy is tested in analyticalbenchmarking problems and real world applications.

� MA-41Monday, 8:30-10:00HH-Heyerdahl

Evolutionary Multiobjective Optimization

Stream: Multiobjective OptimizationInvited sessionChair: Juergen Branke, Warwick Business School, University ofWarwick, CV4 7AL, Coventry, United Kingdom,[email protected]: Arnaud Liefooghe, LIFL - CNRS - INRIA, Université Lille1, Lille, France, [email protected]

1 - Preference-based Evolutionary Algorithm for ManyObjective Optimization ProblemsAnkur Sinha, Department of Information and ServiceEconomy, Aalto University School of Economics, 4.13Chydenia, Runeberginkatu 22-24, 00100, Helsinki, Finland,[email protected], Kalyanmoy Deb, Pekka Korhonen, JyrkiWallenius

Recent contributions towards handling many objective optimizationproblems propose to seamlessly integrate the decision maker with theexecution of an evolutionary algorithm. Preference elicitation from thedecision maker is performed at the intermediate steps of the algorithmto direct the search in preferred regions. In this study, we discuss onesuch technique where an interactive evolutionary algorithm based onimplicitly defined value functions is used to find the most preferredsolution in a fixed number of interactions with the decision maker.

2 - Automatic Design of Multi-Objective OptimizatonMetaheuristicsManuel López-Ibáñez, IRIDIA, Université Libre de Bruxelles,Av. F. Roosevelt 50, CP 194/6, 1050, Bruxelles, Belgium,[email protected], Thomas StützleWe propose to automatically instantiate multi-objective ant colony op-timization (MOACO) algorithms from a flexible framework that en-capsulates most of the design alternatives found in the literature. Thisapproach allows us to generate, with little effort, new MOACO algo-rithms that are designed for specific multi-objective problems. Theautomatically generated MOACO algorithms are able to outperformthe MOACO algorithms from the literature. The proposed approachcan be easily extended to other multi-objective metaheuristics, such asmulti-objective evolutionary algorithms.

3 - Anytime Local-Search Algorithms for Multi-ObjectiveOptimizationJérémie Dubois-Lacoste, IRIDIA, Université Libre deBruxelles, 50, av. F. Roosevelt, 1050, Bruxelles, Belgium,[email protected], Manuel López-Ibáñez,Thomas StützleMany optimization algorithms are designed to deliver the best possibleoutcome after a given computation time. However in many real-lifesituations, the available computation time may be unknown a priori."Anytime" algorithms, on the other hand, are designed to obtain thebest possible results whenever the algorithm is stopped. In this workwe present anytime variants of two algorithms for multi-objective op-timization, Two-Phase Local Search and Pareto Local Search, whichare essential components of state-of-the-art algorithms for well-knownbi-objective problems.

4 - Multi-objective combinatorial benchmark problemsand search space structureArnaud Liefooghe, LIFL - CNRS - INRIA, Université Lille 1,Lille, France, [email protected] first discuss the issues of designing and analyzing multi-objectivecombinatorial optimization problem instances, and we identify exist-ing attempts of defining such benchmark for different domains. Next,we propose a possible alternative in the form of a multi-objective vari-ant of NK-landscapes with objective correlation. We define and ana-lyze a number of problem features that impact the search space charac-teristics, and we give guidelines for the design of evolutionary multi-objective optimization algorithms and other metaheuristics, based onthe main search space features.

� MA-42Monday, 8:30-10:00BW-Amber

Simulation 1Stream: SimulationInvited sessionChair: Michael Manitz, Technology and Operations Management,Chair of Production and Supply Chain Management, University ofDuisburg/Essen, Mercator School of Management, Lotharstr. 65,47057, Duisburg, Germany, [email protected]

1 - A research of computer network systems’ productiv-ity and efficiency with simulation modelingSultan Ceren Salkın, Industrial Engineering, IstanbulTechnical University, Kenan Evren Bulvarı Esencan Sitesi Ablok 2/5, Çukurova, 01170, ÇUKUROVA, ADANA, Turkey,[email protected] networks are the most prevalent systems with manyapplications-like internet sales. Internet commands are working withcomplex algorithms that cause communication troubles. With thisstudy, it is proposed to show the usage efficiency of the computer net-work systems by describing system topology with ARENA simulationmodel. First, web system working principle was analyzed; data wassuited to probability distributions. System variables, attributes, enti-ties were defined and after results, system resource usage ratios werecriticized to make suggestions.

20


2 - The impact of a waiting-time threshold in overflowsystems with impatient customersMichael Manitz, Technology and Operations Management,Chair of Production and Supply Chain Management,University of Duisburg/Essen, Mercator School ofManagement, Lotharstr. 65, 47057, Duisburg, Germany,[email protected], Raik Stolletz

We analyze the performance of call centers with impatient customers,two levels of support, and an overflow mechanism. Waiting calls fromthe front-office queue, if not, reneging are sent to the back office ifat least one back-office agent is available and if a certain thresholdon the waiting time is reached. We approximate such systems via acontinuous-time Markov chain that allows for overflow immediatelyupon arrival. Two different approaches for the derivation of the respec-tive probability of an overflow are developed the reliability of which isshown by numerical experiments.

3 - Value analysis of healthcare asset tracking systemsusing simulationEce Arzu Demircan, Industrial Engineering, TOBB Universityof Economics and Technology, 06560, Ankara, Turkey,[email protected], Nilgun Fescioglu-Unver

Healthcare systems try to improve the quality of patient care and safety,and reduce their costs by use of new technologies such as Radio Fre-quency Identification (RFID) based asset tracking systems. Some hos-pitals abstain from using this technology because they are not sure howmuch value the system will bring. The aim of this research is to gen-erate a value analysis framework for healthcare asset tracking systems,and develop a simulation model which can measure the benefits of as-set tracking in hospitals of different scales and patient arrival-rates.

� MA-43Monday, 8:30-10:00BW-Granite

Multistage Stochastic Programming

Stream: Stochastic ProgrammingInvited sessionChair: Alois Pichler, Statistics and Operations Research,University of Vienna, Universitätsstraße 5, 1010, Wien-Vienna,Vienna, Austria, [email protected]

1 - Ambiguity in multistage stochastic programmingand worst case treesGeorg Pflug, Department of Statistics and Decision SupportSystems, University of Vienna, Universitaetsstrasse 5,A-1010, Vienna, Austria, [email protected]

We consider the multistage ambiguity problem, i.e. a multistage deci-sion model, where the governing probability is only known to lie in aset of models. These ambiguous models are defined through balls w.r.t.the nested distance of tree processes.

The ambiguity problem is formulated and solved as a minimax prob-lem. A subproblem is to find tree processes, which belong to the ambi-guity set, but give worst values for the optimization problem. We givesome algorithm to identify this wort case tree process and show someillustrative examples.

2 - Measures of information in multistage stochasticprogrammingMarida Bertocchi, Department of Mathematics, Statistics,Computer Science and Applications, University of Bergamo,Via dei Caniana 2, 24127, Bergamo, BG, Italy,[email protected], Francesca Maggioni, ElisabettaAllevi

Multistage stochastic programs are usually hard to solve in realisticallysized problems. Providing bounds for their optimal solution, may helpin evaluating whether it is worth the additional computation for thestochastic program versus simplified approaches. In this talk we gen-eralize the value of information gained from deterministic, pair solu-tions and rolling-horizon approximation in the two-stage case to the

multistage stochastic formulation. We show that theorems proved fortwo stage case are valid also in the multistage case. Numerical resultson a case study illustrate the relationships.

3 - Stochastic programming formulations of coherentmultiperiod risk measurementMartin Densing, Energy Economics, PSI, Paul ScherrerInstitut, 5232, Villigen, Switzerland, [email protected]

Coherent risk measurement is related to stochastic optimization. Anexample is the single-period risk measure Conditional-Value-at-Risk(CVaR) that can be formulated as a stochastic linear program. The fa-vorable properties of CVaR led to recent interest in multi-period exten-sions. We consider the case of coherent multi-period risk measurementfor a value process. We discuss an extension of CVaR to the productspace of state and time, and a time-consistent recursive definition. Ina finite setting on a scenario tree, we consider duality relations, andintegration in mean-risk optimization.

4 - Decomposition of Risk MeasuresAlois Pichler, Statistics and Operations Research, Universityof Vienna, Universitätsstraße 5, 1010, Wien-Vienna, Vienna,Austria, [email protected]

We provide a new identity for the multistage Average Value-at-Riskand selected, other risk measures.

The identity is based on the conditional Average Value-at-Risk at ran-dom level, which is introduced. It is of interest in situations, where theinformation available increases over time, so it is – among other appli-cations – customized to multistage optimization. The identity relatesto dynamic programming. We elaborate further dynamic programmingequations for specific multistage optimization problems and derive acharacterizing martingale property for the value function.

� MA-44Monday, 8:30-10:00BW-Marble

Airline Revenue Management

Stream: Revenue ManagementInvited sessionChair: Tatsiana Levina, School of Business, Queen’s University,143 Union str, K7L 3N6, Kingston, Ontario, Canada,[email protected]: Yuri Levin, School of Business, Queen’s University, 143Union str, K7L 3N6, Kingston, Ontario, Canada,[email protected]

1 - Uncertainty in air cargo revenue managementEmily Cookson, Management Science, Lancaster University,LA1 4YX, Lancaster, United Kingdom,[email protected], Kevin Glazebrook, JoernMeissner

The air cargo industry faces some unique challenges including highlyvolatile demand and variable tendering. In the case of mixed carriers,the prioritisation of passengers and their baggage complicates bellyspace allocation. A new dynamic programming formulation has beendeveloped to maximise a carrier’s expected profit from cargo. Capac-ity and demand uncertainty is modelled using probability distributions.Ad hoc acceptance decisions are made using dynamic prices computedusing various numerical analysis techniques. The sales problem oflong-term versus short-term contracts is discussed.

2 - Airline product design based on consumer prefer-ence for ticketing restrictionsDong Myong Lee, Business Administration, Korea AerospaceUniversity, Hwajun-dong, 412-791, Goyang, Gyunggi, Korea,Republic Of, [email protected], Do Hee Han, MeghanaDhungana

Ticketing restrictions are important factors for designing airline prod-ucts. With various types of ticketing restrictions, airlines have offeredmany airline tickets to differentiate their customers. This paper fo-cuses on the product design problem for airlines based on customer

21


preference values. For measuring the values, we apply a choice basedconjoint analysis model. With empirical analysis on the survey datafor air travelers in Korea, we will show airlines can offer various kindsof air ticket products with a high selecting possibility.

3 - A continuous-time dynamic pricing model knowingthe competitor’s pricing strategyKimitoshi Sato, Graduate School of Finance, Accounting andLaw, Waseda University, 103-0027, Tokyo, Japan,[email protected], Katsushige SawakiWe consider a dynamic pricing model for a firm knowing that a com-petitor adopts a static pricing strategy. We establish a continuous timemodel to analyze the effect of the dynamic pricing on the improvementof revenue. Customers arrive to purchase tickets in accordance with ageometric Brownian motion. We derive an explicit closed-form ex-pression for optimal pricing policy to maximize the expected revenue.It is shown that when the competitor adopts a flat rate pricing policy,a dynamic pricing is not always effective in terms of the expected rev-enue compared to the fixed pricing strategy.

4 - Dynamic pricing of flight passesYuri Levin, School of Business, Queen’s University, 143Union str, K7L 3N6, Kingston, Ontario, Canada,[email protected], Mikhail Nediak, HuseyinTopalogluMany airlines have recently introduced a "flight pass" which targetscustomers in a frequent traveler segment. This product permits itsholder to obtain a seat on a specified number of flights at a fixed price.Dynamic pricing of passes is contingent upon pricing of individualflights and how customers choose between passes and individual book-ings. Customers in this setting are inherently forward-looking. Wepresent a stochastic dynamic programming formulation for this prob-lem and analyze the properties of the pricing policy.

� MA-45Monday, 8:30-10:00BW-Water

Advertising decisions in the supply chain

Stream: Revenue ManagementInvited sessionChair: Salma Karray, Faculty of business & IT, UOIT, 2000Simcoe Street North, L1H7K4, Oshawa, Ontario, Canada,[email protected]

1 - Bilateral Cooperative AdvertisingMarcus Kunter, Faculty of Business & Economics, AachenUniversity, Kackertstr. 7, D-52064, Aachen, Germany,[email protected] study channel coordination in a manufacturer-retailer channel withbilateral moral hazard. We argue that existing approaches involve as-sumptions that contradict reality or require instruments that are oftennot adopted by managers (ex-post bargaining, per-unit cost sharing, in-troduction of a third party). As an alternative, we show that channelcoordination is achieved by contracting on bilateral cooperative adver-tising and wholesale pricing. However, the channel may not be coor-dinated by this contract since it is dominated by other contracts fromchannel members’ individual perspectives.

2 - Cooperative Advertising in a Dynamic Retail MarketDuopolySuresh Sethi, School of Management, SM30, University ofTexas at Dallas, 800 W. Campbell Rd., 75080, Richardson,TX, United States, [email protected], Anshuman ChutaniCooperative advertising is a key incentive offered by a manufacturer toinfluence retailers’ promotional decisions. We study cooperative ad-vertising in a dynamic retail duopoly where a manufacturer sells hisproduct through two competing retailers. We model the problem asa Stackelberg differential game in which the manufacturer announceshis shares of advertising costs of the two retailers or his subsidy rates,and the retailers in response play a Nash differential game in choosingtheir optimal advertising efforts over time.

3 - Optimal Sequence of play for channels with cooper-ative advertising programsSalma Karray, Faculty of business & IT, UOIT, 2000 SimcoeStreet North, L1H7K4, Oshawa, Ontario, Canada,[email protected]

Most research about cooperative advertising programs in channels re-lies on the assumption that manufacturers and retailers decide of pric-ing and marketing efforts simultaneously. This paper evaluates thiscentral assumption and investigates the optimal periodicity (sequenceof play) of pricing and marketing efforts (ME) decisions for a distribu-tion channel using a game theoretic analysis.

22

EURO 25 - Vilnius 2012 MB-03

Monday, 10:30-12h00

� MB-01Monday, 10:30-12h00RB-Alfa

IFORS Invited Tutorial: Professor ErhanErkutStream: Keynote, Tutorial and Plenary LecturesTutorial sessionChair: Elena Fernandez, Universitat Politecnica de CAtalunya,Barcelona, Spain, [email protected]

1 - How to Make OR the Most Liked Course in the Cur-riculum?Erhan Erkut, Rector, Ozyegin University, Kusbakisi Cad.No:2, Istanbul, Turkey, [email protected]

As teachers of OR, many of us would like to think that OR is one ofthe most valuable topics in university education. However, past ex-perience shows that some of the students in our classrooms may notagree with this. How do we close this perception gap, and attract thebest students to OR? This is just another problem, and if we think weare the best-trained problem solvers in the world, we should be ableto overcome it easily. Unfortunately, while we may have exceptionaltraining in analytical thinking and quantitative analysis, many of uslack the tools necessary for strategic planning, brand creation, and rep-utation management. Furthermore, most of us lack formal training inteaching, which makes matters worse.

My colleagues at the University of Alberta and I have experimentedwith various methods and tools over two decades, and we estab-lished that an OR curriculum can be very successful even in a busi-ness school– a considerably more challenging environment than anengineering school. The winning recipe includes ingredients such asspreadsheets, real-time modeling ("slow learning"), course manage-ment system, web tools, on-line communication, large classes, labs,student assistants, videos, guest speakers, on-line exams, and groupprojects; all delivered with a heavy emphasis on applications, and witha student-centered pedagogical approach. The results have been verypositive; the reputation of the introductory course became "most use-ful," many of the best students took our electives, all members of ourteam won teaching awards, and through our collaboration with the in-dustry we established a know-how transfer unit called Centre for Ex-cellence in Operations.

In this tutorial, I will describe the problem we faced, and our attemptsto solve it with examples of what worked and what did not, with an em-phasis on what audience members could transfer to their institutions.The focus is not only on making one course popular, but on managingthe entire student supply chain, complete with a student club, compe-titions, conferences, internships, and graduate programs. If you areinterested in improving the perception and image of OR in your insti-tution through teaching (and strategic planning), you are welcome toattend this session.

� MB-02Monday, 10:30-12h00RB-Beta

New scheduling models and algorithms

Stream: SchedulingInvited sessionChair: Dirk Briskorn, University of Siegen, Germany,[email protected]

1 - A tactical approach to skill managementMurat Firat, Mathematics and Computer Science, EindhovenUniversity of Technology, Postbus 513, 5600 MB, Eindhoven,Netherlands, [email protected], Cor Hurkens, Alexandre Laugier

In multi-skill workforce planning, employees are specialized in differ-ent fields. We consider a time horizon in which several projects, eachwith a target makespan, are to be completed. If a makespan is notachievable with available skills, an option is outsourcing some jobsby using a fixed budget. We investigate another option of increasingthe available amount of skills with the aim of saving some near-futureoutsourcing costs. This study addresses a center question: "how toevaluate the value of a skill?".

2 - Integrated planning of jobs and maintenance activi-ties on a single machineDirk Briskorn, University of Siegen, Germany,[email protected], Stefan Bock, Andrei Horbach

We focus on single machine scheduling subject to machine deteriora-tion. The maintenance level specifies the machine’s current mainte-nance state. While jobs are processed the maintenance level drops bya certain - possibly job-dependent - amount. A maintenance level ofless than zero is associated with the machine failure. Consequently,scheduling maintenance activities that raise the maintenance level maybecome necessary in order to prevent maintenance level being becom-ing negative. We present complexity results and approaches consider-ing two types of maintenance activities.

3 - A labeling algorithm for scheduling deliveries with e-vehiclesSleman Saliba, Corporate Research Germany, ABB AG,Wallstadter Str. 59, 68526, Ladenburg, Germany,[email protected], Sven Krumke, André Chassein

In this presentation, we solve a real-world scheduling problem occur-ring for courier services employing electrical vehicles. E-vehicles arerequested to make deliveries during given time windows while theyneed to be flexibly re-charged at certain re-charging points during thedelivery trips. The objective is to make as many deliveries as possi-ble while minimizing the travel cost. This problem can be formulatedas a Resource Constraint Shortest Path Problem (RCSPP). Since theRCSPP is NP-hard, we present a fast Labeling Algorithm that can beemployed in real-world scheduling systems.

4 - Hybrid method for minimizing earliness and tardi-ness penalties in a single-machine problem with acommon due dateChristophe Wilbaut, LAMIH, University of Valenciennes, LeMont Houy, ISTV2, 59313, Valenciennes, France,[email protected], Saïd Hanafi,Rachid Benmansour, Mustapha Ratli, Rita Macedo

In this paper we deal with the single machine earliness and tardinessscheduling problem against common due dates. We propose a hybridmethod mixing heuristics and mathematical programming to solve it.The approach reaches strong upper bounds and lower bounds using:relaxation techniques; a temporary setting of variables to generate andsolve reduced problems; the generation of pseudo-cuts to strengthenthe problem formulation. Heuristics can be integrated at different lev-els to enhance the approach. The efficiency of the method is shownthrough its results on a set of available instances.

� MB-03Monday, 10:30-12h00RB-L1

Business Excellence in Logistics: FutureTransport and Production

Stream: Business Excellence in LogisticsInvited sessionChair: Martin Josef Geiger, Logistics Management Department,Helmut-Schmidt-University, Holstenhofweg 85, 22041, Hamburg,Germany, [email protected]

1 - A Scheduling heuristic for surface treatment processof steel coils and its applicationOzlem Uzun Araz, Industrial Engineering Dept., Celal BayarUniversity, Turkey, [email protected], Ozgur Eski,Ceyhun Araz, Levent Bayoglu

23

MB-04 EURO 25 - Vilnius 2012

The paper deals with the sequencing problem for the surface treatmentprocess of steel coils in a Turkish Fastener Company. Our problemis to sequence the steel coils waiting for surface treatment process andschedule a single crane which performs material handling within stagesfor minimizing the total tardiness of jobs. A scheduling heuristic isproposed that is capable of handling special constraints about mate-rial types and material handling requirements. The performance of theheuristic has been tested by an application on the surface treatment linein Norm Fasteners Co., Turkey.

2 - Future Transport Demand Analysis for TurkeyFusun Ulengin, Industrial Engineering, Dogus University,Dogus Universitesi, Muhendislik Fak. Endustri Muh. Bol,Acibadem, Kadikoy, 34722, Istanbul, Turkey,[email protected], Sule Onsel, Burc Ulengin, OzayOzaydin, Özgür Kabak

The main objective of this study is to conduct a demand analysis toidentify whether the capacity expansion plans will be in line with thelikely development of transport demand, given expected trends of thedrivers of transport demand. An integration of multiple regression andthe Bayesian Net (BN) is used for this purpose. Scenario analysisis also conducted in order to see the impact of different investmentstrategies to reach different target in transport mode demand. The timehorizon is in line with planned durations of concession contracts fortransport infrastructure.

3 - Multi-criteria analysis of transport policy targets withthe help of telematic servicesGideon Mbiydzenyuy, Computing, Blekinge Institute ofTechnology, Biblioteksgatan 4, 374 24, Karlshamn, Sweden,[email protected]

The aim of this paper is to assess how the choices of telematic sys-tems through service offers can enable our understanding of differ-ent possibly competing transport policy goals in the context of roadfreight. Since such goals are aimed at different targets, we consider theachievement of different transport policy goals using different telem-atic systems as a multi-criteria decision analysis problem. Further, thepaper uses a constraint-based approach to determine a satisfactory so-lution. The results suggest the contribution of different telematic ser-vices to different transport policies.

4 - Measuring Cost Effects of Supply Chain Stockouts:An Interactive Approach with Fast Cost FunctionConvergenceSebastian Langton, Logistics Management Department,Helmut-Schmidt-University, 22041, Hamburg, Germany,[email protected], Martin Josef Geiger

Measuring cost effects of stockouts is a crucial aspect in Supply ChainManagement. In our talk, we present an interactive approach for ob-taining cost function estimations from a decision maker/planner. First,a questioning strategy for the fast convergence of the computed costfunctions is implemented in the interactive phase (similar techniquesare known from Robust Ordinal Regression). Second, function ap-proximations are derived by employing Genetic Programming, a mod-ern metaheuristic. A decision support system integrating both aspectsis presented, and numerical results are discussed.

� MB-04Monday, 10:30-12h00RB-L2

Public Transport Perspectives within aChanging Mobility Environment

Stream: Optimization in Public TransportInvited sessionChair: Claus Doll, Sustainability and Infrastructures,Fraunhofer-Institut for Systems and Innovation Research ISI,Breslauer Str. 48, 76139, Karlsruhe, Germany,[email protected]

1 - USEmobility: Why do people switch to environmen-tally friendly modes of transport?Jolanta Skalska, EU projects, Allianz pro Schiene e.V.,Reinhardtstr. 18, 10117, Berlin, Germany, Germany,[email protected]

The EU project USEmobility applies a novel approach that directlyidentifies the reasons for behavioural change of European citizens inlocal and regional transport where rail is the backbone of the mobilitychain. In 6 European countries and 10 regions people who changedtheir individual mobility-mix were interviewed to find out factors thatinfluence the choice of transport mode. Recommendations for politicaldecision makers and transportation companies will be made on how toachieve the change towards eco-friendly mobility and better adapt thetransport services to what the customers want.

2 - Social Costs and Benefits of Sustainable and HealthyMobility PatternsClaus Doll, Sustainability and Infrastructures,Fraunhofer-Institut for Systems and Innovation Research ISI,Breslauer Str. 48, 76139, Karlsruhe, Germany,[email protected]

As many other Industrialised world regions, Europe is facing an agingand declining population, more tight public households and changes inpeoples’ concerns on lifestyles and sustainability. However, more eco-friendly mobility must pay off for the individual user as well as forlocal authorities and public transport providers. In this paper we thuslook at direct costs, time budgets, health impacts, safety, environmen-tal indicators, GDP and employment of alternative forms of mobilitypatterns from the individual’s perspective as well as for the economyin total.

� MB-05Monday, 10:30-12h00RB-L3

Maritime Transportation 1

Stream: Maritime TransportationInvited sessionChair: Harilaos Psaraftis, National Technical University of Athens,Evrota 20, 15451, Neo Psychico, Greece, [email protected]

1 - An Integrated Model for Ship Routing with Transship-ment and Berth AllocationJiyin Liu, School of Business and Economics, LoughboroughUniversity, Ashby Road, LE11 3TU, Loughborough,Leicestershire, United Kingdom, [email protected],King-Wah Pang

We consider the operation of shipping companies that operate con-tainer vessels and company-owned terminals. An integrated model isdeveloped to make coordinated decisions on container ship routing,berthing time allocation at the terminals, as well as transshipment ofcontainers so that the overall cost can be minimized. To obtain a solu-tion efficiently, we propose a decomposition-based heuristic approach.Computation experiment results show that significant benefit can beachieved by integrating the decisions on ship routing, berthing timeallocation, and transshipment of containers.

2 - Stochastic Optimal Positioning of Tramp Vessels: AMarkovian ApproachEvangelos Magirou, Informatics, Athens University ofEconomics, Patission 76, N/A, Athens, Greece, [email protected]

Determining the optimal sequence of ports that maximizes the time av-erage profit of a tramp vessel dates back to Dantzig who solves it forconstant rates. The model was extended to periodic charter rates andthen to rates being random variables independent in time using DP.We relax this assumption by introducing a state of the charter marketMarkov chain, specific rates depending stochastically on the state. Theformulation differs from the traditional MDP models; solution meth-ods are developed and illustrated. The charter selection rules are intu-itive and appealing to practitioners.

24


3 - A taxonomy and survey of speed models in maritimetransportHarilaos Psaraftis, National Technical University of Athens,Evrota 20, 15451, Neo Psychico, Greece,[email protected] speed is important for economic reasons, as it is a key deter-minant of fuel cost, a significant component of the operating cost ofships. Moreover, speed is an important parameter of the overall supplychain. However, as emissions are directly proportional to fuel con-sumed, speed is also very much connected with the environmental di-mension of shipping. This paper presents a taxonomy and survey ofspeed models in maritime transport, that is, models in which speed isone of the decision variables.

4 - Maritime transport as an important link in the globalsupply chainAlgirdas Šakalys, Competence Centre of IntermodalTransport and Logistics, Vilnius Gediminas TechnicalUniversity, Plytins st. 27, LT-10105, Vilnius, Lithuania,[email protected] East West Transport Corridor, connecting northern Europe withRussia and Far East is an excellent platform to develop an innova-tive pilot testing ground where modern technology and informationsystems contribute to increased efficiency, traffic safety and securityas well as reduced environment impact in the corridor. The corridorstands out as green transport corridor which will reflect an integratedtransport concept where Short See Shipping, rail, inland water waysand road complement each other to enable the choice of environmen-tally friendly transport.

� MB-06Monday, 10:30-12h00RB-Gamma

Design and parameter tuning ofmetaheursticsStream: MetaheuristicsInvited sessionChair: Mike Wright, The Management School, LancasterUniversity, Dept. Of Management Science, LA1 5AG, Lancaster,Lancashire, United Kingdom, [email protected]

1 - Scheduling cricket umpires using neighbourhoodsearch — The dramatic impact of a simple changein neighbourhood definitionMike Wright, The Management School, Lancaster University,Dept. Of Management Science, LA1 5AG, Lancaster,Lancashire, United Kingdom, [email protected] paper reports on the dramatic effect of a relatively simple changein definition of neighbourhoods for a cricket umpire scheduling systemused in practice for amateur leagues in England. After further analysisof the results, the neighbourhood definition was changed, with dramat-ically improved effect. Results will be presented for data from bothleagues. These results serve to emphasise the value of good neighbour-hood definition for neighbourhood search methods — probably muchmore important than the specific choice of metaheuristic technique.

2 - Online Parameter Tuning: Another Aspect of Learn-ing for MetaheuristicsArif Arin, Engineering Management & Systems Engineering,Old Dominion University, 241 Kaufman Hall, 23529,Norfolk, VA, United States, [email protected], Ghaith RabadiMetaheuristics are effective approaches in solving large scale opti-mization problems. However, most metaheuristics have several pa-rameters that need to be tuned before they can reach good results. Inthis study an online parameter tuning method where parameters arechanged during search process is proposed. The method memorizesand stores the information about the search process to determine theprobability of selecting each parameter setting for future iterations. Toevaluate the proposed algorithm it will be applied to Meta-RaPS insolving 0-1 multidimensional knapsack problem.

3 - Probabilistic stopping rules for GRASP heuristicsCelso Ribeiro, Department of Computer Science,Universidade Federal Fluminense, Rua Bogari 70, 22471-340,Rio de Janeiro, RJ, Brazil, [email protected], IsabelRosseti, Reinaldo Souza

A drawback of most metaheuristics is the absence of effective stop-ping rules. We show experimentally that the solution values obtainedby GRASP fit a truncated Normal distribution. We use this approxi-mation to obtain an estimation of the number of iterations needed toimprove the best known solution. This estimation is used to proposeand implement effective stopping rules based on the trade off betweensolution quality and the time needed to find a solution. This strategy isvalidated by a computational study reporting results obtained for someGRASP heuristics.

4 - A solution clustering based guidance mechanism forparallel metaheuristic algorithmJianyong Jin, Molde University College, Britvegen 2, N-6411,Molde, Norway, [email protected], ArneLøkketangen, Teodor Gabriel Crainic

We present a solution clustering based guidance mechanism for a par-allel metaheuristic which solves the capacitated vehicle routing prob-lem. The solutions are clustered according to their similarities. Theknowledge extracted from the solution clusters is applied to guide theintensification and diversification of the parallel algorithm. The com-putational experiments show that the proposed guidance mechanismis effective. The presented algorithm has generated solutions to thebenchmark instances that are very competitive or better than the bestsolutions previously reported in the literature.

� MB-07Monday, 10:30-12h00RB-Eta

Neighborhood reduction strategies forvehicle routing

Stream: Vehicle Routing and Logistics OptimizationInvited sessionChair: Wout Dullaert, Faculty of Economics and BusinessAdministration, VU University Amsterdam, De Boelelaan 1105,1081 HV, Amsterdam, Netherlands, [email protected]

1 - A Simple Parameter-free Heuristic for the Fleet Sizeand Mix Problem with Time WindowsOlli Bräysy, Procomp Solutions Ltd, Kiviharjuntie 11,FI-90220, Oulu, Finland, [email protected], WoutDullaert

We suggest a new metaheuristic for the fleet size and mix vehicle rout-ing problem with time windows. The suggested metaheuristic is basedon simple random-sized jumps. The key idea is to set a random maxi-mum percentual limit for each move and a random limit to the numberof accepted moves that worsen the solution. The actual search is basedon known segment relocation operations. Computational testing on thebenchmark problems show that suggested new metaheuristic is com-petitive to previous best approaches, providing a significant number ofnew best-known solutions.

2 - Efficient local search methods for the large-scale Ve-hicle Routing Problem.Onne Beek, Mgt Information Science & Operations Mgt,Ghent University, Tweekerkenstraat 2, 9000, Gent, Belgium,[email protected], Birger Raa, Wout Dullaert

Even after 60 years of research, state-of-the-art (meta)heuristics oftenperform poorly for solving very large instances of the Vehicle Rout-ing Problem. This paper therefore deals with analyzing, redesigningand testing local search methods that are the building blocks of thesemetaheuristics. It focuses on efficient data structures and parallel algo-rithms suited for modern multicore systems, with an emphasis on sim-plicity and scalability, aiming to create new "best practices’ for moreefficiently solving large-scale VRP instances.

25


3 - Incorporating constraint preservation in GA opera-tors for VRPTW problemGintaras Vaira, Vilnius University, Institute of Mathematicsand Informatics, Akademijos St. 4, 08663, Vilnius, Lithuania,[email protected], Olga KurasovaIn this research we investigate the constraint handling approaches usedin genetic algorithm for VRPTW problem. The investigation dealswith feasibility preservation operators in GA. These operators are re-sponsible to identify feasible parts in chromosome and reuse them inthe offspring. Possible chromosome encoding approaches are investi-gated to reduce the complexity of GA operators while identifying andreusing feasible parts. Results of this research are compared to otherGA feasibility handling approaches used in VRPTW including penaltybased and repair based techniques.

� MB-08Monday, 10:30-12h00RB-Epsilon


Stream: Network OptimizationInvited sessionChair: Dimitri Papadimitriou, Bell Labs, Alcatel-Lucent,Copernicuslaan 50, 2018, Antwerp, Antwerp, Belgium,[email protected]

1 - Single Souce Tree Network Design Under ConvexCostsHenrique Luna, Instituto de Computação, UniversidadeFederal de Alagoas, Rua Eng. Mario de Gusmao 1230/1202,Ponta Verde, 57035000, Maceio, Alagoas, Brazil,[email protected], Gilberto Miranda, Ricardo CamargoThe single source tree network design problem under convex costs isa referential problem in networks for materials, energy or data trans-portation. The model is a hard large scale mixed-integer nonlinear pro-gram. In order to overcome the difficulties, two methods are deployed:Generalized Benders Decomposition and a combination of Outer Ap-proximation with Benders Decomposition. The new scheme is effec-tive on solving instances up to 702 edges in a few seconds. Propertiesof price unicity and differentiability are also studied to cope with thecontinuous relaxation of the problem.

2 - SearchCol algorithms for unsplittable multicommod-ity flow problemsFilipe Alvelos, Departamento de Produção e Sistemas,Universidade do Minho, Campus de Gualtar, 4710-057,Braga, Portugal, [email protected], Amaro de Sousa,Dorabella SantosSearchCol, short for "metaheuristic search by column generation",is an algorithmic framework for approximately solving integer pro-gramming / combinatorial optimization problems with a decompos-able structure. We apply SearchCol algorithms to two multicommod-ity flow problems. In the problems addressed, the commodities, as-sociated with origin/destination pairs, must be routed through a singlepath. We consider two different objectives: (i) minimize the total costand (ii) minimize the maximum load of an arc. We present computa-tional results for different variants of the SearchCol algorithms.

3 - Ultra-fast meta-heuristic for the spectrum re-allocation problem in flexgrid optical networksMarc Ruiz, Computer Architecture, Universitat Politècnica deCatalunya, c/ Jordi Girona 1-3, 08034, Barcelona, Spain,[email protected], Alberto Castro, Luis Velasco, JaumeComellasThe performance of future flexgrid optical networks can be improvedreducing spectrum fragmentation. To improve capacity performance,spectrum re-allocation of already established traffic is performed invery short running time (e.g., 1 second), thus minimizing their im-pact over network operation. To produce near-optimal solutions withinthat stringent time period, we propose a randomized multi-constructiveprocedure followed by a path-relinking intensification. Its benefitsagainst single-constructive procedures are evaluated. This researchwas funded by the TEC2011-27310 ELASTIC project.

4 - Exploiting geometry of large-scale topologies forrouting operationsDimitri Papadimitriou, Bell Labs, Alcatel-Lucent,Copernicuslaan 50, 2018, Antwerp, Antwerp, Belgium,[email protected] routing aims at providing a competitive tradeoff betweenthe memory space required to store the routing tables, the computa-tional complexity and the stretch of the routing paths it produces. Thispaper proposes a distributed procedure to characterize the graph curva-ture and to assign node coordinates for geometric routing to properlyoperate. Indeed, existing procedures for verifying sufficient conditionsholds globally results in computational challenges, in particular if weassume that global topology information is not available at each nodeand central processing is inapplicable.

� MB-09Monday, 10:30-12h00RB-Zeta


Stream: Cutting and PackingInvited sessionChair: Maria Antónia Carravilla, INESC TEC, Faculty ofEngineering, University of Porto, R. Dr. Roberto Frias s/n,4200-465, Porto, Portugal, [email protected]

1 - A hybrid metaheuristic for the nesting problemJosé Fernando Oliveira, INESC TEC, Faculty of Engineering,University of Porto, Rua Dr. Roberto Frias, 4200-465, Porto,Portugal, [email protected], Ramon Alvarez-Valdes, AntonioMartinez Sykora, Maria Antónia Carravilla, A. MiguelGomes, Jose TamaritThe 2-dimensional nesting problem involves the placement of a givenset of polygons, not necessarily convex, into a container. In our prob-lem, the container is a strip of fixed width and the objective is mini-mizing the required length. In this work we develop a metaheuristicscheme in which local search and other heuristic tools are combinedwith the use of an exact procedure for solving some specific subprob-lems. The efficiency of the proposed procedure is assessed by solvinga set of instances from literature and comparing the results with themost recent and successful heuristic approaches.

2 - Optimisation Containment Problem of ArbitraryShaped Objects into a Circular ContainerAlexander Pankratov, Department of Mathematical Modelingand Optimal Design, Institute for Mechanical EngineeringProblems of the National Academy of Sciences of Ukraine,61046, Kharkov, Ukraine, [email protected], Yuri Stoyan,Tatiana RomanovaWe consider a containment problem for irregular shapes into a circularcontainer of minimal radius. Each shape is formed by circular arcs andline segments and may be free translated and rotated. As an efficienttool of mathematical modeling of non-overlapping, containment anddistance constraints we use phi-functions. We provide a mathematicalmodel as a nonlinear constraint optimization problem. Each terminalnode of the solution tree corresponds to a system of inequalities involv-ing infinity-differentiable functions. A solution algorithm is outlined.

3 - Tetris-like Item Packing with Balancing and Addi-tional Conditions: an MIP-based Heuristic ApproachGiorgio Fasano, Space Infrastructures & Transportation,Thales Alenia Space Italia, Str. Antica di Collegno 253,10146, Turin, Italy, [email protected] work focuses on the orthogonal packing of tetris-like items insidea convex domain. Additional conditions, such as the presence of sep-aration planes, are taken into account and particular attention is givento the balancing issue, in the presence of filling material. A MixedInteger Programming formulation and a heuristic approach are consid-ered to tackle real-world instances efficiently. Some case studies arelooked into to provide significant insights. In spite of this work origi-nating within the context of space engineering, it can be extended to anumber of different implementations.

26


4 - The Dotted-Board Model, a promising MIP model forthe Nesting ProblemMaria Antónia Carravilla, INESC TEC, Faculty ofEngineering, University of Porto, R. Dr. Roberto Frias s/n,4200-465, Porto, Portugal, [email protected], Franklina Toledo,Cristina Ribeiro, José Fernando Oliveira, A. Miguel Gomes

In the Nesting Problem the aim is to position, without overlap, a set of2-dimentional small pieces of different shapes over a rectangular pieceof material with a known width and infinite length. The objective isto minimize the used length of the rectangular piece. The decisionvariables in the Dotted-Board model are the dots of the board. The1-dimensionality of the dots in the Dotted-Board allows a straightfor-ward use, in the MIP model, of the constraint programming conceptsof "domain of a variable’ and of "domain reduction’.

� MB-10Monday, 10:30-12h00RB-Theta

Performance Analysis and Inventory

Stream: Production and the Link with Supply ChainsInvited sessionChair: Frédéric Dugardin, LOSI, University of Technology ofTroyes, 12, rue Marie Curie, 10010, Troyes, France,[email protected]: Christophe Duhamel, ICD - LOSI, CNRS, 12 rue MarieCurie, BP 2060, 10010, Troyes, France,[email protected]: Olga Nazarenko, NTUU "KPI", Mayakovskogo avenue17c, fl.72, 02225, Kiev, Ukraine, [email protected]

1 - Performance evaluation of a merge system: A distri-bution centre with multiple random suppliersMichael Vidalis, Business Administration, University ofAegean, Sachtouri 19 Agios Dimitrios, 17341, Athens,Greece, [email protected], Stelios Koukoumialos

A two echelon discrete flow merge supply network consists of a num-ber of non identical reliable suppliers that feed a distribution center(DC) with a shared buffer. The active lead times and service time atDC are exponential distributed. The supply network is modeled as acontinuous time Markov process with discrete states. The transitionmatrices of these specific systems are explored and a computationalalgorithm is developed to generate performance measures for differentvalues of system characteristics. The algorithm is used as a design toolto optimize system’s behavior.

2 - Does greater company size lead to better perfor-mance? Evidence from Greek food supply chainEmel Aktas, Brunel Business School, Brunel University, UB83PH, Uxbridge, United Kingdom, [email protected],Michael Bourlakis, George Maglaras, Christos Fotopoulos

We analyze performance differences within the food supply chain (SC)by examining its key members (growers, manufacturers, wholesalersand retailers) with respect to company size (micro, small, medium andlarge). Greek food companies evaluated their food SC performancebased on key indicators (efficiency, flexibility, responsiveness, productquality and total supply chain performance). The findings suggest thatGreek food SC cannot be characterized as agile since the results indi-cate low performance in flexibility indicators. Manufacturers are chainchampions in terms of total SC performance.

3 - Inventory management optimisation in multi-echelonenterpriseOlga Nazarenko, NTUU "KPI", Mayakovskogo avenue 17c,fl.72, 02225, Kiev, Ukraine, [email protected], YuriyPasenchenko

The optimal inventory management is vital for production or tradingenterprise. Distribution network is based on the principle of hierar-chical structure. Its elements are independent companies operating intheir own interests, trying to optimize there own stocks. It is more

effective to optimize inventory in such multi-echelon distribution net-work as integral system through application of VMI approach. Resultsof simulation show behavior strategies of the center, resellers, and re-tailers that lead to an optimal level of inventory in the system as awhole and each of elements in particular.

� MB-11Monday, 10:30-12h00RB-Iota

Transport Modelling

Stream: Transportation and LogisticsInvited sessionChair: Jaume Barcelo, Statistics and Operations Research,Tecnical University of Catalonia, Edifici Omega, Jordi Girona,1-3, 8034, Barcelona, Catalonia, Spain, [email protected]

1 - Logit models incorporating driving styles variablesfor representing gap-acceptance behaviorRiccardo Rossi, Department of Civil, Architectural andEnvironmental Engineering, University of Padova, ViaMarzolo, 9, I35131, Padova, -, Italy, [email protected],Massimiliano Gastaldi, Claudio Meneguzzer

The paper proposes an analysis of Logit models for representing gap-acceptance behavior at priority intersection. Differently from otherstudies, in this work gap-acceptance data were collected from drivingsimulator tests. The effectiveness of the data collected from experi-ments with driving simulators is highlighted: the use of driving stylesvariables not detectable from direct observations (on site) in driver gap-acceptance models can positively affect the accuracy of operationalanalysis (micro simulation, maneuver capacity and LOS assessment).

2 - An integrated fleet assignment model with supply-demand interactionsBilge Atasoy, Transport and Mobility Laboratory, ÉcolePolytechnique Fédérale de Lausanne (EPFL), CH-1015,Lausanne, Switzerland, [email protected], MichelBierlaire, Matteo Salani

We integrate an itinerary choice model into a fleet assignment modelso that price and demand interaction is endogenously included in themodel. The decisions of the integrated model are the fleet assignment,seat allocation for each cabin class, schedule design and the pricingof the itineraries. We analyze the impact of including the demandmodel on the profits and transportation capacity from the airlines pointof view. To deal with the complexity of the resulting mixed integernonlinear problem, we build a heuristic method based on Lagrangianrelaxation and subgradient optimization.

3 - Ship route modelling for traffic pattern analysisRonald Pelot, Industrial Engineering, Dalhousie University,5269 Morris St., PO Box 15000, B3H 4R2, Halifax, NovaScotia, Canada, [email protected], Dong Lin, CaseyHilliard

Ship movement tracking is improving due to various satellite andground-based systems. This point data is sufficient for individual shiptracking or visual traffic patterns. But there are applications wheremaritime traffic analysis requires vector information on typical ship-ping routes and densities along each segment of these routes. We de-veloped a method for aggregating the individual vessel trajectory datainto routes and paths, that can be used to identify outliers in real-time,whether due to deliberate deviance from normal routes or for someaccidental or nefarious purposes.

4 - The sensor location problemJaume Barceló, Statistics and Operations Research,Universitat Politècnica de Catalunya, Jordi Girona salgado1-3, Campus Nord, Edifici C5, Office 215, 08034, Barcelona,Spain, [email protected]

The main input for traffic models are Origin-Destination matrices, de-scribing traffic patterns across the network, critical in Traffic Manage-ment supported by Dynamic Traffic Assignment models. OD matrices

27


are not directly observable, a seed matrix is adjusted from link flowcounts provided by traffic counting stations. The adequacy of the de-tection layout strongly determines the quality of the adjusted OD. Wepresent two alternative formulations as set covering models with sideconstraints exploiting the advantages of new technologies, and proposea tabu search heuristic to solve it.

� MB-12Monday, 10:30-12h00RB-Omicron

Fuzzy Relations & Neural Networks

Stream: Fuzzy Systems, Neural Networks & ArtificialIntelligenceInvited sessionChair: Martin Gavalec, Department of Information TechnologiesFIM, University of Hradec Kralove, Rokitanského 62, 50003,Hradec Kralove, Czech Republic, [email protected]

1 - Eigenproblem of Circulant and Toeplitz matrices inextremal algebraHana Tomaskova, Information Technologies, Faculty ofInformatics and Management, University of Hradec Králové,Rokitanskeho 62, 50003, Hradec Kralove, Czech Republic,[email protected], Martin Gavalec

Eigenproblem of circulant and Toeplitz matrices in extremal algebraare studied. Both types of matrices have a special form. Toeplitz ma-trix is determined by vector of inputs in the first row and first column.Circulant matrix is special case of Toeplitz matrix and is determinedby vector of inputs in the first row. Investigation of eigenvectors inmax-min algebra is important for applications connected with reliabil-ity of complex systems or with fuzzy relations. For special types ofmatrices the computation of the eigenproblem can often be performedin a simpler way than in the general case.

2 - Learning Effect Approach for Scheduling Problemwith Fuzzy Processing Time and Fuzzy Due DateMerve Kayaci Çodur, Industrial Engineering, AtaturkUniversity, Ataturk University Engineering Faculty, IndustrialEngineering, 25200, Erzurum, Turkey,[email protected], Vecihi Yigit

In scheduling problems, by considering the imprecise of the durationsin the reality of manufacturing, processing times and due dates are in-troduced as fuzzy numbers. In most studies, fuzzy processing times ofa job are assumed fixed. However in many realistic settings, the actualprocessing time of a job maybe more or less than its normal process-ing time when it is scheduled later. This phenomenon is known as the"Learning Effect’ So we indicate study in which the uncertainty of ajob processing time not only defines fuzzy numbers, but also definesits scheduled position. According to the complex property of the prob-lem, a heuristic approach of Genetic Algorithm is applied to solve thisNP-hard problem.

3 - Mean-Partial moments models for portfolio selectionwith fuzzy returnsMbairadjim Moussa Alfred, Université Montpellier 1 -Lameta, France, [email protected], Sadefo KamdemJules, Terraza Michel

Since financial risk relates only to loss, lower partial moment is amore suitable risk measure compared to variance. In order to mea-sure the risk of fuzzy portfolio return, a concept of partial moment forfuzzy variable, is defined in this paper. As extensions of the fuzzymean-semivariance model, some mean-partial moments models arepresented and the corresponding variations are discussed. In orderto solve the proposed models, PSO algorithm integrating fuzzy sim-ulation is designed. Finally, several numerical examples are given inorder to illustrate the effectiveness of our modeling idea.

4 - Simulation of Stochastic Linear Queuing SystemsRichard Cimler, Information Technologies, University ofHradec Králové, Rokitanského 62, 500 02, Hradec Králové,Czech Republic, Czech Republic, [email protected],Martin Gavalec

Cost optimization in stochastic linear queuing systems is considered inthe contribution. Simulation approach is used because the exact pre-diction is not possible in stochastic systems. The simulated system islinear, i.e. it consists of a series of servers. The aim of simulationis to find the best way of controlling the server capacities due to sys-tem setup. The optimization is oriented to minimizing the total costs,which are divided into four parts: working costs, idle costs, queuingcosts and change costs.

� MB-13Monday, 10:30-12h00RB-Tau


Stream: Lot-sizing and Related TopicsInvited sessionChair: Christian Almeder, Chair for Supply Chain Management,European University Viadrina, Große Scharrnstr. 59, 15230,Frankfurt (Oder), Germany, [email protected]

1 - The impact of major factors in food processing in-dustry on the capacitated lot-sizing and schedulingproblemG.D.H. (Frits) Claassen, Operations Research and Logistics,Wageningen University, Hollandseweg 1, 6706 KN,Wageningen, Netherlands, [email protected], J.c.Gerdessen, M.t. Patek

Traditional lot sizing and scheduling models violate crucial issues infood processing industry. We develop models that relax all assump-tions regarding changeover matrices including the triangular setup con-ditions; take into account effects of deterioration due to perish abilityof inventory; have the flexibility to take issues for physical distributioninto consideration. Optimal production schemes become significantlydifferent when these issues are considered. We apply a time-baseddecomposition heuristic and show that near-optimal solutions can befound in reasonable time.

2 - A multicommodity lotscheduling model for the softdrink production processSocorro Rangel, UNESP - Sao Paulo State University, RuaCristovão Colombo, 2265, 15054000, S.J. do Rio Preto, SãoPaulo, Brazil, [email protected], Michelli Maldonado

The production of soft drinks involves two main stages: syrup prepara-tion and bottling. Two main approaches have been used in the literatureto model the scheduling decisions of the bottling stage. The first oneis a small bucket approach in which each period is divided into subpe-riods. The second is a big bucket one. The objective of this work is todiscuss the use of a multicommodity formulation for the asymmetrictraveling salesman problem to model the scheduling decisions in thebig bucket approach. The proposed model is studied and compared toothers presented in the literature.

3 - Modeling the multi-level lot-sizing and schedulingproblem - a comparisonRenate Traxler, European University Viadrina Frankfurt(Oder), Germany, [email protected], Christian Almeder

In this work we compare different models from the literature for inte-grating the lot-sizing and the scheduling decisions in multi-stage sys-tems. For that purpose, we show their abilities and limitations in de-scribing relevant aspects of a production environment. Furthermore,we show the computational behavior of the models by applying themto benchmark instances. The structural and numerical comparisonsshow that there are considerable differences between the approaches.They allow us to provide a guideline for selecting the right modelingapproach for different planning situations.

28


� MB-14Monday, 10:30-12h00RB-Omega

Discrete-continuous scheduling

Stream: Scheduling under Resource ConstraintsInvited sessionChair: Jan Weglarz, Institute of Computing Science, PoznanUniversity of Technology, Piotrowo 2, 60-965, Poznan, Poland,[email protected]

1 - Metaheuristics for discrete-continuous projectscheduling with discounted cash flows and variouspayment modelsGrzegorz Waligora, Institute of Computing Science, PoznanUniversity of Technology, Piotrowo 2, 60-965, Poznan,Wielkopolska, Poland, [email protected]

Discrete-continuous project scheduling problems with discounted cashflows are considered, in which the number of discrete resources is ar-bitrary, and there is one continuous, renewable and limited resource.The processing rate of each nonpreemptable activity is a continuous,increasing function of the amount of the continuous resource allottedto the activity at a time. A positive cash flow is associated with eachactivity, and the objective is to maximize the net present value (NPV).Metaheuristic approach to the considered problems is proposed undervarious payment models.

2 - Metaheuristic approaches for power-aware schedul-ing problemRafal Rozycki, Institute of Computing Science, PoznanUniversity of Technology, ul.Piotrowo 2, 60-965, Poznan,Poland, [email protected]

We consider the problem of scheduling power driven jobs on paral-lel identical processors. Jobs are preemptive and related. Each job ischaracterized by the size and the processing speed function. Power andenergy limits have to be respected in a feasible schedule. The problemis to find both: a sequence of jobs on processors as well as an alloca-tion of power to jobs that minimize the schedule length. We proposesome metaheuristic approaches that produce suboptimal solutions ofthe problem in a reasonable time. The effectiveness of the approacheshas been compared experimentally.

3 - Multimode resource-constrained project schedulingproblem with setup costsMarek Mika, Institute of Computing Science, PoznanUniversity of Technology, ul Piotrowo 2, 60-965, Poznan,Poland, [email protected]

In the multimode resource-constrained project scheduling problem aset of activities has to be scheduled in such a way that precedence andresource constraints are met. Each activity can be performed in oneof several execution modes differing between themselves in resourcerequirements and processing times. Some of the resources requiredby an activity have to be set up before the execution of this activity.Both the setup time and the setup cost depend on the execution mode.The objective is to minimize the total cost of all setups. A heuristicapproach is proposed to solve the problem.

4 - Optimal resource allocation in virtual datacenters —a multidimensional continuous bin packing approachHolger Schrödl, Otto-von-Guericke University Magdeburg,Germany, [email protected]

The optimized used of resources in IT datacenters is of vital interestfor IT service providers. Abundance of resource is cost-intensive, re-ducing IT resources leads to bottlenecks. Resource factors in datacen-ters are processor schedules, storage space and computing time. Exist-ing approaches are limited how these resource factors are considered.First, the authors study multidimensional bin packing models to de-scribe resource allocation in IT datacenters. Second, resource factorsare described through general continuous functions which representthe behavior of certain resource classes.

� MB-15Monday, 10:30-12h00RB-2101

Nonlinear Optimization and Applications 1

Stream: Nonlinear ProgrammingInvited sessionChair: Simone Göttlich, School of Business Informatics andMathematics, University of Mannheim, A 5, 6, Mannheim, 68131,Germany, [email protected]

1 - Range Approximation of Multivariate Polynomials forGlobal OptimizationMartin Stöcker, Fakultät Mathematik, ProfessurWirtschaftsmathematik, Technische Universität Chemnitz,Reichenhainer Str. 41, Zimmer 727, 09107, Chemnitz,Germany, [email protected]

We consider polynomials in an arbitrary number of variables whichfrequently occur in optimization problems. A common challenge isto find bounds for the range of polynomials. We introduce two ap-proaches which generate lower and upper bounds of multivariate poly-nomials over an n-dimensional box. The first method is based on theestimation of the contribution of the polynomial’s summands, whilethe second one uses an expansion into Bernstein polynomials to obtainthe desired bounds. We apply the methods to a real-world problemwhich occurs in the development process of vehicle transmissions.

2 - Linear least squares with different types of con-straints: effectiveness of methods.Arnaud Vandaele, Mathematics and Operations Research,University of Mons, rue de Houdain, 9, 7000, Mons, Belgium,[email protected]

In this work, we study different approaches used to solve the well-known Linear Least Squares problem, used in many fields of OR.Specifically, we focus on this problem with two types of constraints onthe variables: either they must be positive, or they must be located be-tween bounds. Different technics exist to solve these problems. In thiswork, we compare these methods on different points of view, payinga particular attention to their effectiveness of determining the optimalactive set, which is the set of variables reaching one of their bounds atthe optimal solution.

3 - Optimization Techniques: Combinatorial Optimiza-tion and Adjoint CalculusSimone Göttlich, School of Business Informatics andMathematics, University of Mannheim, A 5, 6, Mannheim,68131, Germany, [email protected]

The focus is on optimization problems arising in the continuous mod-eling of network applications such as production systems and trafficflow. We present an adjoint calculus for solving the optimality systemand discuss suitable discretizations of this system. For particular dis-cretizations the resulting problem is in fact a mixed-integer program.To speed up the solution procedure, we introduce presolve techniquesthat are inspired by the continuous dynamics.

4 - An improved approach to subset selection for param-eter estimation in online applicationsDiana López, Chair of Process Dynamics and Operation,Berlin Institute of Technology, Sekr. KWT-9, Str. Des 17.Juni 135, 10623, Berlin, Germany,[email protected], Tilman Barz, HarveyArellano-Garcia, Günter Wozny

Subset selection is a systematic approach to system identification innonlinear least squares parameter estimation. Ill-conditioned parame-ters are fixed at prior estimates and a well-conditioned parameter esti-mation problem is solved to determine the active parameters. The pro-cedure permits to determine the parameters whose effects are uniqueand have a strong effect on the output measurements. The experimen-tal results correspond to a biological system (ion-exchange high per-formance liquid chromatography column for protein separation) wheretransport and adsorption parameters are determined.

29


� MB-16Monday, 10:30-12h00RB-2103

Hybrid Algorithms & NonsmoothOptimization

Stream: Nonsmooth OptimizationInvited sessionChair: Angel A. Juan, Computer Science, Open University ofCatalonia, Rambla Poblenou, 156, 08018, Barcelona, Spain,[email protected]: Albert Ferrer, Dpt. of Applied Mathematics I,Technological University of Catalonia, Av. Doctor Marañon,44-50, 08028, Barcelona, Catalunya, Spain,[email protected]

1 - Successive Approximations of the SimuRoute Pro-cedure to solve the Heterogeneous Vehicle RoutingProblemJavier Faulin, Department of Statistics and OR, PublicUniversity of Navarre, Los Magnolios Builing. First floor,Campus Arrosadia, 31006, Pamplona, Navarra, Spain,[email protected], Angel, A. Juan, Alba Agustín,Miguel Angel LlorenteWe propose a Successive Approximations Method (SAM) for solvingthe Heterogeneous Vehicle Routing Problem (HVRP) in real scenar-ios. Our approach solves, in each iteration, the Capacitated VehicleRouting Problem (CVRP) for the non-served nodes and the maximumcapacity of unused vehicles. We used the so-called SimuRoute Pro-cedure via Generalized Clarke and Wright (SR-GCW) to solve the as-sociated CVRP. Thus, the largest routes which are generated from theCVRP solution are saved as a partial solution for the HVRP. Finally,competitive results have been obtained in reasonable computing times.

2 - Solving non-smooth problems through biased ran-domization of heuristicsAngel A. Juan, Computer Science, Open University ofCatalonia, Rambla Poblenou, 156, 08018, Barcelona, Spain,[email protected], Albert Ferrer, Javier Faulin, HelenaRamalhinho LourençoWe discuss the use of probabilistic algorithms for solving combinato-rial optimization problems with non-smooth objective functions. Ourapproach employs non-uniform probability distributions to add a bi-ased random behavior to well-known heuristics. By doing so, a largeset of alternative good solutions can be quickly obtained in a naturalway and without complex configuration processes. This approach doesnot need to assume any particular property of the objective function.The procedure can be especially useful in problems where propertiessuch as non-smoothness or non-convexity lead to a highly irregularsolution space.

3 - A Randomized Algorithm for the HeterogeneousFixed Fleet Vehicle Routing ProblemJose Caceres Cruz, IN3 - Computer Science, Open Universityof Catalonia, Roc Boronat, 117, 7th floor, Mediatic, 08018,Barcelona, Spain, Spain, [email protected], Angel, A. Juan,Daniel Riera, Helena Ramalhinho LourençoThe academic point of view for the Vehicle Routing Problem (VRP)has undergone significant theoretical advances. Several techniques areenabling for solving real-world scenarios like the heterogeneous fixedfleet VRP. In this study, we present an approach for the HFFVRP usinga hybrid algorithm combining a randomized Clarke & Wright heuris-tic, a memory, and a random policy for preassignment of vehicles.The Java implementation is compared over classical datasets againstother saving heuristic implementation. So far, the obtained results arepromising and other policies are being considered.

4 - Solving a real application of the Time-Constraint Ca-pacitated Open Vehicle Routing ProblemHelena Ramalhinho Lourenço, Department of Economics andBusiness Q-5850017D, Universitat Pompeu Fabra, R. TriasFargas 25-27, 08005, Barcelona, Spain,[email protected], Luciana Pessoa, Alex Grasas,Angel A. Juan, Mauricio Resende

This work considers a real application of the Capacitated Time-constrained Open Vehicle Routing Problem in the health care sector,the blood sampling collection problem. The problem is COVRP withan additional constraint: the time between the first collecting point andthe laboratory must be smaller or equal that a maximum time. Themain objective is to minimize the number of vehicles used. We pro-pose a Bias Random Key-Genetic Algorithm, and present results forreal instances and classical ones. For the real instances, we could re-duce significantly the number vehicle significantly.

� MB-17Monday, 10:30-12h00RB-2105

Location and GISStream: Applications of Location AnalysisInvited sessionChair: Burcin Bozkaya, Sabanci School of Management, SabanciUniversity, Orhanli, Tuzla, 34956, Istanbul, Turkey,[email protected]

1 - A fast spatial query algorithm for location analysisHonora Smith, School of Mathematics, University ofSouthampton, Highfield, SO17 1BJ, Southampton,Hampshire, United Kingdom, [email protected]

The use of new spatial query capabilities in relational databases com-bined with specific heuristics has allowed us to create an algorithm thatcan process very large location modeling problems at high speed. Theprogram was developed to tackle a healthcare modeling problem inSouth Africa that had over 3,200 demand points and 220 possible ser-vice locations. Using a version of Microsoft R© SQL Server, possiblesolutions for this problem have been found in a few minutes. Visu-alization of travel to nearest service locations is facilitated by GoogleMaps route finding capabilities.

2 - Location Survey and GIS (Geographic InformationSystem) for Shopping Centers Across the Country(Turkey)Ceren Erdin Gundogdu, Business Administration, YildizTechnical Univercity, Barboros Bulvari Besiktas, 34349,Istanbul, Turkey, [email protected]

Being a part of social life, Shopping Centers (SCs) become more pop-ular and represent the environments where consumption comes intoprominence. In this study a model based on geographical informationsystem was created and decision making process was analyzed withthe purpose of location survey of SCs to be opened across the coun-try by taking population, economic and social data, geographical andspatial data. In analysis, ESRI, ArcGIS software pack, ArcCatalog,ArcMap, ArcToolbox tool boxes and software required for topologydevelopment, data conversion and geographical analysis were used.

3 - Locating Temporary Shelter Areas After a Large-Scale DisasterBurcin Bozkaya, Sabanci School of Management, SabanciUniversity, Orhanli, Tuzla, 34956, Istanbul, Turkey,[email protected], Firat Kilci, Bahar Yetis Kara

In this study, we develop a mathematical model that decides on thelocation of shelter areas and the population assignment to these areas.Our formulation is based on Turkish Red Crescent’s criteria on shel-ter area selection. We employ a GIS-based decision support systemto implement our model. The system retrieves the distance data fromArcGIS and obtains a solution with the help of Gurobi optimizationsoftware. The resulting solutions are visualized on a map that is gen-erated by ArcGIS. We perform a case study on Kartal, a district ofIstanbul, Turkey, using our decision support system.

30


� MB-18Monday, 10:30-12h00RB-2107

Risk Analysis and Options

Stream: Financial Modelling & Risk Management (con-tributed)Contributed sessionChair: Susanne Griebsch, School of Finance and Economics,University of Technology Sydney, PO Box 123 Broadway, 2007,Sydney, NSW, Australia, [email protected]

1 - Designing a combinatorial auction for real estateDries Goossens, Operations Research and Business Statistics(ORSTAT), KU Leuven, Naamsestraat 69, 3000, Leuven,Belgium, [email protected], Frits Spieksma,Onderstal Sander

Real estate auctions are not uncommon, but typically involve a sequen-tial and individual auctioning of properties, with bidders facing the ex-posure problem. Further, in a real estate setting, there may be variousconstraints on the allocation, and bidders are often inexperienced. Tak-ing this into account, we discuss the design of a combinatorial auction(choices with respect to e.g. pricing, feedback, and activity rules), re-sulting from a simulation with students. We also present complexityresults for the winner determination problem.

2 - A Stochastic Volatility Cointegration Model in Contin-uous TimeSusanne Griebsch, School of Finance and Economics,University of Technology Sydney, PO Box 123 Broadway,2007, Sydney, NSW, Australia, [email protected],Kay Pilz

We investigate the valuation of spread options written on futures con-tracts which exhibit co-movement over large time periods. In general,co-moving futures display certain characteristics when deviating fromtheir normal spread level due to changes from usual market conditions,e.g. the mean-reversion of spreads. However, we detect a new propertybetween co-moving futures in the empirical data and propose a modelthat allows for a representation of this property. In a second step, wederive pricing formulas for European and spread options. We concludewith a discussion of numerical results.

� MB-19Monday, 10:30-12h00RB-2111

Energy and Environmental markets

Stream: Commodities and Financial ModelingInvited sessionChair: Silvana Stefani, Quantitative Methods for Economics andBusiness Sciences, Università Milano Bicocca, Piazza AteneoNuovo 1 U7-4023, I-20126, Milano, Italy,[email protected]

1 - The impact of grandfathering on emission certifi-cates for the electricity sectorPaolo Falbo, Department of Quantitative Methods, Universityof Brescia, Contrada Santa Chiara, 50, 25122, BRESCIA BS,Italy, [email protected], Daniele Felletti, Silvana Stefani

We focus on the impact of EUAs on the optimal policy of a compet-itive electricity producer. The effect of grandfathering is consistentlyshown to introduce significant distortions to the system. It is theoreti-cally shown that there is a threshold value of carbon price so that, forprices above this, the EUA becomes an incentive for reduced produc-tion rather than a penalty for inefficient producers. These theoreticalresults are supported by the data of one producer from Italy and onefrom Germany.

2 - An inspection of Energy and Agricultural Markets af-ter FukushimaAngelica Gianfreda, Florence School of Regulation, EUI,Italy, [email protected], Giacomo Scandolo

After the earthquake occurred in Fukushima in March 2011, the hy-pothesis of replacing nuclear generation with alternative sources hasbeen explored. Then considering the need to mitigate climate changeand reduce carbon emissions, renewable sources have attracted an in-creasing interest. However, wind, solar and wave energy show seri-ous limitations, hence biofuels are becoming an interesting issue withimplications on agricultural commodities and consequently on worldpopulation. Therefore, this contribution is aimed to understand the im-pact of Fukushima on energy and agricultural markets.

3 - Forecasting Volatility with Interday or Intraday Data:A Case Study on Istanbul Stock Exchange 30 Na-tional IndexAlper Inkaya, Middle East Technical University, Institute ofApplied Mathematics, METU Üniversiteler Mah. DumlupınarBlv. No:1, Çankaya, 06800, Ankara, Turkey,[email protected]

Finding a proper volatility model that forecasts true volatility ac-curately lies in the heart of risk management and asset valuation.We compare forecasting performance of time series models, such asARCH type models and ARFIMA model, by estimating one-day aheaddaily volatility of the Istanbul Stock Exchange 30 National Stock In-dex using historical and realized volatility measurements. We controlwhether ISE 30 daily close series has long memory, and compare theforecasting ability of long memory models by several penalty criteriawith traditional models.

� MB-20Monday, 10:30-12h00RB-2113

Data Mining and Risk Analysis

Stream: Data Mining in the Financial SectorInvited sessionChair: Marcus Hildmann, Information Technology and ElectricalEngineering, ETH Zurich, ETL G 24.2, Physikstrasse 3, 8092,Zürich, Switzerland, [email protected]

1 - Estimation of Stochastic Volatility Models for Com-modities with few OptionsGonzalo Cortazar, Ingenieria Industrial y de Sistemas,Pontificia Universidad Catolica de Chile, Vicuna Mackenna4860, Santiago, Chile, [email protected], Federico Alonso

Stochastic volatility specifications are required to adequately model thedynamics of commodity prices. To obtain good parameter estimates,prices for non-linear derivatives with different maturities are necessary,but some commodities trade only futures and short-term options. Thispaper proposes a three-factor multicommodity model with stochasticvolatility and shows how to use long-term options written on one com-modity to help estimate the dynamics of another. The model is imple-mented for Brent and WTI commodities with excellent results.

2 - Data cleansing of implied volatility surfacesDejan Stokic, tbd, 60386, Frankfurt, Germany,[email protected]

Data cleansing of implied volatility surfaces is one of the most im-portant topics in managing the risk of options portfolios. We set thefocus on the static arbitrage checks and assume that implied volatil-ity surfaces are already cleaned of outliers and missing data points.We present a non-parametric method for correcting the surface, so thatresulting surface is convex in moneyness dimension and monotonicin time to expiry dimension, while at the same time keeping the dis-tribution of implied volatility returns of the surface points minimallychanged.

31


3 - A Linear Programming Model for Enhanced Indexa-tion based on Strong Stochastic DominanceRenato Bruni, Dip. Ingegneria Informatica, Automatica eGestionale, University of Roma "Sapienza", Via Ariosto 25,00185, Roma, Italy, [email protected], FrancescoCesarone, Andrea Scozzari, Fabio TardellaIn the field of Portfolio Optimization, Enhanced Indexation is the prob-lem of selecting a portfolio that generates excess return with respect toa benchmark index. In this work, we propose a linear programmingmodel for Enhanced Indexation that selects an optimal portfolio ac-cording to a generalization of strong stochastic dominance. Since ourmodel has an exponential number of constraints, we solve it througha constraint generation procedure. Some experimental results are pre-sented for well-known financial data sets showing good out-of-sampleperformance of our model.

4 - Pump Storage Hydro Plant Valuation under Risk Con-straintsMarcus Hildmann, Information Technology and ElectricalEngineering, ETH Zurich, ETL G 24.2, Physikstrasse 3, 8092,Zürich, Switzerland, [email protected], FlorianHerzog, Sebastiano RossiFor power producers, the valuation of pump storage hydro plants be-comes an important task for investment decisions, capacity market par-ticipation and operation planning. We propose a scenario based linearprogramming approach, which maximizes the profit by calculating theoptimal power plant operation schedule on hourly basis with cVaR asrisk constraint. In addition, capacity markets, such as control energymarkets can be taken into account. The presented method allows valu-ing pumped storage hydro plants, including risk and control energyreserves in a computational efficient way.

� MB-21Monday, 10:30-12h00RB-2115

Financial Mathematics and OR 2Stream: Financial Mathematics and ORInvited sessionChair: Tansel Avkar, Department of Mathematics, Middle EastTechnical University, 06531, Ankara, Turkey,[email protected]

1 - Recently crisis, credit boom and nonperformingloans in CEE countries-panel data evidenceBlanka Škrabic Peric, Department of Quantitative Methods,Faculty of Economics, Matice hrvatske 31, Split, Croatia,[email protected], Josip ArnericThe paper investigates impact of credit boom during 2000th and re-cently crisis on nonperforming loans in CEE counties. Namely, thebeginning of 2000th was characterized by many features as the foreigncapital inflow, entry of foreign banks and economic growth. The rapidcredit growth and lenient credit standard in the period before the crisishave caused an increase in nonperforming loans during the crisis. Dy-namic panel data is used to analyze the impact of rapid credit growthon nonperforming loans in 16 CEE countries during period 1994-2010on aggregated data for each country.

2 - A Technology Credit-Scoring Model for Service In-dustrySo Young Sohn, ISE, Yonsei university, Shinchon-dong, 134,Sudaemoon-gu, 137-049, Seoul, Korea, Republic Of,[email protected], Mooyeob LeeA technology credit guarantee fund is established to support the growthand innovation of small and medium enterprises (SMEs) based on theirtechnology. A technology credit-scoring model has been introduced toselect applicants in a manner intended to reduce their loan default rate.However, this model was heavily influenced by the manufacturing in-dustry, although the role of the service industry has been growing. Inthis paper, we propose a new technology credit-scoring model for theservice industry by applying a logistic regression model to empiricaldata.

3 - Properties and Calculation of MVaR and MCVaRJinwook Lee, RUTCOR, Rutgers University, 640Bartholomew Rd., 08854, Piscataway, New Jersey, UnitedStates, [email protected], Andras Prekopa

In a recent paper by A. Prekopa (2010) a survey is presented in con-nection with Multivariate Value-at-Risk (MVaR) that has been knownfor some time under the name of p-Level Efficient Point (pLEP) and anew multivariate risk measure, called Multivariate Conditional Value-at-Risk (MCVaR), was introduced. The purpose of this paper is tofurther develop properties of MVaR and MCVaR. This includes thepresentation of a new methodology to numerically calculate MCVaR,by the use of the discrete moment problem. Numerical examples arepresented.

4 - A bank’s attractiveness as described by a cuspcatastrophe modelVassilis Angelis, Business Administration, University of theAegean, Michalon 8, 82100, Chios, Greece,[email protected], Katerina Dimaki

In the present paper we introduce a bank’s Image as a measure givingthe true picture of its current state of development and future prospects.This picture when received and perceived by its customers will deter-mine their attitude. Based on this Image existing/potential customerswill decide to stay/cooperate or not with the given bank. However, theperception and reaction times to any changes of a bank’s Image aredifferent for the various groups of its customers. Our objective is todefine a bank’s Image, build a model for it and apply it to a real case.

� MB-22Monday, 10:30-12h00RB-2117

Multiple-criteria evaluation anddecision-making under fuzziness: theoryand applications

Stream: Fuzzy Sets and Soft ComputingInvited sessionChair: Jana Talasova, Dept. of Mathematical Analysis andApplications of Mathematics, Faculty of Science, PalackyUniversity Olomouc, tr. 17. listopadu 1192/12, 771 46, Olomouc,Czech Republic, [email protected]

1 - A multi-agent multiple-criteria approach to fraud de-tection under fuzzinessMario Fedrizzi, Department of Computer and ManagementSciences, Universita di Trento, Via Inama 5, 38122, Trento,Italy, [email protected], Alessandro Buoni, AlessandroBuoni

We introduce a multi-agent system (FIDES) aiming at offering to ateam of auditors an innovative way to manage fraud schemes, assum-ing that the judgments of auditors are represented by fuzzy preferencerelations. First we introduce a dynamical group decision model toachieve a consensual design of the attack tree, the basic componentof the system. Then, starting from a linguistic valuation of the nodesof the attack tree, we show how to propagate the values up the treethrough an aggregation process based on multiple-criteria techniques.

2 - Multiple criteria evaluation of creative work out-comesJana Talasova, Dept. of Mathematical Analysis andApplications of Mathematics, Faculty of Science, PalackyUniversity Olomouc, tr. 17. listopadu 1192/12, 771 46,Olomouc, Czech Republic, [email protected], JanStoklasa

The Register of Artistic Performances is currently being developed inthe Czech Republic that contains information on works of art originat-ing from creative activities of art colleges. Outcomes in various fieldsof artistic production are divided into 36 categories. The abstract cate-gories are scored according to the significance, size, and international

32


reception of corresponding works of art. The Saaty’s matrix of pair-wise comparisons is applied to determine the scores of categories. Anew view to the scores calculation and a notion of week consistency ofthe Saaty’s matrix were introduced.

3 - Linguistically oriented model for academic staff per-formance evaluationJan Stoklasa, Dept. of Mathematical Analysis andApplications of Mathematics, Faculty of Science, PalackyUniversity Olomouc, tr. 17. listopadu 1192/12, 771 46,Olomouc, Czech Republic, [email protected], JanaTalasova, Pavel Holecek

The information system for academic staff performance evaluation (ISHAP) currently used at Palacky University in Olomouc will be pre-sented. The core of IS HAP is a multiple criteria evaluation modelbased on linguistic fuzzy rule bases. The apparatus of linguistic fuzzymodelling enables us to describe complex aggregation functions in away comprehensible to the evaluators and academic staff members thatare being evaluated. Innovative ways of interpreting the outputs of theevaluation and using them for classification purposes will also be pre-sented.

4 - Multicriteria analysis of teacher’s managerial compe-tenciesHelena Brozova, Dept. of Systems Engineering, CzechUniversity of Life Sciences, Kamycka 129, 165 21, Prague 6 -Suchdol, Czech Republic, [email protected]

This article deals with identification and evaluation of the students’expectance of teacher’s competencies at the Faculty of Economics,Czech University of Life Sciences in Prague. In this research we donot intend to evaluate the teacher’s scientific ability which can’t besubject of students’ evaluation. The key characteristics of the man-agerial competencies of teachers are set according to the Casselmanntypology of teacher’s roles. Than the students pairwise comparisons ofvarious teacher’s characteristics and competencies were analysed usingthe Analytic Hierarchy Process.

� MB-23Monday, 10:30-12h00RB-Delta

Optimality Conditions and Stability

Stream: Generalized Differentation and OptimizationInvited sessionChair: Alexander Kruger, School of Science, InformationTechnology & Engineering, University of Ballarat, UniversityDrive, Mount Helen, P.O. Box 663, 3353, Ballarat, Victoria,Australia, [email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - First and Second Order Optimality Conditions forVector Optimization Problems with Non-solid Posi-tive ConeValentin Gorokhovik, Institute of Mathematics, NationalAcademy of Sciences of Belarus, ul. Surganova, 11, 220072,Minsk, Belarus, [email protected]

The talk deals with a nonlinear vector optimization problem with non-solid ordering cone. The main purpose is to present a unified approachto deriving first and second order optimality conditions, both necessaryand sufficient, for solutions of this problem. Our approach includestwo stages. At first, we reduce the vector optimization problem to thevariational system consisting of the scalar inequality and the operatorequality, and then, analyzing this system with variational (convex andnonsmooth) methods, we obtain optimality conditions.

2 - Risk-Sensitive Average Optimality in Finite and De-numerable Markov Decision Chains

Karel Sladky, Department of Econometrics, Institute ofInformation Theory and Automation, Academy of Sciences ofthe Czech Republic, Pod Vodarenskou vezi 4, 18208, Prague8, Czech Republic, [email protected]

The contribution is devoted to the risk-sensitive optimality criteria infinite or denumerable state Markov Decision Chains, where expecta-tion of the stream of one-stage rewards generated by a Markov chain isevaluated by exponential utility function. Then even for unichain mod-els risk-sensitive average optimality may depend on the starting stateand be optimal only if the process starts in transient state. Explicit for-mulae and optimality conditions for the growth rates of expected utilityas well as for mean value of the corresponding certainty equivalent willbe discussed.

� MB-24Monday, 10:30-12h00CC-A11

Centralized Inventory and Production

Stream: Game Theory and LogisticsInvited sessionChair: Ana Meca, Operations Research Center, UniversidadMiguel Hernández, Avda. Universidad s/n, Edificio Torretamarit,03202, Elche, Alicante, Spain, [email protected]

1 - Cooperation in Assembly Systems: the Role ofKnowledge Sharing NetworksFernando Bernstein, The Fuqua School of Business, DukeUniversity, 100 Fuqua Drive, 27708, Durham, NC, UnitedStates, [email protected], Gurhan Kok, Ana Meca

We consider an assembly system with a buyer purchasing componentsfrom several first-tier suppliers. Suppliers incur variable inventorycosts and fixed setup production costs, and invest in process improve-ment activities which have a direct impact on their fixed costs. Theassembler establishes a knowledge sharing network, modeled as a co-operative game among suppliers. As a result of cooperation, all suppli-ers achieve a level of cost reduction equal to that of the most efficientsupplier. We investigate the resulting impact on the investment processand cost structure for the assembly system.

2 - Cost allocation in inventory transportation systemsIgnacio García-Jurado, Department of Mathematics, CorunaUniversity, Faculty of Computer Science, Campus de Elvina,15071, Coruna, Spain, [email protected], Ma GloriaFiestras-Janeiro, Ana Meca, Manuel Alfredo MosqueraRodríguez

We deal with the cost allocation problem arising in an inventory trans-portation system with a single item and multiple agents that place jointorders using an EOQ policy. The fixed order cost of each agent is de-fined and the assumptions are stated. For these inventory transportationsystems we introduce and characterize a rule which allows us to allo-cate the costs generated by the joint order. We show that our rule hasgood properties from the point of view of stability.

3 - Capacity sharing: capacity or labor division? A co-operative game approachShoshana Anily, Faculty of Management, Tel Aviv University,69978, Tel Aviv, Israel, [email protected]

In this talk we discuss two types of capacity sharing approaches tomodel various service/production systems as cooperative games withtransferrable utility, and we analyze their core. The literature on co-operative games usually assumes that total capacity is divisible andallocable additively. This premise is reasonable in some settings, butin others it may yield unacceptable solutions. We suggest another ap-proach, which is aligned with the concept of division of labor, wheretotal processing time, rather than total capacity over all resources, isallocated additively.

33


4 - k-norm Cost Games with Critical PlayersAna Meca, Operations Research Center, Universidad MiguelHernández, Avda. Universidad s/n, Edificio Torretamarit,03202, Elche, Alicante, Spain, [email protected], GreysSosicIn this talk, we present some special classes of cost games that satisfyk-norm property and possess players who contribute to cost reductionof all members of the alliance that they belong to (which we call crit-ical players). This type of games can be found in knowledge-sharinggames, inventory games, holding cost games, etc. While papers ana-lyzing these games study specific examples of k-norm cost functionsand analyze how players can reduce their cost through cooperation, ourgoal is to provide an analysis of stability for more general classes ofk-norm games with critical players.

� MB-25Monday, 10:30-12h00CC-A23

Practical Issues of EnvironmentalManagement

Stream: Energy, Environment and Climate 1Contributed sessionChair: Carmen Anido, Economic Analysis: Quantitatve Economy,Autonoma University of Madrid, Faculty of Economics,Cantoblanco., 28049, Madrid, Spain, [email protected]

1 - Dynamic Modelling of Climate Change AdaptationPathways: The Case of Surf Life Saving AustraliaOz Sahin, School of Engineering, Griffith University, GoldCoast Campus, 4222, Griffith University, Qld, Australia,[email protected], Marcello Sano, Russell Richards,Shauna SherkerSurf Life Saving Australia (SLSA) is Australia’s major water safetyand rescue authority whose assets and operations are threatened by ex-treme events and sea level rise. This research aims to explore adap-tation pathways and enhance the capacity of SLSA to tackle thesethreats. To achieve this aim, systems dynamics is used to test concep-tual models developed by stakeholders. These models are designed toconnect climatic drivers, impacts on assets, operations and communi-ties, and strategies to improve adaptive capacity. The initial outcomesof 4 case studies across Australia will be presented.

2 - Materials and Technologies for Industrial HeritageSustainabilityMichal Ganobjak, Institute of History and Theory ofArchitecture and Monument Restoration, Slovak university oftechnology, Faculty of architecture, Námestie slobody 19, 81245, Bratislava, Slovakia, Slovakia, [email protected],Eva KralovaWe study the efficient use of brownfield sites, buildings of manufac-turing process converting to the new alive features. The adaptation ofthe industrial heritage to the new features while maintaining the val-ues must be chosen thoughtfully. Our study shows the optimizationof intervention choices through simulation and visualization of localresources and their interactions. It comprehensively sees the issue ofsocial sustainability, ecology of abandoned industrial environment andits economic aspect.

3 - Impacts of Climate Change on the Slovak EconomyTomas Domonkos, Institute of Economic research, SlovakAcademy of Sciences, Sancova 56, 81105, Bratislava,Slovakia, [email protected], Ivan LichnerClimate change represent one of the most challenging issues humanityhas to face, because they could critically affect all areas of our life. Weaim to analyze and determine the impact of climate change on the Slo-vak economy. Integrated Assessment Model which comprise of threesub-models was designed. Basic principles of the AD-DICE modeldeveloped by de Bruin, Dellink and Tol (2007) are incorporated into arecursive-dynamic CGE model which analyzes the impact of climatechange under various adaptation intensive policies.

� MB-26Monday, 10:30-12h00CC-A24

Recent Research Advances in the OilSectorStream: OR in the Oil and Gas SectorsInvited sessionChair: Irina Dolgopolova, Economics and AdministrativeSciences, Middle East Technical University, Odtu Kent, KonukEvi 1, B Block, 108, Ankara, Turkey, 06420, Ankara, Turkey,[email protected]

1 - From a decoupled to an integrated planning ap-proach in the supply chain of specialty oilsMario Guajardo, Finance and Management Science, NHHNorwegian School of Economics, NHH, Institutt forForetaksøkonomi, Helleveien 30, 5045, Bergen, Norway,[email protected], Martin Kylinger, Mikael RönnqvistWe propose linear models for decoupled and integrated planning in adivergent supply chain for specialty oils. The optimization probleminvolves decisions on production, inventory, transportation, sales anddistribution. The integrated objective is to maximize contribution. Inthe decoupled approach, an internal price system pursues to align sell-ers with this objective. While the total contribution is higher in theintegrated solution, the sellers’ contribution may be lower. We suggestcontribution sharing rules to make both the company and sellers betteroff under the integrated planning.

2 - Dynamic pricing application in the energy marketGamze Ogcu, Industrial engineering, Fatih University,Büyükçekmece, 34500, Istanbul, Turkey,[email protected], Ozlem CosgunThe recent rise in world oil prices and the consequent run-up in gaso-line prices provide a useful analog for illustrating the role of dynamicpricing in electricity markets. In this study, we consider a dynamicpricing problem in one of the private energy companies in Turkey.In the competitive market electricity prices change very frequently,so pricing is the most important tool to response demand. We resortstochastic dynamic programming to decide on the optimal policy. Theresults show that the dynamic pricing policy provides more revenuethan the policy applied currently.

3 - Revenue management and pricing: A case study fora natural gas supplierOzlem Cosgun, Industrial Engineering, Fatih University, FatihUniversitesi endustri muhendisligi bolumu, HadımkoyKampüs Buyukcekmece, 34500, buyukcekmece, Istanbul,Turkey, [email protected], Emre Meterelliyoz,Abdulkadir Kıvrak, Yunus Sartık, Seniye Ümit Oktay FıratIn this study, the dynamic pricing problem of a natural gas supplier inIstanbul is considered. The natural gas consumption is varied accord-ing to the months and cities since the weather condition and revenuelevels of the people in these cities are important factors in consump-tion amount. Clustering analysis is used for determining clusters whichshow similar consumption behavior. Our goal is to find optimal pricesfor each cluster by using stochastic dynamic programming. We try tosearch whether the revenue levels increase by applying dynamic pric-ing policy instead of fixed pricing or not.

� MB-27Monday, 10:30-12h00CC-A25

MCDA and Decision ProcessesStream: Decision ProcessesInvited sessionChair: K. Nadia Papamichail, Manchester Business School,University of Manchester, Booth Street East, M15 6PB,Manchester, United Kingdom, [email protected]

34


1 - Design and evaluation of MCDA interventionsK. Nadia Papamichail, Manchester Business School,University of Manchester, Booth Street East, M15 6PB,Manchester, United Kingdom, [email protected],Theodor Stewart

This paper presents a framework for analyzing and evaluating multi-criteria decision analysis (MCDA) interventions. The framework isapplied to two case studies to illustrate the process of designing andimplementing such interventions. We reflect on the process, the wayit was managed, the composition of the decision body, the tools em-ployed, the information exchanged between actors and the duration ofthe process. We discuss the outcomes and impact of the two MCDAinterventions and provide lessons learned.

2 - Criteria definition and weight assignment in MCDA: areview on environmental problemsMiguel Morgado, INESC Coimbra and Faculty of Economics- University of Coimbra, Portugal, [email protected],Luis C. Dias

Environmental decisions processes involve usually a large number ofstakeholders with conflicting interests. Multi-Criteria Decision Aiding(MCDA) methods have demonstrated to provide more insight and in-crease the transparency of the decision process. We present a literaturereview to analyze and understand how social actors are being involvedin such decision processes, the roles of the experts and the roles of theresearchers involved. We focus on two questions: who and how de-fines the criteria to evaluate the alternatives, and who and how assignsthe weights to each criterion.

3 - Introduction to a new evidential reasoning rule fordecision analysisJian-Bo Yang, Manchester Business School, The University ofManchester, Booth Street East, M15 6PB, Manchester, UnitedKingdom, [email protected]

In the paper, a recently discovered ER rule is reported that provides ageneral process to combine multiple pieces of independent evidence inthe form of belief distribution under the common condition that everypiece of evidence plays a limited role equal to its relative weight. Itgenerates total belief from combination of any two pieces of evidenceas the addition of the bounded average of individual support from eachof them and the orthogonal sum of joint support from both of them. Anumerical example is examined and its significance for decision anal-ysis and processes are discussed.

4 - Supporting the restoration of historical portable or-gan with a new hybrid MCDM modelTian-Jong Hwu, Business Management, National UnitedUniversity, 36003, Miaoli, Taiwan, [email protected],Mei-Chen Lo, Jerzy Michnik, Tadeusz Trzaskalik, MaciejNowak, Gwo-Hshiung Tzeng

The physical state of any historical object exacerbates with time. Animpression it makes on the public becomes effete and its value de-creases. To counteract the influence of time, the conservation andrestoration of historical objects are undertaken. Before starting con-servation, the research to determine its range and to choose the bestmethod should be undertaken. An aim of this study is to apply thenew hybrid MCDM model DANP (DEMATEL-based ANP) to assess-ing the relative values of the historical portable organ and to rank thepotential methods and range of conservation and restoration.

� MB-28Monday, 10:30-12h00CC-A27

OR in Quality Management 2

Stream: OR in Quality ManagementInvited sessionChair: Gulser Koksal, Industrial Engineering, Middle EastTechnical University, Inonu Blvd., 06800, Ankara, Turkey,[email protected]

1 - Optimum 2k-p factorial experiment sequencing: A bi-objective mathematical modelVitoria Pureza, Production Engineering, Universidade Federalde Sao Carlos, Via Washington Luiz, km 235, 13564-090, SaoCarlos, Sao Paulo, [email protected], Pedro Oprime,Antonio Branco Costa

Sequencing experiments with good statistical properties and low costsis a key challenge for both researchers and practitioners. We addressthe problem by proposing a mathematical programming formulationthat aims to generate minimum cost run orders with the best statisti-cal properties for 2k full-factorial and fractional-factorial designs. Re-sults for designs of up to 64 experiments indicate that the approach iscapable of producing optimal or sub-optimal solutions regarding theobjectives addressed for a given design matrix.

2 - Observer scheduling for quality monitoringSatyaveer S. Chauhan, John Molson School of Business,Concordia University, MB 11-317, 1455 de Maissoneuve,Blvd. west, H3G 1M8, Montreal, Quebec, Canada,[email protected], Anjali Awasthi, AnnaGuinzbourg

We propose a decision tool for scheduling observers to access the ser-vice quality in the field. The location of service professional changesdynamically but known. An observer should spend a predefined timewith the service professionals. Services are provided 24 hrs basis butthe observers are scheduled during the daytime. The goal is to observerevery staff for a given number of times during the planning horizon.The overall problem is similar to travelling salesman problem and wedevelop a heuristic approach and tested on the real data set. The exper-imentation results will be presented.

3 - A two-stage stochastic programming model for p-chart designElif Elcin Gunay, Industrial Engineering, Sakarya University,Sakarya, Sakarya, Turkey, [email protected], UfukKula, Tülin Aktin

Car manufacturing is performed in three shops called body, paint andassembly. To control the painting operations it is common to use an at-tributes (p-chart). Determination of chart parameters affects the paintshop objectives (minimizing the color-changeover, or maximizing thecapacity). Therefore, it is important to consider paint shop constraintsand objectives in the design of a p-control chart. Problem formulatedas a two-stage stochastic programming problem, and solved by usingsample approximation algorithm. We provide managerial insights onthe problem by performing a numerical study.

� MB-29Monday, 10:30-12h00CC-A29

Robust Optimization and Non-ConvexInteger Programming

Stream: Mixed-Integer Non-Linear ProgrammingInvited sessionChair: Emiliano Traversi, Fakultät für Mathematik, TechnischeUniversität Dortmund, Germany, [email protected]

1 - Cutting Planes for the Network Loading Problem withStochastic DemandsKonstantinos Kaparis, Managemenst Science, Lancaster,United Kingdom, [email protected], Adam Letchford,Stein W. Wallace

We consider a stochastic variation of the Network Loading Problem.The demand is uncertain and its realisation depends on a finite numberof scenarios. We model this problem as a 2-stage stochastic mixed-integer program with complete recourse.We introduce the new classof probabilistic metric inequalities and we describe a heuristic sepa-ration algorithm.Finally, we report our computational results from theapplication of the proposed scheme on a set of test instances.

35


2 - Robust Integer Non-Linear OptimizationRuth Hübner, Institut für Numerische und AngewandteMathematik, Georg-August-Universität Göttingen,Lotzestraße 16-18, 37083, Göttingen, Germany,[email protected], Anita SchöbelFinding robust solutions of an optimization problem is relevant for ap-plications since for most practical problems the input data is not knownexactly but depends on some unknown scenario. The concept of strictrobustness asks for the best solution which is feasible for every sce-nario. This is in particular hard for integer optimization problems. Weinvestigate the level sets of the robust problem and identify which prop-erties of the original problem still hold for its robust counterpart. Inparticular, we show in which cases an optimal integer solution can beobtained by rounding techniques.

3 - A concave programming approach for Sparse Princi-pal Component Analysis and Sparse representationof signalsFrancesco Rinaldi, Dipartimento Informatica e Sistemistica,Sapienza, via Ariosto, 25 Roma, 00185, Rome, Italy,[email protected], Giampaolo LiuzziWe consider Sparse Principal Component Analysis (SPCA) and sparserepresentation of signals. SPCA is important for data analysis andcompression and has applications in science and engineering. Thegoal is obtaining principal components with only a few nonzero co-efficients, which are usually easier to interpret. The second problemis studied in signal analysis and consists in finding a sparse represen-tation of a signal in terms of a given dictionary. We propose an al-gorithm based on concave programming formulations of the problemsand present numerical results on test problems and real data.

4 - Separable underestimators for Quadratic Combina-torial Optimization.Emiliano Traversi, Fakultät für Mathematik, TechnischeUniversität Dortmund, Germany,[email protected], Christoph BuchheimWe propose a method to obtain separable underestimators forQuadratic Combinatorial Optimization Problems. By exploiting sep-arability we can provide lower bounds by solving an Integer LinearProblem and use them in a Branch and Bound scheme. This is usefulin practice when the underlying linear counterpart is easy to solve. Weinvestigate the tightness of the bounds and their effect on the runningtime of the algorithm. Computational results are provided concerningthe Quadratic Binary Unconstrained Problem and the Quadratic Span-ning Tree Problem.

� MB-30Monday, 10:30-12h00CC-A31

Emerging Applications in Finance

Stream: Emerging Applications in Finance and Logis-ticsInvited sessionChair: Amparo Soler-Dominguez, Universitat Jaume I, Spain,[email protected]

1 - Assessing the role of managers as determinants ofmutual fund efficiency: a nonparametric approachAmparo Soler-Dominguez, Universitat Jaume I, Spain,[email protected], Emili Tortosa-Ausina, Juan CarlosMatallin-SaezThis study contributes to an incessant intent of achieving conclusionsfrom the manager decision-making and its implications for the port-folio efficiency. We apply non-parametric techniques for measuringmutual fund’s performance and we extend the methodology in order toisolate the manager’s influence on the degree of efficiency reached. Re-sults shed light both on investors and practitioners’ concerns becausedifferences among managers do actually arise. This work provides aguide for evaluating managers’ forecasting ability and how they per-form covering ex ante investors’ expectations.

2 - Optimal stopping rules for the random horizon dura-tion problemsMitsushi Tamaki, Business Administration, Aichi Univeristy,370 Kurozasa Miyoshi, 470-0296, Nishikamo, Aichi, Japan,[email protected]

As a variation of the secretary problem, Ferguson et al. considered aproblem called duration problem whose objective is to maximize thetime of possession of a relatively best object when a known numberof objects appear. We here generalize this problem into two directionsby allowing the number N (of objects) to be a random variable witha known upper bound n and also allowing the objects to appear in ac-cordance with Bernoulli trials. We give a sufficient condition for theoptimal rule to be a threshold rule.

3 - The Effects of Topology and Behavioral Canon onNetwork EquilibriaAbhijit Deshmukh, School of Industrial Engineering, PurdueUniversity, 241 Grissom Hall, 47907, West Lafayette, IN,United States, [email protected], Brandon Pope

Network topology has far ranging implications on the presence of con-tagion effects and existence of equilibria in networks of agents. Behav-ioral canon, which determines how decisions are made and the evolu-tion of states in the network, is defined by the interaction mechanismbetween agents through the network topology. This presentation con-siders two behavioral canon paradigms, bargaining and diffusion. Westudy the impact of each paradigm on the speed and quality of the equi-libria reached on networks. Specifically, we study hybrid networks ofPoisson random graphs and scale-free networks.

� MB-31Monday, 10:30-12h00CC-A33

Game Theory and Social Networks 2

Stream: Game Theory and Social NetworksInvited sessionChair: Elisenda Molina, Statistics, Universidad Carlos III deMadrid, Madrid, Spain, [email protected]

1 - Social Capital. A Game Theoretic Approach.Enrique González-Arangüena, Statistics and OperationResearch III, Complutense University of Madrid, Avda.Puerta de Hierro s/n, 28040, Madrid, Madrid, Spain,[email protected], Anna Khmelnitskaya, Conrado M.Manuel

The notion of social capital appeared initially to highlight the impor-tance of social networks relations as a valuable resource for socialand/or economic affairs. We define an index of social capital usinggame-theoretical concepts. We assume that interests of individuals arepresented by means of a cooperative game which take into accountpossibly different players abilities. The network of players relationsis modeled by a (undirected) graph. The social capital of each actoris then measured as the difference between his Myerson value and hisShapley value.

2 - A Game-theoretic Approach to NetworksMartha Saboya, Análisis Económico: Economía Cuantitativa,Universidad Autónoma de Madrid, Paseo de la Chopera 206,1o-B, 28100, Alcobendas (Madrid), Madrid, Spain,[email protected], Guillermo Owen, Susana Lopez

We consider teams whose members are represented as nodes of agraph. By cooperation, the several members can accomplish muchmore than by acting individually. This collaboration is however onlypossible if the several members can communicate, and this requireslinks in the graph. The links can be more or less effective; being ef-fective links generally more costly. We use the Myerson approach togames on graphs to represent the amount of work that can be done, andthe method of multilinear extensions to represent more or less efficientlinks. This allows us to look for optimal links.

36


3 - Average Forest Value for Directed Graph RestrictedGamesElisenda Molina, Statistics, Universidad Carlos III de Madrid,Madrid, Spain, [email protected], Ramon Flores,Anna Khmelnitskaya, Juan Tejada

We propose a value for cooperative TU games with partial action, wheninfluence relations are directional: the Average Forest Value. Based onthis concept, we also propose an index of global Social Network Effi-ciency which measures the ability of the network to promote efficientcoalition structures to form. We analyze their properties and show theirbehavior by means of some illustrative applications (hierarchies, sym-metric games, ...).

� MB-32Monday, 10:30-12h00CC-A34

Understanding the Practice of Soft ORInterventions IStream: Soft OR and Problem Structuring MethodsInvited sessionChair: L. Alberto Franco, Hull University Business School,University of Hull, Management Systems, Cottingham Road, HU67RX, Hull, United Kingdom, [email protected]

1 - The emergence of agreements within facilitated mod-elling workshopsThanos Papadopoulos, Hull University Business School,University of Hull, HU6 7RX, Hull, United Kingdom,[email protected], L. Alberto Franco

The paper focuses on the emergence of agreements in facilitated mod-elling workshops. We address the literature gap concerning the dy-namics and mechanisms that underpin trajectories and outputs of softOR interventions using Actor-Network Theory (ANT). Our analysissuggests that the trajectories and outcomes of such interventions aresocially constructed, emerging from associations of humans and non-humans. We also argue that in practice ANT offers "translation’ to ex-plain the emergence of agreement in the workshop, whereas the work-shop offers models and space for translation to occur.

2 - Exploring dialogue in problem structuring work-shopsElena Tavella, Institute of Food and Resource Economics,University of Copenhagen, Denmark, [email protected]

The use of Problem structuring methods (PSMs) is claimed to havepositive effects on the quality of dialogue. In this presentation I explorethis by examining the dialogue from a PSM workshop held with mem-bers of a local food network in the UK. Drawing on Tsoukas’ (2009)Dialogical Model of Organizational Knowledge Creation, I discuss thecharacteristics of productive dialogue, and assess the extent to whichthey were present in the workshop and how they related to the outputs.Implications of the use of a dialogue lens for research into and practiceof PSM workshops will also be discussed.

3 - Exploring the communicative impact of ProblemStructuring Methods: Results of an interaction-analysis based research.Orestis Afordakos, Warwick Business School, University ofWarwick, CV4 7AL, Coventry, United Kingdom,[email protected]

This paper presents the results of an interaction analysis based re-search, examining non-experimental groups engaging in strategic con-versations, while employing a facilitated PSM. The different elementsof the research framework will be examined with particular emphasison the methodological aspects of interaction analysis. Results in theareas of conflict management and model appropriations will be pre-sented. The presentation will conclude with a discussion on the inter-pretation of the results.

4 - On the roles, uses, and effects of models in Soft ORworkshopsL. Alberto Franco, Hull University Business School,University of Hull, Management Systems, Cottingham Road,HU6 7RX, Hull, United Kingdom, [email protected] presentation pays attention to how groups use models within thecontext of soft OR interventions. Models are conceptualised as arte-facts with specific roles, uses and effects according to the relative com-plexity of different ’boundaries’ faced by workshop participants dur-ing interaction. This conceptualisation is then illustrated via a series ofcase vignettes drawn from an ongoing research programme studyingthe impact of soft OR interventions in multi-organisational settings.The conceptual and prescriptive value of the conceptual framework isthen discussed, as well as its implications for research into the evalua-tion of soft OR interventions.

� MB-33Monday, 10:30-12h00CC-A37

OR and Modelling for SustainableDevelopment

Stream: OR for Development and Developing Coun-triesInvited sessionChair: Subhash Datta, Institute of Management Technology, 35Km Milestone, Katol Road, 441502, NAGPUR, Maharashtra,India, [email protected]

1 - OR/MS Models for Sustainable Development in IndiaSubhash Datta, Institute of Management Technology, 35 KmMilestone, Katol Road, 441502, NAGPUR, Maharashtra,India, [email protected] problems of sustainability can be resolved through OR/MS meth-ods. However, there are major constraints like corruption, infrastruc-ture inadequacies, bureaucratic deficiencies and long gestation periodsin implementation. We discuss the possibility of future transforma-tion of the society through continuous interactions and result orientedmethodologies based on DSS, simulation and connectivity. Once thecommon people are ensured about the benefits in terms of more pro-duction, distribution and employment they have no hesitations in im-plementation of the OR/MS models.

2 - Multicriteria decision making in comparison of objec-tive and subjective criteria of companies’ successBranka Marasovic, Faculty of Economics Split, 21000, Split,Croatia, [email protected], Ivana TadicAlthough financial indicators usually are taken as the most importantindicators of success, lately authors argue about the same relevance inusage of subjective indicators. The aim of this paper is to researchwhether the most successful companies regarding financial ratios (ob-jective group of criteria) show similar results by other groups of crite-ria, as human resource management evaluation and evaluation of entirecompanies’ success (subjective groups of criteria). Selected companieswill be ranked by each group of criteria using multicriteria decisionmaking method PROMETHEE II.

3 - Revised ICI as a leading indicator of Croatian indus-trial production during a recessionMirjana Cizmesija, Faculty of Economics, Dept. of Statistics,University of Zagreb, J.F.Kennedy Square 6, 10000, Zagreb,Croatia, [email protected], Natasa Erjavec, VlastaBahovecWith the aim to improve its forecasting properties, especially duringa recession, we modified a standard composite leading indicator (ICI)in Croatia. Two revised ICIs are proposed and their predictive powerswere analysed on the bases of several econometric models with vari-ables: standard ICI, two revised ICIs and Croatian industrial produc-tion (y-o-y growth rate). As for a Croatian industrial production duringthe recession it is recommended to forecast only one quarter ahead andrevised leading indicator, calculated with negative manager’s answers,showed the best forecasting properties.

37


4 - Flexible approach in planning of infrastructure devel-opmentJerzy Paslawski, Civil and Environmental Eng., PoznanUniversity of Technology, Instytut Konstrukcji Budowlanych,ul. Piotrowo 5, PL 60-965, Poznan, Poland,[email protected]

Flexible approach seems very useful in planning and design when en-vironment is very difficult to predict and forecast. A special attentionmust be paid in design and planning of infrastructure in developingcountries. Analysis of different types of infrastructure (highways, air-ports, bridges) demonstrates high potential to expand but real demand(traffic) is very difficult to forecast in typical time periods (years). Thesudden development of low cost airline carriers is an example of thisproblem. In countries in transition a special attention must be paid oninvestment decision.

� MB-34Monday, 10:30-12h00CC-A39

Cooperative games: dynamics,robustness and related topics

Stream: Cooperative Game TheoryInvited sessionChair: Ehud Lehrer, Tel Aviv University, 69978, Tel Aviv, Israel,[email protected]

� MB-35Monday, 10:30-12h00CC-A41

Assortment and Replenishment Planning

Stream: Demand and Supply in Consumer Goods andRetailingInvited sessionChair: Michael Sternbeck, Supply Chain Management andOperations, Catholic University of Eichstaett-Ingolstadt, Auf derSchanz 49, 85049, Ingolstadt/Donau, Germany,[email protected]

1 - Joint product assortment, inventory and price opti-mizationArgyro Katsifou, TOM, École Polytechnique Fédérale deLausanne (EPFL), Odyssea 1.16 - Station 5, 1015 Lausanne,Switzerland, 1015, Lausanne, Switzerland,[email protected], Ralf W. Seifert, Jean-SébastienTancrez

We study a retailer’s joint problem of product assortment, pricing andinventory management optimization given limited shelf space. Theproduct assortment is composed of "standard" and "variable" products.We propose an iterative heuristic to find the combined product assort-ment as well as the inventory level and the price for each product thatmaximize retailer’s overall profit. The results of our study are illus-trated by numerical examples.

2 - Selecting order packaging units for store delivery ingrocery retailingMichael Sternbeck, Supply Chain Management andOperations, Catholic University of Eichstaett-Ingolstadt, Aufder Schanz 49, 85049, Ingolstadt/Donau, Germany,[email protected], Heinrich Kuhn

In store operations are responsible for the highest share of operationalcosts within the internal retail network. The order packaging unit(OPU) per stock keeping unit is regarded as one driver of in store ef-ficiency. This is the number of consumer units that are combined to

one order and distribution unit for supplying the individual stores. Weconsider a replenishment cycle system that is used by grocery retailersfor non-perishable, non-seasonal and non-promotional items. For thatsetting, we present an approach with which appropriate OPUs can beselected to support packaging decisions.

3 - Scheduling in-store replenishments under service-level and capacity constraintsRob Broekmeulen, OPAC, TU Eindhoven, P.O. Box 516, Pav.E10, 5600 MB, Eindhoven, -, Netherlands,[email protected]

Almost half of the shelf allocations in planograms of retail stores areinsufficient, resulting in overflow stocks in the backroom and frequentadditional replenishments after the order delivery from the outside sup-plier. We propose a procedure to reduce the number of additional re-plenishment trips by clustering the products in schedules for joint in-store replenishment based on the service level target, the allocated storespace on the shelves and the available handling capacity. The proce-dure can also assist planogramming decisions from a cost perspective.

� MB-36Monday, 10:30-12h00CC-A43

DEA and Performance Measurement:Methodology 2

Stream: DEA and Performance MeasurementInvited sessionChair: Dimitris Despotis, Department of Informatics, Universityof Piraeus, 80, Karaoli & Dimitriou Street, 18534, Piraeus,Greece, [email protected]

1 - Two-stage data envelopment analysis: A slacks-based measure procedure for the upper bound per-formanceFuh-Hwa Liu, Industrial Engineering & Management,National Chiao Tung University, 1001 Ta Shueh Road, Dept.IE&M, 300, Hsin Chu City, Taiwan, Taiwan,[email protected], Yu-Cheng Liu

We introduce a three-phase procedure to measure upper performancebound of 2-stage DEA problems. Phase I uses a slack-based measure(SBM) model to assess the process of converting inputs into outputs,and treats the intermediate items as non-discretionary. The interme-diate items are then classified as to-be-decreased (TBD), not-to-be-changed (NC), and to-be-increased (TBI). Phase II employs anotherSBM model to assess the process of converting of inputs and outputs.Phase II repeats until no NCs are identified. Phase III is a SBM model,similar to Phase II but without NCs.

2 - Cost efficiency when prices are not fixed: Disentan-gling quantity and price efficiencyEmmanuel Thanassoulis, Aston Business School, AstonUniversity, Operations and Information Management, B47ET, BIRMINGHAM, United Kingdom,[email protected], Maria Portela

We propose an approach to compute and decompose cost efficiency incontexts where both input quantities and their prices can be adjusted tominimise the aggregate cost of the outputs produced. The models de-veloped can accommodate situations where the degree of control overinput prices is minimal. We introduce a novel decomposition of theoverall measure of cost efficiency into a quantity and a price efficiencycomponent. The efficiency measures are expressed in terms of per-centage cost savings that can be attained through changing prices andquantities towards the overall minimum cost target.

3 - Value-based DEA: A piece-wise linear programmingapproachDimitrios-Georgios Sotiros, Department of Informatics,University of Piraeus, 18534, Piraeus, Greece,[email protected], Dimitris Despotis

38


Value-based DEA is a general approach to incorporate preferences inDEA efficiency assessments. One among the different methods pro-posed is to convert the raw input/output data into values (utilities). Inthis context, we present a piece-wise linear programming approach toDEA efficiency assessments, based on a data transformation-variablealteration technique and assurance-region constraints. Our develop-ments are applicable in both oriented and non-oriented DEA models.We illustrate our approach with a data set drawn from the literature.

4 - An alternative approach to two-stage DEA with a se-ries relationship between the stagesGregory Koronakos, Informatics, University of Piraeus,KAPETAN ZAXARIAN 39, 18757, KERATSINI, Greece,[email protected], Dimitris Despotis

The two fundamental approaches to efficiency assessment in two-stageDEA are the multiplicative and the additive. Both approaches assumea series relationship between the two stages and differ in the way theyconceptualize the decomposition of the overall efficiency to the effi-ciencies of the individual stages. We present in this paper an alterna-tive approach to additive efficiency decomposition in two-stage DEA.We test our models with data sets taken from previous studies and wecompare the results with those reported in the literature.

� MB-37Monday, 10:30-12h00CC-Act

Bioinformatics IIStream: Computational Biology, Bioinformatics andMedicineInvited sessionChair: Aleksandra Swiercz, Institute of Computing Science,Poznan University of Technology, Piotrowo 2, 60-965, Poznan,Poland, [email protected]: Pawel Wojciechowski, Poznan University of Technology,60-965, Poznan, Poland, [email protected]

1 - Highly efficient parallel algorithm for de novo assem-blyAleksandra Swiercz, Institute of Computing Science, PoznanUniversity of Technology, Piotrowo 2, 60-965, Poznan,Poland, [email protected], Piotr Gawron, Jacek Blazewicz,Marta Kasprzak

De novo assembly is the process of reconstructing an unknown DNAsequence from shorter fragments. Short fragments of an organism areread by sequencers, which can produce huge amount of fragmentsin short time. There is a need for the time-efficient algorithm whichcould assembly these fragments together. The proposed hybrid paral-lel method allows using two-level hierarchy: computations in threads(one node with many cores) and computations on different nodes in acluster. Tests carried out on data of Prochlorococcus marinus showedhigh accuracy of the results and short time of computations.

2 - The fast method for contig extraction from the over-lap graphPiotr Gawron, Institute of Computing Science, PoznanUniversity of Technology, ul. Piotrowo 2„ 60-965, Poznan,[email protected]

In this study a method for fast searching of the overlap graph was pro-posed. In presented approach the heuristic algorithm looks for thelongest path in a graph. This path corresponds to the longest con-tig in the overlap graph representation of a DNA assembly problem.It bases on an exact algorithm for minimum path coverage in acyclicgraph. Because of the fact that overlap graph generally is not acyclictransformation of a graph to the acyclic form has been also proposed.The transformation uses specific properties of the overlap graph.

3 - A greedy randomized adaptive search procedurewith Path Relinking for the Shortest SuperstringProblem

Theodoros Gevezes, Mathematical, Physical andComputational Sciences, Aristotle University of Thessaloniki,51124, Thessaloniki, Greece, [email protected]

The Shortest Superstring Problem (SSP) is an NP-hard CombinatorialOptimization Problem (COP) which has attracted the interest due toits applications in DNA sequencing. GRASP is a heuristic that pro-duces good quality solutions for a variety of COP. Path Relinking (PR)integrates intensification and diversification strategies in search for op-timum solutions. An implementation of GRASP with PR for solvingthe SSP is developed for the first time. Computation results indicatethat the new method finds the optimum solution in many cases and ifnot, its solutions are very close to the optimum ones.

4 - A lattice model approach to the molecular distancegeometry problemCristina Requejo, Mathematics & CIDMA, University ofAveiro, 3810-193, Aveiro, Portugal, [email protected],Agostinho Agra, Rosa Maria Figueiredo, Carlile Lavor,António Pereira, Nelson Maculan

The Distance Geometry Problem (DGP) consists in finding an embed-ding in Rn of a given weighted undirected graph such that edge weightsare equal to the corresponding Euclidean distances in the embedding.In order to consider distances that might be affected by experimen-tal errors, presented in real applications, we propose a lattice modelapproach applied to DGP’s related to determination of molecular con-formations.

� MB-38Monday, 10:30-12h00HH-Colombus

Duality in Multi-objective Optimization

Stream: Multiobjective OptimizationInvited sessionChair: Andreas Löhne, Institut für Mathematik, MLUHalle-Wittenberg, Theodor-Lieser-Straße 5, 06099, Halle (Saale),Germany, [email protected]

1 - Alternative generalized Wolfe type and Mond-Weirtype vector dualityEmilia-Loredana Pop, Faculty of Mathematics and ComputerScience, Babes-Bolyai University, 1 M. Kogalniceanu,400084, Cluj-Napoca, Romania,[email protected], Sorin-Mihai Grad

To a general vector optimization problem we attach two new vectorduals by means of perturbation theory. These vector duals are con-structed with the help of the recent Wolfe and Mond-Weir scalar du-als for general optimization problems, proposed by R.I. Bot and S.M.Grad, by exploiting an idea due to W. Breckner and I. Kolumbán. Thenwe attach vector dual problems of Wolfe type and Mond-Weir type byusing different vector perturbation functions to both constrained andunconstrained vector optimization problems, seen as special cases ofthe general vector optimization problem.

2 - Lagrange necessary conditions for Pareto minimiz-ers in Asplund spaces and applicationsChristiane Tammer, Mathematics and Computer Science,Martin-Luther-University Halle-Wittenberg,Theodor-Lieser-Str. 5, D-06120, Halle, Germany,[email protected], Truong Q. Bao

In this talk, new necessary conditions for Pareto minimal points ofsets and Pareto minimizers for constrained multiobjective optimiza-tion problems are established. Our approach is based on a version ofthe separation theorem for nonconvex sets and the subdifferentials ofvector-valued and set-valued mappings. Furthermore, applications inmathematical finance and approximation theory are discussed.

39


3 - Applications of duality in multiobjective optimizationAndreas Löhne, Institut für Mathematik, MLUHalle-Wittenberg, Theodor-Lieser-Straße 5, 06099, Halle(Saale), Germany, [email protected]

Two applications of duality in multi-objective optimization are dis-cussed. First, duality is essential to develop dual algorithms to solvelinear and convex problems. In particular, we focus on the potentialof dual algorithms to solve certain classes of multi-objective combina-torial problems efficiently. The second part deals with applications inMathematical Finance. Multi-objective optimization and duality playa role in market models with transaction costs.

� MB-39Monday, 10:30-12h00HH-Cousteau

MADM Applications II


1 - Evaluating Information Quality and Validity of ValueLine Stock RatingsShin-Yun Wang, Department of Finance, National Dong HwaUniversity, Taiwan, [email protected], Po-Lung Yu

Investment in stock market involves many decision rules and variables;hence investors are increasingly relying on ratings provided by rat-ing agencies to guide their stock selections. However, do these stocksratings have information value? Are rating agencies’ ratings valid?We research these questions by analyzing the performance of ratingagencies. We provide empirical evidence on the performance of rat-ing services by analyzing the performance of Value Line. We intro-duce the concepts of domination cones, domination structures and non-dominated solutions.

2 - Selecting the Optimal Foreign Dramas for Commer-cial TV Stations in Taiwan Using Delphi and AHPPi-Fang Hsu, Department of Communications Management,Shih Hsin University, No.1, Lane 17, Mu-Cha Road, Sec.1,11604, Taipei, Taiwan, [email protected], Shu-Yu Lin

This study develops a model for selecting foreign dramas based onCommercial TV station manager’s perceptions. First, the proposedmodel adopts the Modified Delphi Method to find out the suitable eval-uative criteria of foreign dramas. Next, the model apply the AnalyticHierarchy Process (AHP) to determine the relative weights of crite-ria, then ranks the alternatives and selects the optimal foreign dramas.The results indicated that concerns think highly of criteria in order aredrama content perspective, cast perspective, advertising budget per-spective, foreign data of origin perspective.

3 - Developing a Selection Model for Newspaper Inte-grated Marketing Communication ServiceYa-Han Lin, Department of Communications Management,Shih Hsin University, No.1, Lane 17, Mu-Cha Road, Sec.1,11604, Taipei, Taiwan, [email protected], Pi-Fang Hsu

This research develops a model for selecting newspaper integratedmarketing communication (IMC) service. The proposed model com-prises two parts, with the first part using the modified Delphi methodto identify suitable evaluative criteria of IMC service, while the secondpart applies Analytic Hierarchy Process (AHP) to calculate and rankthe optimal IMC service. Additionally, the model is then tested by anewspaper IMC service selection case for advertisers. The proposedmodel helps advertisers to effectively select newspaper IMC service,making it highly applicable for academia and commerce.

4 - Measuring the International Backpackers TravellingRiskPin-Ju Juan, Department of International TourismManagement„ Tamkang University, No.180, Linwei Rd.,Jiaoxi Township„ 26247, Yilan County, Taiwan,[email protected], Shin-Yi Lin

This study intends to explore a decision model for international back-packers by developing a travelling risk selection. The factors and crite-ria for the evaluation decision model are derived from exhaustive liter-ature review and experts interviews. By using the experts’ interviews,content analysis and modified Delphi method, the study aims to obtainfactors and criteria for travelling risk. Finally, a recommended solu-tions are presented and analyzed based on the strategic plan and canprovide valuable insights for tourism administrators and academics ininternational tourism market.

� MB-40Monday, 10:30-12h00HH-Livingstone

MADM Applications VI


1 - A qualitative model for collective intelligence analy-sisJiang-Liang Hou, Dept of Ind. Eng. & Eng. Mgt., NationalTsing Hua Univ., No. 101, Section 2, Kuang-Fu Road, 300,Hsinchu, Taiwan, [email protected], Hung Lung Lin,Shih-Ting Yang, Yu-Min Chiang

Based on the concept of CI, a three-stage tendency determinationmodel of community opinions is proposed to develop a platform forgathering public opinions, determining tendency of the communityopinions and supporting decision makers with decision conclusions.The three stages include "analysis of opinions contents’, "representa-tion of formatted opinion contents’, and "analysis of opinions tendencyand representation of conclusions’. The proposed model is to extracttext-based opinions of individuals into a generally accepted decision-making reference for the decision maker.

2 - Taiwan Health Communication Research: Applica-tion and Impact of New MediaRose Hung, Department of Communications Management,Shih Hsin University, Taiwan, [email protected], Pi-FangHsu

With the rise of social networks and mobile media, health communica-tion models also come with each passing day, compared to traditionalmedia, new media more effectively to achieve the dissemination ofhealth knowledge and help people or patients to implement lifestyleadjustment? This study analysis of the Taiwan government and privatehealth communication activities in recent years, through the combina-tion of interviews and questionnaires, the results are analyzed to findout the health communication effect and construct a health dissemina-tion pattern.

3 - Hybrid MCDM Approach for Company Growth Strat-egy Selection in MarketingTuncay Gürbüz, Industrial Engineering, GalatasarayUniversity, Ciragan cad. No.36, Ortakoy, 34357, Istanbul,Turkey, [email protected], Y. Esra Albayrak

In today’s competitive markets, one can easily see that the importanceis not producing goods and services but being able to sell what oneproduces. Nowadays, the concept of marketing is customer-orientedand aims to meet the needs and expectations of customers to increase

40


their satisfaction. While creating a marketing strategy, an enterprisemust consider many factors and decide to perform either target deter-mination or company growth strategies. In this study, company growthstrategies are analyzed with a hybrid MCDM method consisting of An-alytic Network Process and Choquet Integral.

� MB-41Monday, 10:30-12h00HH-Heyerdahl

AHP/ANP 1Stream: Analytic Hierarchy Process, Analytic NetworkProcessInvited sessionChair: Josef Jablonsky, Dept. of Econometrics, University ofEconomics Prague, W.Churchill sq. 4, 13067, Prague 3, CzechRepublic, [email protected]

1 - Determinants of customers satisfaction in the Nige-rian aviation industry using AHP modelAdebola Adekoya, Business Administration, University ofLagos, Room 111, Dept. of Bus. Admin., Faculty of Bus.Admin., University of Lagos, Akoka, Yaba„ 234, Akoka,Yaba, Lagos, Nigeria, [email protected], B.e.a.Oghojafor, Emmanuel O. Oyatoye

The airline industry in Africa’s most populous nation has witnessedexplosive growth in recent years with older domestic airlines being incompetition with new entrants. The expansion has given Nigerians awider choice of airlines. This research is designed to apply the Ana-lytic Hierarchy Process to find out the determinants of customers satis-faction in the Nigerian aviation industry. This is an exploratory studywith focus on primary data which will be collected using conveniencesampling and snowballing techniques. A quantitative approach is usedto analyze data provided.

2 - Performance evaluation of inmetro: a contribution ofmulticriteria decision aidLorena Berzins, ibmec, Rua Ipanema, 99/902, 22631390, Riode Janeiro, RJ, Brazil, [email protected], LuisRangel

A performance appraisal program, when well structured, is a usefultool to measure organization performance. The performance of an em-ployee is measured under multicriteria, and this work uses AHP, a Mul-ticriteria Decision Aiding tool, to structure PA. The study was testedin a governmental organization in Brazil, INMETRO. The method isflexible because it can be adjusted to the needs of each department. Butat the same time has the same structure and criteria to the whole orga-nization, so it is possible to evaluate the performance of the companyand compare results between each department.

3 - Combine ANP and Fuzzy integral approaches forsupplier evaluation and selectionJames Liou, Industrial Engineering and Management,National Taipei University of Technology, No. 1 Chung-HsaoE. Rd. Sec. 3, 106, Taipei, Taiwan, [email protected],Yen-Ching Chuang, Gwo-Hshiung Tzeng

In this study, we propose a novel fuzzy integral-based model, which ad-dresses the dependent relationships between the various criteria and thenon-additive weighted gap analysis. The relations-structure among thecriteria and the weights of the criteria are built with the aid of the Deci-sion Making Trial and Evaluation Laboratory and Analytical NetworkProcess (DANP) method. The fuzzy integral is then used to aggre-gate the gaps and weights. The proposed model remedies prior short-comings and should be more reasonable. The method is demonstratedusing data from a Taiwanese company.

4 - Comparison of methods for deriving priorities in theanalytic hierarchy processJosef Jablonsky, Dept. of Econometrics, University ofEconomics Prague, W.Churchill sq. 4, 13067, Prague 3,Czech Republic, [email protected]

A crucial problem in the analytic hierarchy/network process is how toderive priorities form pairwise comparison matrices. The most popu-lar methods for deriving priorities are eigenvector, proposed originallyby T. Saaty, logarithmic least square and least square method.The pa-per deals with other alternative approaches using methodology of goalprogramming — one is based on minimization of sum of absolute orrelative deviations and another on minimization of maximum devia-tion. Results are compared on a set of randomly generated matrices ofdifferent sizes and different levels of consistency.

� MB-42Monday, 10:30-12h00BW-Amber

EURO Doctoral Dissertation Award (EDDA2012)

Stream: EURO Doctoral Dissertation Award (EDDA2012)Invited sessionChair: Jacques Teghem, Mathro, Faculté Polytechnique/UMonss,9, rue de Houdain, Rue de Houdain 9, 7000, Mons, Belgium,[email protected]: Silvano Martello, DEIS, University of Bologna, VialeRisorgimento 2, 40136, Bologna, Italy, [email protected]: Hartmut Stadtler, Institute for Logistics and Transport,University of Hamburg, Von-Melle-Park 5, 20146, Hamburg,Germany, [email protected]: Hartmut Stadtler, Uni Hamburg, Von-Melle-Park 5, 20146,Hamburg, Germany, [email protected]: Ahti Salo, Systems Analysis Laboratory, Aalto UniversitySchool of Science and Technology, P.O. Box 11100, Otakaari 1 M,00076, Aalto, Finland, [email protected]: Stéphane Dauzere-peres, Manufacturing Sciences andLogistics, Ecole des Mines de Saint-Etienne, 880 avenue de Minet,13541, GARDANNE, France, [email protected]

� MB-43Monday, 10:30-12h00BW-Granite

Stochastic Programming - Stability,EstimationStream: Stochastic ProgrammingInvited sessionChair: Petr Lachout, Dept. Probability and Statistics, CharlesUniversity in Praha, Sokolovská 83, 186 75, Praha, CzechRepublic, [email protected]

1 - Empirical Estimates via Heavy TailsVlasta Kanková, Econometrics, Institute of InformationTheory and Automation of ASCR, Pod Vodárenskou vezí 4,Praha 8, CZ 18208, Prague, Czech Republic, Czech Republic,[email protected]

Optimization problems depending on a probability measure corre-spond to many economic and financial applications. Since this mea-sure is often unknown, a solution has to be usually determined on thedata basis, underlying distribution function is replaced by empiricalone. Many efforts has been paid to the investigation of the correspond-ing empirical estimates of an optimal value and an optimal solution.It was mostly done for the underlying distribution with thin tails. Theaim of the contribution is to focus on the case of heavy tails, includingshape parameter in (1,2).

41


2 - Stochastic program with additional informationPetr Lachout, Dept. Probability and Statistics, CharlesUniversity in Praha, Sokolovská 83, 186 75, Praha, CzechRepublic, [email protected]

Treating a stochastic program we usually possess an additional infor-mation about the subject and background which should be incorpo-rated in the program. Typically, there are historical observations ofthe system, recently observed data, expert suggestions and forecasts,knowledge about smoothness of the program, etc. We intend to de-scribe structure of such programs together with a stability discussion.

3 - Universal Confidence Sets - Estimation and Relax-ationSilvia Vogel, Mathematics and Natural Sciences, IlmenauUniversity of Technology, Postfach 100565, 98684, Ilmenau,Thuringia, Germany, [email protected]

The derivation of universal confidence regions for constraint sets, op-timal values and solutions sets of decision problems usually assumesconcentration-of-measure properties for the objective and/or constraintfunctions and some knowledge about the true problem, such as valuesof a growth function. If these values are not available, one can estimatethem relying on the concentration-of-measure inequalities for the truefunctions. We show how such estimates can be derived and incorpo-rated. Furthermore, we investigate confidence sets which are obtainedvia relaxation of certain inequalities.

4 - Stability and convexity of chance-constrained prob-lems with structural dependenceMichal Houda, Econometrics, Institute of Information Theoryand Automation of the ASCR, Pod Vodárenskou vezí 4, Praha8, CZ 18208, Prague, Czech Republic, [email protected]

In chance-constrained problems, the convexity of the constraint setis important. It is known that a suitable combination of a concav-ity property of the probability distribution and concavity of constraintmappings are sufficient conditions to the convexity of the constraintset. We assume furthermore a structural dependence assumption: con-straint rows not required to be independent. The resulting constraintset is non-convex but can be approximated by a convex one. We showthat, applying stability results on optimal values and optimal solutions,these remain stable under common assumptions.

� MB-44Monday, 10:30-12h00BW-Marble

Inventory, Pricing, and CustomerBehaviorStream: Revenue ManagementInvited sessionChair: Qing Li, Dept. of ISOM, Hong Kong University of Scienceand Technology, Clear Water Bay, Kowloon, Hong Kong,[email protected]

1 - Group buying mechanismsRachel Zhang, IELM, Hong Kong UST, Clear Water Bay,00000, Kowloon, Hong Kong, [email protected]

When a seller offers quantity discounts, interested buyers may eitherself-organize or rely on a third party to aggregate their purchasingquantities to obtain lower prices. This paper concerns group buyingamong heterogeneous buyers given a seller’s quantity discount sched-ule under both the uniform price and non-uniform price group buy-ing mechanisms. We show that the latter have a direct impact on thebuyers’ payments, purchase quantities, and the seller’s revenue. Wealso compare buyers’ purchasing quantities and surplus under differ-ent group buying mechanisms and when buyers buy alone.

2 - Optimal pricing policy for recyclable short life-cycleproductsHsiao-Fan Wang, IEEM, Natl Tsing Hua University, 101Sec.2, Kuang-Fu Rd.„ 30043, Hsinchu, Taiwan,[email protected]

Because of the environmental degradation, producing green products isnecessary. Since green products are market driven, proposing a methodfor pricing on a short-life, green product in market so that an enterpriseor a retailer can make profit from producing and selling the green prod-ucts is our primal intention. Under a given market scale and recyclingprinciple, the optimal pricing policy based on Game Theory was pro-posed by a developed DP model for two kinds of products which havedifferent degrees of greenness from different manufacturers sold byone retailer.

3 - Managing perishable goods under different cus-tomer preferencesQing Li, Dept. of ISOM, Hong Kong University of Scienceand Technology, Clear Water Bay, Kowloon, Hong Kong,[email protected], Xiaoli Wu

The shelf lives of perishable goods are influenced by decisions madealong the supply chains and on shelves. How should firms coordi-nate the shelf life decisions and inventory decisions? We show thatthe answer depends critically on customers’ sensitivity to freshness.We consider two classes of customers and measure sensitivity to fresh-ness by the percentage of customers who purchase only new items, andwhether the customers who accept both new and aged items retrieveitems FIFO or LIFO.

� MB-45Monday, 10:30-12h00BW-Water

Applications of Pricing and RevenueManagement

Stream: Revenue ManagementInvited sessionChair: Luce Brotcorne, INRIA, Parc Scientifique de la HauteBorne, 40 avenue Halley, Bat a Park Plaza, 59650, Villeneuved’Ascq, France, [email protected]

1 - Revenue management in sequential distribution: Anapplication to the Hindi Movie IndustryMegha Sharma, Operations Management, Indian Institute ofManagement Calcutta, A-306, Academic Block, IIM Calcutta,Joka, DH Road, Kolkata, 700104, Kolkata, West Bengal,India, [email protected], Sumanta Basu

With about 125 movie releases in a year, Hindi movie industry is oneof the largest producers of movies in the world. Over the years the lifecycle of movies has shortened significantly and a movie is typicallypremiered on TV within 4 months of its release. While a movie’s earn-ings from TV premier are significant and decrease sharply with time,an early release cannibalizes revenue from other channels. This workattempts to determine the optimal timing of TV premier using statisti-cal methods and mathematical modeling in light of video piracy, andmovie reviews through social media.

2 - Dynamic pricing with predefined tariff constraintsKris Meyers, Vlerick Leuven Gent Management School andGhent University, Reep 1, 9000, Gent, Belgium,[email protected], Mario Vanhoucke, Behzad Samii

We develop a discrete dynamic pricing model for electricity (but alsofor applications like mobile communication, car rental) to maximizeprofit under fluctuating demand, in a discrete finite time horizon. Theseller has an agreement with the buyers to sell at preset tariffs. He cancharge the highest tariffs for a limited amount of periods. Cost is apiecewise linear function of the quantity sold. We propose an IP for-mulation, test the model with numerical experiments and develop al-gorithms for several extentions like multiple customer segments, costfor changing tariffs, stochastic demand.

42

EURO 25 - Vilnius 2012 MC-02

3 - An efficient heuristic approach for a dynamic pricingproblem in telecommunicationsJean-robin Medori, OrangeLabs, France,[email protected], Mustapha Bouhtou, MichelMinoux

We consider a new pricing problem where a carrier wants to set dy-namic prices to maximize daily revenue from a voice service. A MixedInteger Quadratic Program (MIQP) is proposed to model this prob-lem. Because of network load constraints, finding an optimal solutionis challenging. To bypass this computational limitation we propose aheuristic which solves easier MIQP sub-problems. The heuristic guar-antees a feasible solution whenever the original pricing problem is fea-sible. Experimentations show that it consistently produces solutionsclose to exact optimum within reduced computation times.

4 - A Bilevel Approach for a Long Haul Freight Trans-portation Pricing ProblemLuce Brotcorne, INRIA, Parc Scientifique de la Haute Borne,40 avenue Halley, Bat a Park Plaza, 59650, Villeneuved’Ascq, France, [email protected], Diaby Moustapha,El-ghazali Talbi

In this talk we consider a pricing problem involving two long-haul fullload freight carriers (A,B) operating similar markets. We assume thatcarrier B cannot serve all the demand addressed to him with its owntransportation fleet. He thus needs to use outsourcing: carrier A orits competitors. Carrier A, which unused capacity, has thus to defineprices to carrier B transportation requests to maximize revenue and bal-ance the load. This sequential and non-cooperative decision-makingprocess is represented as a bi-objective bilevel program. Numericalresults are given.

Monday, 12:30-14:00

� MC-01Monday, 12:30-14:00RB-Alfa

Keynote Lecture: Professor Ceyda Oguz

Stream: Keynote, Tutorial and Plenary LecturesKeynote sessionChair: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - Computational Biology and Operations ResearchCeyda Oguz, Department of Industrial Engineering, KocUniversity, Rumeli Feneri Yolu, Sariyer, 34450, Istanbul,Turkey, [email protected] the scientific understanding of organisms through compu-tation has an ever-increasing importance that necessitates more re-search into fundamental problems of molecular biology. Research di-rection for these problems involves better analytical tools and moreefficient algorithms due to massive volume of data. Hence compu-tational biology can benefit from operations research area for devel-oping mathematical models and devising efficient algorithms. In theliterature, while integer programming and network models are usuallyemployed at the modeling stage, dynamic programming, branch-and-bound algorithm, heuristics and metaheuristics are the prevalent com-putational methods employed for the problems. In this talk, I will givean overview of application of operations research methods to severalcomputational problems that are faced in molecular biology.

� MC-02Monday, 12:30-14:00RB-Beta

Scheduling Approaches for ComplexManufacturing Systems

Stream: SchedulingInvited sessionChair: Lars Moench, FernUniversität in Hagen, 58097, Hagen,Germany, [email protected]

1 - A GRASP Approach for the Job Shop SchedulingProblem with Total Weighted Tardiness ObjectiveJens Kuhpfahl, School of Economics and Business, Chair ofProduction and Logistics, MLU Halle-Wittenberg, Germany,[email protected], Christian BierwirthSupply chain cooperation of companies requires a complex coordina-tion of distributed but interdependent production processes. A con-sequence of this development is that production scheduling is oftenmore concerned with keeping due dates than realizing short flow timesof the jobs. In our talk we consider the job shop scheduling problemwith minimizing the total weighted tardiness. We present a greedyrandomized adaptive search procedure that includes a powerful con-struction heuristic, an effective selection of neighbourhoods within theimprovement phase and a navigation concept for the search.

2 - Heuristics for Integrated Process Planning andSchedulingOleh Sobeyko, University of Hagen, 58097, Hagen, Germany,[email protected], Lars MoenchWe consider an integrated process planning and scheduling problemfor a flexible job shop. The performance measure is the total weightedtardiness. A mixed integer programming formulation is presented. Be-cause the problem is NP hard, we propose an iterative procedure. Avariable neighborhood search scheme is used to select appropriate pro-cess models, i.e., bill of materials and routes. A shifting bottleneckheuristic is used to schedule the jobs given a process model. We presentthe results of computational experiments.

43

MC-03 EURO 25 - Vilnius 2012

3 - Scheduling Flowline Manufacturing Cells with Inter-cellular Moves: The Cell Scheduling ProblemJanis Sebastian Neufeld, Industrial Management, TUDresden, TU Dresden, Fakultät Wirtschaftswissenschaften,01062, Dresden, Germany,[email protected], Udo BuscherThe scheduling of multiple manufacturing cells with intercellularmoves has not received much attention so far. Analogously to groupscheduling, for the so called cell scheduling problem a sequence ofjobs within each cell as well as a sequence of part families are de-termined. We point out some major differences between the cell andgroup scheduling problem. Furthermore, specific problem structuresare identified and two simulated annealing algorithms are presentedto solve the permutation as well as non-permutation cell schedulingproblem more effectively than existing heuristic approaches.

4 - A New Hybrid Algorithm for Solving the Resource-Constrained-Project-Scheduling-Problem (RCPSP).Luis Moreno, Sistemas, Universidad Nacional de Colombia,Cra. 80 65-223, Bloque M8A of. 207, Medellin, Antioquia,Colombia, [email protected], Javier Diaz, Gloria PeñaThe RCPSP is an NP-hard problem, for which numerous heuristicmethods have been proposed for a solution. Our research presents analgorithm that is a combination of a known exact branch and boundalgorithm, with a new heuristic manner of "uniformly’ visiting, a pre-defined number of times, all of the areas for the feasible space of so-lutions, and skipping branches in the solution tree, while making ad-ditional search efforts in promising areas. The algorithm is evaluatedbased on 480 different instances of 30 activities from the PSPLIB li-brary which is available online.

� MC-03Monday, 12:30-14:00RB-L1

Business Excellence in Logistics:Advanced Vehicle and Inventory Routing

Stream: Business Excellence in LogisticsInvited sessionChair: Martin Josef Geiger, Logistics Management Department,Helmut-Schmidt-University, Holstenhofweg 85, 22041, Hamburg,Germany, [email protected]

1 - Cyclic Delivery Schedules for Inventory RoutingProblemGultekin Kuyzu, Industrial Engineering, TOBB University ofEconomics and Technology, Sogutozu Caddesi No:43,Sogutozu, 06560, Ankara, Turkey, [email protected], AliEkici, Okan OzenerWe consider an inventory routing problem where a common vendor isresponsible for replenishing the inventories of several customers over aperpetual time horizon. The objective of the vendor is to minimize thetotal cost of transportation of a single product from a single depot toa set of customers with deterministic and stationary consumption rateover a planning horizon while avoiding stock-outs at the customer loca-tions. We focus on constructing a repeatable (cyclic) delivery schedulefor the product delivery.

2 - Multi-Objective Inventory Routing: Recent AdvancesMartin Josef Geiger, Logistics Management Department,Helmut-Schmidt-University, Holstenhofweg 85, 22041,Hamburg, Germany, [email protected], ThibautBarthelemy, Marc SevauxWe consider a multi-objective generalization of the inventory routingproblem, in which inventory levels and routing costs are not combinedinto an overall evaluation function but treated separately. The prob-lem is solved by local search, and numerical results are computed andreported. Particular emphasis has been laid on the representation ofsolutions from a practical point of view. In detail, individual frequencyvalues are derived for each customer, implementing a recurring deliv-ery policy. Our findings show that there is great potential for tradeoffsbetween the two objectives.

3 - A Multi-Stage Optimisation Approach for DeliveryPlanning and SchedulingVitaly Bolshakov, Department of Modelling and Simulation,Riga Technical University, Kalku Street 1, LV-1658, Riga,Latvia, Latvia, [email protected], Galina Merkuryeva,Michael Affenzeller, Erik Pitzer

Product delivery planning has a high priority in the modern transporta-tion logistics. In distribution centres, this task is related to tactical plan-ning of different types of product deliveries, taking into the accountpredetermined time constraints, transportation costs and variable de-mand. Real-life applications also require solution of multiple tasks. Amulti-stage optimisation approach for delivery planning and schedul-ing is proposed. Optimisation solutions for separate stages are builtwithin HeuristicLab environment and adjusted for a specific businesscase.

4 - Selective and Periodic Inventory Routing ProblemÖzge Tüncel, Industrial Engineering, Koc University, KocUniversity Rumelifeneri Yolu, Sarıyer, 34450, Istanbul,Turkey, [email protected], Sibel Salman, Onur Kaya, DenizAksen

A biodiesel production facility collecting waste vegetable oil fromsource nodes uses the collected waste as raw material for biodieselproduction. The manager needs to decide which of the source pointsto include in the collection program, which of them to visit on eachday, which periodic routing schedule to repeat over an infinite horizonand how many vehicles to operate such that the total cost is minimizedwhile the production requirements and operational constraints are met.A heuristic model is constructed in order to find near optimal solutionsfor small and large sets of instances.

� MC-04Monday, 12:30-14:00RB-L2

Optimal Control I

Stream: Optimal ControlInvited sessionChair: Erik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39, 85577,Neubiberg, Germany, [email protected]: Ekaterina Kostina, Department of Mathematics andComputer Science, University of Marburg, Hans-Meerwein-Str.,35032, Marburg, [email protected]

1 - Direct Optimal Control Methods for a Centralized Ap-proach to Separation ManagementEkaterina Kostina, Department of Mathematics and ComputerScience, University of Marburg, Hans-Meerwein-Str., 35032,Marburg, [email protected]

A key element of future Air Traffic Management concepts are automa-tion tools helping to separate aircrafts in a tactical environment. Thiswork investigates optimal control approach to separation. Our opti-mization framework has full information of the traffic situation in anairspace volume and hence generates conflict-free trajectories for allaircrafts within a regarded airspace. Real-time simulations showed thatcompared to human solutions the method could yield a significant de-crease of delay, flight track excess, as well as fuel savings. Joint workwith M. Poppe and L. Walter.

2 - Optimal control models of renewable energy produc-tion under fluctuating supplyGernot Tragler, OR and Control Systems, Vienna Universityof Technology, Argentinierstr. 8/105-4, A-1040, Vienna,Austria, [email protected], Elke Moser, Dieter Grass,Alexia Prskawetz

We consider optimal control models with infinite horizon, in whichfossil and renewable energy cover production-related energy demand.The main focus lies on the investigation how prices and policy inter-ventions (e.g. taxes, subsidies, quotas) can affect the attractiveness of

44


renewable energy and whether a switch towards a more sustainable en-ergy supply can be stimulated. The assumption of fluctuating supplyof renewable energy sources implies non-autonomy, which requiressophisticated numerical solution methods. Parameter values are sup-ported by data and sensitivity analyses are carried out.

3 - Proposing and simulating a new suboptimal hy-brid embedding-PMP technique for tumor anti-angiogenesis optimal controlAlireza Fakharzadeh, Mathematics Dept., Shiraz University ofTechnology„ 71555-313, Shiraz, Iran, Islamic Republic Of,[email protected], Zahra KhoshgamSeveral mathematical models exist on describing tumor angiogenesisprocess. A famous one the models is proposed by Hahnfeldt et al.which Ledzewics gained a complete solution for that optimal controlmodel based on the Pontryagin Maximum Principle (PMP). This proto-col is not applicable. We introduce a new suboptimal protocol (hybridembedding-PMP technique). The simulation results and comparisonbetween the analytical solution and the hybrid embedding-PMP proto-col, for the same initial condition, indicate that our suggested protocolis easier, practical and acceptable.

� MC-05Monday, 12:30-14:00RB-L3

Liner shipping optimization

Stream: Maritime TransportationInvited sessionChair: Fernando Alvarez, Goteborggata 9, 0566, Oslo, Norway,[email protected]

1 - A Multi-period Planning Model for Liner Shipping’sBunker ManagementXiaoyu Wang, School of Civil and EnvironmentalEngineering, Nanyang Technological University, School ofCEE, NTU, 50 Nanyang Avenue, Singapore, 639798,Singapore, Outside US, Singapore, [email protected],Chee Chong TeoThis paper considers bunker risk management for container lines. Anonlinear finite horizon optimization program is developed to simul-taneously optimize liner network planning and bunker hedging. Themodel captures the uncertainty of bunker price and the associated tol-erance for bunker risk exposure. As a rolling-horizon model, it allowsthe updating of the market information and forecast, such as bunkerprices, over the planning horizon. The insights gained from our anal-ysis would be useful for the container shipping industry in improvingtheir bunker risk management.

2 - The Vessel Schedule Recovery Problem - disruptionmanagement in liner shippingBerit Dangaard Brouer, DTU Management Engineering,Technical University of Denmark - DTU, Bygning 426, 2800,Kongens Lyngby, Denmark, [email protected], ChristianEdinger Munk Plum, Bo Vaaben, David Pisinger, JakobDirksenMaritime transportation is the backbone of world trade and is account-able for around 3% of the worlds CO2 emissions. We present the Ves-sel Schedule Recovery Problem (VSRP) to evaluate a given disruptionscenario and to select a recovery action balancing the trade off be-tween increased bunker consumption and the impact on the remainingnetwork and the customer service level. The model is applied to 4real cases from Maersk Line. Solutions are comparable or superior tothose chosen by operations managers. Cost savings of up to 58% maybe achieved.

3 - Decision Support to Overcome Traffic Congestion inConstrained Transport CorridorsStefan Voss, Wirtschaftsinformatik/Information Systems,University of Hamburg, Von-Melle-Park 5, 20146, Hamburg,Germany, [email protected], Silvia Schwarze,Xiaoning Shi

To cope with increasing traffic, network resource extension may betraded against congestion based costs. We focus on game theoreticmethods applied on scenarios for constrained transport corridors, likecanals in maritime shipping or similar bottlenecks in other traffic sys-tems. Real-world traffic corridors are analyzed to obtain informationon restrictions at bottlenecks and corresponding behavior of their users.In particular, the change of delay and costs when switching from de-centralized to centralized models is of interest as well as an analysis ofthe interaction between both positions.

4 - Competition and efficiency in the container linershipping industryChrista Sys, Transport and Regional Economics, University ofAntwerp, Prinsstraat 13, 2000, Antwerp, Belgium,[email protected]

The present paper examines a model-based measure of competition.This measure of competition based on Boone’s theoretical work quan-tifies the impact of marginal costs on performance. Boone shows thatwhen profit differences are increasingly determined by marginal costdifferences, this indicates increased competition. The evolution ofcompetition will be calculated using a pooled data set of 20 liner op-erators. After correcting for industry-specific effects (i.e. number ofservices, average ship size), the findings suggest that over the 2000-2008 period the competition has intensified.

� MC-06Monday, 12:30-14:00RB-Gamma

Production, planning and facilty location

Stream: MetaheuristicsInvited sessionChair: Maria Eugénia Captivo, Departamento de Estatística eInvestigação Operacional, Universidade de Lisboa, Faculdade deCiências and Centro de Investigação Operacional, Campo Grande,Bloco C6, Piso 4, 1749-016, Lisboa, Portugal, [email protected]

1 - A Dynamic Scheduling Approach for Solving the Sin-gle Machine Total Weighted Tardiness ProblemSener Akpinar, Industrial Engineering, Dokuz EylulUniversity, 35160, Izmir, Buca, Turkey,[email protected], Alper Hamzadayi, Simge YelkenciKose, Hanifi Okan Isguder, Atabak Elmi

In this paper we aim at solving the single machine total weighted tar-diness problem (SMTWT) by combining simulation via genetic algo-rithm (GA). In the classical SMTWT all jobs are available at time zeroto be processed on the machine. In this study jobs enter the systemaccording to an arrival process and processed on a single machine withregard to a priority and a due date assigned to each job. The problemconsists of a set of independent jobs with distinct processing times,weights and due dates to be scheduled on a single machine. The solu-tion approach is based on GA and simulation.

2 - A competitive genetic algorithm for single row facil-ity layoutRavi Kothari, Production & Quantitative Methods, IndianInstitute of Management Ahmedabad, c/o SAO Office,Dorm-21, Room-16, Vastrapur, IIM Ahmedabad, 380015,Ahmedabad, Gujarat, India, [email protected],Diptesh Ghosh

The single row facility layout problem (SRFLP) is the NP-hard prob-lem of arranging facilities on a line, so as to minimize the weightedsum of the distances between all pairs of facilities. We present a ge-netic algorithm to solve large SRFLP instances. Our computationalexperiments show that an appropriate selection of genetic operatorscan yield high quality solutions in spite of starting with an initial pop-ulation that is largely randomly generated. Our algorithm improves onthe previously best known solutions for 24 of 43 benchmark instancesand is competitive for the remaining ones.

45


3 - A genetic approach for planning elective surgeries ina Portuguese hospitalMaria Eugénia Captivo, Departamento de Estatística eInvestigação Operacional, Universidade de Lisboa, Faculdadede Ciências and Centro de Investigação Operacional, CampoGrande, Bloco C6, Piso 4, 1749-016, Lisboa, Portugal,[email protected], Inês Marques, Margarida PatoReduced budgets in the healthcare sector pressure health institutions toan efficient use of resources. The operating theatre represents a greatproportion of the hospital budget. Furthermore, it is a central servicewith connections and implications in the service of many other hos-pital units. A more efficient use of the operating room becomes ofgreat relevance within the hospital administration. This talk presentsa genetic approach developed to address the elective surgery planningproblem arising in a hospital in Lisbon. Some results obtained with thehospital’s data are presented.

4 - A Hybrid Metaheuristic-Dynamic Programming Algo-rithm for Integrated Optimization of Manual OrderPicking and Sorting Warehouse OperationsAthanassios Nikolakopoulos, Chemical Engineering, NationalTechnical University of Athens, Heroon Polytechneiou, 9,15780, Athens, Greece, [email protected], IoannisGanas, Ioannis Konstantaras, Konstantina SkouriThis work introduces an ILP formulation of the problem of throughputtime minimization of order picking and sorting warehouse operations.It includes due times, sorting times and TSP sub-problems with binpacking constraints. An Evolutionary Threshold Accepting algorithmand a Dynamic Programming part are combined into a hybrid solutionmethod. Results are produced for artificial and real life instances andcompared to the results of the algorithmic simulation of conventionalpractices. Computational experience shows a 10% reduction of aver-age total makespan in affordable computational times.

� MC-07Monday, 12:30-14:00RB-Eta

Applications of Vehicle Routing

Stream: Vehicle Routing and Logistics OptimizationInvited sessionChair: Johan Oppen, Molde University College, P O Box 2110,6402, Molde, Norway, [email protected]

1 - A Real-World VRP with Uncertain Delivery QuantitiesBirk Baumbach, Operations Research, Fraunhofer IVI,Zeunerstrasse 38, 01069, Dresden, Germany,[email protected], Axel SimrothAn LPG logistics service provider with a fleet of 200 tank trucks sup-plies some 150,000 customers. Daily up to 2,000 shipments at a totaldistance of 40,000 km are handled. The main challenge of this real-world VRP is the dynamics due to uncertainty in delivery quantities.We present a practical solution approach that is twofold: (1) Additionalinformation relevant to planning is provided by the use of telematics.(2) Uncertainty is considered by specialised algorithms that enable apermanent dynamic route planning in real-time. As a result the dis-tance travelled could be decreased by 9.8%.

2 - A two phase multi-objective algorithm for the bi-objective VRPJorge Oyola, Molde University College, 6411, Molde,Norway, [email protected], Arne LøkketangenThis work deals with a bi-objective vehicle routing problem. This in-cludes the balance of the tour lengths as an objective, in addition to theusual minimization of the total tour length. A two phase algorithm isproposed to approximate the set of solutions to the problem. In phase Ia tabu search heuristic is used to find the solutions that optimize each ofthe objectives independently. In the second phase, more solutions arefound either through an iterative process that combines path-relinkingand local search or a GRASP ruin-and-recreate algorithm. Computa-tional results will be presented

3 - Milk Collection in Western Norway Using Trucks andTrailersArild Hoff, Molde University College, P.O.Box 2110, 6402,Molde, Norway, [email protected]

Milk collection is a problem which is well known in rural areas allaround the world. This talk considers a real world problem for a Nor-wegian dairy company collecting raw milk from farmers. Most farmsare inaccessible for a large truck carrying a trailer. Thus the routes areorganized as a main tour between larger parking spots where the traileris left behind, and the truck drives subtours from this spots to visitthe actual farms. The talk will present heuristics for constructing suchtours and computational results comparing our result with the currentplan of the company.

4 - Intermodal flexible liner shippingJohan Oppen, Molde University College, P O Box 2110,6402, Molde, Norway, [email protected]

We present a maritime transportation problem dealing with the trans-port of frozen

fish from Russia and Norway to terminals in Norway, the Netherlandsand the UK. The problem can be viewed as a flexible liner shippingproblem, as both the first and the last port, as well as some terminalsalong the route, are fixed, while other terminals may also be servedby truck via another terminal to save time, or to relieve vessel capac-ity restrictions. We present a mathematical model and show that smallproblem instances can be solved to optimality using standard software.

� MC-08Monday, 12:30-14:00RB-Epsilon


Stream: Network OptimizationInvited sessionChair: M. Emre Keskin, Industrial Engineering, BogaziciUniversity, Bogazici Universitesi Endustri Muhendisligi, Hisarustu/ Besiktas, 34470, Istanbul, Turkey, [email protected]

1 - DiffServ Aware Routing Optimization in Telecommu-nication NetworksAmaro de Sousa, Universidade de Aveiro, Instituto deTelecomunicações, Campus Universitário de Santiago, VATNr. PT502854200, 3810-193, AVEIRO, Portugal,[email protected], Carlos Lopes, Dorabella Santos, Filipe Alvelos

Differentiated Services (DiffServ) aim to provide different QoS levelsto different service classes. We propose the implementation of band-width based DiffServ where traffic flows of higher QoS classes arerouted by less loaded network links. We define the routing problem asan Integer Programming Model, which let us obtain lower bounds, andpropose heuristics to compute feasible solutions. We present compu-tational results showing that the heuristics compute good routing solu-tions and that proper routing optimization can efficiently achieve ser-vice differentiation.

2 - Optical Fiber Network Design with Real Life Con-straints and Cost FunctionStanislas Francfort, CORE/M2V, Orange Labs, 38-40 rue dugeneral leclerc, issy les moulineaux, France,[email protected], Frederic Moulis, CédricHervet, Matthieu Chardy

The deployment of Fiber To The Home (FTTH) is crucial to telecom-munication operators for both economical and quality of service rea-sons. This presentation deals with a Passive Optical Network (PON)design problem focusing on real-life optical cabling constraints. Alarge set of expenses, such as labor expenses, is also considered intothe cost function we wish to minimize. This optimization problem isformulated as an integer linear program (ILP) and several solving ap-proaches are designed. Tests performed on real instances assess theefficiency of the proposed solution algorithms.

46


3 - The Package Server Location ProblemArnaud Malapert, Laboratoire I3S - UMR7271 - UNS CNRS,2000, route des Lucioles - Les Algorithmes - bât. Euclide B,BP 121, 06903, Sophia Antipolis, France,[email protected], Mohamed Rezgui, Jean-CharlesRégin, Jean Parpaillon, Yvan ManonWe introduce the package server location problem. A number of pack-age servers are to be located at nodes of a network. Demand for thesepackage servers is located at each node. The objective is to minimizethe number of package servers while maximizing the efficiency andthe reliability of the broadcast of packages to clients. These objec-tives are clearly contradictory: the broadcast becomes more efficientas the number of servers increases. The problem is formulated andanalyzed as a multi-objective optimization problem. Results for exactapproaches using MILP are reported.

4 - An integrated approach to lifetime maximization inwireless sensor networksM. Emre Keskin, Industrial Engineering, Bogazici University,Bogazici Universitesi Endustri Muhendisligi, Hisarustu /Besiktas, 34470, Istanbul, Turkey,[email protected], I. Kuban Altinel, Necati Aras,Cem ErsoyMathematical programming provide a very utilizable tool for opti-mizing some Wireless Sensor Network (WSN) performance metricsby concentrating on some of the decision instruments such as, sen-sor placement, sensor scheduling, data and mobile sink routing. Inthis work, we introduce two new mathematical programming models,which integrate all the decision instruments. We quantify the effect ofour integrated approach by comparing results of our models with theresults of some mathematical models from the literature.

� MC-09Monday, 12:30-14:00RB-Zeta


Stream: Cutting and PackingInvited sessionChair: Ramon Alvarez-Valdes, Statistics and Operations Research,University of Valencia, Faculty of Mathematics, Doctor Moliner50, 46100, Burjassot, Spain, [email protected]

1 - An analysis of the problem instances and the datasets used for computational experiments in the Iden-tical Item Packing ProblemElsa Silva, INESC TEC, Porto, Portugal,[email protected], José Fernando Oliveira, GerhardWäscherThe two-dimensional rectangular Identical Item Packing Problem(IIPP) consists in placing the maximum number of identical rectangu-lar items onto a single rectangular pallet. This problem has been widelystudied in the last forty five years, and different problem instances anddata sets had been used for the computational experiments of the dif-ferent approaches proposed to solve the problem. In this work, theseproblem instances and data sets are analyzed and the hardest and un-solved problem instances pointed out, highlighting the direction forfuture research in the IIPP.

2 - Optimizing Shelf Space in Retail StoresTeresa Bianchi-Aguiar, INESC TEC, Faculty of Engineering,University of Porto, Portugal, [email protected], MariaAntónia Carravilla, José Fernando OliveiraIn retail stores, the number of available products is constantly increas-ing, leading to a continuous need for space management. To do so,retailers rely on planograms, designs of fixtures (e.g. shelves) showingthe desired products allocation. Driven by the case of a Portuguese su-permarket chain, we will present a hierarchical approach to planogramgeneration, a Placement C&P Problem known in literature as ShelfSpace Allocation Problem. The proposed approach splits the probleminto two main levels: allocation of space to each product and placementof product items into each allocated space.

3 - A local optimization switch algorithm for packingnon-identical circlesGeorgiy Yaskov, Department of Mathematical Modeling andOptimal Design, Institute for Mechanical EngineeringProblems. National Academy of Sciences of Ukraine, 2/10Pozharsky St., 61046, Kharkov, Ukraine, [email protected],Yuri Stoyan

We consider a problem of packing non-identical circles into a rectan-gle of minimal length. The problem is reduced to a problem of packingcircles of variable radii. A mathematical model of the problem is con-structed and its characteristics are studied. Based on the characteristicswe offer a solution strategy of the problem including a number of sub-problems. For searching for local minima and solving the subproblemsa gradient method of the second order is used. A novel way of switch-ing from a local minimum to another one is applied. We compare 146numerical benchmark examples.

4 - Determining the best shipper sizes for sending prod-ucts to customersRamon Alvarez-Valdes, Statistics and Operations Research,University of Valencia, Faculty of Mathematics, DoctorMoliner 50, 46100, Burjassot, Spain, [email protected],Maria Teresa Alonso Martínez, Francisco Parreno, JoseTamarit

A distribution company in Spain has to send products, packed intoshipper boxes, from the store to the retail shops. The problem is todecide the sizes of the shipper boxes to keep at the store so as to mini-mize the cost of packing all the forecasted demands along the planninghorizon. In this work we describe several integer linear programmingformulations for the problem and a metaheuristic algorithm based onreducing and increasing the sizes of shipper boxes. A computationalstudy conducted on real instances provided by the company is pre-sented and discussed.

� MC-10Monday, 12:30-14:00RB-Theta

Non-deterministic modelsStream: Production and the Link with Supply ChainsInvited sessionChair: Farouk Yalaoui, Institut Charles Delaunay, ICD LOSI,University of Technology of Troyes, 12, Rue Marie Curie BP2060, 10000, Troyes, France, [email protected]: De Souza Mauricio, Departamento de Engenharia deProdução, Universidade Federal de Minas Gerais, Av. PresidenteAntônio Carlos, 6627 , Belo Horizonte, Brazil,[email protected]: Y. Ilker Topcu, Industrial Engineering, Istanbul TechnicalUniversity, Istanbul Teknik Universitesi, Isletme Fakultesi, Macka,34357, Istanbul, Turkey, [email protected]

1 - Manufacturing flexibility in an oligopoly competitionwith demand uncertaintyLiu Yang, Business School, University of InternationalBusiness and Economics, Business School, University ofInternational Business and Economics, No. 10, Hui Xin DongJie, Chao Yang District, Beijing, 100029 China, 86-010,Beijing, China, [email protected], Chi To Ng

This paper investigates manufacturing flexibility in an oligopoly com-petition with demand uncertainty. We consider a general oligopolymodel consisting of n firms who adopt either flexible capacity strategyor in-flexible capacity strategy to compete in a same market. We char-acterize the equilibrium of the competition. We find that the equilib-rium is classified into four situations, when FCS is dominant; IFCS isdominant; two strategies coexist; and none is profitable, respectively.We also conduct sensitive analyses of the key influential factors andprovide some managerial insight.

47


2 - Designing a robust supply chain for competitionagainst existing competitors in retailer levelGhasem Rajabloo, Industrial Engineering, Urmia Universityof Technology, Band Ave, Urmia, Iran, Urmia, WestAzarbayjan, Iran, Islamic Republic Of,[email protected], Shabnam Rezapoor, MaghsudSolimanpur

We present a model for robust design of the network of an entrant sup-ply chain in markets with price dependent demands and a pre-existingcompeting chain. The model assumes dynamic competition betweennew and existing supply chains, probabilistic customers’ behavior andprobability of disruption in facilities and transportation links. We con-sider extra capacity, strategic stock and substitutable facility strategiesto mitigate the effects of possible disruptions and the impact of facil-ity location decisions. We illustrate and discuss the obtained resultsthrough several numerical examples.

3 - A decision analysis on cooperative advertisingSebnem Burnaz, Management Engineering, Istanbul TechnicalUniversity, ITU Isletme Fakultesi, Macka, 34357, Istanbul,Turkey, [email protected], Esin Ekmekci, Y. Ilker Topcu

Today competition takes place among supply chains. Therefore, it iscritical for the companies to sustain collaborative relationships withtheir stakeholders. At this point, cooperative practices gain more im-portance to build competitive advantage. Manufacturers use cooper-ative (co-op) advertising to strengthen the image of their brands andto motivate consumers in their buying decisions. This study aims toanalyze the changes in the portions of manufacturer and distributor insharing co-op advertising in sectors represented by different levels ofsubstitution rate and competition intensity.

� MC-11Monday, 12:30-14:00RB-Iota

Transport Planning

Stream: Transportation and LogisticsInvited sessionChair: Hani Mahmassani, Transportation Center, NorthwesternUniversity, 600 Foster St, 60208, Evanston, Illinois, United States,[email protected]

1 - New Algorithm for the Stochastic Generalized Trans-portation ProblemMarcin Anholcer, Department of Operations Research,Poznan University of Economics, Al. Niepodleglosci 10,61-875, Poznan, Poland, [email protected]

Generalized Transportation Problem (GTP) is a variant of Transporta-tion Problem where the amounts (or values) of good change during thetransportation process. We consider the stochastic version of the prob-lem, where the values of demand of the destination points are randomvariables. The goal is to minimize the expected value of the total cost.We present new exact method for the GTP.

2 - The Dynamic Driver-Routing ProblemNiels Agatz, Decision and Information Science, RSMUniversity, Burg. Oudlaan 50, 3000DR, Rotterdam,Netherlands, [email protected], Jordan Srour, Johan Oppen

The dynamic dial-a-driver service is an innovative concept in the realmof passenger transportation in which the customer requests a driver totransfer him/her from one location to another in the customer’s owncar. A particularly dynamic application of this service focuses on peo-ple who must get themselves and their car home safely after drinkingtoo much. In this study, we consider the problem of assigning drivers tocustomer requests and company chase cars to driver requests with theobjective to minimize the total distance traveled by the chase vehicles.

3 - Development of an optimisation framework forscheduling of street worksRahman Pilvar, Institute for Transport Studies, University ofLeeds, LS2 9JT, LEEDS, WEST YORKSHIRE, UnitedKingdom, [email protected]

Urban transportation networks are subject to various capacity-degrading events due to road maintenance and utility works activities.All so often a conflict of interest prevails among the stakeholders re-garding the underlying street works details such as the timing, suit-able level of capacity occupation, etc. The present article provides aNetwork Design Problem framework for optimum scheduling of streetworks in a transportation network. A multi-objective bi-level program-ming formulation is presented to minimise the total costs associatedwith delays to road users and utility companies while ensuring that theaccessibility level of the network would be minimally affected.

4 - Optimal Bus Route Frequency Allocation in a Large-Scale Urban SystemHani Mahmassani, Transportation Center, NorthwesternUniversity, 600 Foster St, 60208, Evanston, Illinois, UnitedStates, [email protected], Ismail Omer Verbas

This study develops new formulations for the bus frequency allocationproblem. Existing bus frequency allocation in the literature maximizesridership and waiting time savings subject to budget, fleet, policy head-way and maximum load constraints by solving for route frequencies.The formulations consider routes with multiple service patterns, andinclude cost minimization with minimum ridership constraint, bus sizeselection as an additional decision variable, and maximum load as apenalty function. The approach is demonstrated in an actual large-scale system.

� MC-12Monday, 12:30-14:00RB-Omicron

Application of fuzzy logic and softmethodologies in decision-making andfinancial modelling

Stream: Fuzzy Sets and Soft ComputingInvited sessionChair: Robert Fullér, IAMSR, Åbo Akademi University,Joukahainengatan 3-5, 20520, Åbo, Finland, [email protected]: Mikael Collan, Department of Business Economics andLaw, Lappeenranta University of Technology, Raivaajankatu 5 as.3, 24100, Salo, Finland, [email protected]

1 - Fuzzy Models and StatisticsReinhard Viertl, Statistics and Probability Theory, ViennaUniversity of Technology, Wiedner Hauptstr. 8, 1040, Wien,Austria, [email protected]

In statistics there appear different kinds of uncertainty. One which isfrequently neglected is fuzziness of data. In order to obtain realistic re-sults from statistical inference it is necessary to describe the fuzzinessof data quantitatively. This is possible using fuzzy models. Gener-alized statistical methods are available and related software is underdevelopment. Another kind of fuzziness is the uncertainty of a-prioridistributions in Bayesian inference. This uncertainty can be describedby so-called fuzzy probability distributions. Related generalized pro-cedures will be given.

2 - An Integration of TOPSIS Method and Interval Type IIFuzzy Sets to Evaluate Municipality Investment Alter-natives: The Case of Istanbul MunicipalityNazmi Sener, Industrial Engineering, University of Yalova,Muhendislik Fakultesi Sehit Ömer Faydali Cad., YalovaAtaturk Stadyumu karsisi Merkez, 77100, Yalova, Turkey,[email protected], Hasan Huseyin Turan

48


Due to the limited capital sources, municipalities have to be carefulabout defining the investment priorities. Thus, in this study the ex-tension of TOPSIS method and interval type II fuzzy sets is suggestedto guide municipalities. A survey conducted on strategic planning ex-perts is used as initial information for the proposed model. As a casestudy suggested method is tested to rank several alternatives in Istan-bul. The results reveal insights for feasibility and effectiveness of pro-posed methodology. Finally, sensitivity analysis is performed to testthe stability of the rankings.

3 - Quantifying expected benefits from investmentprojects with the fuzzy pay-off methodMikael Collan, Department of Business Economics and Law,Lappeenranta University of Technology, Raivaajankatu 5 as.3, 24100, Salo, Finland, [email protected]

The pay-off method is a simple new method for framing the invest-ment profitability analysis problem and for enhancing the support forinvestment decision making. The method is based on directly derivinga fuzzy number from a set of cash-flow scenarios that is then used as abasis for derivation of descriptive numbers. These descriptive numbersinclude, e.g., the real option value. The method is a complement to theprofitability analysis toolpack of managers and is very robust and canbe used for all types of investments, it is "excel compatible" and fullytransparent all the way.

4 - Numerical Patent Analysis with a Fuzzy CompoundReal Option ModelXiaolu Wang, Åbo Akademi University, Finland,[email protected]

Patent valuation has always been a difficult task due to the great un-certainties involved in patent exploitation process, which provides uswith the incentive to use fuzzy estimator for the volatility of the ex-pected future monopoly profits. It is also recognizable that investing inpatent application buys the investor an right, but not an obligation, tocommercialize the innovation, we therefore present how the proposedfuzzy compound real option model can be used to generate decision-support for deciding whether to go ahead with a patent application ornot.

� MC-13Monday, 12:30-14:00RB-Tau


Stream: Lot-sizing and Related TopicsInvited sessionChair: Franklina Toledo, Applied Mathematics and Statistic,ICMC - USP, Av. Trabalhador Saocarlense, 400, CP 668,13.560-970, Sao Carlos, Sao Paulo, Brazil, [email protected]

1 - An approach of implementing the Kalman filter in thecalculation of lot sizesKatharina Amann, Chair of Production and Logistics,Georg-August-University Goettingen, Platz der GöttingenSieben 3, 37073, Göttingen, Lower Saxony, Germany,[email protected], Jutta Geldermann

Nowadays, Production Planning is an important assignment in indus-trial companies. In fact, many companies use simple approaches withseveral restrictions to plan their production. The presentation intro-duces an advanced approach in Production Planning: It includes theKalman filter as an estimator of the customers demand in the calcula-tion of lot sizes. A forecasting competition between the Kalman filterand well-know forecasting methods is performed. Based on these re-sults, the possible impact of forecasting methods on lot sizes and stocklevels as well as service levels is analyzed.

2 - Hybrid relax-and-fix heuristics for the lot-sizing andscheduling problem in the industry of consumergoods for personal hygieneDebora Ronconi, Production Engineering, University of SaoPaulo, Av. Prof. Almeida Prado,128, 05508070, Sao Paulo,Sao Paulo, Brazil, [email protected], Márcio Kawamura

This work addresses the integrated lot sizing and scheduling problemthat occurs in a very competitive market in which the good servicelevel show up in the competition for the clients. The operational en-vironment is composed of unrelated parallel machines with sequence-dependent setup times and costs. Backordering is allowed but it is ex-tremely undesirable. We propose several resolution methods based onthe relax-and-fix heuristics where hybrid partitioning strategies are an-alyzed. The computational experiments were conducted on randomlygenerated instances and real problems.

3 - A hybrid metaheuristic for production planning in thespinning industryVictor Camargo, Institute of Mathematics and ComputerSciences, University of São Paulo, Av. do Trabalhadorsão-carlense, 400, 13566-590, São Carlos, SP, Brazil,[email protected], Franklina Toledo, BernardoAlmada-Lobo

We investigate a two-stage lot-sizing and scheduling problem aris-ing at a spinning industry. A new hybrid metaheuristic called HOPS(Hamming-Oriented Partition Search) are proposed to solve the prob-lem. It is a branch-and-bound procedure that incorporates a fix-and-optimize improvement method. An innovative partition choice for thefix-and-optimize is developed. The computational tests with instancesgenerated based on real data attest the HOPS to be a good alterna-tive to solve MIPs with recognized partitions such as the lot-sizing andscheduling case.

� MC-14Monday, 12:30-14:00RB-Omega

Resource allocation and scheduling inpractice: further complexity

Stream: Scheduling under Resource ConstraintsInvited sessionChair: Sacramento Quintanilla, Economia Financiera ymatematica, University of Valencia, Avda. Naranjos s/n, 46071,Valencia, Spain, [email protected]

1 - Balancing the use and distribution of warehouse re-sources: a scheduling pre-processing procedureM. Angeles Pérez, Mathematics for Ecomomy, University ofValencia, Avda Naranjos s/n, 46021, Valencia, Spain,[email protected], Francisco Ballestin, M.Pilar Lino,Sacramento Quintanilla, Vicente Valls

Scheduling warehouse storage/retrieval orders is a complex processthat assigns to each order a location, a forklift and a start time, tak-ing into account pallet specifications, traffic restrictions and forkliftcapabilities. Locations are previously selected maximizing total freestorage space but, their spatial distribution and forklift requirementsinfluence the total time required to complete the orders. We present amulti-objective heuristic algorithm that re-assigns locations balancingspatial distribution and forklift workloads. Computational results arealso presented.

2 - Simultaneous allocation of medical resources andscheduling of visits in a Hospital at Home serviceSacramento Quintanilla, Economia Financiera y matematica,University of Valencia, Avda. Naranjos s/n, 46071, Valencia,Spain, [email protected], Francisco Ballestin, M.PilarLino, M. Angeles Pérez, Vicente Valls

In a Hospital at Home Service, patients are at home and have to be vis-ited regularly by a team comprising a doctor and a nurse. The Servicehas a pool of doctors and nurses with different specializations. Theteam that visits a patient has to have the expertise required to treat thepatient. The Service daily determines the teams and vehicles that willvisit patients and also establishes the sequence of visits for each teamwith the objective of cost minimization. We present a solution proce-dure for this resource allocation and task scheduling problem and somecomputational results.

49


3 - Approximate algorithms for simultaneous projectscheduling and resource allocation with controllableprocessing timesJose M. Framinan, Industrial Management, School ofEngineering, University of Seville, Avda. de losDescubrimientos s/n, Seville, Spain, [email protected], VictorFernandez-Viagas Escudero

This paper presents several heuristics for an integrated ProjectScheduling and Staff Assignment (PSSA) problem where tasks haveprecedence relations and release times. The processing time of eachtask depends on the amount of resources assigned (over- and under-coverage is allowed). Furthermore, the employee can perform certaintask only feasible if he/she possesses the appropriate skill. The ob-jective is to minimize the makespan of the project. A GRASP and anIterated Greedy algorithm have been designed and successfully testedon a test-bed based on the well-known library PSPLIB.

4 - Project Resource Leveling and Robustness Opti-mization using ANPElena Rokou, Industrial Management & OR, MechanicalEngineering, National Technical University of Athens,Heroon Polytechneiou 9 9, Athens, Greece, Greece,[email protected], Konstantinos Kirytopoulos

The paper presents a DSS for multi-objective project scheduling un-der precedence and resource constraints. It handles both preemptiveand not scheduling, multiple performing modes and partial or full al-location of resources to activities. Project performance criteria aremakespan minimization, resource leveling and robustness. Firstly thedecision maker sets the weighting factors for each component usingANP and a linear value function is formed. Secondary, backward andforward SGS along with priority rules generate initial feasible sched-ules that are further exploited by an adaptive GA.

� MC-15Monday, 12:30-14:00RB-2101


Stream: Nonlinear ProgrammingInvited sessionChair: Edite M.G.P. Fernandes, Algoritmi R&D Centre, Universityof Minho, School of Engineering, Campus de Gualtar, 4710-057,Braga, Portugal, [email protected]

1 - A multi-layers method to improve optimization algo-rithms. Application to the design of bioreactors forwater treatment.Benjamin Ivorra, Matematica Aplicada, UniversidadComplutense de Madrid, Plaza de ciencias, 3, 28040, Madrid,Spain, Spain, [email protected], Juan M. Bello Rivas,Jérôme Harmand, Angel Manuel Ramos, Alain Rapaport

We envision global optimization as finding a suitable initial guess of aconsidered optimization algorithm. One can imagine that this possibil-ity improves the existing optimization methods. To do so, we present amethodology that consists in the resolution of a sub-optimization prob-lem by using a multi-layers line search algorithm. We check the effi-ciency of this approach by considering benchmark and industrial prob-lems. In particular, we focus on the optimization of the shape and hy-drodynamic properties of a bioreactor in which a wastewater substrateis degraded by a microbial ecosystem.

2 - Comparing two MINLP models for the HydroScheduling Problem: The case of a head-dependentcascaded reservoir system in PortugalJavier Diaz, Sistemas e Informatica, Universidad Nacional deColombia, Cra 80 65-223, Facultad de Minas Bloque m8Aoficina 212, 1, Medellin, Antioquia, Colombia,[email protected], Luis Moreno

Two mixed-integer non-linear programming (MINLP) models for theoptimal short-term hydro-scheduling (STHS) are compared. These twomodels include the following features: head-dependency, start/stop ofunits, discontinuous operating regions, and discharge ramping con-straints. Furthermore, they consider hydroelectric power generationas a nonlinear function of both the water discharge and the net head. Inorder to compare assumptions and results of both models, we adjustedthe first one in order to apply it in the case of the second, the maincascaded hydro systems in Portugal.

3 - Continuous optimization using a biologically-motivated multilevel systemJongChen Chen, Information Management, National YunLinUniversity of Science and Technology, 123 University Road,Section 3, Douliu, Taiwan, Taiwan, [email protected]

Our group had previously developed a multilevel system operatingacross a wide spectrum of organizational levels. The interactionsamong different levels of information processing expand its problemsolving capability. To address the continuous optimization problems,we used the system to control the motions of a walking robot. Theexperimental results showed that the robot was capable of learning ina continued manner in a physical environment. We then made somechange in the robot’s structure, and the results showed the system pos-sessed continuous optimization capability.

4 - Nonlinear Optimal Control of Spacecraft Attitude Us-ing Internal Momentum RotorsM. Navabi, Shahid Beheshti University, GC, Iran, IslamicRepublic Of, [email protected], Mina Tavana

In this paper, attitude of a satellite is controlled by internal momen-tum rotors utilizing a nonlinear optimal control method. This methodis based on State Dependent Riccati Equation. It is difficult to solvea nonlinear system optimally with state and control constraints. Thecapabilities of State Dependent Riccati Equation technique are illus-trated in producing control design for spacecraft attitude control withquadratic constraint on quaternion. The simulation results show thatthe mentioned method have desired accuracy for spacecraft attitudenonlinear control problems.

� MC-16Monday, 12:30-14:00RB-2103

Numerical methods of nonsmoothoptimization

Stream: Nonsmooth OptimizationInvited sessionChair: Albert Ferrer, Dpt. of Applied Mathematics I,Technological University of Catalonia, Av. Doctor Marañon,44-50, 08028, Barcelona, Catalunya, Spain,[email protected]

1 - Hyperbolic smoothing function method for minimaxproblemsAdil Bagirov, School of Science, Information Technology &Engineering, University of Ballarat, University Drive, MountHelen, P.O. Box 663, 3353, Ballarat, Victoria, Australia,[email protected]

In this talk, we discuss an approach for solving finite minimax prob-lems. This approach is based on the use of hyperbolic smoothing func-tions. In order to apply them we reformulate the objective function inthe minimax problem. Based on these results an algorithm for solvingthe finite minimax problem is proposed. We present results of numer-ical experiments and compare the proposed algorithm with the algo-rithm that uses the exponential smoothing function as well as with thealgorithm based on nonlinear programming reformulation of the finiteminimax problem.

50


2 - Nonsmooth nonconvex optimization via smooth op-timizationGurkan Ozturk, Industrial Engineering, Anadolu University,AU-MMF-Industrial Engineering, Iki Eylul Campus, 26480,Eskisehir, Turkey, [email protected], Adil Bagirov

In this paper, an algorithm for minimizing locally Lipschitz functionsis developed. Descent directions in this algorithm are computed bysolving a system of linear inequalities. In order to solve this system wereduce it to the minimization of the convex piecewise linear functionover the unit ball. Then the hyperbolic smoothing function is appliedto approximate this problem by the sequence of smooth problems.

3 - A Truncated Codifferential Method: Using hypogra-dients from current and previous iterationsAli Hakan Tor, Department of Mathematics, Middle EastTechnical University, Inonu Bulvari, Cankaya, 06531, Ankara,Turkey, [email protected], Adil Bagirov

The propesed method uses the concept of codifferential. To constructthe whole set of codifferential is too complicated, so a few elements ofthis set are used to find the descent directions. Moreover, the methodemploys not only a few elements from the codifferential at current iter-ation but also some codifferentials from previous iterations for compu-tation of descent directions. The convergence of the proposed methodis proved. By using performance profiles, results of numerical exper-iments addressing a set of convex and nonconvex test problems arereported by comparing some other methods.

4 - Iterative local search algorithm for solving non-smooth flow shop problemsAlbert Ferrer, Dpt. of Applied Mathematics I, TechnologicalUniversity of Catalonia, Av. Doctor Marañon, 44-50, 08028,Barcelona, Catalunya, Spain, [email protected], Angel,A. Juan, Helena Ramalhinho Lourenço

This work discusses the use of an Iterative Local Search (ILS) algo-rithm for solving flow-shop problems with non-smooth objective func-tions. The ILS framework has been adapted to perform a parameter-free local-global search process. The parameter-free property distin-guishes our approach from other similar approaches, which perfor-mance strongly depends upon several parameters. A set of experimentshave been carried out. The results obtained so far suggest that our ap-proach can be successfully applied in a variety of non-smooth schedul-ing problems.

� MC-17Monday, 12:30-14:00RB-2105

Reverse Logistics Network Design

Stream: Applications of Location AnalysisInvited sessionChair: Sibel A. Alumur, Industrial Engineering Department,TOBB University of Economics and Technology, Sogutozu cad.No:43, Sogutozu, 06560, Ankara, Turkey, [email protected]

1 - Sustainable Reverse Logistics Network Design forMulti-modal Transport of Household Plastic WasteJacqueline Bloemhof, Operations Research and Logistics,Wageningen University, Hollandseweg 1, 6700 EW,Wageningen, Netherlands, [email protected],Xiaoyun Bing, Jack van der Vorst

Household plastic waste can be collected, separated and treated to beused as recycled plastics again. This is both an economically and en-vironmentally attractive option, although transportation of the plasticwaste also causes extra environmental and economic costs. This pa-per studies the reverse logistics network of plastics in the Netherlandsand adopts multi-modality transportation to deal with future increaseof plastic waste. We use an MILP model-based scenario study takinginto account transportation, processing and emission costs to developa sustainable logistics network design.

2 - Integrating forward and reverse flows: a genericmodel for remanufacturing activityNeslihan Demirel, Industrial Engineering, Gazi University,Turkey, [email protected], Hadi Gökçen

The importance of the recovery of used products has been widely rec-ognized in the literature and practice. This is mainly due to the envi-ronmental deterioration and the limited capacities of disposal areas. Toplan a closed loop supply chain strategically and to coordinate produc-tion, remanufacturing, disposal and outsourcing operations efficiently,we proposed a multi-level, multi-product, multi-period mixed integerlinear programming (MILP) model. We illustrated the model on a nu-merical example and carried out sensitivity analyses.

3 - Multi-Period Reverse Logistics Network DesignSibel A. Alumur, Industrial Engineering Department, TOBBUniversity of Economics and Technology, Sogutozu cad.No:43, Sogutozu, 06560, Ankara, Turkey,[email protected], Stefan Nickel, FranciscoSaldanha-da-Gama, Vedat Verter

In this paper, we propose a profit maximization modeling frameworkfor reverse logistics network design problems. We present a mixed-integer linear programming formulation that is flexible to incorporatemost of the reverse network structures plausible in practice. The pro-posed general framework is justified by a case study in the context ofreverse logistics network design for washing machines and tumble dry-ers in Germany. We conduct extensive parametric and scenario anal-ysis to illustrate the potential benefits of using a dynamic model, andalso to derive a number of managerial insights.

� MC-18Monday, 12:30-14:00RB-2107

Financial Modelling

Stream: Financial Modelling & Risk Management (con-tributed)Contributed sessionChair: Tatjana Slijepcevic-Manger, Faculty of Civil Engineering,University of Zagreb, Fra Andrije Kacica-Miosica 26, 10000,Zagreb, Croatia, [email protected]

1 - Moment-matching method with monomial approachAlessandro Staino, Business Science Department, Universityof Calabria, Via Pietro Bucci Cubo 3c, 87036, Arcavacata diRende (CS), Italy, [email protected], AndreaConsiglio

Moment-Matching methods are often used to find an approximated so-lution to stochastic programming problems where the uncertainty ismodeled through continuous distributions. In this paper we present amoment-matching method coming from the monomial method whichis a technique to solve systems of algebraic equations iteratively.Specifically, we show how to generate distributions of financial returnsso that some statistics are matched. A very valuable feature of ourmethod is its capability of generating arbitrage-free financial markets.

2 - Stratified Sampling Implementations in FinancialSimulationIsmail Basoglu, Industrial Engineering, Science Institute,Industrial Eng. Dept. Engineering Faculty, BogaziciUniversity South Campus Bebek, 34342, Istanbul, Turkey,[email protected], Wolfgang Hörmann

Monte Carlo methods are applicable to most of the rare event financialproblems. Yet, the yielding variance is large for this type of stochasticevents. One can use stratified sampling, which is based on the de-composition of the sampling domain and conditional sampling withineach stratum and simulation results are combined throughout the strata.Our research is mainly focused on optimal construction of strata struc-tures and implementing stratified sampling in an automatic and effi-cient manner for different type of financial problems.

51


3 - Modeling Hedge Fund Performance Using NeuralNetwork ModelsJoseph Paradi, Chemical Engineering and Applied Chemistry,Univeresity of Toronto, 200 College Street, M5S3E5, Toronto,Ontario, Canada, [email protected], MarinosTryphonas

New neural network model evaluates monthly mutual fund using back-wards elimination regression for explanatory variable selection. Thecase study on 293 N/A hedge funds between 1998 and 2010. 25 riskfactors including equity, fixed income, foreign exchange, and com-modity factors were employed as explanatory variables. Two mainconclusions: 1) the new neural network model delivered better out-of-sample performance compared to linear regression models; 2) risk fac-tor selection based on backwards elimination regression consistentlyimproved the model performance across the whole time period.

4 - Ruin probability functions with N intersectionsTatjana Slijepcevic-Manger, Faculty of Civil Engineering,University of Zagreb, Fra Andrije Kacica-Miosica 26, 10000,Zagreb, Croatia, [email protected]

In this paper we show that for any positive integer N we can constructtwo spectrally negative Levy processes such that their ruin probabil-ity functions have at least N positive intersections. This means that insome cases the optimal ruin strategy can be very unstable.

� MC-19Monday, 12:30-14:00RB-2111

Quantitative approaches to bankperformance and asset evaluation

Stream: Operational Research and Quantitative Mod-els in BankingInvited sessionChair: Enrique Ballestero, Escuela Politecnica Superior de Alcoy,Technical University of Valencia, 03801, Alcoy (Alicante), Spain,[email protected]: Mila Bravo, ALCOY SCHOOL, TECHNICALUNIVERSITY OF VALENCIA, 03801, ALCOY, Spain,[email protected]

1 - Ranking funds offered by the Spanish CaixaBank: Acompromise approach to performanceEnrique Ballestero, Escuela Politecnica Superior de Alcoy,Technical University of Valencia, 03801, Alcoy (Alicante),Spain, [email protected]

The CaixaBank offers funds to customers. To evaluate their perfor-mance a linear-quadratic compromise programming model is devel-oped. This measure can be extended to many investment criteria whilethe traditional ratio-based and leverage-based measures are essentiallylimited to profitability and risk. It also avoids negativity in the ratiovalues. Sophisticate compromise programming proves superior to thesimple use of ratios. Customers’ profiles of The CaixaBank are de-fined by weighting systems. Each one leads to coherent results whichare compared to the results from domination analysis.

2 - Evaluating artistic assets: A multicriteria methodPaloma Pareja, Independent consultant, Spain,[email protected], Sonia Zendehzaban

To assess painting assets of bank collections in terms of quality, a mul-ticriteria model is developed relying on domination analysis and mod-erate pessimism decisions under uncertainty. As preferences for theartistic criteria change from a decision maker to another, they are con-sidered unknown. The criteria are features such as theme originality,drawing and color. As a result, the assets are ranked by quality in-dexes, which is a previous step to determine quality to price ratios.Data from collections as well as the computational process and resultsare tabulated and analysed.

3 - Multicriteria ranking of Spanish banks from the 2011European Banking Authority wide stress testMila Bravo, ALCOY SCHOOL, TECHNICALUNIVERSITY OF VALENCIA, 03801, ALCOY, Spain,[email protected], Antonio BenitoThis paper aims at ranking 25 Spanish banks from external informa-tion published by the Bank of Spain, which provides financial criteriafrom the European Banking Authority (EBA)-wide stress test in 2011.As a previous step, each bank is characterized from the criteria as non-dominated or dominated by convex combinations of the other banks.Weighted goal programming (which requires assuming preferences forthe criteria) and the moderate pessimism decision making model un-der uncertainty (which does not require assumptions on preferences)are used to rank the non-dominated banks.

� MC-20Monday, 12:30-14:00RB-2113

Risk Analysis and Credit Scoring

Stream: Data Mining in the Financial SectorInvited sessionChair: Cristian Bravo, Department of Industrial Engineering,University of Chile, Republica 701, Santiago, 8370439, Santiago,Chile, [email protected]

1 - Improving Credit Scoring by Differentiating DefaulterBehaviorCristian Bravo, Department of Industrial Engineering,University of Chile, Republica 701, Santiago, 8370439,Santiago, Chile, [email protected], Lyn Thomas,Richard WeberIn this work, we formalize the common notion that there are two typesof defaulter, those who do not pay because of cash flow problems,and those that do not pay because of lack of willingness to pay. Weproposes to differentiate them using a game theory model to describetheir behavior, and from this model construct a set of constraints fora semi-supervised constrained clustering algorithm that creates a newtarget variable. We benchmark several multiclass supervised modelsto test the approach, finding significant improvement in classificationaccuracy, and strong insights about defaulters.

2 - Intensity models and transition probabilities forcredit card loan delinquenciesJonathan Crook, University of Edinburgh Business School,University of Edinburgh, Credit Research Centre, 50 GeorgeSquare, EH8 9AL, Edinburgh, Lothian, United Kingdom,[email protected] estimate the probability of delinquency and default for a sampleof credit card loans using intensity models, via semi-parametric mul-tiplicative hazard models with time-varying covariates. These modelsallow for insights into the factors that affect movements towards (andrecovery from) delinquency, and into default (or not). Results indi-cate that different types of debtors behave differently while in differentstates. The probabilities estimated for each type of transition are thenused to make out-of-sample predictions over a specified period of time.

3 - Comparative Analysis of Dynamic Models Specifica-tion for Behavioral Scoring in a Microfinance Portfo-lioJose Pizarro, Department of Industrial Engineering,University of Chile, Santiago, Chile, [email protected],Cristian BravoTraditional approach in the analysis and estimation of the probabilityof default is based on static frameworks. It has been shown that thesemodels can be improved using panel data, in which the model is builtconsidering the observed dynamic of the included variables. We studyif the traditional dynamic logistic model can be improved if we modifythe specification using a neural network model, and test if the dynamicaspect of the model improves classification capacity. We conclude ana-lyzing if these changes improve the prediction capability of the model.

52


4 - Dynamic affordability assessmentKatarzyna Bijak, School of Management, University ofSouthampton, Highfield, SO17 1BJ, Southampton, UnitedKingdom, [email protected], Lyn ThomasIn the credit decision making process, both applicant’s creditwor-thiness and affordability should be assessed. Affordability is oftenchecked on the basis of declared current income, assumed standardexpenditure and existing debts. In contrary, a dynamic affordabilityassessment is proposed by us. The affordability at a given moment intime is defined as a total amount of money which the applicant can useto service their debts. It depends on the applicant’s income and expen-diture which both vary in time. The suggested model can contribute toresponsible lending.

� MC-21Monday, 12:30-14:00RB-2115

Financial Mathematics and OR 3Stream: Financial Mathematics and ORInvited sessionChair: Tansel Avkar, Department of Mathematics, Middle EastTechnical University, 06531, Ankara, Turkey,[email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - A Rank-based Approach to Cross-Sectional AnalysisLudovic Cales, HEC - Department of Finance, University ofLausanne, Dorigny, 1015, Lausanne, Switzerland,[email protected], Monica Billio, Dominique GueganThis paper studies the cross-sectional effects present in the market us-ing a new graph theoretic framework. We model the evolution of adynamic portfolio, i.e., a portfolio whose weights change over time,as a function of cross-sectional factors where the predictive ability ofeach factor is described by a variable. This modeling permits us tomeasure the marginal and joint effects of different cross-section fac-tors on a given dynamic portfolio. Associated to a regime switchingmodel, we are able to identify phases during which the cross-sectionaleffects are present in the market.

2 - Bayesian approach for stochastic models with condi-tional heteroscedasticity in the analysis of the returnvolatility of an agricultural commoditySandra Cristina Oliveira, Business Administration, UNESP -Univ. Estadual Paulista, Av. Domingos da Costa Lopes, 780,17602-496, Tupã, São Paulo, Brazil, [email protected] volatility of returns generated by commodities has implicationsin formulation of policies for the country’s economic performance.Given the relevance of ARCH family models in the solution of eco-nomical and financial problems, this work compares the Bayesian es-timates for the parameters of AR-ARCH processes with normal andStudent’s t distributions for returns of coffee beans prices. An infor-mative prior distribution was adopted and posterior summaries wereobtained by Monte Carlo Markov Chain simulation. Results showedthat AR-ARCH process with t distribution adjusted better to the data

3 - Solving Large Two-Stage Stochastic OptimizationProblems with Portfolio SafeguardStan Uryasev, Director of Risk Management and FinancialEngineering Lab, University of Florida, FL 32611-6595,Gainesville, 303 Weil Hall, Florida, United States,[email protected] the solving two-stage stochastic optimization problems iswell developed, there are few practical examples of applications.Mostly, example calculations were done by researchers in stochasticoptimization area to exemplify new efficient mathematical program-ming approaches. Few engineers have such extensive expertise. Thesecomplications lead to the development of special meta-commands inPortfolio Safeguard (PSG) software for efficient coding. To be lessconservative than maximum loss, but more conservative than average,you can use the CVaR operator.

4 - Corporate Control with Cross-OwnershipMarc Levy, CEB, ULB, 50, av. F.D. Roosevelt, 1050,Brussels, Belgium, [email protected], Ariane Szafarz

In cross-ownership structures, existing methods are unable to evalu-ate control stakes. This is because cross-ownership induces circularity,and makes it impossible to identify winning coalitions among share-holders. To fill this gap, this paper proposes a novel theoretical ap-proach. This approach generalizes the Banzhaf index from game the-ory and combines it with Markov voting chains. From a governancestandpoint, we show that cross-ownership may favor shareholder ex-propriation. To illustrate this point, we revisit the case of the GermanAllianz Group. This paper also puts forward an algorithm.

� MC-23Monday, 12:30-14:00RB-Delta

Generalized Convexity and Optimization


1 - On the pseudoconvexity of a wide class of general-ized fractional functionsLaura Carosi, Statistics and Applied Mathematics, Universityof Pisa, Via Ridolfi, 10, 56124, Pisa, Italy, [email protected],Laura Martein, Alberto Cambini

Even though pseudoconvexity plays an important role in optimization,very few classes of pseudoconvex functions are known. We consider awide class of generalized fractional functions: the sum between a lin-ear one and a ratio which has an affine function as numerator and thep-th power of a positive affine one as denominator; p can be any pos-itive number different from 1. Provided nonnegativity conditions andthese functions to be not pseudoconvex, we study pseudoconvexity onthe nonnegative orthant. We derive necessary and sufficient conditionsexpressed in terms of the initial data.

2 - Continuous-time nonlinear programming under gen-eralized type I invexityAndreea Madalina Stancu, Institute of Mathematical Statisticsand Applied Mathematics, The Romanian Academy, Calea 13Septembrie, nr 13, RO-050711, Bucharest, Romania,[email protected], Ioan Stancu-Minasian

We consider a continuous-time programming problem (P) with nonlin-ear operator equality and inequality constraints. This problem belongsto the area of generalized convex optimization. In fact, we establishsufficiency optimality criteria under generalized type I invexity condi-tions.

3 - Weak pseudo-invexity, generalized Fritz-John opti-mality and duality in nonlinear programmingHachem Slimani, Laboratory of Modeling and Optimizationof Systems LAMOS, Computer Science Department,University of Bejaia, 06000, Bejaia, Algeria,[email protected], Mohammed Said Radjef

We use a generalized Fritz-John condition and weak invexity to deriveoptimality conditions and duality results for a constrained nonlinearprogramming. A new concept of Fritz-John type stationary point is de-fined and a characterization of solutions is established. Moreover, weprove several duality results for a Mond-Weir type dual. It is shown,that the introduced generalized Fritz-John condition combining withthe invexity with respect to different functions are especially easy inapplication and useful in the sense of sufficient optimality conditionsand of characterization of solutions.

53


� MC-24Monday, 12:30-14:00CC-A11

OR in Human ResourcesStream: OR in Human ResourcesInvited sessionChair: Irina Kazina, Telematika and Logistika, TTI, Lomonosovastr.1, Riga, Latvia, [email protected]: Majda Bastic, Faculty of Economics and Business,University of Maribor, Razlagova 14, 2000, Maribor, Slovenia,[email protected]

1 - The influence of individual factors and work environ-ment on creativityMajda Bastic, Faculty of Economics and Business, Universityof Maribor, Razlagova 14, 2000, Maribor, Slovenia,[email protected], Gabrijela Leskovar-Špacapan

The main aim of the present study was to answer the question whichfactors significantly contribute to individual commitment to creativ-ity. Questionnaire was administered to MBA and part-time studentswith working experience in Slovenian organizations. Methods usedincluded instrument development, an exploratory analysis, a confirma-tory factor analysis, and a test of a structural model. The results showthat individual characteristics and leadership have a positive affect oncreativity. The relationship between knowledge management and cre-ativity was not found as significant.

2 - Functionality of self-assessment in 360 degree feed-back methodGözde Kubat, Kazan Vocational School, Baskent University,Ankara, Turkey, [email protected], Sükran Öktem, HakanTurgut, Ismail Tokmak

The importance of human factors is increasing in quality improvement.Companies are directed to Total Quality Management to establish astandard structure. In this research, human resources researched invarious aspects. A questionnaire has been applied in various sectorsin Ankara-Turkey as a dimension of individual performance measure-ment as the size of the self assessment. Information from multi—sector employees has been obtained with the Snowball Method. Self-assessment of the data obtained as a result of the analysis is aimed toreveal the size of the 360 (DFM) functionality.

3 - Research of control systems within the limits of theeducational programIrina Kazina, Telematika and Logistika, TTI, Lomonosovastr.1, Riga, Latvia, [email protected], RostislavKopytov

The purpose of the present work consists in showing as it is possibleto conduct research of a control system on the basis of studying of itscharacteristics. In process of studying extend representation about acontrol system. That representation is formed at the expense of systemacceptance as single whole, and being in somebody a system environ-ment. The offered technique is spent on comparison of degree of thecompetence of discipline to program requirements by the account of72 characteristics. The received results testify to high degree of coin-cidence.

4 - Is there a place for OR in the HR Metrics World?Gregory Lee, Wits Business School, University of theWitwatersrand, #2 St David’s Place, Parktown, 2000,Johannesburg, Gauteng, South Africa, [email protected]

HR metrics has taken the human resources world by storm, with an ex-plosion of business analytics, data, and financial analysis. The HR met-rics field has largely been shaped through industrial psychology. Doesmodern operational research have new insights to offer HR metricsover and above the current standards? I argue that OR does have muchto offer, including the shaping of metrics systems through problem-based methods, the incorporation of data mining and simulation tech-niques, optimisation possibilities, and others. Implications for humanresource thinking and implementation are suggested.

� MC-26Monday, 12:30-14:00CC-A24

OR in Agriculture I

Stream: OR in Agriculture, Forestry and FisheriesInvited sessionChair: Sara Veronica Rodriguez-Sanchez, Graduate Program inSystems Engineering, Universidad Autonoma de Nuevo Leon,66450, San Nicolas de los Garza, Nuevo Leon, Mexico,[email protected]

1 - Surveillance systems in livestock production chains:An approach for quantitative economic optimizationXuezhen Guo, Business Economics Group, WagenignenUniversity, Hollandseweg 1, WAGENINGEN, 6700EW,wageningen, Netherlands, [email protected], G.D.H.(Frits) Claassen, Alfons Oude Lansink, Helmut Saatkamp

A novel conceptual approach is presented aimed at economic opti-mization of portfolios of surveillance systems for livestock hazards,operated by e.g. a Food Safety Authority (FSA). Subjective appraisalof surveillance criteria by a multitude of stakeholders is major fac-tor in estimating the portfolio surveillance performance, eventually de-termining surveillance resource allocations. This approach comprises(1) linear optimization, (2) stakeholder elicitation and (3) subjectiveweighing of surveillance performance. The approach is illustrated withresults regarding application by the Dutch FSA.

2 - A mixed integer linear programming model for pro-duction planning of supply chains in the Spanish pigindustryLluisM Pla, Mathematics, University of Lleida, JaumeII,73,25001, Lleida, Spain, [email protected], EsteveNadal

This study introduces a preliminary formulation of a mixed integer lin-ear programming model to optimize the production planning of a pigsupply chain in Spain. This model takes into consideration the costsassociated to each type of farm (breeding-growing-finishing). As re-sult occupancy of facilities is maximised and corresponding flow ofanimals from farm to farm derived. All-in all-out management is con-sidered in fattening (finishing) farms while continuous management isconsidered for the rest. Furthermore, constraints represent a typicalSpanish supply chain delivering pigs to an abattoir.

3 - How to improve profitability of the beef herd ? -the application of the multi-level hierarchic Markovmethod.Anna Stygar, Department of Cattle Breeding, University ofAgriculture in Krakow, al. Mickiewicza 24/28, 30-059,Cracow, Poland, [email protected], Anders RinggaardKristensen, Joanna M. Makulska

In this paper the authors present the optimization and simulation modelof beef heifer management that could be used to support the econom-ically optimal decisions made by beef producers. The model was for-mulated as 3-level hierarchic Markov process. The optimized deci-sions concerned: feeding level in suckling and weaning period, wean-ing and breeding strategy. Optimal rearing strategy was found bymaximizing total discounted net revenues from the predicted lifetimeproductivity measured by the cumulative standard weight of weanedcalves.

4 - Production planning in a Beef meat packing plantSara Veronica Rodriguez-Sanchez, Graduate Program inSystems Engineering, Universidad Autonoma de Nuevo Leon,66450, San Nicolas de los Garza, Nuevo Leon, Mexico,[email protected], Alysson M. Costa, Roger Z.Rios-Mercado

The production planning of a beef meat packing plant involves the taskof determining the number of cattle to use and how to cut them up inorder to meet a given demand. There are several cutting patterns. Eachpattern yields a specific set of products. The major challenge faced bythe top management is to balance the benefits of selling products fromone part of the carcass when there is less demand for products from

54


other parts. We develop a mixed integer linear programming modelfor this problem and illustrate how the use of this model has a positiveimpact in the company revenue.

� MC-27Monday, 12:30-14:00CC-A25

Decision Analysis and Group DecisionMaking

Stream: Decision ProcessesInvited sessionChair: Jeffrey Keisler, Management Science & InformationSystems, University of Massachusetts Boston, 100 MorrisseyBlvd, M/5-249, 02125, Boston, MA, United States,[email protected]

1 - Decision analysis with function-valued variablesJeffrey Keisler, Management Science & Information Systems,University of Massachusetts Boston, 100 Morrissey Blvd,M/5-249, 02125, Boston, MA, United States,[email protected] models typically include uncertain real and discrete state vari-ables. In principle, state variables could themselves be functions. Thisis useful for incorporating frameworks from various fields of knowl-edge. While theoretically allowed, this extension presents practicalchallenges throughout the DA process. Model structure determinesrequired assessments and calculations. The latter may involve simpli-fying assumptions, e.g., restricting to the space of quadratic functions.To implement this idea, we must balance precision of results againstease and speed of assessment and computation.

2 - On the notion of dictatorship in group utility func-tionsLuis C. Dias, Faculdade de Economia / INESC Coimbra,INESC Coimbra, Rua Antero de Quental, 199, 3000-033,Coimbra, Portugal, [email protected], Paula SarabandoKeeney proposed a cardinal utility axiomatization for the problem ofaggregating utility functions of members of a group. One of the as-sumptions in this axiomatization was nondictatorship. We proposestronger formulations for this condition introducing the concept ofstrategic dictator and presenting the corresponding characterization ofcompliant group cardinal utility functions. We also characterize anextension to extend this notion of dictator to groups of individuals.An example using robustness analysis will illustrate the application ofthese nondictatorship principles.

3 - Individual heuristics and group decision effective-nessMatthias Seifert, Operations & Technology, IE BusinessSchool, Maria de Molina 12, 5th Floor, 28006, Madrid, Spain,[email protected], Manel BaucellsWe use reduced ordered binary decision diagrams (ROBDDs) to studythe link between individual heuristic processes and group decision ef-fectiveness. Exploiting the concept of cumulative dominance in multi-attribute choice contexts, we start out by analyzing the probability thatindividuals identify dominant alternatives if they make errors duringthe evaluation of the choice set. We then demonstrate how such in-dividual inefficiencies influence the effectiveness of group decisionsrelying on majority voting.

� MC-28Monday, 12:30-14:00CC-A27

Nonsmooth Optimization I

Stream: Structured Optimization and ApplicationsInvited sessionChair: Dominik Dorsch, Dept. Mathematics, RWTH Aachen

University, Templergraben 55, 52056, Aachen, NRW, Germany,[email protected]

1 - Lower Level Chance Constraints in Robust PortfolioOptimizationMarcel Sinske, Institute of Operations Research, KarlsruheInstitute of Technology, Kaiserstraße 12, 76131, Karlsruhe,Germany, [email protected]

We combine the well-known mean-variance approach by H.M.Markowitz and the robust counterpart for a convex optimization prob-lem, introduced by A. Ben-Tal and A. Nemirovski, to achieve a robustportfolio optimization problem. Further we consider for the uncertainparameters a minimum volume ellipsoid as the uncertainty set. Suchproblems can be formulated as bilevel problems with a chance con-straint in the lower level problem. We show some properties of thistype of problem and develop an approach to solve it.

2 - Generalized derivatives of the projection onto thecone of positive semidefinite matricesAnna-Laura Wickström, Universität Zürich, Switzerland,[email protected]

We are interested in sensitivity and stability analysis of solution sets ofoptimization problems under set or cone constraints. A main motiva-tion behind our work is the analysis of semidefinite programs (SDPs).We wish to explore the sensitivity analysis of SDPs with help of gen-eralized derivatives.

3 - Nonsmooth versions of Sard’s theoremTomas Bajbar, Institute of Operations Research, KarlsruheInstitute of Technology, Kaiserstraße 12, 76131, Karlsruhe,Germany, [email protected], Vladimir Shikhman

We present a comparison between some versions of Sard’s Theoremwhich have been proven recently for special function classes with dif-ferent definitions of critical points. Considering the class Ck, the classof min-type or min-max functions, the motivation for the definition ofcritical point is the topological structure of the inverse image. We com-pare topological critical points and critical points defined via metricregularity in the class of min-type and min-max functions. We illus-trate the whole problematic by some examples.

4 - Express Line Queues and Space-Time GeometryEitan Bachmat, Computer Science, Ben-Gurion U., 1 BenGurion Ave, 84105, Beer Sheva, Israel,[email protected]

We consider queues in which jobs are sent to servers according to sizecutoffs, as with express lines in supermarkets. We show that the opti-mal choice of cutoffs is governed by an equation which is similar to thegeodesic equation in space-time geometry. We consider the queueingtheoretic consequences of this relation and exhibit a queueing theoreticanalogue of curvature.

� MC-29Monday, 12:30-14:00CC-A29

Algorithms and Methods for MINLPproblems

Stream: Mixed-Integer Non-Linear ProgrammingInvited sessionChair: Sonia Cafieri, Lab. MAIAA, Dept. de Mathematiques etInformatique, Ecole Nationale d’Aviation Civile, 7 Ave. EdouardBelin, 31055, Toulouse, France, [email protected]: Claudia D’Ambrosio, LIX, CNRS - Ecole Polytechnique,route de Saclay, 91128, Palaiseau, France,[email protected]

55


1 - Mixed-integer programming under the presence ofuncertainty: On the solution of multi-parametricMILP problemsMartina Wittmann-Hohlbein, Imperial College London,United Kingdom, [email protected],Efstratios PistikopoulosWe consider multi-parametric MILP problems with uncertainty simul-taneously present in the coefficients of the objective function and theconstraints. We present two-stage methods for the approximate solu-tion such that the computational burden to derive explicit solutions isreduced. The first approach combines robust optimization and multi-parametric programming. In the second approach we exploit the spe-cial structure of the problems and employ suitable relaxations of bilin-ear terms. In particular, we embed novel MILP-relaxations of bilinearterms in the proposed procedure.

2 - GloMIQO: Global Mixed-Integer Quadratic OptimizerChristodoulos Floudas, Chemical Engineering, PrincetonUniversity, 08544, Princeton, NJ, United States,[email protected], Ruth MisenerThe Global Mixed-Integer Quadratic Optimizer, GloMIQO, is anumerical solver addressing mixed-integer quadratically-constrainedquadratic programs (MIQCQP) to epsilon-global optimality. Algorith-mic components are presented for: reformulating user input, detect-ing special structure, generating tight convex relaxations, dynamicallygenerating cuts, partitioning the search space, bounding the variables,and finding feasible solutions. We extensively test GloMIQO using asuite of 399 problems from process networks applications, computa-tional geometry problems, and standard performance libraries.

3 - A new Branch and Bound algorithm for MIQPsAmélie Lambert, CEDRIC, CNAM, 292 rue saint martin,75141, paris, France, [email protected], AlainBillionnet, Sourour ElloumiLet (MQP) be a MIQP that consists in minimizing a quadratic func-tion subject to linear constraints. Our approach to solve (MQP) is firstto consider (MQP’), an equivalent MIQP that has a convex objectivefunction, additional variables and constraints, and additional quadraticconstraints. Then, we propose a new Branch and Bound based on therelaxation of the quadratic constraints to solve (MQP’). We performexperiments on pure-and mixed-integer instances of medium size, andshow that their solution times are improved by our Branch and Boundin comparison with two existing approaches.

4 - Optimistically Approximating Non-linear Optimiza-tion Problems through MILPClaudia D’Ambrosio, LIX, CNRS - Ecole Polytechnique,route de Saclay, 91128, Palaiseau, France,[email protected], Andrea Lodi, SilvanoMartello, Riccardo RovattiWe present a new piecewise linear approximation of nonlinear func-tions. It can be seen as a variant of classical triangulations with moredegrees of freedom to define any point as a convex combination of thesamples. For example, in the case of functions of 2 variables, eachpoint is approximated as convex combination of 4 samples instead of3 like in the traditional triangulation approach. When embedded inMILP models it results in a smaller number of binary variables. Weshow theoretical properties of the approximating functions and providecomputational evidence of the impact of the method.

� MC-30Monday, 12:30-14:00CC-A31

Emerging Applications in ProductionManagement

Stream: Emerging Applications in Finance and Logis-ticsInvited sessionChair: Jakob Krarup, Dept. of Computer Science, University ofCopenhagen, Ydervang 4, DK-3460, Birkeroed, Denmark,[email protected]

1 - Assigning experts to tasks in a projectStanislaw Walukiewicz, Systems Research Institute, PolishAcademy of Sciences, ul. Newelska 6, 01-447, Warsaw,Poland, [email protected], JakobKrarup

The problem of assigning experts to tasks in a project can be mod-eled with the so-called Virtual Production Line (VPL) proposed byWalukiewicz in 2006. VPL appears to be a natural extension of theClassical Production Line (CPL) dating back to the car manufacturerFord in 1913. Our approach is based on the concepts of social capitaland proximity where social capital is defined as formal or informal re-lations between at least two experts. The proximity refers to specificinterrelations linking experts on VPL. Comparisons of the assignmentsof experts to VPL and workers to CPL are considered.

2 - An inventory model with time dependent demandrate, trade credits and partial backloggingVikram jeet Singh, MATHEMATICS, Lovely ProfessionalUniversity, Jalandhar-Delhi G.T Road(NH-1), 144402,Phagwara, Punjab, India, [email protected]

In this paper we developed an inventory model for deteriorating itemswith time dependence of demand. Many inventory models in past weredeveloped under the assumption that the holding cost is constant for theentire inventory cycle. But this is particularly false in the storage ofdeteriorating and perishable items such as food products. The holdingcost is assumed to be varying over time in only few inventory models.Partially backlogged shortages are permitted and backlogging rate iswaiting time for the next replenishment.

3 - Whole Life Costing (WLC) model in asset manage-mentWenjuan Zhang, Warwick Business School, WarwickUniversity, ORMS Group, CV4 7AL, Coventry, UnitedKingdom, [email protected]

A Whole Life Cost (WLC) approach has been developed for deter-mining capital expenditure requirements for asset management. Thisapproach has addressed the problem caused by uncertainty associatedwith the maintenance cost and their timing by modelling asset degrada-tion (timing to performance loss and boundary condition interventionstates) capital and maintenance costs, risks and their probabilities, us-ing Bayesian statistics, for each Maintainable Item (MI).

� MC-31Monday, 12:30-14:00CC-A33

Mathematical methods for decisionsupport in energy and industrial sectors

Stream: Iterative Methods for Economic ModelsInvited sessionChair: Elisabetta Allevi, Quantitative Methods, University ofBrescia, Contrada Santa Chiara, 50, 25122, Brescia, Italy,[email protected]: Adriana Gnudi, Mathematics, Statistics, Computer scienceand Applications, University of Bergamo, Via dei Caniana, 2,24127, Bergamo, Italy, [email protected]: Igor Konnov, University of Kazan, Kazan, RussianFederation, [email protected]

1 - The steel industry: a mathematical model under en-vironmental regulationsRossana Riccardi, Quantitative Methods, University ofBrescia, Contrada Santa Chiara, 50, 25122, Brescia, Italy,[email protected], Elisabetta Allevi, Adriana Gnudi,Francesca Bonenti, Claudia Avanzi

In this work, a spatial equilibrium problem is formulated for analyzinghow the price of electricity can influence investment decisions in thesteel sector under EU ETS. In this light, two main processes for steelmaking have to be considered: integrated mills (BOF) and Electric

56


Arc Furnace (EAF) in minimills. Integrated mills produce high levelsof direct emissions while emissions in EAF are indirect emissions inthe form of electricity. A technological representation of the marketwill be introduced and the possibility of electricy auto-production inintegrated mills will be analyzed.

2 - Evaluating the Impacts of the EU-ETS on Prices, In-vestments and Profits of the Italian Electricity MarketGiorgia Oggioni, Department of Quantitative Methods,University of Brescia, Italy, Contrada Santa Chiara, 50,25122, Brescia, Italy, Italy, [email protected], FrancescaBonenti, Elisabetta Allevi, Giacomo Marangoni

This paper investigates the economic impacts of the European Emis-sion Trading Scheme (EU-ETS) on the Italian electricity market by apower generation expansion model where generators behave either asCournot or perfectly competitive players. The developed models areused to measure the effects of the EU-ETS on electricity prices and de-mand, investments and generators’ profits taking also into account theimplications of the third phase. The solution of the overall system isfound by exploiting the mixed complementarity theoretical frameworkand the implementation is conducted in GAMS.

3 - Electricity markets under linear price/demand andquadratic costs: Equilibrium à la Cournot and Sup-ply Function EquilibriumFrancesca Bonenti, Quantitative Methods, University ofBrescia, Contrada Santa Chiara 50, 25122, Brescia, Italy,Italy, [email protected], Magalì E. Zuanon

The Supply Function Equilibrium approach in an electric market takesplace whenever firms compete in offering a schedule of quantities andprices. On the other hand, in the Cournot equilibrium model each firmchooses the optimal output by maximizing its own profit. Bonenti andZuanon presented the explicit expressions characterizing analyticallythe Cournot equilibrium and the Supply Function equilibrium valuesfor the produced quantities, the market-clearing price and the profitsin a particular version of the models under linear price/demand andquadratic costs.

4 - Splitting Methods for Dynamic Auction ProblemsWith Set Valued MappingsAdriana Gnudi, Mathematics, Statistics, Computer scienceand Applications, University of Bergamo, Via dei Caniana, 2,24127, Bergamo, Italy, [email protected], ElisabettaAllevi, Igor Konnov

We present a spatial price equilibrium model for the dynamic case withset valued cost functions. The model describes the behavior of a systemof auction markets joined by transmission lines subject to joint balanceand capacity flows constraints for a certain time period. We treat thismodel as an extended primal-dual system of variational inequalities.Some splitting type methods are proposed to find its solution with in-exact and exact optimization algorithms.

� MC-32Monday, 12:30-14:00CC-A34

Understanding the Practice of Soft ORInterventions IIStream: Soft OR and Problem Structuring MethodsInvited sessionChair: Ashley Carreras, Leicester Business School, De MontfortUniversity, Marketing, The Gateway, LE19BH, Leicester, UnitedKingdom, [email protected]

1 - A Code of Best Practice for Soft Operational AnalysisDiederik J.D. Wijnmalen, Dept. for Behavioural and SocietalSciences, TNO Organisation for Applied Scientific Research,P.O. Box 96864, 2509 JG, The Hague, Netherlands,[email protected]

The purpose of the ’NATO Guide for Judgement-based OA in DefenceDecision Making’ is to create an understanding of what judgement-based OA is; to clarify what this type of analysis can do to help addressproblematic situations; and, to provide guidance on how a judgement-based OA study should be carried out to maximise the validity, cred-ibility and acceptance of such a study and its outcomes. In three vol-umes the Guide addresses analysts, clients and executives. The presen-tation will focus on its Code of Best Practice volume which is consid-ered to be relevant to a general analyst audience.

2 - Planning Soft Workshops: getting the right people inthe roomSue Merchant, Blue Link Consulting, 4,Shepherds Way, WD37NJ, Rickmansworth, Hertfordshire, United Kingdom,[email protected]

Soft OR has many techniques available now, along with much practicaladvice, for the successful running of a range of different types of work-shop or decision conference. A critical aspect of planning a workshopinvolves making sure that the right people attend the workshop in thefirst place. This presentation looks at the questions which organisersneed to ask themselves regarding participants when arranging a work-shop, notes some of the advice available in the literature, and reportson good practice recently distilled from a group of practitioners.

3 - Enhancing the Dimensions of Procedural Justice inFocus Group Workshops using Facilitated Group De-cision SoftwareParmjit Kaur, De Montfort University, LE19BH, Leicester,United Kingdom, [email protected], Ashley Carreras

This paper examines the way in which causal mapping aided by groupdecision software adheres to the tenets of procedural justice. The work-shops utilise a dual facilitation process that allows a more authenticparticipant voice to be heard. We seek to explain why this processof investigation surfaces more authentic qualitative data by aligningthe process of investigation with the principles of procedural justice asfound in organisational justice literature.

4 - Mapping Collective IntentionsAshley Carreras, Leicester Business School, De MontfortUniversity, Marketing, The Gateway, LE19BH, Leicester,United Kingdom, [email protected]

An integral part of Strategy Workshops utilising causal mapping is themovement towards acollected set of actions that participants in thatworkshop commit themselves to.This paper outlines one approach tounderstanding this degree of collectiveness by utilising the concept ofintentionality, proposed by Searle and refined by Velleman, to matchcontent of one person’s agreement to action with another’s. After out-lining this approach to collective intentions the paper looks at examplesof such agreements in the maps of workshops facilitated by the author.

� MC-33Monday, 12:30-14:00CC-A37

Humanitarian Logistics and DisasterManagement

Stream: OR for Development and Developing Coun-triesInvited sessionChair: Begoña Vitoriano, Estadística e Investigación Operativa I,Fac. Matemáticas, Universidad Complutense de Madrid, Plaza deCiencias, 3, Ciudad Universitaria, 28040, Madrid, Spain,[email protected]

1 - Structured classification of disasters and emergen-ciesJ. Tinguaro Rodriguez, Department of Statistics andOperational Research, Complutense University of Madrid,Plaza de Ciencias, 3, 28040, Madrid, Spain,[email protected], Javier Montero, Begoña Vitoriano

57


A fully precise numerical evaluation of disasters’ effects is unrealisticin the time-pressured, highly uncertain decision context taking placejust after a disaster strike. Instead of a numerical evaluation, here itis considered that it is rather more plausible and realistic to classifythe severity of the consequences of a disaster in terms of the relevantscenarios for NGO’s decision makers. However, some features of thatcontext entail the necessity of considering and assuming an structureover the set of classes, somehow modelling those features inside of theclassification model.

2 - Uncertainty and Robustness in Humanitarian SupplyChains for Disaster ReliefAlistair Clark, Dept of Engineering Design and Mathematics,University of the West of England, Frenchay Campus,Coldharbour Lane, BS16 1QY, Bristol, United Kingdom,[email protected]

A bidirectional supply chain network model to get humanitarian aidquickly to needy recipients in a natural disaster is formulated mathe-matically, and then discussed in the context of uncertainty about trans-portation resources, aid availability, and the suddenness and degree of"demand", poor information feedback and multiple decision-makers indifferent aid agencies. How can robust and stable plans be maintainedand disruption limited when new events or information necessitate therevision and updating of existing plans and decisions?

3 - Decision Aid Models on Preparedness Operations onDisaster ManagementBegoña Vitoriano, Estadística e Investigación Operativa I,Fac. Matemáticas, Universidad Complutense de Madrid,Plaza de Ciencias, 3, Ciudad Universitaria, 28040, Madrid,Spain, [email protected], M. Teresa Ortuño, FedericoLiberatore, Celeste Pizarro Romero, Clara Simon de Blas

Natural disasters can strike a population suddenly and unexpectedly orbeing slower and repeated. Especially in this second case that forecast-ing is possible, it becomes crucial for an effective response having pre-pared population, emergency protocols, resources location, and plan-ning operations. Decision aid models into this context should considernot only the immediate response but also the subsequent treatment ofvictims. A multicriteria model for shelters location taking into accountthe evacuation operations and the later distribution of humanitarian aidwill be presented.

4 - Model to validate the quality of data sources in aubiquitous crowd-sourcing contextJose Robledo, Sistemas de Información, Universidad deGuadalajara, PERIFERICO NORTE 799 MOD L 305, LOSBELENES, 45100, ZAPOPAN, JALISCO, Mexico,[email protected], Laura Plazola Zamora

Gathering data by a crowd-sourcing method through the use of ubiqui-tous computing is a relatively new approach for acquiring and process-ing data related to real issues such as vehicle traffic, epidemic spreadand in general incidence of alerts. A concern that arises when data isacquired by this manner in order to be used to feed a decision supportsystem is to ensure that data collected complies with certain quality cri-teria. This work proposes the use of a model to validate the quality ofdata sources through mechanisms of redundancy, trust and reputationin order to meet the set requirements.

� MC-34Monday, 12:30-14:00CC-A39

Topics in Cooperative Game Theory

Stream: Cooperative Game TheoryInvited sessionChair: Manuel A. Pulido Cayuela, Statistics and OperationsResearch, University of Murcia, Campus de Espinardo. Facultadde Matemáticas, 30100, Murcia, Murcia, Spain, [email protected]: Josep Freixas, Applied Mathematics 3, TechnicalUniversity of Catalonia, Av. Bases de Manresa, 61-73, E-08242MANRESA. Spain, 08242, Manresa, Spain,[email protected]

1 - Enumerations of voting systems: Fibonacci se-quences and the gold numberJosep Freixas, Applied Mathematics 3, Technical Universityof Catalonia, Av. Bases de Manresa, 61-73, E-08242MANRESA. Spain, 08242, Manresa, Spain,[email protected], Sascha KurzVoting systems are structures which are numerable as a function of thenumber of voters. The Fibonacci sequences appear regularly for gameswith some few types of equivalent players and many of them just dif-fer asymptotically by a multiplicative factor, which is the gold numberor its square. The paper summarizes enumerations of voting struc-tures which follow some paterns. Additionally, we also point out someproblems, related to values of cooperative games, where the numberof possible rankings for the Shapley value in different games is thek-dimensional Catalan numbers.

2 - Finding extremal voting systems via integer linearprogrammingSascha Kurz, Mathematics, Physics and Computer Science,University of Bayreuth, Universitätsstraße 30, 95440,Bayreuth, Bavaria, Germany, [email protected] types of yes/no voting systems are frequently studied. Us-ing integer linear programming we determine extremal voting systems.As examples we consider voting systems whose Shapley-Shubik vec-tor has minimal distance to a given power distribution; whose Pub-lic Good Index maximally violate local monotonicity; or which arefarthest away from weighted voting games with respect to a recentlyintroduced hierarchy of simple games. We present the general under-lying ideas and computational results for instances where exhaustiveenumeration of all voting systems is infeasible.

3 - Success and decisiveness on symmetric gamesMontserrat Pons, Applied Mathematics 3, TechnicalUniversity of Catalonia, Av. Bases de Manresa 61-73, 08242,Manresa, Spain, [email protected], Josep FreixasThe notion of success has usually been considered just a secondaryaspect of decisiveness, due partially to the fact that success and de-cisiveness show a linear relationship when all vote configurations areequally probable.In our work we analyze the relationship between success and deci-siveness for anonymous and independent probability distributions onsymmetric games.

4 - A model for players’ tendencies when playing a co-operative gameM. Albina Puente, Applied Mathematics III, TechnicalUniversity of Catalonia, Av. Bases de Manresa-61-73,Manresa, Spain, [email protected], FrancescCarrerasWe present the theory for multibinary probabilistic values on cooper-ative games. As they are defined by tendency profiles, new tools areprovided to encompass situations due to the influence of players’ per-sonality when playing a given game. When the profile is positive, themultilinear extension is a potential function and gives a computationaltool. We also adapt to these values some properties, concerning nulland nonnull players, balanced contributions, dominance and mono-tonicity, and regularity arises as a convenient condition. The subjectfalls within the topic of restricted cooperation.

� MC-35Monday, 12:30-14:00CC-A41

Engineering Optimization

Stream: Engineering OptimizationInvited sessionChair: Wolfgang Achtziger, Department of Mathematics,University of Erlangen-Nuremberg, Chair of Applied Mathematics2, Cauerstrasse 11, 91058, Erlangen, Germany,[email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

58


1 - Development of a multi-layer ANFIS model for theprediction of tertiary scale formation in the steel in-dustryJonathan Kennedy, Specific, Swansea University, Baglan BayInnovation & Knowledge Centre, Baglan Energy Park,Central Avenue, SA12 7AX, Baglan, Port Talbot, Wales,United Kingdom, [email protected] is an oxide that forms on steel during the hot mill process that hasa detrimental effect on the surface quality of surface critical products.Scale formation is dependent on numerous factors such as thermo-mechanical processing routes, chemical composition, thickness androlls used. This paper describes the early development of a 3 layerAdaptive Neuro Fuzzy Inference System (ANFIS) model, which pre-dicts the formation of scale on strip steel from various input criteria.The ANFIS model deployed in this work is purely spreadsheet basedwith Excels Solver used for optimising the weights.

2 - Design of a RFID Based Real Time Warehouse Man-agement System for Steel IndustryOzgur Eski, Industrial Engineering Dept., Celal BayarUniversity, 53.Sok. No:6 Daire:20, Baki Aktuglu Apt.Bornova, 35040, IZMIR, Turkey, [email protected],Ceyhun Araz, Levent Bayoglu, Filiz TuranIn steel industry, the efficient storage of steel coils is important. Intraditional systems, coils are randomly placed on the floor and the per-formance of the storage operations depends on operator’s experience.This results in inefficient use of warehouse and mistakes. In this studywe proposed an RFID based real-time warehouse management system(RWMS). RWMS helps the warehouse managers to achieve better in-ventory control and to improve the efficiency of operations. We alsodeveloped a heuristic for efficient placement of steel coils. A casestudy is conducted in Norm Fasteners Co, Turkey.

3 - Damage characteristic of hybrid composite plaquesin 2m/s speed impactLevent Bozkurt, Ilgin Vocational School of Higher Education,42600, Ilgın/ Konya, Turkey, [email protected] composite can be defined as having more than one staple withinsame composite. However, in stratified composite materials, generallylow velocity impacts are considered as impacts which are lower than100 m/s. In this study, characteristics of hybrid composite stratifieswhich are exposed to low velocity impact are studied. Low velocityimpact experiments are done in testing device which is developed forweight fall. After experiments which are done under different borderconditions, dynamical answer of the material, rising damages and elas-tical specifications are searched.

4 - Conjoint analysis for Concept Design of CommercialDiesel EngineJeasu Jeon, Information&Industrial Engineering, YonseiUniversity, 134 Shinchon-dong, Seoul, Korea, Republic Of,[email protected], So Young SohnCommercial diesel engine is a main component for various applica-tion domains such as excavator, commercial vehicle, generator, etc.The regulation concerning green management along with various cus-tomers’ needs has to be reflected on the design of diesel engine. How-ever, diesel engine manufacturers rarely produced customer orientedproducts. In this paper, we apply adaptive conjoint analysis to identifythe most sensitive design attribute for commercial diesel engine designand forecast the ideal product for China market which has the highestshare in the world.

� MC-36Monday, 12:30-14:00CC-A43

DEA and Performance Measurement:Applications 1

Stream: DEA and Performance MeasurementInvited sessionChair: Armando Zeferino Milioni, Departamento de Organizacao,Instituto Tecnologico de Aeronautica, CTA ITA IEMB,12228-900, Sao Jose dos Campos, Sao Paulo, Brazil,[email protected]

1 - Measuring the efficiency of a pharmacy chain bymeans of Advanced DEALudmila Neumann, Institute of Management Control andBusiness Accounting, TU Braunschweig, Pockelsstraße 14,38106, Braunschweig, Lower Saxony, Germany,[email protected], Heinz Ahn

The Advanced DEA is a methodological enhancement of the tradi-tional DEA that allows the inclusion of a hierarchical goal structureinto the analysis. This approach does not only consider inputs and out-puts, but also their impact on the relevant objectives. The present studyprovides an application of Advanced DEA to measure the efficiency ofa European pharmacy chain with 49 DMUs. The results show to beparticularly appropriate for identifying partial inefficiencies that arenot recognized by traditional DEA models.

2 - Benchmarking on Spanish University librariesClara Simon de Blas, Statistics & Operations Research, ReyJuan Carlos University, Departamental II, Desp. 042,C/Tulipan s/n, 28933, Mostoles, Madrid, Spain,[email protected], Jose Simon Martin, Alicia Arias

In 47 Spanish university libraries, efficiency is computed in the use ofhuman resources in the execution of internal library processes. DEAand outlier detection methods are combined to assess management onbenchmarking efficiency. DMUs that are primarily identified as effi-cient are submitted to a complementary analysis to detect the numberof times they perform as peer.If the number of reference units is low,then other evidences for establishing the superiority of its performanceis realized through the analysis of results excluding an output, as theroll of the output in the variables set.

3 - Service Benchmarking in the Portuguese Hospitals:An Application of Data Envelopment AnalysisRicardo A. S. Castro, Faculdade de Engenharia, Universidadedo Porto, Portugal, [email protected], MariaPortela, Ana Camanho

In this study we present an evaluation model of hospital services inPortugal, using an efficiency of care perspective. Multiple inputs andoutputs were chosen based on their impact on the total costs incurredby all clinical services. A DEA model was applied to a specific serviceand constraints on the weights were imposed to avoid efficient unitswith an injudicious choice of them. Results show the factors wherethe biggest possible savings can be made. A comparison between best-practice services and inefficient services is also presented, revealingthe reasons for a superior performance.

4 - Using DEA to evaluate Brazilian Institutions ofHigher EducationArmando Zeferino Milioni, Departamento de Organizacao,Instituto Tecnologico de Aeronautica, CTA ITA IEMB,12228-900, Sao Jose dos Campos, Sao Paulo, Brazil,[email protected], Diego Geraldo, Joyce Teixeira, Luciene Alves

Every year Brazilian institutions of higher education are evaluated andthe final result of the evaluation is a number called GIC - general indexof courses. The GIC is calculated from the weighted sum of the re-sults of the evaluation of undergraduate and postgraduate (masters anddoctorate) courses of each institution. The weighting factors, whichtake into account the number of enrolled students, are often an objectof long discussions. We show that, indeed, the ranking of institutionsaccording to GIC can vary widely according to the weighting factors.We use DEA to discuss this issue.

� MC-37Monday, 12:30-14:00CC-Act

Bioinformatics IIIStream: Computational Biology, Bioinformatics andMedicineInvited sessionChair: Marta Szachniuk, Institute of Bioorganic Chemistry, PAS,Noskowskiego 12/14, 61-704, Poznan, Poland,[email protected]

59


Chair: Jacek Blazewicz, Institute of Computing Science, PoznanUniversity of Technology, ul.Piotrowo 2, 60-965, Poznan, Poland,[email protected]

1 - An optimization framework for intensity modulatedproton therapy treatment planningGino Lim, Industrial Engineering, University of Houston,E206, Engr. Bldg 2, 77204, Houston, TX, United States,[email protected], Wenhua Cao, Xiaodong Zhang

Intensity modulated proton therapy (IMPT) has rapidly emerged asan advanced type of radiation therapy for cancer patients because itcan achieve precise dose to the tumor volume while greatly sparingthe normal tissues. However, use of mathematical programming tech-niques has not been widely addressed for optimizing treatment param-eters. Therefore, we introduce an optimization framework to optimizebeam angles, spot positions and corresponding spot intensities simul-taneously. Benefits of proposed methods are demonstrated by severalcase studies on clinical prostate cancer patients.

2 - On the selection of images and determining the ageof acquisition of words for a recognition test usingexploratory data analysisGastão Gomes, Métodos Estatísticos, Universidade Federal doRio de Janeiro, Rua Dois de Dezembro, 15/804, Flamengo,Rio de Janeiro, 22220-040, Rio de Janeiro, Rio de Janeiro,Brazil, [email protected], Sergio Camiz, Christina Gomes,Fernanda Senna

We are concerned about the evaluation of images to be used in a testwith aphasic patients. Correspondence Analysis (CA) was applied toevaluate 38 subjects, in order to understand to what extent they identifythe image. As a result 161 words were selected. These new selectedwords were now submitted to two sets of judges with differently sug-gested scales of age of acquisition: a free-scale one and another withthe suggested slots of ages. The effect of the scale was studied byPCA, Multiple CA and Multiple Factor. The results indicate a goodagreement between the two scales.

3 - Benchmarking and proposed improvements of thedatabase of Clusters of Orthologous Genes (COG)Farzana Rahman, Computational Biology Group, Universityof Glamorgan, United Kingdom, [email protected],Mehedi Hassan, Tatiana Tatarinova, Alexander Bolshoy

We developed a non-parametric Bayesian method for validation of con-sistency of COG database. We were motivated by the observation thatfor at least 35% clusters from COG database, the distribution of genelength cannot be approximated using hierarchical Poisson-Gamma dis-tribution, but as a mixture of distributions. In the framework of Non-parametric Bayesian approach, model parameter distributions consistof multiple discrete ’support’ points, up to one per subject in the pop-ulation. Each support point is a set of point estimates of each modelparameter value, plus probability of that set.

� MC-38Monday, 12:30-14:00HH-Colombus

Multicriteria Decision Support Systems inIndustry

Stream: Multiobjective OptimizationInvited sessionChair: Karl-Heinz Kuefer, Optimization, Fraunhofer ITWM,Fraunhofer-Platz 1, D-67663, Kaiserslautern, Germany,[email protected]: Kai Plociennik, Optimization, Fraunhofer ITWM,Germany, [email protected]

1 - Efficient Approximation of Convex and Non-ConvexPareto-Frontiers with Application to ChemicalProcess-Optimization

Uwe Nowak, Optimization, Fraunhofer ITWM,Fraunhofer-Platz 1, 67663, Kaiserslautern, Germany,[email protected], Michael Bortz, Karl-HeinzKüfer

In many practical simulation-based optimization problems, an efficientapproximation of the (possibly non-convex) Pareto frontier is highlydesirable. We consider this problem for very involved simulationsof chemical processes. We present different Pareto set approxima-tion schemes: The weighted sum sandwiching algorithm for convex isvery fast. In contrast, the slower Pascoletti-Serafini hyperboxing algo-rithm is applicable to non-convex Pareto sets. Furthermore, we presenta combination of both algorithms, suited for practical problems withonly small non-convex parts of the Pareto frontier.

2 - A multicriteria optimization approach for radiofre-quency ablation (RFA) planningKatrin Teichert, Optimization, Fraunhofer ITWM,Fraunhofer-Platz 1, 67663, Kaiserslautern, Germany,[email protected], Philipp Süss, Karl-HeinzKuefer

In RFA treatment planning, high-frequent current is applied to locallyheat up and destroy cancerous tissue. A placement of the needle-shaped applicator is considered good if it leads to complete tumor de-struction and also avoids specific treatment risks. In the modeling ofthis multicriteria optimization problem a main difficulty lies in the dis-connectedness of the feasible set, as the applicator must not intersectwith bones or organs. We investigate how the Pareto set is affectedand, based on our findings, propose a method to interactively navigateon the set of non-dominated solutions.

3 - Multicriterial decision support for photovoltaic powerplant designKai Plociennik, Optimization, Fraunhofer ITWM, Germany,[email protected], Hendrik Ewe, Ingmar Schüle,Karl-Heinz Kuefer

Finding a good photovoltaic power plant design is a complex taskwith many decisions to be made. The right electrical component typesand amounts (PV modules, inverters, cables) and the best componentlayout and electrical design have to be found. Since a multitude of(partially conflicting) objective functions like "minimize constructioncosts’ or "maximize energy yield’ is present, the problem is stronglymulticriterial. We present software which proposes a set of reasonableplant designs, from which the engineer chooses the best in a multicri-terial decision support system.

4 - Multicriteria Rolling Wave Planning and SchedulingBastian Bludau, Optimization, Fraunhofer ITWM, Germany,[email protected], Sebastian Velten, Karl-HeinzKuefer

In real-life industrial applications of production planning and schedul-ing, new tasks and unforeseen events force a once computed scheduleto be adjusted over and over again. A common way to deal with suchdynamic problems in practice is rolling wave scheduling. We present aperiodic rescheduling model that uses flexible bounding constraints tosimultaneously control multiple objective criteria such as resource uti-lization, adherence to delivery dates, and schedule stability. Particularemphasis is given on the adjustment of the bounding constraints.

� MC-39Monday, 12:30-14:00HH-Cousteau

MADM Applications III


60


1 - Identifying Reasons of Customer WaitingChih-Chin Liang, Department of Business Administration,National Formosa University, Taipei, Taiwan,[email protected] may become angry and leave after waiting an intolerablylong time, leading to marketing opportunities being lost. Once a ser-vice sector understands the factors driving customers to join a line,that company might adopt marketing strategies targeting waiting cus-tomers. To identify factors, this study uses principal component anal-ysis to identify the importance of each factor. Finally, the weight ofeach component is also found and assigned to help understand the im-portance of each component in decisions about joining a queue.

2 - Identifying Hierarchical and Simultaneous Multi-player Game-solution for (Non-) cooperative Adver-tising in Manufacturers and Retailers Using MOPSO-CD or NSGA IIChie-bein Chen, International Business, National Dong HwaUniversity, 1, Sec. 2, Da-hsueh Rd. Shou-feng, 974, Hualien,Taiwan, [email protected], Jung-Ho Lu, YuanchauLiourThis research is to solve the vertical (non-)cooperative advertisingproblem under different market response functions using MOPSO-CDor NSGA II integrated the Stackelberg game or Nash game to identifytheir equilibriums (or solution) by considering the polices in long-termbranding investments by manufacturers and in short-term promotionefforts by retailers. Finally, this research will implement some casesand their numerical results will demonstrate the feasibility and theirdifferent numerical results of the vertical cooperative advertising insupply chain.

3 - Application of Fuzzy Subsethood Measure as the Fil-ter for Outliers in Likert’s Questionnaire AnalysisCheng-Chuang Hon, Department of Industrial Engineeringand Management, Minghsin University of Science andTechnology, 1 Hsin-Hsing Road, Hsin-Fong, Hsinchu,Taiwan., 304, Hsinchu, Taiwan, [email protected],Ling-Lang TangIn order to avoid deviations caused by autocratic and subjective atti-tudes in decision making process, bringing decision makers togetherto obtain group consensus is one of the best choices for achieving ac-curate performance. For collection of individual judgment was usedLikert Scale questionnaire. We propose Fuzzy Subsethood Measureas the "Outlier Elimination’ method for fuzzy data analysis expressedin types of triangular fuzzy number, which can adjust the calculatedresults escaping from the bias. Also an empirical example, data collec-tion through Likert’s questionnaire is provided.

4 - An Evaluation Model of Implementing RFID for 3PL’sService Agility and ValueLing-Lang Tang, School of Management, Yuan Ze University,135 Yuan-tung road„ 320, Chung-Li, Taoyuan, Taiwan,[email protected], Cheng-Chuang Hon,Ming-Tsang LuThis paper builds up an evaluation model for company using RFIDsystem and finds how RFID affects organization’s agility and servicevalue. We use system dynamics to evaluate and explain the compli-cated nonlinear and dynamic business process. Through this model,we found that corporation’s agility, advance visibility of RFID and ITassistants could bring higher business benefits and reduce customerswaiting time. It proves that adopting RFID system has ability to boostagility capability. The 3PL implementing RFID makes process moreefficient and save more cost than before.

� MC-40Monday, 12:30-14:00HH-Livingstone

MADM Applications VII

Stream: MCDA: New Approaches and ApplicationsInvited sessionChair: Chin-Tsai Lin, Department of Business Administration,Ming Chuan University, 5F, No.130, Jihe Rd., Shihlin District„11111, Taipei, Taiwan, [email protected]

Chair: Chie-bein Chen, International Business, National DongHwa University, 1, Sec. 2, Da-hsueh Rd. Shou-feng, 974, Hualien,Taiwan, [email protected]

1 - Efficiency and Productivity of Human Resources inScience and Technology in Innovation and EfficiencyDriven NationsYing-Chyi Chou, Department of Business Administration,Tunghai University, 181 Taichung-kang Rd., Sec. 3, Taichung,Taiwan, R.O.C., 407, Taichung„ Taiwan,[email protected], Hsin-Yi Yen

Data Envelopment Analysis (DEA) was applied to assess the relativeefficiency of HRST and determine productivity variation of each nationusing the Malmquist model. The empirical results demonstrate that theefficiency of HRST differs among nations, depending on the level ofeconomic development. From Malmquist model estimates, productiv-ity indices were calculated and decomposed into two categories: tech-nological change and efficiency change of productivity variation. Abenchmark for nations aiming to improve their HRST efficiency andproductivity has been established.

2 - An Assessment of Taiwan’s Solar Water Heater Sub-sidy Policy Using Logistic Diffusion CurvesPao-Long Chang, Department of Business Administration,Feng Chia University, 100, Wenhwa Road, Seatwen,Taichung,Taiwan, R.O.C, 407, Taichung, Taiwan,[email protected], Chiung-Wen Hsu

As many countries are implementing policies to achieve carbon reduc-tion, assessing the effectiveness of measures is important. Using in-stallation area data from the non-subsidized period (1992—1999) andsubsidized period (2000—2010), logistic curves are used to model theadoption of solar water heating and assess the effectiveness of Taiwan’ssolar water heater subsidy policy. The result showed that the govern-ment’s average annual investment of NT $1,052 for solar water heatersled to a 1-ton carbon emission reduction in 10 years. Thus, only 57%of the goal by 2020 can be achieved.

3 - Investigation of feature dimension reduction basedGLCM/SVM for color filter defect classificationYu-Min Chiang, Industrial Management, I-Shou University,No.1, Sec. 1, Syuecheng Rd., Dashu District„ 84001,Kaohsiung City, Taiwan, Taiwan, [email protected],Yao-Chang Lin, Jiang-Liang Hou, Shih-Ting Yang

This paper presents an efficient color filter (CF) defect recognition sys-tem, based on gray-level co-occurrence matrix (GLCM) and supportvector machines (SVM). GLCM is used to extract textural informationfirst. The extracted features are further selected by F-score, informa-tion gain and individual feature accuracy selection methods. The ideais to reduce dimensionality of CF defect space. The selected featuresare used as features for the SVM classification process. Experimen-tal results indicate the proposed method achieves outstanding perfor-mance for CF defect classification.

4 - Environmental Fuzzy Multi-Attribute Decision-Making in Integrated Wastewater ManagementHojjat Mianabadi, Faculty of Civil Engineering andGeosciences, TU Delft, Room 4.92.2, Building 23, Stevinweg1, 2628 CN, Delft, Netherlands, [email protected],Mehrdad Mirabi, Erik Mostert, Mohammad Bagher Sharifi

One of the important challenges of modern human life is increasingproduction of urban wastewater in metropolises. Hence, selection thebest method to collect and reuse of wastewater is important. There aredifferent methods to run sewage systems; the best method of whichshould be selected considering structural conditions, economic, social,and political criteria. Wastewater management of the Niasar city, Iran,is investigated as a real case study using two risk based MADM meth-ods. The results show that MAMD methods are necessary to reachsustainable development in waste water management.

61


� MC-41Monday, 12:30-14:00HH-Heyerdahl

AHP/ANP 2Stream: Analytic Hierarchy Process, Analytic NetworkProcessInvited sessionChair: Tihomir Hunjak, The Faculty of Organization andInformatics, Croatia, [email protected]

1 - Combination of DEMATEL and ANP for the cargoshipping company selection problemNilsen Kundakcı, Business Administration, PamukkaleUniversity, Pamukkale University, Faculty of Economics andAdministrative Sciences, Department of BusinessAdministration, 20070, Denizli, Turkey,[email protected], Esra Aytaç, Aysegül Tus Isık

Cargo shipping is a vital part of logistic functions but selecting the rightcargo shipping company is a difficult task for companies. This selec-tion can be handled with the help of multiple criteria decision making(MCDM) methods. This paper attempts to solve the cargo shippingcompany selection problem by adopting MCDM model that combinesDecision Making Trial and Evaluation Laboratory (DEMATEL) andAnalytical Network Process (ANP). DEMATEL is employed to com-pute the effects between selection criteria and ANP is used to deter-mine the best cargo shipping company considering all alternatives.

2 - Utilization of Multi-Criteria Decision Making Meth-ods at the Phase of Compare and Contrasting of theCourse of Actions within the Military Decision Mak-ing Process.Erman Atak, war academy, Turkey, [email protected]

It is concluded that using Analytic Hierarchy Process (AHP) and An-alytic Network Process (ANP), one of the methods of Multi-CriteriaDecision Making, will provide an important benefit at the phase ofcompare and contrasting the course of actions, evaluated as the poorpoint of AKVES. The criteria which have to be used for AHP andANP, being applied in a military operation, have been established. Inthe light of established criteria, how a model of ANP can be set up, hasbeen showed.

3 - Analysis of course selection using ahp and discretechoice analysisAnamarija Jazbec, Mathematic and Technical Basics, Facultyof Forestry, University of Zagreb, Sveto�imunskac.25, 10000, Zagreb, Croatia, [email protected]

The main objective of this paper is to determine the importance of thecriteria in the selection of courses at the graduate level at Faculty usingthe AHP method and discrete choice analysis, compare the obtainedresults. Students (n = 22) were asked about the criteria for choos-ing courses: 1) usefulness, 2) interesting program, 3) easiness, 4) rec-ommendation of senior colleagues. AHP and discrete choice analysisproduced very similar results: usefulness of knowledge looking for joband recommendations of senior colleagues has proved to be equallyimportant criteria for selection of courses.

4 - An AHP/DEA hybrid model for the evaluation of Croa-tian research institutes performancesTihomir Hunjak, The Faculty of Organization andInformatics, Croatia, [email protected]

The purpose of the analysis was to compare the research institutes fromtwo aspects; accountability toward their mission and their autonomy.For research outputs assessment, an indicator was developed using theappropriate AHP model and in consultations with an expert for bib-liometrics. As a measure of autonomy, the ratio between own incomeand the amount of funding gained from the state budget was used. Asa part of the research, cross efficiency analysis was conducted and thesensitivity analysis was performed by weight restrictions.

� MC-42Monday, 12:30-14:00BW-Amber

EEPA 2012 - 1Stream: EURO Excellence in Practice Award 2012Invited sessionChair: Michel Bierlaire, ENAC INTER TRANSP-OR, ÉcolePolytechnique Fédérale de Lausanne (EPFL), GC B3 454, Station18, CH-1015, Lausanne, Switzerland, [email protected]

1 - Clearance Pricing Optimization at ZaraFelipe Caro, UCLA Anderson School of Management, 110Westwood Plaza, Suite B420, 90095, Los Angeles, CA,United States, [email protected], Jérémie GallienUntil 2007, Zara used a manual process for determining price mark-downs. This is a challenging problem due to the variety of unsoldarticles with few price data points. In collaboration with Zara’s pricingteam, we developed an alternative process relying on a formal forecast-ing model feeding a price optimization model. As part of a controlledfield experiment conducted in Belgium and Ireland during the 2008Fall-Winter season, this new process increased clearance revenues byapproximately 6%. Zara is currently using this process worldwide forits markdown decisions during clearance sales.

2 - Scheduling the German Basketball LeagueStephan Westphal, Institute for Numerical and AppliedMathematics, University of Goettingen, Lotzestr. 16-18,37083, Göttingen, Germany,[email protected] season, the German First Division Basketball League (BekoBBL) is faced with the problem of scheduling the games among theirteams. In addition to the typical constraints that arise in creating aschedule, the league is faced with several special constraints. As theclassical schedules and models applied so far have caused numerousproblems, resulting in postponed games, lost ticket sales and unbal-anced standings, we developed a new approach taking all these re-quirements into account. This approach was first used by the BBLto generate the schedule for the season 2011/2012.

3 - Optimization of advertisement revenue for theFrench TV group TF1Thierry Benoist, e-lab, Bouygues SA, 32 avenue Hoche,75008, PARIS, France, [email protected], FrédéricGardi, Antoine JeanjeanBouygues’operations research team (the e-lab) has been working withthe leading French TV channel TF1, for the past 15 years. This talkpresents our achievements in working with this advertising broadcasterand identifies the practical keys to success in this partnership. Theselected projects, which cover the Internet and television businesses,gave TF1 a competitive advantage by allowing it to provide quickerand better answers to advertisers’requests and to better use its limitedand perishable airtime inventory. TF1 estimates the resulting revenueincrease at 20 million euros per year.

� MC-43Monday, 12:30-14:00BW-Granite

Stochastic Programming: Applications tothe energy sector

Stream: Stochastic ProgrammingInvited sessionChair: Marida Bertocchi, Department of Mathematics, Statistics,Computer Science and Applications, University of Bergamo, Viadei Caniana 2, 24127, Bergamo, BG, Italy,[email protected]: Maria Teresa Vespucci, Department of InformationTechnology and Mathematical Methods, University of Bergamo,via Marconi, 5a, 24044, Dalmine (BG), Italy, [email protected]

62


1 - Operating fully renewable power systems: biomassand wind caseRuth Dominguez, Electrical Engineering, Universidad deCastilla - La Mancha, Avda. Carlos III, s/n, 45071, Toledo,Spain, Spain, [email protected], Miguel Carrión,Antonio J. Conejo

Renewable energy sources play an increasingly significant role in cur-rent power systems. Despite of the numerous advantages of renewablesources, most of them are non-dispatchable and cannot be continuouslycontrolled by the system operator. This complicates the operation ofpower systems with large renewable integration. We propose an op-erational tool to be used in power systems including wind farms andbiomass units. Opposite to wind units, biomass plants are dispatchableand can be used to counteract the wind power uncertainty. An exampleis presented to illustrate the proposed model.

2 - A multi-stage stochastic programming model for in-vestments in natural gas infrastructureAsgeir Tomasgard, Applied economics and operationsresearch, Sintef Technology and society, Trondheim, Norway,[email protected]

We present a multistage stochastic model that evaluates investmentsin natural gas infrastructure, taking into account existing and planneddesign. The uncertainty facing the decision makers include both up-stream and downstream uncertainty, such as; reservoir volumes, thecomposition of the gas in new reservoirs, market demand and pricelevels. The model integrates analysis of long-term and short-term un-certainty.

3 - A multistage stochastic model for the electric powergeneration capacity expansion problem under differ-ent risk measures.Maria Teresa Vespucci, Department of InformationTechnology and Mathematical Methods, University ofBergamo, via Marconi, 5a, 24044, Dalmine (BG), Italy,[email protected], Marida Bertocchi, Laureano FernandoEscudero, Stefano Zigrino

We consider the electric power generation capacity expansion prob-lem, over a long horizon of a price-taker power producer, who hasto choose among thermal power plants and power plants using renew-able energy sources (RES), while considering regulatory constraints onCO2 emissions, incentives to generation from RES and risk. The max-imization of the expected profit will be subject, alternatively, to CVaR,to first-order stochastic dominance constraints (SDC), and to second-order SDC. Results of a computational comparison between the aboverisk measures will be reported.

� MC-44Monday, 12:30-14:00BW-Marble

Incorporating uncertainty in EnergyManagement Optimization Problems: Along and winding road

Stream: Stochastic ProgrammingInvited sessionChair: Sandrine Charousset-brignol, OSIRIS, EDF, 1 Avenue duGénéral de Gaulle, 92141, CLAMART Cedex, France,[email protected]: Wim van Ackooij, OSIRIS, EDF R&D, 1 Avenue duGeneral de Gaulle, 92141, Clamart, France,[email protected]

1 - Incorporating uncertainty in large scale Energy Man-agement Optimization Problems in practice.Sandrine Charousset-brignol, OSIRIS, EDF, 1 Avenue duGénéral de Gaulle, 92141, CLAMART Cedex, France,[email protected]

Uncertainty plays a key role in Energy management optimization prob-lems. The difficulty of incorporating it in decision making tools residesin the intractability of fully general stochastic optimization problems.Even stochastic dynamic programming struggles for realistically sizedmodels. For this reason many heuristics are used in practice. In thistalk, we will discuss some of these and emphasize the importance ofproperly integrating uncertainty.

2 - Towards a Multi-stage Robust Formulation for theNuclear Reactor Outage Scheduling ProblemNicolas Dupin, univ Bordeaux1 - EDF R&D, 33405, Talence,France, [email protected], MarcPorcheron, Pascale Bendotti

Each month EDF staff computes nuclear reactor outage dates and refu-eling quantities over a multi-year forecast. The schedule has to complywith various constraints. Problem data is subject to uncertainty in gen-eration capacity as well as outage duration, and monthly updates leadto uncontrolled changes in the forecast not compatible with stability re-quirements. A multi-stage robust formulation with "adaptive" recourseon dates is proposed for a simplified problem involving uncertainty onoutage duration.

3 - Modeling the residential electrical load curve: astochastic algorithm for realistically starting the do-mestic appliancesArnaud Grandjean, Enerbat, EDF R&D, Site des Renardières,Avenue des Renardières, Bâtiment W 49-51, 77818, Moret surLoing, France, [email protected]

The electrical load curve is an essential tool in the electricity supplyfield. The power demand curve of the domestic sector is particularlyinteresting because a series of technical and behavioural breaks are go-ing to impact on it in a near future. In order to enhance the knowledgeof the residential load curve, we built a bottom-up, technically explicitreconstitution model. One of its key points is the way electrical end-uses are launched within the day. We propose a new stochastic algo-rithm that realistically starts the domestic appliances without "simply’reproducing metered data.

� MC-45Monday, 12:30-14:00BW-Water

Management accounting andmanagement control 1

Stream: Simulation in Management Accounting andManagement ControlInvited sessionChair: Stephan Leitner, Dept. for Controlling and StrategicManagement, Alpen-Adria-Universität Klagenfurt,Universitätsstrasse 65-67, 9020, Klagenfurt, Austria,[email protected]

1 - Agent-based simulation analysis of performancemeasurement systems considering uncertainties ofa learning modelYusuke Goto, Dept. of Software & Information Science, IwatePrefectural University, 152-52, Sugo, 020-0193, Takizawa,Iwate, Japan, [email protected], Shingo Takahashi

The agents’ learning model is considered highly contingent on orga-nizational culture, various environmental factors, individual character-istics, etc. It is difficult to specify the agents’ learning model by anempirical investigation into the real organization to be modeled, whilethe agents’ learning model is an essential module of the agent-basedmodel to analyze the effectiveness of performance measurement sys-tems (PMSs) under a given situation. We conduct scenario analysisof PMSs considering uncertainties of agents’ learning model. Severallearning models are to be analyzed.

63

MD-01 EURO 25 - Vilnius 2012

2 - A computational tool for performing production andwarehouse budgets in an environment with uncer-tainty and seasonalityJairo Coronado-Hernandez, Ingeniería Industrial, UniversidadTecnológica de Bolívar, 130001, Cartagena de indias, Bolívar,Colombia, [email protected], José P.Garcia-Sabater

Our problem is framed in a multi-product, -period and a limited pro-duction and storage capacity. We intend that the generated budget isrobust to fluctuations and seasonality of demand subject to the limita-tions of system and requirements. We propose a method of 3 stages:i. the demand is time-disaggregated; ii. objective stock levels are cal-culated with a fuzzy linear programming model with the aim of pro-viding a long-term vision to the third stage models; iii. developing thebudget structured in a set of scenarios as a consequence of a simula-tion/optimization in rolling horizon.

3 - Simulation comparison of relative and absolute op-erators for the inference of criteria weights from dis-cordance related informationAndrej Bregar, Informatika, Vetrinjska ulica 2, 2000, Maribor,Slovenia, [email protected]

Several methods/operators exhibiting various levels of relativeness areintroduced for the automatic derivation of criteria weights according toselective effects of veto. They are applicable to outranking and MAUTbased methods, whereby in the case of MAUT a special criterion typeis proposed to allow for veto modeling in the form of global discor-dance functions. Absolute and relative measurements are studied witha simulation based experiment, which considers sensitivity to inputs,rank reversals, richness of information and ability to discriminate con-flicting alternatives.

4 - Financial structure optimization by using an adaptedgoal programming methodTunjo Peric, Bakeries Sunce, Komedini 1, 10090, Zagreb,Croatia, [email protected], Zoran Babic

This paper proposes an adapted goal programming method for solv-ing multicriteria fractional linear programming models. The proposedmethodology is tested on the example of company’s financial struc-ture optimization. The obtained results indicate the possibility of effi-cient application of the proposed methodology for company’s financialstructure optimization as well as for solving other multicriteria frac-tional programming problems.

Monday, 14:30-16:00

� MD-01Monday, 14:30-16:00RB-Alfa

Tutorial Lecture: Professor KennethSörensenStream: Keynote, Tutorial and Plenary LecturesTutorial sessionChair: Marc Sevaux, Lab-STICC - CNRS UMR 3192, Universitéde Bretagne Sud - UEB, Centre de Recherche - BP 92116, 2 rue deSaint Maude, 56321, Lorient, France, [email protected]

1 - Metaheuristics - the metaphor exposedKenneth Sörensen, Faculty of Applied Economics, Universityof Antwerp, Prinsstraat 13, 2000, Antwerpen, Belgium,[email protected] recent years, the field of combinatorial optimization has witnesseda true tsunami of "new" metaheuristic methods, most of them basedon a metaphor of some natural process. The behavior of virtually anyspecies of insects, the flow of water, musicians playing together, itseems that no ideas are too far-fetched to serve as inspiration to launchyet another metaheuristic. In this tutorial, we will critically investigatewhether the resulting methods are truly original, or just a repackagingof old ideas. In general, the usefulness of metaphor-based methodswill be put to the test.On the other side of the spectrum, researchers in metaheuristics arepicking up ideas traditionally found in exact methods only, such as anintelligent decomposition of the problem and the use of exact methodsto solve subproblems. In some problem domains such as vehicle rout-ing or scheduling, a consensus is starting to condense on which ideasand which methods work and which do not. This is the result of manyyears of development of ever more powerful metaheuristics, combinedwith a careful study of the combinatorial properties of the problem.Additionally, deconstructing of a method, combined with statisticaltesting of the components and parameters of a metaheuristic can re-veal those components and parameter settings that truly contribute tothe performance.The main aim of this talk is to provide some guidelines on how to de-velop an efficient and effective metaheuristic. We will show that theway forward is through the rigorous use of scientific methods, ratherthan through the indiscriminate introduction of yet another metaphor.

� MD-02Monday, 14:30-16:00RB-Beta

Scheduling models in Logistics

Stream: SchedulingInvited sessionChair: Dirk Briskorn, University of Siegen, Germany,[email protected]

1 - Determining crane areas for balancing workloadamong interfering and non-interfering cranesNils Boysen, Lehrstuhl für ABWL/ Operations Management,Friedrich-Schiller-Universität Jena, Carl-Zeiß-Str. 3, 07743,Jena, Germany, [email protected], Malte Fliedner,Simon EmdeThis paper treats the problem of subdividing an area for storing con-tainers, such that the workload is evenly shared among the cranes op-erating the resulting subareas. We consider two crane sets: while non-crossing constraints between cranes of the same set need to be ob-served, cranes of different sets do not interfere. This problem is, forinstance, relevant for scheduling the (un-)loading of vessels by parallelquay cranes operating on opposing berths. We formalize the resultingoptimization problem, prove computational complexity, and presentexact and heuristic solution procedures.

64

EURO 25 - Vilnius 2012 MD-04

2 - Packing chained items into bins with applications tocontainer handling and project schedulingMalte Fliedner, TU Darmstadt, Hochschulstrasse 1, 64289,Darmstadt, Hessen, Germany, [email protected],Dirk Briskorn

In this work we study a variant of the well-known bin packing prob-lem where items are subject to chaining constraints. The problem isencountered in rail container terminals, but can also be seen as a basicscheduling problem under partially renewable resources. We providecomplexity results for its decision version and develop exact solutionapproaches with (pseudo-)polynomial runtime for specific subprob-lems. We further investigate two optimization versions of the problemand provide efficient approximation algorithms.

3 - A dynamic programming approach to the aircraftlanding problem with aircraft classesAlexander Lieder, Chair of production management,University of Mannheim, Schloss, Schneckenhof Ost, 68131,Mannheim, Germany, [email protected], RaikStolletz, Dirk Briskorn

The static aircraft landing problem is to find a minimum-cost scheduleof aircraft landings while assuring that the necessary separation timebetween two landings is kept. By assuming a limited number of aircraftclasses with common separation times and cost functions, the problemcan be solved to optimality in polynomial time. We present a dynamicprogramming approach which is able to solve medium-sized probleminstances (30-50 aircraft), that take hours to be solved by a MIP solvercould be solved within seconds. To cope with larger instances, a CPS-heuristic has been implemented.

4 - Scheduling of the batch annealing process in a fas-tener companyCeyhun Araz, Industrial Engineering Dept., Celal BayarUniversity, Turkey, [email protected], Ozgur Eski,Levent Bayoglu, Filiz Turan

This study deals with the scheduling problem for the batch anneal-ing process in the heat treatment of steel coils. It is a parallel-machinemulti-product scheduling problem with constraints on products includ-ing material type and size. The multiple objectives in scheduling thebatch annealing process are to minimize total tardiness of steel coilsand to maximize coil stacks throughput. The scheduling algorithm de-veloped was applied to the batch annealing shop in Norm FastenersCo., Turkey. The industrial application shows the effectiveness of thedeveloped scheduling algorithm.

� MD-03Monday, 14:30-16:00RB-L1

Logistics, Transportation, Traffic - 1

Stream: Logistics, Transportation, TrafficContributed sessionChair: Thierry Benoist, e-lab, Bouygues SA, 32 avenue Hoche,75008, PARIS, France, [email protected]

1 - Two-stage vehicle routing problem with dynamic se-lection of entrepotsJuraj Pekár, Department of Operations Research andEconometrics, University of Economics, Dolnozemska 1,85235, Bratislava, Slovakia, [email protected], Ivan Brezina,Zuzana Cicková

The paper presents two-stage vehicle routing problem with dynamicselection of entrepots. The goal is to carry the goods from a centralstore to the entrepots with the use of different types of vehicles as fromentrepots to final customers. The problem could not be decomposedinto two separate one-stage problems, since the selection of entrepotsis realized dynamically from the selected set of nodes. The object is toconstruct a mathematical model, so that the total costs of delivery ofgoods were as low as possible.

2 - The Line Traveling Salesman with partial orderingThierry Benoist, e-lab, Bouygues SA, 32 avenue Hoche,75008, PARIS, France, [email protected], AntoineJeanjean, Vincent Jost

The Line Traveling Salesman with Partial Order is a variant of theTraveling Salesman Problem, where the cities to be visited lie on aline, the distance between two cities is the absolute difference betweentheir abscissae and a partial ordering is given on the set of cities. Weprove that this problem, encountered on linear construction schemes,is NP-complete. Using key dominance properties and lower bounds,we design a call-based dynamic program able to solve instances withup to 450 cities.

3 - Degrees of freedom in calculating inventory-carryingcosts: a simulation studyChristoph Siepermann, Faculty of Mathematics andEconomics, Ulm University, Helmholtzstr. 22, 89081, Ulm,Germany, [email protected]

The determination of inventory-carrying costs causes serious problemswhich result amongst others from different ways of calculating the av-erage capital tied up in storage. The paper tries to estimate, by meansof a simulation study, the extent of differences between the variousways of determining the average value of inventory as they can oc-cur in practice. It will be shown that, under certain circumstances, thedifferent formulas for calculating the average value of inventory canlead to quite different results and therefore to degrees of freedom indetermining inventory-carrying costs.

4 - The Target Visitation ProblemAchim Hildenbrandt, Computer Science,Ruprecht-Karls-University Heidelberg, Im Neuenheimer Feld368, Room 104, 69120, Heidelberg, Baden-Würtenberg,Germany, [email protected],Gerhard Reinelt

The target visitation problem (TVP) is concerned with finding a routeto visit a set of targets starting from and returning to some base. Inaddition to the distance travelled a tour is evaluated also by taking intoaccount preferences which address the sequence in which the targetsare visited. The problem is thus a combination of two well known op-timization problems: the travelling salesman and the linear orderingproblem. In our talk we point out some properties of the complex butnevertheless interesting polyhedral structure and show approaches tosolve the TVP to optimality.

� MD-04Monday, 14:30-16:00RB-L2

Optimal Control II

Stream: Optimal ControlInvited sessionChair: Gernot Tragler, OR and Control Systems, ViennaUniversity of Technology, Argentinierstr. 8/105-4, A-1040,Vienna, Austria, [email protected]

1 - Normality of the maximum principle for the Bolzaproblem under state and end point constraintsDaniela Tonon, Institut de mathématiques de Jussieu,Universite Pierre et Marie Curie Paris 6, 4 Place Jussieu,75252, Paris, France, [email protected], HeleneFrankowska

It is well known that every strong local minimizer of the Bolza problemunder state constraints satisfies a constrained maximum principle. Itsnormality is very useful for investigating the Lipschitz continuity of lo-cal minimizers, for regularity of the adjoint state and optimal controlsand for derivation of higher order optimality conditions. We presentsufficient conditions for normality of optimal trajectories in the pres-ence of general state and end point constraints. This is accomplishedthanks to a new inward pointing condition applicable to unboundedcontrol systems.

65


2 - Network based computing environment for solvingoptimal control problemsRadoslaw Pytlak, Mechatronics, Institute of AutomaticControl and Robotics, Warsaw University, of Technology,02-525, Warsaw, Poland, [email protected], TomaszTarnawski, Mariusz Kaleta, Tomasz SliwinskiThe aim of the presentation is to show the current state of work on com-puting environment IDOS (Interactive Dynamic Optimization Server)which enables solving optimal control problems by using Internet ser-vices. The computing environment is equipped with dynamic opti-mization modeling language — DOML which is an extension of Mod-elica language. The presentation will focus on optimization solversfor optimal control problems with integer valued controls; elements ofDOML language; mechanism of adding new solvers to the environ-ment. Also the use of the environment IDOS will be discussed.

3 - Dynamic Programming Approach to Optimal ControlProblems with Pure State ConstraintsDaniel Hoehener, Institut de Mathématiques de Jussieu,Université Pierre et Marie Curie, 4 place Jussieu, 75005,Paris, France, [email protected], Giancarlo FacchiWe consider an optimal control problem with pure state constraints. Itis well-known that the dynamic programming principle allows to provethat the value function is a generalized solution of the HJB equation.By using viability techniques, we show that the value function is ac-tually the unique solution of the HJB equation, even if the dynamicsare discontinuous in the time variable. The main novelty of the proofis an approximation result of feasible trajectories by trajectories lyingin the interior of the constraints. Such approximations have also otherapplications in control theory.

4 - Determining the optimal strategies for stochasticcontrol problem with finite time horizonDmitrii Lozovanu, Institute of Mathematics and ComputerScience, Academy of Sciences of Moldova, Academiei 5,IMI, MD-2020, Chisinau, Moldova, Moldova, Republic Of,[email protected] stochastic version of classical discrete optimal control problemwith finite set of states and finite time horizon is formulated and stud-ied. In the considered problem we assume that the dynamic systemmay admit states in which the vector of control parameters is changedrandomly and we are seeking for a strategy that minimize the expectedtotal cost during a given number of state transitions. We show thatthe optimal solution of the stochastic problem can be found using thebackward dynamic programming method.

� MD-05Monday, 14:30-16:00RB-L3


Stream: Maritime TransportationInvited sessionChair: Henrik Andersson, Department of Industrial Economicsand Technology Management, Norwegian University of Scienceand Technology, Gløshaugen, Alfred Getz vei 3, NO-7491,Trondheim, Norway, [email protected]

1 - Maritime transport as an important link in the globalsupply chainAlgirdas Šakalys, Competence Centre of IntermodalTransport and Logistics, Vilnius Gediminas TechnicalUniversity, Plytins st. 27, LT-10105, Vilnius, Lithuania,[email protected] East West Transport Corridor, connecting northern Europe withRussia and Far East is an excellent platform to develop an innova-tive pilot testing ground where modern technology and informationsystems contribute to increased efficiency, traffic safety and securityas well as reduced environment impact in the corridor. The corridorstands out as green transport corridor which will reflect an integratedtransport concept where Short See Shipping, rail, inland water waysand road complement each other to enable the choice of environmen-tally friendly transport.

2 - Heuristics for Stochastic and Dynamic MaritimePickup and Delivery ProblemsLars Magnus Hvattum, Dept of Industrial Economics andTechnology Management, Norwegian University of Scienceand Technology, Alfred Getz veg 3, Sentralbygg 1, N-7491,Trondheim, Norway, [email protected], GregorioTirado, Kjetil Fagerholt, Jean-François Cordeau

Recent years have shown that static and deterministic vehicle routingproblems can be solved to near optimality with acceptable compu-tational times using metaheuristics. However, many real world ap-plications are dynamic and include stochastic aspects, such as un-known future customer requests. Explicitly taking into account avail-able stochastic information may yield benefits in these cases. Here,three different heuristics are considered and evaluated in terms of theirability to minimize transportation costs in a dynamic and stochasticmaritime planning problem.

3 - Modeling ballast water in container stowage plan-ningDario Pacino, IT-University of Copenhagen, Rued LangaardVej 7, 2300, Copenhagen, Denmark, [email protected], AlbertoDelgado, Rune Jensen

Automated stowage planning has come to the attention of the industry.Published work in this area assumes constant displacement of the ves-sels, which simplifies the linearization of the hydrostatic calculation.In those models ballast tanks are not taken into account. Since ballastwater can contribute up to 25% of the total displacement, ignoring itcould become a great source of error. We present a model for stowageplanning with ballast tanks and show how to handle the non-linearitiesthat variable displacement forces on the model. Preliminary resultsshow the feasibility of this approach.

4 - Fleet deployment with speed optimizationHenrik Andersson, Department of Industrial Economics andTechnology Management, Norwegian University of Scienceand Technology, Gløshaugen, Alfred Getz vei 3, NO-7491,Trondheim, Norway, [email protected], KjetilFagerholt, Kirsti Hobbesland

We will present a rolling horizon heuristic for a fleet deployment prob-lem with speed optimization. The heuristic is tested on real world in-stances from one of the largest car transporters of the world and showsgood results.

� MD-06Monday, 14:30-16:00RB-Gamma

Vehicle routing problems

Stream: MetaheuristicsInvited sessionChair: Cédric Verbeeck, Industrial management, Ghent University,Technologiepark 903, B-9052, Zwijnaarde, Belgium,[email protected]

1 - A local search heuristic algorithm for the CVRP withopen routesR. Aykut Arapoglu, Industrial Engineering Department,Eskisehir Osmangazi University, Meselik Campus, 26480,Eskisehir, Turkey, [email protected], Abdurrahman Yildiz

We present a variant of the classical capacitated VRP that allows closedroutes ending at the depot as well as open routes that end at customerdemand node(s). The problem arises when the company’s current fleetof vehicles cannot meet the total demand. Closed routes are operatedby the vehicles of existing fleet whereas open routes are run by ex-tra vehicles. The objective is to find a route for each vehicle used,minimizing the total distance traveled. A local search improvementalgorithm is implemented. The performance of the algorithm is testedon data sets adapted from the literature.

66


2 - Comparing metaheuristics for the Time-DependentOrienteering ProblemCédric Verbeeck, Industrial management, Ghent University,Technologiepark 903, B-9052, Zwijnaarde, Belgium,[email protected], Pieter VansteenwegenThis research focuses on developing and comparing efficient solutionmethods to solve time-dependent orienteering problems (TD-OP) inreal time. A time dependent local search move and local evaluationmetric have been successfully implemented together with three basicmetaheuristic frameworks. Test instances were developed in a way tosimulate real life congestions patterns. Comparing the results provesthat obtaining solutions in real-time is feasible. Currently, instancesare being developed so that the optimal solution is known, providingmore insight in the obtained results.

3 - A Tabu Search Method for a Cement Delivery Prob-lemMarc Uldry, Département d’Informatique, Université deFribourg, Boulevard de Pérolles 90, 1700, Fribourg,Switzerland, [email protected], Alain Hertz, MarinoWidmerWe consider a cement delivery problem with a heterogeneous fleet ofvehicles and several depots. The orders are often larger than the vehiclecapacity; consequently, most customers are visited several times. Thisis a split delivery vehicle routing problem with additional constraints.A Tabu Search resolution method is first described. Experiments onreal life instances are then performed. Comparisons between these re-sults, those obtained using a MILP formulation and those given by thecement supplier company are also presented. Tabu Search producesbetter results in a few minutes.

4 - A multi-objective inventory routing problem for sus-tainable waste management under uncertaintyPamela Nolz, Department of Information Systems andOperations, Vienna University of Economics and Business,Nordbergstrasse 15, 1090, Vienna, Austria,[email protected], Nabil Absi, Dominique FeilletWe formulate a stochastic inventory routing problem for infectiousmedical waste collection minimizing two criteria: (i) distribution costscomprising routing costs and fixed costs incurred each time a touris planned, (ii) expected inventory costs reflecting social objectives,specifically the satisfaction of pharmacists and patients, as well asthe minimization of public health risks. We develop a hybrid multi-objective solution approach integrating an adaptive large neighborhoodsearch into an evolutionary algorithm. Computational experiments areperformed for real-world problem instances.

� MD-08Monday, 14:30-16:00RB-Epsilon


Stream: Network OptimizationInvited sessionChair: Axel Werner, Optimization, Zuse Institut Berlin (ZIB),Takustr 7, D-14195, Berlin-Dahlem, Germany, [email protected]

1 - Bundling Telecommunication Services with NetworkCostsJuan Perez, Industrial Engineering, Universidad de Los Andesde Chile, Av. San Carlos de Apoquindo 2200, Comuna lasCondes, Santiago, Región Metropolitana, Chile,[email protected], Alejandro JofreWe propose a non-linear model for bundling telecommunication ser-vices explicitly considering the costs and capacities of the network.We consider bundles composed of: mobile telephone and internet. Thedecision variables are the fixed prices for bundles and the quantitiesof each service. We analyze the following aspects: users that have al-ready have a bundle with a company, the effect of considering a cost forusers when they buy a new bundle, the optimum time between offers,and that targeting segments of users not only depends on it reservationprices but also on the cost function.

2 - The effects of inaccurate information on the queuejoining policyDavid Raz, Management of Technology, Holon Institute ofTechnology, Holon, Israel, [email protected]

The efficiency aspects of the queue joining policy were thoroughlystudied, starting probably with Winston (1977). It is widely acceptedthat the Shortest Queue (SQ) policy is optimal, with some counterex-amples noted by Whitt in 1986. However, the body of research as-sumes information is fully available and accurate, which is not alwaysthe case. In this work we study the effects of inaccurate or missinginformation on the efficiency and fairness of the queue joining policy.The results show that SQ may not always be the optimal policy to use,if the information is incomplete.

3 - Free Riding Solution for Hybrid P2P NetworksAmad Mourad, LAMOS, Bejaia University, Bejaia, Bejaia,Algeria, [email protected], Djamil Aïssani, MoumenHamouma

Free riding is an important problem in P2P networks in which nodestake without giving, they benefit from shared resources without shar-ing their own. The objective of our study is to ensure the collectiveparticipation of P2P members, by encouraging them to be more active.For this, we introduce a new solution against free riders; it is based onfriend’s lists in each peer, adding only peers that share interesting con-tents. A queuing mechanism with priority is also used at each superpeer. This priority is calculated based on the number of peer’s friendsand their free times.

4 - Optimizing energy consumption of telecommunica-tion networksAxel Werner, Optimization, Zuse Institut Berlin (ZIB),Takustr 7, D-14195, Berlin-Dahlem, Germany, [email protected]

Already today, a considerable part of energy worldwide is spent onthe operation of telecommunication networks. With steadily grow-ing internet traffic this share will further increase, unless energy con-sumption of future networks is optimized and adapted to switched traf-fic instead of provided capacity. We present mixed integer program-ming models for the optimization of multi-layer telecommunicationnetworks using load-adaptive equipment and dynamic energy manage-ment. We discuss preliminary results of computations, based on sndlibinstances, as well as networks by Deutsche Telekom.

� MD-09Monday, 14:30-16:00RB-Zeta


Stream: Cutting and PackingInvited sessionChair: Ozgur Kulak, Çorlu Engineering Faculty, Namık KemalUniversity, N.K.Ü. Çorlu Mühendislik Fakültesi, SilahtarMahallesi, Sinan Dede Mevkii Çerkezköy Yolu 3. km, Çorlu,59860, Tekirdag, Turkey, [email protected]

1 - Linearization of Non-oriented Polygonal PlacementOptimization ProblemMarina Novozhilova, Department of Economic Cyberneticsand Information Techologies, Kharkov National University ofConstruction and Architecture, 61046, Kharkov, Ukraine,[email protected], Igor Chub

The non-oriented convex polygon placement optimization problem isconsidered in the linearized statement. The additional properties of theproblem linearized feasible region are studied. The method of search-ing a problem local minimum based on the active set idea is proposed.The scheme for determining the descent direction along the frontier ofthe connected component of the linearized feasible region has been de-veloped. A new technique for constructing inequality constraints sys-tem for current solution on each iteration is considered. The stoppingcriterion of iterative process is defined.

67


2 - A Circle Covering approach to the Nesting problemPedro Rocha, INESC TEC, Faculdade de Engenharia,Universidade do Porto, 4200-465, Porto, Portugal,[email protected], Rui Rodrigues, A. Miguel Gomes

Circle Covering (CC) problems aim to completely cover a region withthe minimum number of circles. With an efficient coverage, the regioncan be accurately represented by the set of circles. In Nesting prob-lems, pieces need efficient collision detection for their placement. Wepropose to solve Nesting problems with an approach based on circlecovering to represent irregular shapes and non-linear models to com-pute placement positions.

3 - Packing spherical caps in high dimensional spacesCristiano Torezzan, School of Applied Sciences, University ofCampinas, Brazil, [email protected], SueliCosta

We consider the problem of packing a large number of equal spher-ical caps on the surface of the unit sphere. This problem is closelyrelated to the kissing number problem and applications have appearedin different areas. Although good approaches are known, most of thembecome infeasible when either the number of caps or the dimensiongrows. We present a new modeling which allows to reduce the prob-lem to the one of packing circles in a box in the half of the dimension.Several examples are presented, some of them are the biggest knownconstructive solutions in high dimensions (e.g. 16, 24).

4 - A comparative study of 2D irregular object geome-try modelling solutions for cutting and packing prob-lemsOzgur Kulak, Çorlu Engineering Faculty, Namık KemalUniversity, N.K.Ü. Çorlu Mühendislik Fakültesi, SilahtarMahallesi, Sinan Dede Mevkii Çerkezköy Yolu 3. km, Çorlu,59860, Tekirdag, Turkey, [email protected]

Industrially applicable packing and cutting solutions need to achieveboth a tightly fitting irregular object geometry modeling and a fastplacement optimization by computationally effective intersection,translation and rotation operations. Modeling scalability is also im-portant for different approximation requirements of concurrent opti-mization algorithms. In this study, classical object boundary geometrymodeling and alternative quad tree modeling technique are comparedwith respect to representation quality, scalability, processing time perintersection, rotation and translation operation.

� MD-10Monday, 14:30-16:00RB-Theta

Inventory Management 1

Stream: Production Management & Supply ChainManagement (contributed)Contributed sessionChair: Amin Chaabane, Departement of AutomatedManufacturing Engineering, École de Technologie Supérieure,1100, Notre Dame Street Ouest, H3C1K3, Montreal, Quebec,Canada, [email protected]: Marthy Stívaliz García Alvarado, Departement ofAutomated Manufacturing Engineering, École de technologiesupérieure, 1100, rue Notre-Dame Ouest, H3C 1K3, Montreal,Quebec, Canada, [email protected]

1 - A Serial Inventory System with Information Ex-changeNima Yazdan Shenas, University of Science and Culture, Iran,Islamic Republic Of, [email protected]

A supply chain model consisting of one product, a supplier and a re-tailer is considered. Demands at the retailer are generated by a Poissonprocess. Not covered demands are backordered. The retailer carriesinventory and replenishes stock according to a (Q, R) policy. Basedon available information, the supplier’s policy is defined. Recently, theoptimal policy was obtained, where order size (Q) is given. We extend

the exact cost function and develop an algorithm to find the optimalpolicy, when order size is as a control parameter. We show that ouralgorithm is faster than the previous one.

2 - A single-period inventory placement problem for asupply system with the satisficing objectivePiotr Stalinski, Department of Quantitative Methods inManagement, WSB-NLU, ul. Zielona 27, 33-300, NowySacz, Poland, [email protected], Chia-Shin Chung,James Flynn, Roelof Kuik

Consider the inventory placement problem in an N-stage supply systemfacing a stochastic demand for a single planning period. Each stage isa stocking point holding some form of inventory that after a suitabletransformation can satisfy demand. Stocking decisions are made be-fore demand occurs. The objective is to maximize the probability ofachieving a given profit level. We prove the existence of optimal stock-ing decisions that utilize at most three stages in the supply system. Wecharacterize properties of the optimal stocking decisions and providean efficient algorithm for computing them.

3 - An inventory model with partial backorders assum-ing uniform demandJoaquín Sicilia, Dpt. de Estadística, Investigación Operativa yComputación, Universidad de La Laguna., Facultad deMatemáticas, Avda. Astrofísico Francisco Sánchez s/n,38271, La Laguna. Tenerife., Canary Islands, Spain,[email protected], Valentín Pando, Luis A. San-José, JuanGarcia-Laguna

In this work a stochastic inventory problem with additional order isstudied. If the system is out of stock at the selling period, the vendororders an extra lot size to satisfy a percentage of unmet demand. Weassume the backorder rate depends on the amount of shortage and thefraction of demand served with delay is described by a co-sinusoidalnon-increasing function of the unmet demand. Our objective consistsof determining the optimal order size that maximizes the expected totalprofit when demand is uniformly distributed. Sensitivity analysis andnumerical examples are also presented.

4 - Inventory Control Under The Carbon Emission Trad-ing SchemeMarthy Stívaliz García Alvarado, Departement of AutomatedManufacturing Engineering, École de technologie supérieure,1100, rue Notre-Dame Ouest, H3C 1K3, Montreal, Quebec,Canada, [email protected],Marc Paquet, Amin Chaabane

Inventories have proven to be crucial in logistics. While new trendsforce the integration of sustainable development into supply chainoperations, the carbon emission trading scheme seems to be a goodway to achieve sustainable development efficiently. This integrationhas raised new questions related to how inventory methods should beadapted to include carbon trading. In this talk will discuss the chal-lenges of inventory modeling. Moreover, through a stochastic inven-tory model, we show the direct link between the sustainability objec-tive and inventory control policies.

� MD-11Monday, 14:30-16:00RB-Iota

Traffic Management

Stream: Transportation and LogisticsInvited sessionChair: Livia Mannini, Dep. of Civil Engineering, Roma TreUniversity, Italy, [email protected]

1 - Traffic management and emissions estimation in awithin day dynamic frameworkSimone La Spada, Department of Civil Engineering, RomaTre University, 00146, Rome, Italy, Italy,[email protected], Marialisa Nigro, Stefano Gori, LiviaMannini

68


Reduction of vehicle emissions is related to the improve of life qualityand sustainability. To estimate emissions on large scale urban network,approaches based on aggregated traffic estimates are followed, withoutconsidering temporal variations of traffic conditions. The study wantsto develop a method to estimate pollutant emissions taking into accountthe within-day variations of traffic conditions and parameters related tothe different state of vehicles and that can easily be applied to evaluatethe results of off-line and on-line traffic management strategies in largeurban network.

2 - Implementation of variable speed systems in an ac-cess road (c 31 n) to barcelonaM. Antonia de los Santos, Mechanical Engineering, UPC,Diagonal 647, 08028, Barcelona, Spain,[email protected], David Gallegos, Francisco Liesa

The city of Barcelona was the first in Spain to introduce variable speedsystems as a strategic tool to improve traffic management in 2009, inthe roads C32 and C31S. The C31N is a main access road to Barcelonawhere these systems could be installed. The present work evaluatespotential benefits that the implementation of these intelligent systemsinvolves. In order to achieve this target, a virtual model of the road iscreated, analysing and defining the traffic flow running in the road. Tocheck global potential benefits, 3 more cases are analysed includingspecific traffic jam situations.

3 - Green timing and scheduling: an approach based ongenetic algorithmsGiulio Erberto Cantarella, Dept of Civil Engineering,University of Salerno, via Ponte Don Melillo, 1, 84084,Fisciano (SA), Italy, [email protected], Roberta Di Pace

This paper proposes an approach to traffic light optimization at a sin-gle junction through genetic algorithms. Decision variables are greentimes and their scheduling. Maximization of capacity and minimiza-tion of delay are considered as objectives, also taking into accountmaximum queue length constraints. This approach seems relevantsince it can be extended to signal synchronization with optimizationof green scheduling, differently from current practice. This research isa part of a project aiming at urban network design including optimiza-tion of traffic lights and street lane allocation.

4 - Estimation of traffic flow conditions through data fu-sion techniquesLivia Mannini, Dep. of Civil Engineering, Roma TreUniversity, Italy, [email protected], Ernesto Cipriani,Stefano Gori

Starting from the application of the procedure reported in Wang, Papa-georgiou (2005) based on the correction through the Extended KalmanFilter (EKF) of the second order traffic model, also probe vehicles mea-surements have been taken into account, added to the conventionalfixed ones. Several simulations have been carried out with Matlab.Different data fusion techniques have been analyzed, specifically, thefusion of measurements, rather than the fusion of estimations, has re-vealed to be more effective in improving EKF estimation.

� MD-12Monday, 14:30-16:00RB-Omicron

Decision-making and modelling underfuzzinessStream: Fuzzy Sets and Soft ComputingInvited sessionChair: Mario Fedrizzi, Department of Computer and ManagementSciences, Universita di Trento, Via Inama 5, 38122, Trento, Italy,[email protected]

1 - Boundary properties of inconsistency indices ingroup preference relationsMatteo Brunelli, Åbo Akademi University, Finland,[email protected], Michele Fedrizzi

Consistency of preferences is usually a desirable situation in decisionmaking processes based on preference relations, e.g. the AHP. A rel-atively large number of inconsistency indices has been proposed inliterature in order to estimate inconsistency of preference. We proposea study of inconsistency with respect to consensus where we analyzehow the inconsistency of the consensual preference relation is affectedby the inconsistency of the single decision makers. In particular, wewill reason and dwell on the positive effects of consensus on the con-sistency of preferences.

2 - Error based decision making on approximation meth-ods under fuzzinessSvetlana Asmuss, Department of Mathematics, University ofLatvia, Zellu street 8, LV-1002, Riga, Latvia,[email protected], Vecislavs Ruza

We present a methodology of decision making on approximation meth-ods under uncertainty given by fuzzy sets. We consider approxima-tion of real valued functions when a set we approximate over is fuzzy,meaning that our interest is more focused at some parts of the set. Inorder to estimate the quality of algorithms, we introduce an appropriatefuzzy analogue of a norm using a fuzzy valued integral and describe amethod of its evaluation. In this context, for approximation methodswe obtain fuzzy valued error bounds for some classes of functions anddiscuss criteria of the error optimal decision.

3 - Indirect sociomappingCyril Höschl, QED GROUP, Czech Republic,[email protected], Radvan Bahbouh

Indirect sociomapping enables comparison of people as to similarityof their data profiles. The degree of belonging to the fuzzy set of sim-ilarity is determined by the p-values for the statistical test comparingboth of the individual data profiles. Results are represented in the formof a landscape, with the distances corresponding to the degree of sim-ilarity and the heights showing the overall similarity to the other unitsin the system. This landscape can be used as an interactive interface toformulate statistical tests and visualize their results.

4 - Sociomapping of communicationRadvan Bahbouh, QED GROUP, Czech Republic,[email protected], Cyril Höschl

Sociomapping makes it possible to analyse and visualise communica-tions in teams and organizations. This method is based on representingthe communication interconnectedness through the degree of belong-ing to the fuzzy sets of communication nearness. The obtained fuzzymodels can be mutually compared; this allows capturing of stable com-munication patterns, as well as of developmental trends. Subsequently,the data are depicted in the form of a landscape. Sociomapping is usedin preparing teams for foreign military missions, in experiments simu-lating flight to Mars, and in management teams, etc.

� MD-13Monday, 14:30-16:00RB-Tau


Stream: Lot-sizing and Related TopicsInvited sessionChair: Luis Guimarães, Industrial Engineering and Management,Faculty of Engineering of Porto University, FEUP, Portugal,[email protected]

1 - The integrated lot sizing and scheduling problem inthe brewery: model and MIP-based heuristicMaristela Santos, Department of Applied Mathematics andStatistics, University of Sao Paulo, Av. TrabalhadorSão-carlense, 400 - Centro, Caixa Postal: 668, 13560-970,São Carlos, São Paulo, Brazil, [email protected], TamaraBaldo, Reinaldo Morabito, Bernardo Almada-Lobo

This paper addresses the lot-scheduling problem in the brewery in-dustry. This problem is composed of two stages: the preparation andbottling of the liquids. The first stage may span over several days due

69


to the fermentation process which complicates the synchronization ofboth stages. We propose a new mixed integer model for it consider-ing multiple tanks and bottling lines. The model differs from those ofthe literature on the first stage decisions and requirements. A MIP-based heuristic is developed to obtain solutions capable of addressingthe process and preliminary results are presented.

2 - Dynamic lot sizing problem with minimum orderquantities and remanufacturing optionIrena Okhrin, Juniorprofessur in Information & OperationsManagement, European University Viadrina, GrosseScharrnstrasse 59, 15230, Frankfurt (Oder), Germany,[email protected], Knut Richter

We focus on a special uncapacitated lot sizing problem where a man-ufacturer can meet demand by either producing new items or by re-manufacturing used items. Additionally, a minimum order quantityrestriction, instead of the setup cost, guarantees a certain level of pro-duction lots. A general model of this problem is formulated and severaluseful properties of the optimal solution structure are investigated. Fi-nally, we develop a one-dimensional dynamic programming procedurefor the originally two-dimensional dynamic model that generates theoptimal production plan for some special cases.

3 - Simultaneous production-distribution planning tocoordinate multi-plant supply chainsLuis Guimarães, Industrial Engineering and Management,Faculty of Engineering of Porto University, FEUP, Portugal,[email protected], Diego Klabjan, BernardoAlmada-Lobo

Inspired by a real case, we present a model and solution approach forproduction-distribution planning in multi-plant, multi-warehouse sup-ply chains. The model assigns and schedules production lots to a setof machines, available at each plant, determining overtime, inventoryand transportation quantities. The objective is to minimize overall costwhile satisfying a deterministic demand triggered at the warehouses.The solution approach is based on the model using VNS to fix and im-prove lots allocation guided by the sensitivity analysis of the LP solvedto determine the remaining decisions.

� MD-14Monday, 14:30-16:00RB-Omega

Scheduling with variable parameters I

Stream: Scheduling under Resource ConstraintsInvited sessionChair: Mikhail Y. Kovalyov, United Institute of InformaticsProblems, National Academy of Sciences of Belarus, Surganova 6,220012, Minsk, Belarus, [email protected]: Adam Janiak, Institute of Computer Engineering, Controland Robotics, Wrocław University of Technology, Janiszewskiego11/17, 50372, Wroclaw, Poland, [email protected]

1 - Complexity analysis of an assignment problem withcontrollable assignment costs and its implications inschedulingDvir Shabtay, Dept. of Industrial Engineering andManagement, Ben-Gurion University of the Negev, POB 653,81045, Beer Sheva, Israel, [email protected], Liron Yedidsion,Moshe Kaspi

We extend the linear sum assignment problem to the case where theassignment cost is a function of the amount of resource allocated tothe agent. A solution is defined by the assignment of agents to tasksand by a resource allocation vector. The quality of a solution is mea-sured by two criteria: the total assignment cost and the total weightedresource consumption. We address these criteria via four problem vari-ations and prove that three of them are NP-hard. In addition, we findthat our problem is equivalent to a large set of scheduling problemswhose complexity has been an open question.

2 - Parallel machine scheduling with a variable commondue windowMikhail Y. Kovalyov, United Institute of InformaticsProblems, National Academy of Sciences of Belarus,Surganova 6, 220012, Minsk, Belarus,[email protected], Adam Janiak, WladyslawJaniak, Erhan Kozan, Erwin PeschWe study problems of scheduling jobs on identical parallel machines,with a common due window to be assigned to all jobs. The objective isto minimize the sum of job earliness, job tardiness, due window loca-tion and size costs, which are are nondecreasing job independent func-tions. Solution properties are established and dynamic programmingalgorithms are derived. Polynomial algorithm is given for the singlemachine case, in which the due window size cost is a discretely con-vex or concave nondecreasing function and all the other cost functionsare linear.

3 - Structural properties of time-dependent schedulingproblems with the lp norm objectiveStanislaw Gawiejnowicz, Faculty of Mathematics andComputer Science, Adam Mickiewicz University,Umultowska 87, 61-614, Poznan, Poland,[email protected], Wieslaw KurcWe consider structural properties of single-machine time-dependentscheduling problems with linearly deteriorating jobs and the lp normobjective. We prove that the V-shape property, previously known onlyfor p=1, still holds for infinite many p>1, while the symmetry propertydoes not hold for p » 1. We show that for p>1 the problems possessa kind of matrix convexity. Finally, we express the complexity of theproblems as a function of index p of the norm.

4 - Scheduling jobs with values dependent on their com-pletion timesTomasz Krysiak, Institute of Computer Engineering, Controland Robotics, Wroclaw University of Technology,Janiszewskiego 11/17, 50-372, Wroclaw, Poland,[email protected], Adam JaniakWe consider two scheduling problems with job values and losses ofjob values (costs) dependent on job completion times. In the first prob-lem, we consider scheduling jobs with stepwise values deterioratingover time. We prove strong NP-hardness of a single processor caseand construct a pseudo-polynomial algorithm for a special case withunrelated processors. The second problem is a single processor onewith piecewise linear losses of job values increasing over time. Weprove strong NP-hardness of the problem and existence of a pseudo-polynomial algorithm for its special case.

� MD-15Monday, 14:30-16:00RB-2101


Stream: Nonlinear ProgrammingInvited sessionChair: Max Demenkov, Institute of Control Sciences, Lab 16,Russian Academy of Sciences, 65 Profsoyuznaya str., 117997,Moscow, Russian Federation, [email protected]

1 - Multistart Coupled with a Derivative-Free Filter LocalSearch for Locating Multiple SolutionsFlorbela P. Fernandes, Department of Mathematics - ESTiG,Polytechnic Institute of Bragança, Bragança, Portugal,[email protected], M. Fernanda P. Costa, Ana I. Pereira, EditeM.G.P. FernandesA multistart technique coupled with a derivative-free filter local searchalgorithm for locating all the optimal solutions of a nonconvex con-strained optimization is presented. To reach fast convergence to op-timal solutions, the local search procedure is based on descent direc-tions. The filter-set concept is introduced to handle the constraints ofthe problem. The direction vector is descent for the objective functionif the point is feasible; otherwise, it is descent for the constraint viola-tion. Numerical experiments with benchmark problems and a compar-ison with other methods are included.

70


2 - Bilevel optimization in stability analysis of nonlineardynamical systemsMax Demenkov, Institute of Control Sciences, Lab 16,Russian Academy of Sciences, 65 Profsoyuznaya str., 117997,Moscow, Russian Federation, [email protected] consider region of attraction for a dynamical system - a set of pointsfrom which system trajectories go to a stable equilibrium. One can es-timate it via the sublevel sets of a positive definite Lyapunov function,which is decreasing along any trajectory in the region. We formulatethis task as a bilevel optimization over the function parameters andsystem state variables. On the lower level we have minimization of thefunction with nonconvex equality constraint. On the upper level a sizeestimation of a sublevel set needs to be maximized. Local improve-ment algorithm is discussed.

3 - Numerical solution of nonlinear optimal control prob-lems, A comparative studyM. Navabi, Shahid Beheshti University, GC, Iran, IslamicRepublic Of, [email protected], E. MeshkinfamThe aim of this paper is to compare four different iterative numericaltechniques using in determination the optimal control law. The meth-ods are: Steepest Descent (SD), Variation of Extremals (VE), Quasi-linearization (QL) and Collocation (CL).The Maximum radius orbit transfer in a given time problem is solvedutilizing these methods and comparisons are made in the terms of ac-curacy, the rate of convergence, sensitivity to initial guess and the timeof computer processing. The results show that, CL method has theleast error, VE has the most sensitivity to initial guess.

4 - Performance Analysis of Partial Use of Local Optimi-sation Operator on Genetic Algorithm for TSPMilan Djordjevic, UP DIST, UP FAMNIT, Glagoljaska 8,Slovenia, 6000, Koper, State, Slovenia,[email protected], Andrej BrodnikWe study the influence of hybridization of a genetic algorithm witha local optimizer on instances of a TSPLIB. In tests we applied hy-bridization at various percentages of genetic algorithm iterations. Lessfrequent application of hybridization decreased the average runningtime, whereas the quality of solution on average deteriorated only from0.21% till 1.40% worse than the optimal solution. We studied at whichiterations to hybridize. We applied it at random iterations, at the initialiterations, and the ending ones, and proved where being the best.

� MD-16Monday, 14:30-16:00RB-2103

Uncertainty Analysis in Energy and WaterResourcesStream: Uncertainty and Perturbations in Optimizationand in Environmental and Energy ModellingInvited sessionChair: John Boland, School of Mathematics and Statistics,University of South Australia, Mawson Lakes Blvd., 5095,Mawson Lakes, South Australia, Australia,[email protected]

1 - Reconciling rainfall modelling on differing timescalesJohn Boland, School of Mathematics and Statistics,University of South Australia, Mawson Lakes Blvd., 5095,Mawson Lakes, South Australia, Australia,[email protected], Julia PiantadosiRainfall perform well for the time scales on which they are producedbut for other time scales, a model’s performance can be erratic. Previ-ous experience in estimating five minute variance of wind farm outputderived from modelling ten second output series (Agrawal et al 2010)has led us to be able to perform a similar exercise with modelling dailyrainfall series in a fashion that preserves the monthly variance. We de-velop an AR(3) model for daily rainfall (after multiplicative deseason-ing), and through standard transformations for stationary time series,develop estimates for variances for monthly rainfall series that matchthe observed monthly variances.

2 - Catching uncertainty of windYulia Gel, Statistics and Actuarial Science, University ofWaterloo, 200 University Ave., N2L 3G1, Waterloo, ON,Canada, [email protected] wind energy has long been recognized as one of the mostattractive options but its generation is still found to be much easierthan its integration into liberalized electricity markets due to a highlyvolatile nature of wind. In this talk we propose a new approach toprobabilistic wind power forecasting which delivers competitive fore-casts while imposing minimal requirements on observed data. This isespecially critical as lack of expertise and related costs in wind farmsite selection still remain one of the key barriers to untapping windresources in developing countries.

3 - SVM Models and Auction Protocols in Electricity Mar-ketsCarlo Lucheroni, School of Science and Technologies,University of Camerino, via M. delle Carceri 9, 62032,Camerino (MC), Italy, [email protected], Renato DeLeoneModelling of hourly electricity prices is mostly made by stochasticdifference equations, assuming a causal relation between succeedinghours. Because of the power market structure, in fact this assumptionis true only at the level of succeeding days, since un-causal autocor-relations can appear within each day, due to technical constraints en-forcement by system operators. Support Vector Machines used as au-toregressive estimators on daily price hourly profiles offer a new wayfor price modelling, able to take into account intra-day correlationswithout mixing them with inter-day correlation.

4 - Mathematical methods for rainfall modellingJulia Piantadosi, Centre for Industrial and AppliedMathematics, School of Mathematics and Statistics,University of South Australia, Mawson Lakes Campus,Mawson Lakes Boulevard, Mawson Lakes, 5095, Adelaide,South Australia, Australia, [email protected] major challenge for future management of water resources is to op-timize economic, social and environmental outcomes subject to sus-tainable usage in a changing and highly variable climate. Historicalrecords provide only one realisation and generation of realistic syn-thetic data is an important objective to assess the impact of rainfallvariability on water systems. The aim of this work is to propose newmathematical models for rainfall simulation to inform analysis of cur-rent rainfall patterns and enable better projections for a comprehensiverange of future scenarios.

� MD-17Monday, 14:30-16:00RB-2105

Further Applications of Location Analysis

Stream: Applications of Location AnalysisInvited sessionChair: Mercedes Landete, Departamento de Estadística yMatemática Aplicada, University Miguel Hernández of Elche,Avda. del Ferrocarril s/n, 03202, Elche, Alicante, Spain,[email protected]

1 - Assessing Optimal Routes in the Natural Park ofDoñana (Spain)Eva Barrena, Applied Mathematics I, University of Sevilla,Avda. Reina Mercedes s/n, 41012, Sevilla, Spain,[email protected], Francisco A. Ortega Riejos, Isabel Ternero,Miguel Angel PozoThis paper explores the compatibility between a successful manage-ment of environmental conservation and a sustainable developmentthrough rural tourism by designing optimal routes along environmen-tal units (cells) in a Natural Park. Two types of scenarios are assumed:Safaris route, where the objective is to minimize the total distance trav-elled during supervising all the cells (appropriate for ranger’s route),and Zookeeper’s route, where a representative sample of cells must bevisited (for a touristic route) (supported by Spanish/FEDER projectsMTM2009-14243 and P09-TEP-5022).

71


2 - A polyhedral study on location models with knap-sack constraintsRoberto Javier Cañavate Bernal, Métodos Cuantitativos eInformáticos, Technical University of Cartagena (UPCT),Calle Real, 3, Facultad Ciencias Empresa, 30201, Cartagena,Murcia, Spain, [email protected], Mercedes LandeteIn this paper we consider set packing location problems with knap-sack constraints. We propose to study the intersection polyhedron ofthe two well-known problems: the classical knapsack problem (KP)and the set packing problem (SP). We give conditions for some validinequalities of the KP polytope and the SP polytope to be facets forthe KP-SP problem. The obtained results also generalize polyhedralproperties of some knapsack related problems in the literature.

3 - Delocation Models for Closing and Resizing Redun-dant Branches during Bank RestructuringDiego Ruiz-Hernandez, Quantitative Methods, CUNEF, c/Serrano Anguita 9, 28004, Madrid, Spain, [email protected],David Delgado-Gomez, Laureano Fernando Escudero,Joaquin Lopez-PascualDuring restructuring processes due to mergers, banks face the prob-lem of redundant branches competing with each other for customers.As important as it is, the problem of closing down branches has notbeen fully addressed in the literature. In this paper we introduce a newbranch closing and resizing model based on the capacitated facilitylocation problem. The model considers closing down and costs, andaddresses the problem of resizing open branches to accommodate cus-tomers displaced from closed branches. We motivate the problem withan example from the Spanish saving banks sector.

4 - Different ways of locating spanning trees in a net-workMercedes Landete, Departamento de Estadística yMatemática Aplicada, University Miguel Hernández of Elche,Avda. del Ferrocarril s/n, 03202, Elche, Alicante, Spain,[email protected], Alfredo MarínIn this work we present different formulations for the ordered mini-mum spanning tree problem. We propose one formulation based onthe flow variables and another based on the Miller-Tucker-Zemlin subtour elimination constraints. We distinguish between positive and neg-ative values for the elements of the weight function. Finally, we de-velop a computational study for comparing both formulations. It il-lustrates how much the difficulty of the problems differs depending onthe weight function. The Conclusions section gives some advises forchoosing the best formulation at each case.

� MD-18Monday, 14:30-16:00RB-2107

Analysis of Price Dynamics and HerdBehaviorStream: Financial Modelling & Risk Management (con-tributed)Contributed sessionChair: Frederik Druart, MOSI, Vrije Universiteit Brussel,Twaalfmaandenstraat 9/12, 2000, Antwerpen, Belgium,[email protected]

1 - Stock price dynamics and distribution of returns —before, during and after the financial crisisTibor Kis, Quantitative Methods in Economy, Faculty ofEconomics, Segedinski put 9-11, 24000, Subotica, Serbia,[email protected], Marija Cileg, Gabriela VickoIn this paper we study the characteristics of return distributions for anumber of stocks between 1995-2011. The analysis relates to separatesub periods distinguishing the year(s) of the late-2000s financial cri-sis searching for a shift in skewness, kurtosis and autoregressive char-acteristics of return distributions in sub periods and between differ-ent stocks, particular sectors and stock exchanges. Return distributioncharacteristics are investigated on different time levels: daily, weeklyand monthly.

2 - Herding behavior of agents as a background of finan-cial fluctuationsAleksejus Kononovicius, Institute of Theoretical Physics andAstronomy, Vilnius University, Lithuania,[email protected], Vygintas Gontis,Bronislovas KaulakysWe focus on the universal nature of herding behavior leading to themacroscopic description by nonlinear stochastic differential equationsapplicable to the finance and other complex systems. Approach withtwo groups of agents appears to be too crude to reproduce the detailsof financial markets. Thus we extend Kirman’s model to include threegroups of interacting agents. This improvement provides more op-portunities to reproduce sophisticated power law statistics of return infinancial markets.

3 - Bayesian statistical analysis of herding behavious:an application to the Spanish equity mutual fundsPilar Gargallo, Facultad de Economía y Empresa, Universidadde Zaragoza, Gran Vía 2, 50005, Zaragoza, Spain,[email protected], Laura Andreu, José Luis Sarto, ManuelSalvadorThis paper proposes a Bayesian rolling window estimation procedureapplied to the three factor model of Fama and French (1993) to anal-yse the herding behaviour in the style exposures of mutual funds. Thisprocedure let dynamically select the length of the estimation windowby means of weighted likelihood functions that discount the loss ofinformation due to time. The method is very flexible and allows usto consider different approaches to detect herding behaviour by takinginto account the uncertainty associated to the estimation of the stylecoefficients.

� MD-19Monday, 14:30-16:00RB-2111

Financial Decision Making: Application ofMathematical Optimization

Stream: Operational Research and Quantitative Mod-els in BankingInvited sessionChair: Katsumasa Nishide, International Graduate School ofSocial Sciences, Yokohama National University, 79-4 Tokiwadai,Hodogaya-ku, 2408501, Yokohama, Japan, [email protected]

1 - Competition and the Bad News Principle in a RealOptions FrameworkKatsumasa Nishide, International Graduate School of SocialSciences, Yokohama National University, 79-4 Tokiwadai,Hodogaya-ku, 2408501, Yokohama, Japan,[email protected], Kyoko YagiWe consider the investment timing problem where each firm knowswhen the economic condition changes in advance. When firm facesan investment competition, the strategy significantly differs depend-ing on whether the firm is a leader or a follower. More concretely,a follower firm follows the so-called bad news principle proposed byBernanke(1983) near maturity, while a leader firm optimally invest inthe project just based on the current and future economic conditions.

2 - Equilibrium in the Presence of Transaction CostsMasaaki Kijima, Graduate School of Social Sciences, TokyoMetropolitan University, 1-1 Minami-Ohsawa, 192-0397,Hachiohji, Tokyo, Japan, [email protected] consider an equilibrium model in the presence of transaction costs.Such problems have been considered quite extensively in the completemarket setting (no transaction costs). In the presence of proportionaltransaction costs, however, we need to know the quantities of buyingand selling assets. For the case of single risky asset, it will be shownthat the buy price increases and the sell price decreases as the cost in-creases under some reasonable conditions. The trading volume is alsodecreasing with respect to the cost.

72


3 - Optimal Life Insurance Coverage and Annuities withBorrowing and a Leverage ConstraintTeruyoshi Suzuki, Graduate School of Economics, HokkaidoUniversity, Nishi 7 Kita 9, Kita-ku, 060-0809, Sapporo,Hokkaido, Japan, [email protected]

This paper presents the optimal policies of life insurance coverageand annuities for a retired agent using the investment/consumptionmodel with borrowing. One of features in our model is imposing non-negativity constraints on the insurance coverage and annuities consid-ering positive premium loading in the insurance market. It was foundfrom our analytical result that there is an interval of the bequest inten-sity where the agent has no demand for both insurance and annuities.

4 - Robust Portfolio Optimization with CopulasIakovos Kakouris, Computing, Imperial College London,Imperial College London, Exhibition Road, SW7 2AZ,London, United Kingdom, [email protected], Berc Rustem

Conditional value-at-risk (CVaR) is widely used in portfolio optimiza-tion as a measure of risk. CVaR is clearly dependent on the underlyingprobability distribution of the portfolio. We show how copulas can beintroduced to any problem that involves distributions and how they canprovide solutions for the modelling of the portfolio. We use this toprovide the copula formulation of the CVAR of a portfolio. Given thecritical dependence of CVaR on the underlying distribution, we use arobust framework to extend our approach to Worst Case CVaR throughthe use of rival copulas.

� MD-20Monday, 14:30-16:00RB-2113

Energy consumption and priceforecasting

Stream: Data Mining in the Financial SectorInvited sessionChair: Richard Weber, Department of Industrial Engineering,University of Chile, Republica 701, 2777, Santiago, Chile,[email protected]

� MD-21Monday, 14:30-16:00RB-2115


1 - A Parallel Procedure for Dynamic Multi-objective TSPWeiqi Li, School of Management, University ofMichigan-Flint, 303 East Kearsley Street, 48502, Flint,Michigan, United States, [email protected]

This paper proposes a new parallel search procedure for dynamicmulti-objective TSP. We design a multi-objective TSP in a stochas-tic dynamic environment. The proposed procedure first uses parallelprocessors to identify the extreme solutions of the search space foreach of k objectives individually at the same time. These solutions aremerged into a matrix E. The solutions in E are then searched by paral-lel processors and evaluated for dominance relationship. The proposedprocedure was implemented in two different ways: a master-workerarchitecture and a pipeline architecture.

2 - Foreign Direct Investment Trends and ComparativeAnalysis for Chosen Transition CountriesSnjezana Pivac, Faculty of Economics, University of Split,Department of Quantitative Methods in Economics, Maticehrvatske 31, 21000, Split, Croatia, [email protected]

Foreign direct investments (FDI) have high positive impact on eco-nomic impetus especially in transition countries. So in the focus ofthis paper are the structure and forms of FDI and their influence oncrucial macroeconomic variables. In the FDI inflow analysis the coun-try risk rating estimation has been essential. FDI efficiency analysisis enriched with correlation between Incremental Capital Output Ratio(ICOR) and GDP increase. Multivariate classification of ten transitioncountries has been carried out according to all regarding indicators.

3 - A mathematical solution to the Nigerian petroleumproducts supply crisisAdewoye Olabode, mathematics, yaba college of technology,yaba, lagos, +234, yaba, lagos state, Nigeria,[email protected]

Nigeria is the largest oil exporter in Africa and rank sixth in the world.The country for more than two decades has found itself in petroleumproducts supply crisis. With thorough investigation, necessary param-eter is developed to run the policy iteration techniques on continuoussemi — Markov Decision Process. The solution to the model is ob-tained by selecting an arbitrary stationary policy that governs the prob-lem.

4 - Stochastic Dominance in economic and financialcontext, overview and applicationsElena Almaraz Luengo, Estadística e Investigación Operativa,Facultad de Ciencias Matemáticas (UCM), c/Arroyo de laMedia Legua 68, 2oB, 28030, Madrid, Spain,[email protected], Eduardo Almaraz Luengo

We analyze different types of Stochastic Dominance (SD) and its prin-cipal applications in different knowledge areas, with special interest ineconomic and financial context. Some points discussed are: SD con-cepts in connection with particular and interesting random variables inEconomy; Consistency of Mean-Variance rules and its application inweighted random variables; SD in particular economical and financialstochastic processes; SD in context of assets valuation. We finish withthe concept of Almost Stochastic Dominance and its possible applica-tions in economic and financial context.

� MD-23Monday, 14:30-16:00RB-Delta

Various Advances in GeneralizedDifferentiation and Optimization


1 - Interior Epigraph Method for Nonsmooth and Non-convex Optimization via Generalized Augmented La-grangian DualityWilhelm Passarella Freire, Mathematics, Federal Universityof Juiz de Fora, Campus Universitário - Bairro Martelos. Juizde Fora - MG, Brasil, 36036-900, Juiz de Fora, Minas Gerais,Brazil, [email protected], Regina Burachik, CYalcin Kaya

73


We propose a new method (IEM) for constrained nonsmooth and non-convex optimization which uses a generalized augmented lagrangianduality scheme. The new method takes advantage of the special struc-ture of the epigraph of the dual function. We present numerical ex-periments by using test problems from the literature. In particular,we study several instances of the kissing number problem. Our ex-periments show that the quality of the solutions obtained by IEM iscomparable with those obtained by other solvers.

2 - A Novel Approach for Solving the Geometric Dis-tance ProblemHelder Venceslau, Federal University Rio de Janeiro - UFRJ,Brazil, [email protected], Leonardo Nascimento,Adilson Elias Xavier, Gerhard-Wilhelm Weber

The geometrical distance problem in graphs is characterized by de-termining the nodes’ positions in a Euclidian space, according to thegiven distances associated with the arcs. It is a non-convex and non-differentiable problem. It has a relevant application in the determina-tion of proteins’ geometrical structures. It is solved by the HyperbolicSmoothing Technique connected to a speed-up extrapolation scheme.Computational results obtained in the resolution of large instances ofa difficult canonic problem show the efficiency and robustness of themethod.

3 - Generalized Solutions and Optimal Investment in In-finite Horizon Control Problem in Capital Accumula-tion ModelMarta Kornafel, Department of Mathematics, CracowUniversity of Economics, Poland,[email protected]

Control problem in capital accumulation model with vintage structurein Neoschumpeterian framework is presented. Model is described bytransport equation on bounded domain. Coefficients of the state equa-tion are allowed to be discontinuous functions, what differs our ap-proach from the considerations in the existing literature. Problem ofexistence and uniqueness of solution for the state equation is discussedin our presentation. It is shown that among generalized solutions thereexists viscosity one to the state equation on extended domain. Optimalinvestment is obtained via direct method

� MD-24Monday, 14:30-16:00CC-A11

Pioneers in Operations Research

Stream: Pioneers in Operations ResearchInvited sessionChair: Graham Rand, Dept. of Management Science, LancasterUniversity, The Management School, LA1 4YX, Lancaster,Lancashire, United Kingdom, [email protected]

1 - Von Neumann and Vazsonyi: OR pioneers born in Bu-dapestGraham Rand, Dept. of Management Science, LancasterUniversity, The Management School, LA1 4YX, Lancaster,Lancashire, United Kingdom, [email protected], SaulGass

The 2011 volume "Profiles in Operations Research", written, compiledand edited by A. Assad and S. Gass, provides biographies of 43 distin-guished OR pioneers. Three of these were born in Budapest, but madetheir pioneering contributions elsewhere. This paper complements thefollowing one by outlining the fascinating life and times of John vonNeumann, who made seminal contributions to, inter alia, game theory,utility theory and simulation, and Andrew Vazsonyi, who was at theforefront of OR’s move into business and industry after World War II.

2 - Steven Vajda, European OR pioneerJakob Krarup, Dept. of Computer Science, University ofCopenhagen, Ydervang 4, DK-3460, Birkeroed, Denmark,[email protected]

Several European OR pioneers are listed among the 43 distinguishedscholars portrayed in the 2011 volume "Profiles in Operations Re-search’, written, compiled and edited by A. Assad and S. Gass. Theseinclude Steven Vajda (1901-95), born in Budapest, but known as theBritish father of linear programming (LP) and the person who intro-duced LP to both Europe and Asia. An account of Steven’s life andwork is provided.

3 - U.S. OR Pioneers: Engagements in European ORArjang Assad, School of Management, University at Buffalo,SUNY, 160 Jacobs Management Center, Buffalo, 14260-4000,Buffalo, NY, United States, [email protected] developments of Operational Research (OR) owed much to ex-changes between OR pioneers in North America and Europe. This pre-sentation focuses on key engagements of OR pioneers from the U.S. inEuropean OR. Examples are drawn from the earliest periods throughthe 1980s to illustrate the different natures of these engagements andtheir impact on the field.

� MD-25Monday, 14:30-16:00CC-A23

Renewable Energies

Stream: Energy, Environment and Climate 2Invited sessionChair: Mario Ragwitz, FhG-ISI, 76139, Karlsruhe, Germany,[email protected]

1 - A new perturbing parameter for interior point meth-ods applied to short term hydroelectric schedulingAurelio Oliveira, Computational & Applied Mathematics,State University Of Campinas, DMA IMECC Unicamp, C. P.6065, 13081-970, Campinas, SP, Brazil,[email protected], Lilian CarvalhoThe short term hydroelectric scheduling problem can be formulated asa network flow model and efficiently solved employing interior pointmethods. In this work, a new approach to compute the perturbing pa-rameter in the primal-dual interior point method is presented. Numeri-cal results for short term hydroelectric scheduling applied to IEEE testsystems and actual Brazilian systems are presented and compared tothose obtained early. The experiments show that both approaches areattractive for both: running time and robustness and that the new onerequires less iterations to achieve convergence.

2 - Optimal Underground Pumped Hydroelectric StorageDesignAmir José Daou Pulido, Chair of Energy Systems and EnergyEconomics, Ruhr - Universität Bochum, IB 4 - 32,Universitätsstr. 150, 44801, Bochum, Germany,[email protected], Hermann-Josef Wagner, Marco K.Koch, Eugen Perau, Ulrich Schreiber, André NiemannOptimal design of utility scale energy storage systems has beenstrongly researched in the last decades. However, there are few stud-ies addressing Underground Pumped Hydroelectric Storage (UPHS).In Germany, UPHS has become an interesting technology for both en-ergy system flexibility and mining sector. We propose an optimiza-tion model for UPHS design, i.e., reservoirs, waterways and turbo ma-chinery configuration. The model aims to find the UPHS design thatmaximizes long-term profit, given a certain power market developmentscenario.

3 - Investment Incentives under the Influence of Capac-ity Aggregation in Regulated IndustriesSabine Pallas, TU München, Germany, [email protected] purpose of this paper is to investigate the influence of cost basedremuneration on a regulated company’s investment behavior when ca-pacity aggregation occurs and investments overlap. To this end adiscrete-time optimization model is solved analytically for both par-ties. For an expansion of renewable energy large investments in thegrid are necessary. Network operators are subject to a governmentalcost-based remuneration, which can set disincentives to invest. How-ever, this effect has not been investigated for two-stage overlappinginvestments with stepwise capacity expansion.

74


4 - Pollution permits, Strategic Trading and DynamicTechnology AdoptionLuca Taschini, Grantham Research Institute, London Schoolof Economics, Houghton St, WC2A 2AE, London, UnitedKingdom, [email protected] paper analyzes the dynamic incentives for technology adoptionin a transferable permits system, which allows for strategic tradingon the permit market. Technology adoption and allowance prices aregenerated endogenously and are inter—dependent. It is shown thatthe non—cooperative permit trading game possesses a pure—strategyNash equilibrium, where the allowance value reflects the level of un-covered pollution (demand), the level of unused allowances (supply),and the technological status. The proposed policy generates a pricefloor and restores the dynamic incentives to invest.

� MD-26Monday, 14:30-16:00CC-A24

OR in Forestry I

Stream: OR in Agriculture, Forestry and FisheriesInvited sessionChair: Concepcion Maroto, Applied Statistics, OperationsResearch and Quality, Universitat Politecnica de Valencia, Caminode Vera S/N, 46071, Valencia, Spain, [email protected]

1 - A comparison of multiple criteria methods for groupdecision making in forest managementConcepcion Maroto, Applied Statistics, Operations Researchand Quality, Universitat Politecnica de Valencia, Camino deVera S/N, 46071, Valencia, Spain, [email protected],Marina Segura Maroto, Concepción Ginestar, BaldomeroSegura, Juan UriolAnalytic Hierarchy Process, Goal Programming and Outranking Meth-ods have been the approaches most applied to problems of sustainableforest management. It is well known that the results depend on themethod used and one of the main uses of these approaches is to ob-tain consensus or as negotiation tools. Thus, robust results or conclu-sions need to be derived from the use of several techniques. This paperpresents a comparison of several methods for aggregating stakeholderpreferences as applied to the regional planning of Mediterranean for-est. Finally, we will discuss the results obtained.

2 - Multi-criteria optimization method for designing aharvesting and cableway layout for a given road net-work for wood extractionLeo Bont, ETH Zurich, CH-8092, Zürich, Switzerland,[email protected], Hans Rudorf Heinimann, RichardChurchCable-based technologies are a backbone when designing forest har-vesting systems for steep slopes. However, layout and placement ofthe cableways is challenging. To guarantee that wood harvesting costsare minimized and that the process has the least environmental impacton the remaining forest, one must carefully locate both cable towersand cableways. We developed an approach of mixed integer linearprogramming for planning an optimal harvesting and cableway layout.The model delivered for terrain units up to 35 ha results within a rea-sonable timeframe.

3 - EcoForest — A fictive problem for teaching spatially-explicit, multi-objective optimization in forestryJochen Breschan, Department of Environmental Sciences,ETH Zurich, CHN K73.1, Universitaetstr. 22, 8092, Zurich,Switzerland, [email protected] present the EcoForest problem, designed to teach spatially-explicit,multi-objective optimization. A fictive parcel of open land is repre-sented as a grid raster of 8x8 cells. 40 cells must be re-forested suchthat the resulting pattern produces lots of ecological niches. Forestedges and forest core area indicate ecological niches. The correspond-ing (multi-objective Integer Linear) optimization model identifies theoptimal pattern by concurrently maximizing forest edges and forestcore area. Non-Inferior Set Estimation method is applied to computean approximate Pareto-frontier.

4 - Impact of carbon sequestration over the optimumforest rotation problemYavuz Gunalay, Faculty of Economics and AdministrativeSciences, Bahcesehir University, Besiktas, 34353, Istanbul,Turkey, [email protected], Erhun Kula

The threat of climate change, which is also backed by the Kyoto Pro-tocol, has given rise to the importance of the forestry sector. One ofthe important problems of the sector is to determine the optimum ro-tation age the forest. The problem is first introduced in 1849 and stillkeeps its challenge among economists and silviculturists. Based uponSamuelson’s work the optimum felling age occurs at a time when thenet marginal benefits fall below the current rate of interest. We incor-porate the carbon sequestration benefits and re-calculate the optimumcutting age using a multiple rotation model.

� MD-27Monday, 14:30-16:00CC-A25

Agent-based Modeling of DiffusionProcessesStream: Decision ProcessesInvited sessionChair: Christian Stummer, Department of BusinessAdministration and Economics, Bielefeld University, Chair ofInnovation and Technology Management, PO Box 10 01 31,33501, Bielefeld, Germany, [email protected]

1 - The impact of customer dissatisfaction and negativeword-of-mouth on the diffusion of innovations andthe repurchasing behaviour: an agent based simu-lationMarkus Günther, Department of Business Administration andEconomics, Bielefeld University, Universitaetsstr. 25, 33615,Bielefeld, Germany, [email protected], JanaSonnberger

Customer expectations towards a new product, which are composedof individual demands, product information via personal communica-tion and marketing measures, form the basis of customer satisfactionor discontent. We introduce an agent-based model which contributesto a better understanding of discontented clients communicating neg-ative product information and their impact on the diffusion of an in-novation. Additionally, repurchasing decisions, which differ from theinitial adoption, positive network effects and individual agent traits likeextraversion and neuroticism are taken into account.

2 - New product diffusion and supply chain production-sales policies: a comparative studyTina Wakolbinger, WU (Vienna University of Economics andBusiness), Nordbergstraße 15, Vienna, Austria,[email protected], Mehdi Amini, Michael Racer,Mohammad G. Nejad

This paper analyzes the impact of alternative production—sales poli-cies on the diffusion of a new product and the generated NPV of profit.The key features of the ABMS model are: (i) supply chain capacity isconstrained; (ii) consumers’ new product adoption decisions are influ-enced by marketing activities as well as positive and negative word-of-mouth (WOM) between consumers; (iii) interactions among con-sumers taking place in the context of their social network are capturedat the individual level; and (iv) the new product adoption process isadaptive.

75


3 - De-commoditization of a convenience product: anagent-based simulation of its market diffusionChristian Stummer, Department of Business Administrationand Economics, Bielefeld University, Chair of Innovation andTechnology Management, PO Box 10 01 31, 33501,Bielefeld, Germany, [email protected],Elmar Kiesling, Markus Günther, Rudolf Vetschera, Lea M.WakolbingerThe de-commoditization of a convenience product poses an interest-ing challenge for innovation management that has to decide on properstrategies for advertising, pricing, and roll-out of the new product. Inmarkets in which information is largely disseminated through word-of-mouth, an agent-based simulation may provide decision-makers withforecasts of the diffusion process. In this talk we will present suchan approach and illustrate its application for the case of introducing asecond generation biofuel on the Austrian market.

� MD-28Monday, 14:30-16:00CC-A27

Nonsmooth Optimization II

Stream: Structured Optimization and ApplicationsInvited sessionChair: Dominik Dorsch, Dept. Mathematics, RWTH AachenUniversity, Templergraben 55, 52056, Aachen, NRW, Germany,[email protected]

1 - Equilibrium investments in a common value auctionenvironment with endogenous signalsEleftherios Couzoudis, Economics, University of Zurich,Chair for Quantitative Business Administration,Moussonstrasse 15, 8044, Zurich, Switzerland,[email protected], Markus BürgiThe presented normal form game is a common value auction whereeach bidder gathers an independent endogenous private signal aboutthe offered good. We allow signal precision to be scaled continu-ously in correspondence to the investment. In addition the seller canoffer a costly signal to the public, which leads to signal combina-tions. Through reformulation and the use of a Newton-type methodthis model is solved for more than two players. Furthermore, the im-plications of our findings can support the financial regulation and thepolicy-making process.

2 - Local models in equilibrium optimizationVladimir Shikhman, Dept. Mathematics C, RWTH AachenUniversity, Templergraben 55, 52056, Aachen, Germany,[email protected], Hubertus Th. Jongen,Dominik DorschWe study equilibrium optimization problems (EOPs) from a structuralperspective, i.e. up to the smooth coordinate transformations locallyat their solutions. The latter equivalence relation induces classes ofEOPs. We focus on the stable classes corresponding to a dense set ofdata functions. We prove that these "basic classes" are unique. Theirsimplest representatives are called local models. For particular EOPsbasic classes and their local models are elaborated. The latter includebilevel optimization, general semi-infinite programming and Nash op-timization.

3 - On the structure of Nash equilibrium setsDominik Dorsch, Dept. Mathematics, RWTH AachenUniversity, Templergraben 55, 52056, Aachen, NRW,Germany, [email protected], Hubertus Th.Jongen, Vladimir ShikhmanA Nash eaquilibrium problem (NEP) consists of simultaneous opti-mization of N players’ objective functions under constraints. In aGNEP also each players’ constraint functions might depend on the op-posing players’ choices. The number of these common constraints be-comes crucial for a qualitative analysis of the solution sets: for GNEPswithout common constraints and for classical NEPs we show thatgenerically all Nash equilibria are jointly nondegenerate KKT points.Effects like kinks, boundary points, and nonclosedness might occur.Analytical results are given in detail.

4 - Existence of a pure strategy equilibrium in finite sym-metric games where payoff functions are integrallyconcaveTakahiro Watanabe, Graduate School of Social Sciences,Tokyo Metropolitan University, Minamioosawa 1-1,Hachiouji_city, 1920397, Tokyo, Japan,[email protected], Takuya Iimura

In this paper we show that a finite symmetric game has a pure strat-egy equilibrium ifthe payoff functions of players are integrally con-cave (the negative of the integrally convex functions due to Favati andTardella,1990). Since the payoff functions of any two-strategy gameare integrally concave, this generalizes the result of Cheng et al.(2004).A simple algorithm to find an equilibrium is also provided.

� MD-29Monday, 14:30-16:00CC-A29

Selected Topics on Mixed-IntegerNon-Linear Programming

Stream: Mixed-Integer Non-Linear ProgrammingInvited sessionChair: Dennis Michaels, Institute for Operations Research, ETHZuerich, Raemistrasse 101, 8092, Zuerich, Switzerland,[email protected]

1 - Minimizing convex functions over integer pointsMichel Baes, IFOR, ETH, HG.G.22.1, Ramistrasse 101„8092, Zurich, Switzerland, [email protected],Alberto Del Pia, Yurii Nesterov, Shmuel Onn, RobertWeismantel

We consider the problem of minimizing a smooth and strongly convexfunction over integer points in a polytope. We assume to have a black-box procedure that solves some special quadratic integer problems witha constant approximation factor. Despite the generality of our problem,we prove that we can find efficiently, with respect to some assumptionson the encoding of the problem, a feasible solution with an objectivevalue close to the optimum. We describe a few situations where wecan implement the needed black-box procedure efficiently.

2 - Solving MINLPs with SCIPStefan Vigerske, Mathematics, Humboldt University Berlin,Unter den Linden 6, 10099, Berlin, Germany,[email protected]

We discuss recent extensions of the constraint integer programmingframework SCIP for solving mixed-integer nonlinear programs. Non-linear constraints (convex or nonconvex) are handled within an LP-based branch-and-cut algorithm by reformulation, linear relaxation,and domain propagation. In an extensive computational study, wecompare the performance of our implementation with state-of-the-artsolvers for MINLP and analyze the impact of various solver compo-nents on the overall performance.

3 - RedNLP - Heuristic Approach to Large-Scale MINLPsin Gas TransportationRalf Gollmer, Mathematics, University of Duisburg-Essen,Forsthausweg 2, D- 47057, Duisburg, NRW, Germany,[email protected], Rüdiger Schultz, ClaudiaStangl

Detecting feasibility of transportation orders (nominations) in gas net-works is a problem of growing practical interest due to the unbundlingof gas trading and network operating companies. Under planning as-pects transport is viewed in steady state and can be modeled by large-scale non-convex MINLPs. This poses challenging mathematical ques-tions. The talk addresses some structural properties of the constraintsmodeling the sets of admissible points and sketches a heuristic ap-proach for the solution of the problem.

76


4 - Convex underestimation of edge-concave functionsby a simultaneous convexification with multi-linearmonomialsDennis Michaels, Institute for Operations Research, ETHZuerich, Raemistrasse 101, 8092, Zuerich, Switzerland,[email protected], Martin Ballerstein

We address the construction of tight convex underestimators for anedge-concave function over a box. In contrast to common relaxationmethods we derive the underestimators by a simultaneous convexifi-cation of one edge-concave function together with the set of all multi-linear monomials. We show that the tightest underestimators are poly-hedral and give a complete facet-description. A computational casestudy demonstrates the usefulness of the proposed relaxation techniquein global optimization.

� MD-30Monday, 14:30-16:00CC-A31

Emerging Applications in Logistics

Stream: Emerging Applications in Finance and Logis-ticsInvited sessionChair: Paul Trodden, School of Mathematics, University ofEdinburgh, United Kingdom, [email protected]

1 - An MINLP approach to forming secure islands inelectricity networksPaul Trodden, School of Mathematics, University ofEdinburgh, United Kingdom, [email protected],Waqquas Ahmed Bukhsh, Andreas Grothey, Ken McKinnon

Intentional islanding is attracting an increasing amount of attentionas a means of preventing large-scale blackouts in electricity transmis-sion networks. In this talk, a mathematical formulation for islandingis presented, in which suspected unhealthy components of the net-work are isolated while the load shed is minimized. To achieve bal-anced, feasible islands, nonlinear AC power flow equations should beincluded, resulting in an MINLP. In the proposed MILP formulation,these terms are approximated by piecewise linear functions. The ap-proach is demonstrated by results on test networks.

2 - Multiperiod Optimal Power Flow Using Cross Decom-positionAntonio Marmolejo, DAIC, INFOTEC, COPILCO 300.COYOACAN, 04350, CIUDAD DE MEXICO, DISTRITOFEDERAL, Mexico, [email protected]

This work addresses Multiperiod Optimal Power Flow. The mathemat-ical model is a Mixed Integer Non Linear Problem with binary vari-ables. We use Generalized Cross Decomposition to exploit the prob-lem structure and reduce the computational time. Original problem isdivided into a primal subproblem (Non Linear Problem), a dual sub-problem (Mixed Integer Non Linear Problem) and a Master Problem(Mixed Integer Problem). The approach is compared with LagrangeanRelaxation and Generalized Benders Decomposition. We present nu-merical results which show Cross Decomposition performance.

3 - Optimum Renewable Energy Portfolio for RegionalHeatingSecil Ercan, Industrial Engineering, Istanbul TechnicalUniversity, Turkey, [email protected], Zeynep Bektas,Gulgun Kayakutlu

This study analyses the renewable energy resources for constructingan energy portfolio that would best respond to the regional demandand conditions. Heating of an organized site of manufacturing SMEs(small and medium enterprises) is targeted. Because of the continu-ous and non-deterministic structure of the renewable energy sources,nonlinear programming (NLP) is used in optimization. The proposedmodel aims to maximize the energy capacity of renewable energy re-placing the current fossil energy resources. The achievements will beguiding the policy makers in reducing energy dependence.

4 - Strategic Capacity Planning in Knowledge IntensiveOrganizationsManuel Mateo, Departament Business Administration,Universitat Politecnica Catalunya, Avda Diagonal, 647, 7th,E-08028, Barcelona, Spain, [email protected], ErnestBenedito, Rocio de la Torre, Amaia Lusa, Carme Martinez,Marta MasStrategic capacity planning is a key decision in knowledge intensiveorganizations (KIOs). It is basically related to determining the sizeand composition of the workforce in the long term, which means tak-ing decisions on hiring, firing, promotion, transfers and training ofskilled people, considered as the main limiting factor. To the best ofour knowledge, there are no formalized procedures to deal with strate-gic capacity planning in KIOs. The present study aims to define theproblem, identify the most relevant characteristics and propose a clas-sification scheme in order to later develop specific solving methodol-ogy and procedures.

� MD-32Monday, 14:30-16:00CC-A34

Theoretical developments in Soft OR

Stream: Soft OR and Problem Structuring MethodsInvited sessionChair: Jose-Rodrigo Cordoba-Pachon, School of Management,Royal Holloway, University of London, Egham Hill, Egham,Surrey, United Kingdom, [email protected]

1 - Comparing and Contrasting the formal modellingtechniques from Soft Systems Methodology andJOURNEY MakingGiles Hindle, Mangement Systems, Hull University BusinessSchool, University of Hull, Cottingham Road, HU6 7RX,Hull, United Kingdom, [email protected] OR contains a family of approaches that support groups in arange of tasks including tackling complex problem situations and mak-ing strategy. Two popular approaches are Soft Systems Methodology(SSM) and JOURNEY Making. Both approaches use formal mod-elling techniques — purposeful activity system modelling in SSMand cognitive mapping in JOURNEY Making. This paper comparesand contrasts these two modelling techniques. The techniques arefound to have fundamentally different theoretical make-ups, whilstsynchronously allowing overlapping modelling strategies in practice.

2 - Decision Structures in Project ManagementCathal Brugha, Management Information Systems, UniversityCollege Dublin, Quinn School of Business, Belfield, 4, Dublin4, Ireland, [email protected] use meta-theory to show how Project Management structures de-rive from four General Decision Activities: Proposition, Perception,Pull and Push. Project Development is shown to be made up ofthree Committing Phases: Analysis, Design and Implementation, eachof which has three Convincing Others Stages: Self/Technical, Oth-ers/Contextual, World/Situational. These form nine stages, each ofwhich is focused on one of Four Decision Drivers: Committing Self,Convincing Others, Adjusting Others and Adjusting Self.

3 - Critical awareness of worldviews in organisationalchangeMike Yearworth, Systems Centre, University of Bristol,Queens Building, University Walk, BS8 1TR, Bristol, UnitedKingdom, [email protected], Charlotte Dunford,Darren York, Patrick GodfreySystems Practice is the application of a particular way of perceivingand making sense of the world in order to aid understanding. We areconcerned with the problem of organisational change that exists whenattempting to improve the use of Systems Practice in response to meet-ing complex design challenges. However, relying on mandate is notsufficient. We reflect on how change needs to originate from practition-ers that gain a critical awareness of their own worldview and achievea crucial shift towards an appreciation of the different epistemologicaltraditions underpinning Systems Practice.

77


� MD-33Monday, 14:30-16:00CC-A37

OR for Energy and ResourceDevelopment in Developing Countries

Stream: OR for Development and Developing Coun-triesInvited sessionChair: Javier Cano, Rey Juan Carlos University, Spain,[email protected]: Angel Udias, Rey Juan Carlos University, Spain,[email protected]

1 - Expanding the Hydroelectric Power Capacity of theKwanza RiverJavier Cano, Rey Juan Carlos University, Spain,[email protected], Kiombo Jean Marie, David Rios-Insua

The increasing demand and prices of crude oil and its derivatives areforcing authorities to use more efficiently the current sources of energy,and to invest in alternative sources and new infrastructures. We haveanalyzed the situation in Angola, proposing a strategic (and financiallyviable) planning for the expansion of the hydroelectric power capac-ity of the Kwanza River, in order to cope with different and realisticscenarios of future demands, avoiding the collapse of the Angolan En-ergy Reference System (ERS). The decision making problem includesmixed nonlinear-integer terms.

2 - Valorization of riverbanks in the urban areaTomás Hanácek, PhD. Student, Faculty of architecture, STU,Institute of Urban Planning and Land-Use Planning, Lidická3, 91101, Trencín, Slovakia, [email protected]

Degraded river waterfront areas become the subject of revitalization inmany European cities. Many cities dispose of the attribute of the cityon the water. How can we evaluate the potential of the area, whichcombines the natural with the artificial environment? Cultural bene-fit and historical reference of the point based on water mean addedvalue in this mathematical equation. Analytical method SKETCH andGO searches opportunities how to optimize solutions for design urbanspace through multisensual perception of the environment. An im-portant aspect of the method is its multidisciplinary applicability andpossibility of comparison of each research element.

3 - Export Diversification and Resource-based Industri-alization: The Case of Natural GasOlivier Massol, Center for Economics and Management, IFPSchool, 228-232 Avenue Napoléon Bonaparte, 92852,Rueil-Malmaison, France, [email protected]

For a small economy, the ownership of natural gas resources is usuallydescribed as a blessing, but past performances reveal a curse causedby the large variability of export revenues. A modified mean-varianceportfolio model is thus proposed to design a diversification strategycentered on resource-based industries. Using a time series model ofcommodity prices, this model is put to work, to analyze the efficientfrontier, and evaluate the policies implemented in nine economies.

4 - Multicriteria Decision Aid for Equitable Water Distri-bution NetworkJose Luis Castaño Cabrales, Rey Juan Carlos University,28943, Fuenlabrada, Spain, [email protected], Angel Udias,David Rios-Insua, Javier Cano, Hocine Fellag

This work describes an equitable problem, the multiobjective manage-ment (equity vs. efficiency) of water resources in the region of Ka-bila, in Algeria. External factors, as pumping operational schedules,strategic planning, pipes leakages, village’s geographical location, andnetwork constraints make difficult to distribute water fairly in a reli-able and cost-efficient manner. We have developed a decision supportsystem that implements different optimization algorithms (egalitarian,utilitarian, Kalai-Smorodinsky), to help water companies to take reli-able and feasible decisions.

� MD-34Monday, 14:30-16:00CC-A39

Assignment and multicriteria games

Stream: Cooperative Game TheoryInvited sessionChair: Marina Nunez, Mathematical Economics, University ofBarcelona, Av Diagonal, 690, 08034, Barcelona, Spain,[email protected]

1 - Z-Equilibrium for a Mixed Strategic MulticriteriaGameArezki Ferhat, LAboratory of Modeling and Optimization ofSystems (LAMOS), University of Bejaia, Algeria, Universityof Bejaia, Route de Targa-Ouzamour, 06000, Bejaia, Algeria,[email protected], Mohammed Said Radjef

In this paper, we study the Z-equilibrium for a mixed strategic multicri-teria game. This concept is composed of two points: active equilibriumand the Pareto optimality of the active equilibrium. We give the exis-tence conditions of this equilibrium which are weaker than those of thenon-cooperative equilibrium for an n players normal form game. In-deed, we just need the compacity of the strategies sets and the continu-ity of the payoff functions to ensure the existence of the Z-equilibrium.

2 - Bilateral assignment games with the Monge propertyJavier Martinez-de-Albeniz, Matematica Economica,Financera i Actuarial, Universitat de Barcelona, Av. Diagonal,690, 08034, Barcelona, Spain,[email protected], Carles Rafels

We study bilateral assignment games where the assignment matrix sat-isfies the inverse Monge property. In this case, for any 2x2 subma-trix the optimal matching is in its diagonal. Their cores are describedby using only the central tridiagonal band of the elements of the ma-trix. We introduce a suitable notion of matrix monotonicity, calledtridiagonal-monotonicity, and in this case we define the assortative as-signment game, characterize its core in a parametric representation andgive its extreme core points. We give a formula for the nucleolus anda description of the kernel.

3 - A geometric characterization of the nucleolus of theassignment gameMarina Nunez, Mathematical Economics, University ofBarcelona, Av Diagonal, 690, 08034, Barcelona, Spain,[email protected], Francesc Llerena

Maschler et al. (1979) give a geometrical characterization for the inter-section of the kernel and the core of a coalitional game: allocations inboth sets are always the midpoint of certain bargaining range betweeneach pair of players. In the case of the assignment game the kernel canbe determined as those core allocations that satisfy a bisection propertyfor each optimal pair. We now prove that the nucleolus of the assign-ment game can be characterized by requiring this bisection propertybe satisfied not only for optimally matched pairs but also for optimallymatched coalitions.

� MD-35Monday, 14:30-16:00CC-A41


Stream: DEA and Performance Measurement (con-tributed)Contributed sessionChair: Paula Guimaraes, Industrial Engineering, UFF, Paulo SilvaAraujo number 95 apt 302, 20735230, Rio de Janeiro, Rio deJaneiro, Brazil, [email protected]

78


1 - The Efficiency Measure Over Time with UndesirableOutputs: A Case of Auto and Auto Component Indus-try in TaiwanYi-Chiuan Lai, Business Administration, ProvidenceUniversity, 200 Chung-Chi Rd., Shalu District, 43301,Taichung, Taiwan, [email protected], Chin-Min HsuOperation performance is used to measure operation results of a deci-sion making unit. A DEA model that measures efficiency accordingto undesirable outputs is proposed. Several factors are included in themodel. Input items include fixed assets, current assets, R&D cost, andnumber of employees. The desirable output is net sales, and the un-desirable output is defined as loss on inventories. Window analysis isused to discuss the efficiency change over time. Data were collectedfrom auto and auto component industries in Taiwan. We identify vari-ous issues that affect the operation efficiency.

2 - An application of canonical correlation analysis forthe econometric estimation of the stochastic produc-tion frontierJoão Silveira, Applied Social Sciences, University RegionalIntegrated High Uruguay and Missions, URI - SantiagoCampus - Brazil - RS, 97700-000, Santiago, RS, Brazil,[email protected], Attus MoreiraSince the stochastic frontier production model emerged in 1977, inter-est in econometric analysis of productive efficiency has been increas-ing. Yet this kind of analysis is troublesome with regards to multipleoutputs. The multiple-equation approach is not well developed andthe ingleequation approach is subject to special conditions that are notalways fulfilled. A methodological alternative is proposed. The esti-mation process treats each one of the multiple outputs as a independentvariable and can be estimated by canonical correlation analysis.

3 - Evaluation of the efficiency of investments in thehost cities of the Olympic Games using Data Envel-opment Analysis (DEA)Paula Guimaraes, Industrial Engineering, UFF, Paulo SilvaAraujo number 95 apt 302, 20735230, Rio de Janeiro, Rio deJaneiro, Brazil, [email protected], Nissia BergianteThe main idea of the Olympic Games is to create an atmosphere ofglobal solidarity. However, behind it there is a financial plan applied tothe host cities to support generated impacts. As a result is expected anincrement in the number of visitors providing a financial return fromservices related to tourism, in addition to the revenue generated by thegames themselves. In this context, this study aims to evaluate the ef-ficiency of the last five host cities using DEA. As input we chose thetotal investments and for outputs we included visitors (by plane), hotelscapacity, and Games revenue.

4 - A holistic approach for assessing the performance ofmaternity and newborn servicesOlga Li, Economics, UAlg, Campus de Gambelas, 8005-139,Faro, Algarve, Portugal, [email protected], Sérgio Santos,Carla AmadoThe integration of health care services as a potential solution to im-prove their efficiency has attracted considerable interest in recent years.This paper proposes a conceptual framework for integrated maternaland newborn services and explores the use of network data envelop-ment analysis to assess the performance of the holistic system pro-posed. By using data from several countries, an international compar-ison is discussed.

� MD-36Monday, 14:30-16:00CC-A43


Stream: DEA and Performance MeasurementInvited sessionChair: Luis C. Dias, Faculdade de Economia / INESC Coimbra,INESC Coimbra, Rua Antero de Quental, 199, 3000-033,Coimbra, Portugal, [email protected]

1 - The assessment of livability in European citiesAndreia Zanella, Faculdade de Engenharia da Universidade doPorto, Rua Dr. Roberto Frias, s/n, 4200-465, Porto, Portugal,[email protected], Ana Camanho, Teresa Galvão Dias

In this study we address the assessment of livability in European cities.In a first stage, the cities were evaluated taking into account two com-ponents of livability: human well-being and environmental quality.The assessment is based on a Data Envelopment Analysis model spec-ified with a Directional Distance Function. The second stage involvesexploring the exogenous factors that may promote livability in an ur-ban context. A decision tree is used to identify the main factors thatcharacterize the cities and that may explain differences in performance.

2 - Towards sustainability in construction: a DEA as-sessment at an urban levelIsabel Horta, Faculdade de Engenharia, Universidade doPorto, R. Dr. Roberto Frias, 4200-465, Porto, Portugal,[email protected], Ana Camanho, Teresa Galvão Dias

The purpose of this research is to assess construction performance ata municipality level, focusing on resource consumption. The best andthe worst practices are investigated using two types of DEA frontiers:the frontier of the production possibility set and an inverted frontier.In addition, the factors associated with good levels of performance areexplored using truncated regression. In a second stage, the change inperformance over time is analyzed applying the Malmquist index. Toillustrate the methodology developed, the data related to Lisbon mu-nicipalities between 2002-2006 is used.

3 - An Application of Categorical Models of Data Envel-opment AnalysisLuka Neralic, Faculty of Economics and Business, Universityof Zagreb, J. F. Kennedy 6, Stefaniceva 7, 10000, Zagreb,Croatia, [email protected], Dubravko Hunjet, Richard E.Wendell

This paper studies the efficiency of 127 selected towns in the Repub-lic of Croatia using categorical models of Data Envelopment Analysis(DEA). The towns, represented as Decision Making Units (DMUs), arecharacterized into four categories according to their respective popu-lations. Using the number of employed workers and employed assetsas inputs and income as an output, we consider a categorical input-oriented (output-oriented) model with constant (variable) returns-to-scale. The paper presents and analyses computational results on effi-ciency of the towns.

4 - Benchmarking in maintenance and repairing in elec-tricity distribution networks using an additive DEAmodel with incorporation of preferencesMaria Gouveia, IPC/INESCC, Quinta Agricola - Bencanta,3040-316, Coimbra, Portugal, [email protected], Luis C.Dias, Carlos Henggeler Antunes

Benchmarking studies provide a rational for quantifying the differ-ences in performance, grasping the reasons for these differences, com-paring with excellence targets, and developing ways the organizationshould follow to achieve them. This work uses a two-phase methodbased on the weighted additive model that explores links between DEAand Multi-Criteria Decision Analysis incorporating managerial prefer-ences to perform a benchmark study of network areas of an electricitydistribution company concerning maintenance and interruption repair-ing.

� MD-37Monday, 14:30-16:00CC-Act

Bioinformatics IVStream: Computational Biology, Bioinformatics andMedicineInvited sessionChair: Piotr Lukasiak, Institute of Computing Science, PoznanUniversity of Technology, ul.Piotrowo 2, 60-965, Poznan, Poland,[email protected]

79


1 - Protein structure quality assessment and modelingsupport frameworkMaciej Antczak, Institute of Computing Science, PoznanUniversity of Technology, Pl. Marii Sklodowskiej-Curie 5,60-965, Poznan, Wielkopolska, Poland,[email protected], Piotr Lukasiak, JacekBlazewicz, Krzysztof FidelisProtein structure prediction using computational tools is great inter-est of humanity. Moreover, reliable computational methods designedto evaluate protein models quality is relevant in the context of pro-tein structure refinement as currently available computational modelsoutnumber experimentally derived. The proposed method allows toidentify and visualize possible stereochemical, folding, packing incon-sistencies of protein structure. It is also particularly suited to assessapplicability of the target sequence to structural template alignments,a major source of comparative modeling errors.

2 - OR applications for protein structure modelling anddockingMaciej Milostan, Poznan University of Technology, Pl. M.Skłodowskiej-Curie 5, 60-965, Poznan, Poland,[email protected] show a pathway paved by a number of computational extensiveanalysis that lead us through a forest of genomic and structural data to-wards successful models of DCL proteins from Arabidopsis Thalianaand their interactions with RNA. The applied protocol include databasesearches, to find potential homologous sequences, followed by mul-tiple sequence alignments generation, identification of conserved re-gions, identification of structural templates in PDB, secondary struc-ture prediction, MD simulations, protein-RNA docking and evaluationof structures. We put emphasis on OR techniques.

3 - Quality assessment methodologies in analysis ofstructural modelsPiotr Lukasiak, Institute of Computing Science, PoznanUniversity of Technology, ul.Piotrowo 2, 60-965, Poznan,Poland, [email protected] and RNA are the most important molecules in all living organ-isms. In the crowded space of 3D structural models, the developmentand improvement of reliable computational methods designed to eval-uate the quality of protein and RNA models is relevant in the context oftertiary structure refinement. The correct identification of good modelsand recognition of good methodologies against misleading ones gener-ate higher level of confidence in results obtained by different method-ologies. This aspect can influence the quality and usefulness of 3Dstructure prediction in the nearest future.

4 - Rule extraction for pre-eclampsia based on GeneticAlgorithmShohreh Alimohammadi, Perinatology, Medical faculty ofHamedan, Hamedan, Hamedan, Iran, Islamic Republic Of,[email protected], Amene Alimohammadi, MahdiAshrafiIn this article data about pregnant women with pre — eclampsia andmothers without pre-eclampsia were studied in order to determine therules of the disease with the help of Genetic Algorithms and to ob-tain understandable l knowledge on the disease areas. Informationabout the disease, including 543 cases was divided in three categories:healthy, low risk and high risk group. After pre-processing, GeneticAlgorithms have been used for rule extraction of pre-eclampsia. Thethree rules with the help of Genetic Algorithms (for each disease class,a rule) for pre-eclampsia have been extracted.

� MD-38Monday, 14:30-16:00HH-Colombus

Combinatorial Multiobjective Optimization

Stream: Multiobjective OptimizationInvited sessionChair: Martin Josef Geiger, Logistics Management Department,Helmut-Schmidt-University, Holstenhofweg 85, 22041, Hamburg,Germany, [email protected]

1 - Pareto Set Approximation for the Multiobjective SetCovering ProblemMargaret Wiecek, Department of Mathematical Sciences,Clemson University, Martin Hall O-208, 29634, Clemson, SC,United States, [email protected]

The Pareto set of the multiobjective set covering problem is approx-imated within a specified accuracy. Two methods are proposed: onebased on the weighted-sum scalarization and the other based on themax-ordering. The algorithms approximate all (weakly) Pareto solu-tions with the accuracy resulting from the problem size and data. Com-putational results are included.

2 - Multiobjective combinatorial optimization for select-ing treatment regimes in forestryDmitry Podkopaev, Dept. of Mathematical InformationTechnology, University of Jyväskylä, P.O. Box 35 (Agora),FI-40014, University of Jyväskylä, Finland,[email protected], Artti Juutinen, Kaisa Miettinen,Mikko Monkkonen, Pasi Reunanen, Olli-Pekka Tikkanen

We consider an approach to managing large forest landscapes by se-lecting for each small forest stand an individual treatment regime, i.e.,a set of management rules depending on forest’s current characteris-tics. The impact of treatment on the forest is determined through sim-ulation. We consider both economic and nature preservation aspectsand formulate the problem of selecting treatment regimes as a combi-natorial multiobjective optimization problem with linear and quadraticobjectives. We present experiences of solving a computationally chal-lenging real-life problem with 30K forest stands.

3 - A bi-objective approach to reschedule new jobs in aone machine modelJacques Teghem, Mathro, Faculté Polytechnique/UMonss, 9,rue de Houdain, Rue de Houdain 9, 7000, Mons, Belgium,[email protected], Daniel Tuyttens

We consider a one machine scheduling model with a classical objec-tive - either total completion time or maximal tardiness - and two setsof jobs: one with initial jobs already scheduled and one with new jobswhich must be inserted in the schedule. As this reschedule can createa disruption, a second disruption objective is considered, which canbe formulated in four different manners. In this paper the aim is toobtain the set of efficient schedules in regard of the two objectives.Algorithms are provided for the eight possible bi-objective problems.

4 - Immune Algorithm for Multi-Objective Schedule Op-timization in Railway TransportAnatoly Levchenkov, Riga Technical University, 1658, Riga,Latvia, [email protected], Ivars Alps, MikhailGorobetz

This paper presents dual population immune algorithm for multi-objective optimization of railway transport schedule. The target isto minimize the risk of collisions between railway and auto trans-port to reduce the human factor and prevent accidents and to minimizeidle time for road transport on level-crossing by increasing its capac-ity. Mathematical model and target function to prevent collisions andcrashes in railway transport is presented. Immune algorithm is pro-posed for optimization problem solution. Algorithm is tested usingcomputer model of the transport system.

� MD-39Monday, 14:30-16:00HH-Cousteau

MADM Applications IV


80


1 - The study of the university students consumer be-havior tea-shops as exampleYuan-Du Hsiao, Department of Business Administration,Chungyu Institute of Technology, No.40, Yi7th Rd., Keelung,20103 Taiwan, R.O.C., 20103, Keelung, Taiwan, R.O.C.,Taiwan, [email protected], Jen-Chia Chang, Mu-Hui Lai,Liang-Yuan Hsiung

This study main purpose are 1. To discuss the subjects who get drinksfrom consumer behavior. 2. Testing the difference from the consumerof tea-shops. The study finding 1. More than half consumer is choos-ing the shop criterion "important" of the project. 2. Consumer generaldoes not costs at fixed shop. 3. The ones often go shopping at the noon.4. Almost ones always go to "Dou Di Tea-shops". From findings, thesuggestions for the tea-shop boss or studies experts reference related.

2 - The Study on Teachers’ Willingness to AdoptComputer-Aided Instruction in Public VocationalHigh SchoolsJen-Chia Chang, Institute of Technology & VocationalEducation„ National Taipei University of Technology, 10608,Taipei, Taiwan, [email protected], Yuan-Du Hsiao, Mu-HuiLai, Liang-Yuan Hsiung

The study with 126 sampled subject in mid-Taiwan. The reliability andvalidity of the questionnaire have been rendered positive, upon verifi-cation. The mean value of teachers’ willingness to adopt CAI is 3.97,ranged as moderate. Teachers are generally dissatisfied with the CAIsoftware, hardware facilities support and promote CAI in school; how-ever, these three factors have no significant connection with teachers’willingness to apply CAI. The researcher is to offer various sugges-tions, based on this study, to schools and the education administrativedepartments.

3 - A study on relationship model among travel motiva-tion, travel type, travel involvement, and travel ben-efit of Facebook users: A case of Taiwan domestictourismPei-Ting Chen, Travel Management, National KaohsiungUniversity of Hospitality and Tourism, 81267, Kaohsiung,Taiwan, [email protected], Tang-Chung Kan, KuoTzu-hsuan

This study aims to discuss the relation among travel motivation, travelinvolvement, and tourism benefit of Facebook users. It also uses traveltype as the intervening variable to further explore its influence in re-lation to travel pattern, travel involvement and tourism benefit. Weexpect this research result will help understand Facebook users to un-derstand characteristics of Taiwan domestic tourism. This study is ex-pected to build an appropriate social networking users’ travel behaviormodel to develop distinctive and customized community tourism prod-ucts.

4 - Multi-Criteria Comparison of Catering Service Com-panies Using Grey Relational Analysis: The Case ofTurkeyAlper Hamzadayi, Dokuz Eylul University, Turkey,[email protected], Simge Yelkenci Kose, SenerAkpinar, Atabak Elmi, Hanifi Okan Isguder

Catering is the business of providing food and beverage at a remotesite or a site like business meetings and other social occasions. Select-ing catering is an important multi-criteria decision making problem. Inthis study grey relational analysis (GRA) is used for the aim of select-ing the best catering firm providing the most firm satisfaction. Due tothe importance of criteria weights in decision making, Analytic Hier-archy Process is used to find best values of criteria weights in GRAprocess. A case study is demonstrated and evaluated using the devel-oped multi-criteria methodology.

� MD-40Monday, 14:30-16:00HH-Livingstone

MADM Applications VIII


1 - A Radial Basis Function Neural Network (RBFNN)Approach for Multi-label Text Categorization Prob-lemsTai-Yue Wang, Dept. of Industrial and InformationManagement, National Cheng Kung University, 1 UniversityRoad„ 70101, Tainan, Taiwan, [email protected],Hui-Min ChaingThis work presents an automatic text categorization model based onthe RBFNN and utilizes valuable discriminative information in thetraining dataset and incorporates background knowledge during modellearning. The Reuters 21,578 News dataset is utilized to demonstratethe application of the proposed model and to compare its performancewith that of other models. The results vindicate that the performance ofthe proposed model is comparable to that of other models for differentperformance indices.

2 - Fuzzy MCDM technique for planning the wetland en-vironmentVivien Y.C. Chen, Institute of Leisure and Health Business,Taipei Chengshih University of Science and Technology, No.2, Xueyuan Rd., Peitou, 112, Taipei, Taiwan,[email protected], Kuo Tzu-hsuan, Tang-Chung KanThis paper presents a fuzzy multi-criteria analysis approach for select-ing the best plan alternatives in wetland environment. A questionnairewas used to find out from three related groups comprising 15 experts.The subjectivity and vagueness is dealt with the criteria and alter-natives for selection process by using fuzzy numbers with linguisticterms. Incorporated the decision makers’ attitude towards preference,overall performance value of each alternative can be obtained based onthe concept of Fuzzy Multiple Criteria Decision Making.

3 - Improving the quality of leisure and tourism forTheme Parks with a MCDM ModelGwo-Hshiung Tzeng, College of Management, Hsinchu 300,National Chiao Tung University, Kainan University, Taoyuan338, Taiwan, [email protected], Vivien Y.C. ChenThe purpose of this paper is to probe how to use qualitative and quan-titative measurements of Theme Parks to create plan indexes and crite-ria, as well as how to help these indexes achieve the optimal levels foreach criterion. Previous efforts to measure leisure and tourism eval-uations and plans have assumed that the criteria are independent, butthis assumption does not hold in real-world applications. In this study,a hybrid multiple criteria decision-making model is used to addressdependent relationships among various criteria.

4 - Using F-PROMETHEE for evaluating and rankingconstructive projects contractors (A Case Study ofMehr Housing Project of Hashtgerd New City - Iran)Mahdi Nasrollahi, no.82 dehkhoda Ave. masjed St., 36161,hashtgerd new town, tehran, [email protected] this paper a model for evaluating and ranking constructive projectswill be presented. This model is based on six key criteria (good his-tory, Equipment, Management & technical staff, Economic - Finan-cial, experience, and technical capabilities) and a conceptual frame-work. These criteria and their weights were derived from previousstudies. The vagueness of the linguistic terms in the evaluation pro-cess required employment of fuzzy numbers. F-PROMETHEE wasused for the ranking projects. To test the model and as a case study,contractors of housing projects in Iran were ranked.

81


� MD-41Monday, 14:30-16:00HH-Heyerdahl

AHP/ANP 3Stream: Analytic Hierarchy Process, Analytic NetworkProcessInvited sessionChair: William Wedley, Faculty of Business Administration,Simon Fraser University, 8888 University Drive, V5A 1S6,Burnaby, BC, Canada, [email protected]

1 - Correct Derivation Of Ratio Scale Preference ValuesFrom Pairwise Comparison MatricesEng Choo, Faculty of Business Administration, Simon FraserUniversity, 8888 University Drive, V5A 1S6, Burnaby, BC,Canada, [email protected], William Wedley

We consider the problem of deriving ratio scale preference values ofn alternatives from pairwise comparison matrix elicited from the deci-sion maker. We characterize the underlying ratio-relevant information.This leads to a necessary condition for using the ratio-relevant infor-mation in T correctly. The multiplicative form is shown to be the onlysimple row by row method that uses the ratio-relevant information in Tcorrectly. The geometric mean method is the only well-known methodthat satisfies the necessary condition for using the ratio-relevant infor-mation in T correctly.

2 - Using Analytic Network Process (ANP) in Four LinksModelIhsan Yüksel, Busines Administration, Kirikkale University,Kırıkkale University, IIBF, 71450, Kırıkkale, Turkey,[email protected], Erdem Aksakal, Metin Dagdeviren

The aim of this study is to evaluate the strategic environment of an or-ganization within the framework of Lynch’s Four Links analysis by us-ing Analytic Network Process (ANP). In the study, firstly factors whichhave an impact on the organization performance were determined andthen inner dependencies between models and factors were examined.The weights of factors were determined on the basis of these examineddependencies and weights were used for a case study to evaluate thestrategic environment. As result it was seen that analyzing the strategicenvironment can be performed by ANP.

3 - A hybrid MCDM approach to assess the sustainabil-ity of students’ preferences for university selectionKabak Mehmet, Department of Industrial Engineering,Turkish Army Academy, Kara Harp Okulu Dekanlıgı, End. veSis.Müh.Böl., 06654, Ankara, Turkey, [email protected],Metin Dagdeviren

An integrated approach which employs ANP and PROMETHEE to-gether is proposed. It is concerned with criteria influencing stu-dent choice in Turkey to establish a model for predicting students’preference for universities. The ANP is used to analyze the struc-ture of the problem and to determine weights of the criteria, and thePROMETHEE is used to obtain final ranking, and to make a sensitivityanalysis. The results indicate that three factors, future career prospectsand opportunities, scholarship and university’s social life, are at the topin the university selection.

4 - Ahp matrix convergence — a unit interpretationWilliam Wedley, Faculty of Business Administration, SimonFraser University, 8888 University Drive, V5A 1S6, Burnaby,BC, Canada, [email protected], Eng Choo

Raising the paired comparison matrix to powers is efficient for calcu-lating the principal eigenvector. The disadvantage is that higher ordercalculations are difficult to comprehend. This paper suggests a modi-fication - to interpret convergence of cell values as the simple averageof many conversion estimates. Invariance of ratios irrespective of theunit of measure is a trait of ratio scales. The eigenvector achieves thistrait, as the geometric mean that is used in group aggregation and hasmany desirable properties, and the simple mean of unit sum columnshas proved to be quite accurate.

� MD-42Monday, 14:30-16:00BW-Amber

EEPA 2012 - 2Stream: EURO Excellence in Practice Award 2012Invited sessionChair: Michel Bierlaire, ENAC INTER TRANSP-OR, ÉcolePolytechnique Fédérale de Lausanne (EPFL), GC B3 454, Station18, CH-1015, Lausanne, Switzerland, [email protected]

1 - SPRINT: Optimization of Staff Management for DeskCustomer Relations ServicesDaniele Vigo, DEIS, University of Bologna, Via Venezia 52,47023, Cesena, Italy, [email protected], Claudio Caremi,Angelo Gordini, Sandro Bosso, Giuseppe D’Aleo, BeatriceBeleggia

Staff Management (SM) is a critical issue in the design and manage-ment of modern Desk Customer Relation Services (DCRSs) due to thelarge amount of resources involved and the increasing pressure towardsmaintaining a very high level of service. We illustrate a Decision Sup-port System for SM, called SPRINT, developed for a large multi-utilitycompany in northern Italy. After one year of service SPRINT has in-troduced a considerable improvement in the process and achieved rel-evant savings in terms of resource consumption and very high level ofservice.

2 - Logistic planning using DSS FlowOptMikael Rönnqvist, Département de génie mécanique,Université Laval, G1V 0A6, Québec, Canada,[email protected], Patrik Flisberg, Mikael Frisk

Skogforsk is the central research body for the Swedish forestry sec-tor. Skogforsk has developed the decision support system FlowOpt. Acentral part is a powerful optimization engine which has been a keycontribution to the success of the system. It makes use of a flexiblenetwork model with extensions including backhauling and cost shar-ing possibilities. The system has been used in many case studies andimplementations during the last 10 years. It has been used to evalu-ate the performance and suggest improvements at single companies aswell as several companies making collaborative schemes.

3 - Flood Prevention by Optimal Dike HeighteningKees Roos, EWI, TU Delft, Mekelweg 4, 2628CD, Delft, ZH,Netherlands, [email protected], Dick den Hertog, RuudBrekelmans, Carel Eijgenraam

The Dutch government is very much aware of the flooding risks thatThe Netherlands is exposed to. This research tries to answer the fun-damental questions of when and how much to invest in which partsof the constituent segments of a dike ring protecting a certain area ofland. This problem is solved by using Mixed-Integer Nonlinear Pro-gramming techniques. For homogeneous dike-rings explicit formulasfor periodic solutions have been obtained. Dutch government agenciesuse the model to analyze the main dike rings and to propose new safetystandards to be incorporated in the Dutch Water Act.

� MD-43Monday, 14:30-16:00BW-Granite

Probabilistic Programming

Stream: Stochastic ProgrammingInvited sessionChair: Tamas Szantai, Institute of Mathematics, BudapestUniversity of Technology and Economics, Muegyetem rkp. 3.,1111, Budapest, Hungary, [email protected]: Andras Prekopa, RUTCOR, Rutgers University, 640Barholomew Road, 08854-8003, Piscataway, New Jersey, UnitedStates, [email protected]

82


1 - A probabilistic constrained stochastic programmingproblem, where the technology coefficients in thestochastic constraints are normally distributed ran-dom variablesAndras Prekopa, RUTCOR, Rutgers University, 640Barholomew Road, 08854-8003, Piscataway, New Jersey,United States, [email protected], Tamas Szantai

A hybrid algorithm will be presented for the solution, where the (outer)supporting hyperplane algorithm is combined with a (inner) cuttingplane method. We report about the numerical experience in connec-tion with the solution method presented. Special emphasis will be puton the calculation of the function and gradient values.

2 - Solution of probabilistic constrained stochastic pro-gramming problems with discrete random variablesand mixed decision variablesKunikazu Yoda, RUTCOR, Rutgers Center for OperationsResearch, Rutgers University, 640 Bartholomew Rd,08854-8003, Piscataway, NJ, United States,[email protected], Andras Prekopa

We consider the probabilistic constrained stochastic programmingproblem where the decision variables are integers and the randomright-hand side vector follows an r-concave discrete distribution. Use-ful formulations of related problems are presented with the concept ofp-efficient points of a probability distribution. We present a cuttingplane type algorithmic solution to the problem.

3 - Single commodity network design under probabilis-tic constraint with continuous random variablesOlga Myndyuk, Rutcor, Rutgers Center for OperationsResearch, Rutgers, The State University of New Jersey, 640Bartholomew Rd, 08854, Piscataway, NJ, United States,[email protected], Andras Prekopa

We formulate and solve problems to find optimal node and arc capaci-ties in a network, where demands are random, under probabilistic con-straints that ensures reliability on a high level. After elimination of theredundant feasibility inequalities the problem is reduced to an LP thatis solved by the combination of an inner and outer algorithm providingus with both lower and upper bounds for the optimum in each iteration.The possible applications include power systems, water supply, traffic,evacuation networks.

4 - Parallelisation of the so-called Cluster BendersDecomposition algorithm for solving two-stagestochastic linear problemsFrancesc Solsona, Computer Science, University of Lleida,Jaume II, 69, 25001, Lleida, Catalunya, Spain,[email protected], LluisM Pla, Josep Lluis Lerida, JordiMateo

L-shaped is a decomposition method used to solve stochastic linearproblems via scenario analysis. This algorithm is made up of threesteps: the first solves the master problem, the second checks for fea-sible cuts and the third checks for optimality cuts. These cuts are se-quentially added to the master problem to be solved whenever eachnew cut is appended. When there is no more cuts, the optimal solutionis found. The cuts are identified by solving auxiliary submodels de-rived from the scenarios. In this paper we propose the parallelisationof the algorithm named Cluster Benders Decomposition

� MD-44Monday, 14:30-16:00BW-Marble

Topics in Revenue Management

Stream: Revenue ManagementInvited sessionChair: Fredrik Odegaard, Richard Ivey School of Business,University of Western Ontario, 1151 Richmond Street North, N6A3K7, London, Ontario, Canada, [email protected]

1 - Illustrating the need to integrate pricing and supplychain decison-makingPeter Bell, Richard Ivey School of Business, University ofWestern Ontario, N6A 3K7, London, Ontario, Canada,[email protected]

This presentation will use three cases that describe real company situ-ations to illustrate the role of pricing in supply chain management. Iwill try to make the point that good pricing decisions make managinga supply chain much easier: often supply chain issues (such as bot-tlenecks, too much or too little inventory or capacity) can be resolvedthrough more informed pricing.

2 - Optimal pricing and production decision in the pres-ence of substitutionSang Won Kim, College of Business Administration,University of Ulsan, 93 Univ. Road, Mugeo-Dong, Nam-Gu,680-749, Ulsan, Korea, Republic Of, [email protected],Peter Bell

Many firms take into account the fact that the creation of price differ-ences between market segments leads to price-driven substitution. Thiseffect will impact the firms’ price and production decisions. In this pa-per, we consider the impact of price-driven substitution on a firm’spricing and production decisions for a single period. We investigatethe impact of the symmetrical and asymmetrical demand substitutionon optimal prices, production levels and revenue.

3 - New results concerning probability distributions withincreasing generalized failure ratesMihai Banciu, School of Management, Bucknell University,119 Taylor Hall, 17837, Lewisburg, PA, United States,[email protected], Prakash Mirchandani

The generalized failure rate is defined as the extension of the hazardrate of a continuous random variable. If the valuation distribution ofa product has an increasing generalized failure rate (IGFR), then theassociated revenue function is unimodal. Assuming the IGFR propertyis thus very useful in pricing applications. In this talk we prove closureof the property under truncations, propose an extension of the originaldefinition of the property, provide a sufficient condition that guaran-tees closure of the IGFR property under convolution, and extend theconcept to discrete distributions.

4 - Retail Selling with All-Pay AuctionsFredrik Odegaard, Richard Ivey School of Business,University of Western Ontario, 1151 Richmond Street North,N6A 3K7, London, Ontario, Canada, [email protected],Chris Anderson

In the fixed list price channel consumers can purchase the item at p1. Inthe all-pay auction consumers place a bid b, which they forfeit regard-less of the outcome, and the highest bid wins the item. We consider amodification where consumers can use their sunk bid as a credit andbuy the item directly for p2. We characterize a symmetric equilibriumin the strategy and derive optimal prices p1 and p2. We consider boththe situation when there is only one seller operating both channels, andwhen there are two sellers operating the two channels.

� MD-45Monday, 14:30-16:00BW-Water

Management accounting andmanagement control 2

Stream: Simulation in Management Accounting andManagement ControlInvited sessionChair: Stephan Leitner, Dept. for Controlling and StrategicManagement, Alpen-Adria-Universität Klagenfurt,Universitätsstrasse 65-67, 9020, Klagenfurt, Austria,[email protected]

83

ME-01 EURO 25 - Vilnius 2012

1 - Dynamic pricing & seat control problem with cancel-lation and refund policy in airlinesMoon-Gil Yoon, Business Administration, Korea AerospaceUniversity, 200-1 hwajun-dong, 471-712, Koyang-shi,Kyunggi-do, Korea, Republic Of, [email protected], HwiYoung Lee

Changing customer behavior and airline market environments in air-line industries require a new mechanism for improving the revenuemanagement. We consider a dynamic pricing and seat control prob-lem for discrete time horizon with cancellation and refund policy inthe context of the airline industry. Applying the linear approximationtechnique and given the price set for each time, we suggest a mixedInteger Programming model to solve our problem efficiently. Fromthe simulation results, we can find our model makes good performanceand can be expanded to other comprehensive problems.

2 - Product cost distortions in full cost accounting - AMonte Carlo simulation-based analysisCatherine Grisar, Institute of Management Control andAccounting, Hamburg University of Technology,Schwarzenbergstraße 95 D, 20173, Hamburg, Germany,[email protected], Matthias Meyer

This paper presents a systematic approach to quantify the effect ofalternative costing system designs on product costs. Starting with astochastic full cost accounting system, we identify design parametersand their possible specifications that are within the scope of cost ac-countants. The impact of these parameters on product costs is investi-gated via simulation. Through experimental design, we can determinethe main and interaction effect of each design parameter. This papercontributes to research and practice as it illustrates the interdependen-cies within a full cost accounting system.

3 - Interactions among biases in costing systems: Asimulation approachStephan Leitner, Dept. for Controlling and StrategicManagement, Alpen-Adria-Universität Klagenfurt,Universitätsstrasse 65-67, 9020, Klagenfurt, Austria,[email protected]

This simulation study presents results on effects of distortions in rawaccounting data on the quality of information provided by costing sys-tems. Results give insights into interactions among biases and indi-cate that multiple biases do not necessarily affect information qualitynegatively. Surprisingly, in some setups interactions among multiplebiases lead to mitigation, or even compensation among biases. Resultsindicate where (not) to tolerate biases and how to prioritize actionsregarding data quality with respect to accuracy and cost of accuracy.

Monday, 16:30-17:30

� ME-01Monday, 16:30-17:30Opera

Plenary Lecture: Professor Finn Kydland

Stream: Keynote, Tutorial and Plenary LecturesPlenary sessionChair: Marielle Christiansen, Department of Industrial Economicsand Technology Management, Norwegian University of Scienceand Technology, Alfred Getz vei 3, N-7491, Trondheim, Norway,[email protected]

1 - Dynamic Programming and EconomicsFinn Kydland, University of California, Santa Barbara, UnitedStates, [email protected]

All interesting phenomena in macroeconomics are dynamic in nature.An overstatement, perhaps, but only a slight one. As a consequence,over the past three or four decades, dynamic programming has been anessential tool, including in my own research. In quantitative aggregateeconomics, dynamic programming plays a key role from model for-mulation to model calibration to computation of model outcomes. Inthe process, for example, of calibrating a model economy, its steady-state relations are an essential part of what it takes to work backwardsfrom empirical relations among variables to what the model parame-ters must be in order to be consistent with them. With advances inthe development of theory and computers becoming much more pow-erful, the set of interesting questions that can be addressed with suchmodels has expanded dramatically in the past decade or two. This istrue especially for questions that dictate the inclusion of important het-erogeneity across current generations and/or across individuals moregenerally.

In models with a role for the government, anticipated future policy af-fects economic behavior in earlier periods. For example, future taxpolicy affects investment behavior today. If we imagine policy to beselected so as to optimize an objective (welfare) function, the resultingpolicy, in the absence of a commitment mechanism, is generally timeinconsistent. Aside from questions about its implementation, then, aconsequence is that the formulation of the optimal-policy problem isnot recursive. But one can generally convert such a model into a re-cursive structure by introducing an additional pseudo-state variable —a shadow price — which is subject to its own constraint implied bythe model. Similar issues about recursivity, potentially resolved in ananalogous manner, arise in many dynamic contracting environments.

84

EURO 25 - Vilnius 2012 TA-02

Tuesday, 8:30-10:00

� TA-01Tuesday, 8:30-10:00RB-Alfa

Tutorial Lecture: Professor Anita SchöbelStream: Keynote, Tutorial and Plenary LecturesTutorial sessionChair: Gerhard Wäscher, Fakultät für Wirtschaftswissenschaft,Otto-von-Guericke Universität Magdeburg, Postfach 4120, 39016,Magdeburg, Germany, [email protected]

1 - Lines, Timetables, Delays: Models and Trends in Op-timization of Public TransportAnita Schöbel, Institute for Numerical and AppliedMathematics, Georg-August Universiy Goettingen,Lotzestrasse 16-18, 37083, Göttingen, Germany,[email protected]

The talk consists of two parts: In the first part, a state-of-the-art sur-vey about finding lines and timetables in the strategic planning of apublic transportation system will be provided, and the problem of de-lay management, i.e. how to update the system in case of delays willbe explored. The second part will discuss ongoing research questions,namely how the routing of the passengers can be included in the plan-ning phases, how an integration of the planning phases can be simu-lated, and how robustness issues can be integrated. Theoretical andexperimental results on these topics will be shown.

Part 1 gives an overview on line planning, timetabling, and delay man-agement. Given an existing public transportation network with its stops(or stations) and direct connections, the first step in the strategic plan-ning of a transport system is to define lines and their frequencies. Aline is a path in the public transportation network along which regularservice is offered. We discuss various optimization models minimiz-ing either the costs of the lines or maximizing the convenience for thepassengers.

If the lines have been found, the next step is to design a timetable.There are two different models: Periodic timetables which are repeatedon e.g. an hourly basis and aperiodic timetables. The usual aim is tominimize the traveling time for the passengers. Integer programmingformulations will be provided for both problem types and algorithmicapproaches for finding timetables will be reviewed.

Coming to the operational phase, the question of the delay manage-ment problem is how to react in case of delays such that the passen-gers’ travel times do not increase too much. This concerns the decisionif a punctual bus or train should wait for a delayed feeder bus or train orif it should better depart on time. In railway traffic also the limited ca-pacity of the tracks has to be taken into account, i.e. priority decisionshave to be made. Also here, the state-of-the art of delay managementwill be presented.

The three problems discussed in Part 1 can be solved more or lessefficiently. However, the algorithms and approaches are still not suit-able for solving real-world problems. In part 2 we will hence discussfeatures that are missing, identify the resulting research questions andshow some first solution approaches aiming at making the algorithmswork in practice.

A first problem concerns the decision of the passengers. In many pa-pers on line planning, time tabling, and delay management it is as-sumed that it is already known how the passengers travel, i.e. a pas-senger’s weight is assumed to be known for every edge or activity inthe system. This is an unrealistic assumption since the route choiceof the passengers depends on the lines, the timetable and on the delaymanagement strategy. We will discuss how the passengers’ decisionsmay be integrated into the optimization process.

Another issue is robustness. It is not helpful to have an exact optimalsolution that might get totally meaningless if a small disturbance oc-curs. Hence, the aim should be to have a robust transportation systemwhich performs well even if the scenario or the input data changes.We exemplary illustrate different concepts on how robustness can beintegrated for the timetabling phase.

Specialized algorithms for every single planning phase in public trans-portation are known. However, the best line plan is not helpful if itonly allows bad timetables. Hence, some effort should be made to findintegrated solutions.

Using our toolbox LinTim we illustrate how decisions made in earlierplanning steps can influence the subsequent planning phases. Amongothers we show how the quality of a timetable depends on decisionsmade in the line planning phase, how good the approximation of thepassengers’ weights fits to the real behavior of the passengers, and howthe robustness of a public transport system may be simulated.

� TA-02Tuesday, 8:30-10:00RB-Beta

Graph Models in Logistics andScheduling

Stream: SchedulingInvited sessionChair: Maksim Barketau, United Institute of InformaticsProblems, Academy of Sciences of Belarus, 220012, Minsk,Belarus, [email protected]

1 - Minimizing maximal weight of subsets in a bipartitegraphMaksim Barketau, United Institute of Informatics Problems,Academy of Sciences of Belarus, 220012, Minsk, Belarus,[email protected], Erwin Pesch, Yakov Shafransky

In a bipartite graph G=(U,V,E) with weighted edges, set U of vertices ispartitioned into disjoint subsets called components. The aim is to find acomplete matching that minimizes the maximal weight of the compo-nent (sum of the weights of the edges that have ends in the component).We prove the strong NP-hardness of the problem and the absence of apolynomial approximation algorithm with a worst-case ratio less than2 (unless P=NP). We develop heuristics that find solutions that are onaverage 2 percent away from the optimum. The problem is motivatedby a production planning in a railway hub.

2 - Batch scheduling and transfer line design problemsAlexandre Dolgui, IE & Computer Science, Ecole des Minesde Saint Etienne, 158, cours Fauriel, 42023, Saint Etienne,France, [email protected], Sergey Kovalev, Mikhail Y.Kovalyov, Jenny Nossack, Erwin Pesch

We study a transfer line design problem, in which there are severaltypes of operations to be assigned to stations. If a station is assigned atleast one operation of a certain type, it induces a type dependent cost.Precedence relations on the set of operations are given by an arbitrarydigraph. The primary objective is to minimize the number of stationsand the secondary objective is to minimize the total type dependentcost. An equivalent batch scheduling model is presented. Obtainedresults include computational complexity, heuristics, ILP formulationsand computer experiments.

3 - Periodic Scheduling for Wireless Access Networks tothe InternetCelia Glass, Cass Business School, City University, 106Bunhill Row, EC1Y 8TZ, London, United Kingdom,[email protected]

Wireless Access Networks provide low cost internet and broadbandaccess, but have distinct limitations for scheduling. The access points,used e.g. for WiFi, cannot multi-task. This disposes them to perfect pe-riodic scheduling, which offers both reliability of service, and energysaving. We consider binary tree access networks with local transmis-sion. We compare optimised and heuristic solutions, to demonstratethe difficulty of co-ordinating network transmissions without periodic-ity, and the efficiency of perfect periodic scheduling.

4 - New Bounds and Constraint Propagation Techniquesfor the Clique Partitioning ProblemErwin Pesch, FB 5, University of Siegen, Hoelderlinstr. 3,57068, Siegen, Germany, [email protected], FlorianJaehn

85

TA-03 EURO 25 - Vilnius 2012

We consider the Clique Partitioning Problem that has several real lifeapplications such as groupings in flexible manufacturing systems, inbiology, in flight gate assignment, etc.. We present tighter upperbounds for each search tree node than those known from literature, im-prove constraint propagation techniques for fixing edges in each node,and present a new branching scheme. The theoretical improvementsare reflected by computational tests with real life data.

� TA-03Tuesday, 8:30-10:00RB-L1


Stream: Logistics, Transportation, TrafficContributed sessionChair: Erwin van der Laan, RSM Erasmus University, P.O.Box1738, 3000DR, Rotterdam, Netherlands, [email protected]

1 - The efficiency improving of traction drive test benchwith supercapacitor energy storage systemGenadijs Zaleskis, Faculty of Power and ElectricalEngineering, Institute of Industrial Electronics and ElectricalEngineering, Riga Technical University, 1 Kronvalda Blvd.,LV-1010, Riga, Latvia, [email protected], ViestursBrazis

In this article the results of computer modeling for the optimizationof traction drive test bench with supercapacitor energy storage are de-scribed. Test bench operation is considered in cases of the energy stor-age system working at various selected supercapacitor initial voltages.Maximal increasing of possibility of vehicle test bench regenerativebraking with minimal decreasing of autonomous power supply modepossibility is investigated. There is estimated the energy storage sys-tem efficiency improving measures dependence from supercapacitoroperational voltage ranges.

2 - How to place products in a warehouse to minimizeforklifts worktime — a simple data mining based al-gorithmGrzegorz Tarczynski, Department of Operational Research,Wroclaw University of Economics, ul. Komandorska 118/120,53-345, Wroclaw, Poland, [email protected]

Based on historical data, using data mining techniques, probability dis-tributions and conditional probability distributions of goods purchasecan be determined and used to plan the allocation of goods in the ware-house. The Volume-Based Storage Strategies focus on how to putgoods with the highest rotation in locations most easily available towarehouseman. However, often customers buy goods in sets. This pa-per presents an algorithm that takes into account the most purchasedproduct sets. The effectiveness of the algorithm was verified usingsimulation tool.

3 - How the NPV and average cost criterion lead to dif-ferent decisions in the vending machine problemErwin van der Laan, RSM Erasmus University, P.O.Box 1738,3000DR, Rotterdam, Netherlands, [email protected]

The net present value (NPV) criterion is widely seen as an appropriateapproach to deal with financial decision. In literature, however, it ismainly the average cost (AC) criterion that is used. We analyze theNPV and AC approach for the logistics of replenishing vending ma-chines: every R time units the machine is inspected and replenishedup to its capacity, where R should be chosen optimally. We show thatthere may be significant implications of using the AC criterion over theNPV criterion in terms of profit expectations and realizations.

� TA-04Tuesday, 8:30-10:00RB-L2

Air Transportation and Propagation

Stream: Optimization in Public TransportInvited sessionChair: Claus Gwiggner, Information Systems, FU Berlin, Garystr.21, 14195, Berlin, Germany, [email protected]

1 - Optimizing the Staff Schedule for Airport Immigra-tionPhi Doan, Aeronautics and Astronautics, University of Tokyo,4-6-1 Komaba, Meguro-ku, Rcast building 4, room 515,153-8904, Tokyo, Tokyo, Japan, [email protected]

The objective of this research is to shorten the waiting times at an in-ternational airport by optimizing the staff schedule. First, an arrivalmodel predicts the arrival rates at the immigration desks by using theflight schedule. Second, a queueing model estimates the waiting timesfor given service times and staff schedule. Last, an optimization modelthat uses an evolutionary algorithm estimates the number of staff overthe day with the goal of minimizing the staff hours while meeting atargeted maximum waiting time.

2 - Improvement of Static Runway Assignment UsingQueueing ModelRyota Mori, Air Traffic Mangement Department, ElectronicNavigation Research Institute, 7-42-23,Jindaiji-Higashimachi, Chofu, 182-0012, Tokyo, Japan,[email protected]

Congestion is a critical issue at most hub airports. This paper tacklesthis problem occurring at the airport with most passengers in Japan, i.e.Tokyo International Airport. It has four intersecting runways which in-terfere with each other. Although the runway an aircraft uses is deter-mined by its destination or departure airport to reduce air traffic con-trollers’ (ATC) workload, this general rule can increase aircraft waitingtime. To make runway operations more efficient, a queuing model isimplemented and the possible improvement is discussed with keepingminimum increase of ATC workload.

3 - Optimal Trajectory Design by Infinite DimensionalOptimizationDaniel Delahaye, French Civil Aviation University, 7 avenueEdouard Belin, 31055, Toulouse, France,[email protected], Stephane Puechmorel

Trajectories are infinite dimensional objects. There are mainly twoclasses of algorithms for trajectory design. The first class is based onoptimization and the second one on propagation algorithms. In thiswork we present two examples of such algorithms for aircraft trajec-tory design. Both address the conflict resolution problem. The firstone manipulates B-splines and the second one uses a light propaga-tion analogy to generate optimal geodesic trajectories. Results are pre-sented on a real day of traffic over France for which all conflicts aresuccessfully solved.

4 - Sequencing and Swapping Probabilities for TrafficSynchronizationSakae Nagaoka, Air Traffic Management, ElectronicNavigation Research Institute, 7-42-23 JindaijiHigashi-machi, Chofu-shi, 182-0012, Chofu, Tokyo, Japan,[email protected], Claus Gwiggner, Yutaka Fukuda

In air traffic flow planning processes, estimated times of arrival arebased on information from departure times at the origin airport. How-ever, uncertainties about this information vary from aircraft to aircraft.We propose a probabilistic method for adjusting the scheduled time ofarrival for aircraft pairs assuming that rough information on the dis-tributions of estimation errors is available. We discuss the probabilityof swapping aircraft pairs and the conflict probability in relation to thesize of the separation buffer often imposed to the separation minimum.

86


� TA-05Tuesday, 8:30-10:00RB-L3

Maritime Transportation in OffshoreIndustry

Stream: Maritime TransportationInvited sessionChair: Irina Gribkovskaia, Faculty of Economics, Informatics andSocial Sciences, Molde University College - SpecializedUniversity in Logistics, Postbox 2110, 6402, Molde, Norway,[email protected]

1 - A simulation study of the fleet sizing problem for off-shore supply vesselsYauhen Maisiuk, Faculty of Economics, Informatics, andSocial Sciences, Molde University College - SpecializedUniversity in Logistics, Britvegen 2, 6411, Molde, Norway,[email protected], Irina Gribkovskaia

The supply vessels provide offshore installations with supplies froman onshore base according to weekly sailing plans. The goal is to de-cide the cost-optimal fleet of vessels on the long-term hire. The prob-lem is stochastic as sailing speed and service time depend on uncer-tain weather conditions. Due to delays some installations may not beserved within the planned route duration, and not received supplieshave to be delivered later. We propose a discrete-event simulationmodel which evaluates alternative fleet size configurations dependingon several strategies of later delivery of supplies.

2 - Routing and scheduling of offshore supply vesselswith fuel consumption minimization through deter-mining the best discrete value of sailing speed oneach legTatsiana Barysavets, Molde University College - SpecializedUniversity in Logistics, Norway,[email protected], Irina Gribkovskaia, MikalaiMikhailau

We consider the offshore supply vessel routing and scheduling prob-lem with incorporated speed decisions where the aim is to minimizethe fuel consumption. The value of the vessel speed on each sailingleg is assumed to be taken from a given set. The problem is formulatedas a MIP model. Two heuristic approaches are proposed: two-phaseapproach where the set of feasible voyages is generated a priory, andthe problem is solved with a set partitioning like model; and a largeneighbourhood search heuristic where the speed on each sailing leg isdetermined simultaneously with the vessels schedule.

3 - Routing of supply vessels to offshore installationswith deliveries and pickups of multiple commoditiesEugen Sopot, Molde University College, Kirkebakken 27,6413, Molde, Vestland, Norway, [email protected],Irina Gribkovskaia, Gilbert Laporte

Offshore installations are dependent on periodic deliveries of variousgoods from an onshore supply base and continuously generate pickupdemands. We consider the problem of routing supply vessels to off-shore installations as a multi-commodity vehicle routing problem withpickups and deliveries. To guarantee the existence of a feasible solu-tion to the problem it is assumed that each installation may be visitedonce or twice. We propose a variable neighborhood search heuristicalgorithm for the single vessel variant of this problem.

4 - Speed strategies for green supply vessel planningEllen Karoline Norlund, Faculty of Economics, Informaticsand Social Sciences, Molde University College - SpecializedUniversity in Logistics, Britvegen 2, 6402, Molde, Norway,[email protected], Irina Gribkovskaia

We study how to reduce emissions in supply vessel activities throughspeed optimization when planning vessel schedules by solving a MIPmodel for a pre-generated set of voyages. Several speed strategies, de-termining speed for the entire voyage or optimizing speed for each voy-age leg, are proposed. These strategies are applied either a priori dur-ing voyage generation or a posterior to the voyages in the constructed

schedule. Results of tests on several real-life instances show that emis-sions can be significantly reduced when using speed optimization insupply vessel planning.

� TA-06Tuesday, 8:30-10:00RB-Gamma

Engineering and Hybrid search

Stream: MetaheuristicsInvited sessionChair: Saïd Hanafi, ISTV2, LAMIH-SIADE, University ofValenciennes, Le Mont Houy, 59313, Valenciennes, France,[email protected]

1 - Local search methods for conflict-free routing in amulti-processor system on chipMarc Sevaux, Lab-STICC - CNRS UMR 3192, Université deBretagne Sud - UEB, Centre de Recherche - BP 92116, 2 ruede Saint Maude, 56321, Lorient, France,[email protected], Boureima Zerbo, André Rossi,Jean-Charles Creput

We present a combinatorial optimization problem for conflict-freerouting in a Network-on-Chip. Based on time division multiplexingand cyclic emission, the problem consists in finding a set of k-shortestpaths, such that packets will never conflict through the network butcan use shared communication links in an efficient way. On a time-expanded graph, a set of shortest paths computation operators are pre-sented. Then, based on such operators, three iterated local searchschemes are presented and compared to address the problem.

2 - A PSO-based Heuristic for energy-aware schedulingof Workflow applications on cloud computingSonia Yassa, Val d’oise, EISTI, Avenue du Parc, 95000,Cergy, France, [email protected], Rachid Chelouah,Hubert Kadima, Bertrand Granado

We investigate the problem of scheduling workflow applications oncloud computing infrastructures. The cloud workflow scheduling isa complex optimization problem which requires considering variousscheduling criteria. Traditional researches mainly focus on optimizingthe time and cost without paying much attention to energy consump-tion. We propose a new approach based en hybrid PSO heuristic tooptimize the scheduling performance. Our method is based on dy-namic voltage scaling (DVS) technique to minimize energy consump-tion. Simulation results emphasize the robust performance of our ap-proach.

3 - Software Interfaces for Heuristic Solvers and the Fu-ture of HyflexAndrew J. Parkes, School of Computer Science, University ofNottingham, Jubilee Campus, Wollaton Road, NG9 1BB,Nottingham, United Kingdom, [email protected], MatthewHyde, Gabriela Ochoa, Ender Özcan

Stochastic local search is often highly effective; however, many prac-tical applications remain unexploited, not because known algorithmswould fail, but simply because the development cost is too high. Wediscuss recent Computer Science work at the University of Notting-ham to address this challenge; and to support rapid prototyping andcomponent-based development by increasing the modularisation ofheuristic search tools. In particular, we will discuss future extensionsto Hyflex, an API to separate adaptive search techniques from specificdetails of the problem domain.

4 - An Application of Electromagnetic Meta-HeuristicMethod for Solving System of Nonlinear Fredholm In-tegral EquationsHassan Taheri, Mathematics, Khayyam University,91775-1159, Mashhad, Khorasan, Iran, Islamic Republic Of,[email protected]

87


In this paper we proposed an efficient algorithm for solving system ofnonlinear Fredholm integral equations (SNFIEs) by using the Electro-magnetic Meta-Heuristic method (EM). First we choose an approxi-mate solution for SNFIE. Then we will have a system of nonlinear ofequations (SNEs). If we cannot compute this integral because of theintegrands involves some unknowns, then we choose a value for eachconstant. But the question have is how do we select these values toachive a good solution for the SNFIE? To ensure this question we tryto solve the problem by EM method. For solving the SNEs we pro-posed a hybrid of the Newton-GMRES and EM methods.

� TA-07Tuesday, 8:30-10:00RB-Eta

Vehicle routing in practical settings

Stream: New Trends in Vehicle RoutingInvited sessionChair: Jorge E. Mendoza, Institute of Applied Mathematics,Université Catholique de l’Ouest, 3 Place André Leroy, 49008,Angers, France, [email protected]: Victor Pillac, Automatique - Productique, Ecole des Minesde Nantes, 4 Rue Alfred Kastler, La Chantrerie, 44300, Nantes,France, [email protected]

1 - Solving an Industrial Waste Collection Problem usinga Hybrid Column Generation AlgorithmJesper Larsen, Department of Management Engineering,Technical University of Denmark, Produktionstorvet,Building 426, 2800, Kgs. Lyngby, Denmark,[email protected], Kristian Milo Hauge

We present a problem of collection industrial waste. In this problemwaste collection is handled by a fleet of vehicles. Each vehicle cancarry up to eight skip containers of which a maximum of two can befull at the same time. We present an approach to route the vehiclesbased on a combination of column generation (CG) and metaheuristics.The performance is assessed by comparing the results that it achievesto results achieved by an algorithm that is based purely on metaheuris-tics. The results show that the hybrid CG algorithm finds superioirsolutions than the purely heuristic algorithm.

2 - Multi-stage, multi-period distribution planning withinventory and routing considerationsDmitry Ivanov, Supply Chain Management, Berlin School ofEconomics and Law, Germany, [email protected], BorisSokolov, Alexander Pavlov

A real case-study of distribution planning is considered. An origi-nal approach of combining static inventory-routing and optimal con-trol models is proposed to optimal planning of a multi-stage, multi-commodity distribution network in the multi-period mode with adap-tive update of demand, capacity, and supply information. Optimal dis-tributions plans are calculated for optimistic and pessimistic scenar-ios. Subsequently, these plans are analyzed subject to different distur-bances and regarding distribution network design and sourcing plan-ning decisions.

3 - Extending the Periodic Vehicle Routing Problem: ACase Study in Beverage LogisticsDominik Pfeiffer, Chair for Information Systems and SupplyChain Management, University of Münster,Leonardo-Campus 3, 48149, Münster, NRW, Germany,[email protected], Bernd Hellingrath

This paper deals with a variant of the Periodic Vehicle Routing Prob-lem which allows multiple tours to be run per vehicle and day, deliv-ering each customer no more than once a day, respecting driving timeconstraints, and distributing multiple deliveries to the same customerin the planning period as evenly as possible. We develop algorithmssolving the instances of the case studied. The results are validatedagainst historical data from the case at focus. We are able to realizeconsiderable cost reductions and evaluate the algorithms with respectto the solution quality and efficiency.

4 - A real-life Inventory Routing ProblemBenoît Tricoire, Institut de Mathématiques Appliquées,Université Catholique de l’Ouest, 3 Place André Leroy,49008, Angers, France, [email protected], Eric PinsonWe focus on a real-life problem that integrates stock constraints intothe VRP. This problem, motivated by routing problems faced by re-tail companies, can be related to Inventory Routing Problems. Ouraim is to build a planning of pick-ups (to suppliers) and deliveries(to customers), deserved at minimum cost by a set of vehicle activi-ties (routes verifying complex real-world operative constraints). Wepropose a three-step method for our problem, using Lagrangean relax-ation, GRASP, column generation, and tabu search. We compare ourresults to those of end-users

� TA-08Tuesday, 8:30-10:00RB-Epsilon

Supply Chain Design

Stream: Supply Chain PlanningInvited sessionChair: Alparslan Serhat Demir, Industrial Engineering, SakaryaUniversity, Sakarya University Esentepe Kampusu Endustri Muh.Binasi M5 Kat 3-311, 54100, Sakarya, Turkey,[email protected]

1 - Use of decomposition method in forest fuel supplychainAmirhossein Sadoghi, Finance Department, Frankfurt Schoolof Finance & Management, Sonnemannstraße 9-11, 60314,Frankfurt am Main, Germany, [email protected], HeleneLidestamIn this research, we address the capacitated, multi-commodity, multi-period, multi-stage facility location problem regarding production, dis-tribution and allocation in the case that forest residues are to be con-verted into fuel. With increased number of facilities, products and timeperiods, the model becomes large and large scale optimization tech-niques are necessary to get a solution within limited time. A heuristicsolution based on a computationally efficient decomposition methodis proposed. Numerical results and comparison with other possiblemethods are presented.

2 - Modular Supply Chain Optimization in the Construc-tion Industry: An ApplicationPedro Rijo, DECivil, IST, Rua Comandante Cousteau, lote4.04.02.B 3oD, 1990-067, Lisboa, Portugal, [email protected],Amílcar ArantesGlobalization and the need for product differentiation at a fair pricemade supply chain a particularly relevant issue for the construction in-dustry. The companies have to look for decisions that optimize thesupply chain system and integrate in the process their suppliers andclients. Issues like facility location, inventory and transport manage-ment must be considered. In this work we focus on the optimization ofthe supply chain of a complete prefabricated kitchen module using lin-ear programming techniques. The results show guidelines concerningsupply chain configurations and operations.

3 - Robust Supply Chain Network Design by Consider-ing Different Risk Mitigation StrategiesFaeghe Mohamad Doost, industrial engineering, urmiauniversity of technology, Band Ave, Urmia, Iran, Urmia, WestAzarbayjan, Iran, Islamic Republic Of,[email protected], Shabnam RezapourThis paper develops a stochastic mathematical formulation for design-ing the network of a supply chain comprising several production fa-cilities, distribution centers and retailers in markets under uncertainty.In this model, we consider reserving extra capacity, holding strategicstock and keeping substitutable facility strategies to mitigate the effectsof possible disruptions. We use a path-based formulation that helps usconsider supply-side uncertainties. We illustrate the model outputs anddiscuss the results through several numerical examples.

88


4 - Determining Multiple Warehouse Locations and Cus-tomers to be Served by Using Genetic AlgorithmAlparslan Serhat Demir, Industrial Engineering, SakaryaUniversity, Sakarya University Esentepe Kampusu EndustriMuh. Binasi M5 Kat 3-311, 54100, Sakarya, Turkey,[email protected]

In this study, distribution network of a tire manufacturing factory is ad-dressed. In the distribution network, customers are at the different geo-graphical locations and warehouses have to be established in differentplaces to meet the product demand of customers. The problem is thedetermination of the locations where the warehouses are establishedand which warehouse will meet the customers’ demands to minimizetotal transportation cost. The problem is modeled and solved by usinggenetic algorithm technique.

� TA-09Tuesday, 8:30-10:00RB-Zeta


Stream: Cutting and PackingInvited sessionChair: José Fernando Gonçalves, Faculdade de Economia doPorto, Universidade do Porto, Rua Dr. Roberto Frias, s/n,4200-464, Porto, Portugal, [email protected]

1 - A Comparison of Approaches for Solving the 3D Con-tainer Ship Loading Planning Problem by Represen-tation by RulesAnibal Azevedo, Production engineering, State University ofCampinas, Rua dos Aimorés, 480, Apt. P.34, 13081-030,Campinas, São Paulo, Brazil, [email protected], CassildaRibeiro, Galeno Sena, Antônio Chaves, Luiz Salles Neto,Antônio Moretti

The proposed paper presents two contributions: A new 3D ContainerShip Loading Planning Problem modeled as a bi-objective frameworkthat minimizes two criteria: the number of movements and also theinstability measure computed as the distance between gravity and geo-metric center. A new representation that allows a compact and efficientencoding which enables the solution of large-scale problems proper re-lated with real-life scenarios. For example, an instance problem whichdemands a total of 40,545,000 binary variables will need only 30 vari-ables with the proposed approach.

2 - A global optimization approach for three dimensionalrectangular packing problemsPei-Chun Wang, Graduate Institute of Industrial and BusinessManagement, Taipei University of Technology, No. 1, Sec. 3,Chung Hsiao E. Road, Taipei 106, Taiwan, 106, Taipei,Taiwan, [email protected]

The problem addressed in this paper aims to pack a set of rectangularboxes into a large rectangular container in such a way that the occu-pied volume space is minimized. The boxes can be any rectangularstackable objects with different sizes and may be rotated. This papertransforms the packing problem into a mixed-integer linear program byan improved piecewise linearization technique which uses fewer extra0-1 variables and constraints. The reformulated model can be solvedto obtain a global optimum and is more computationally efficient thanprevious studies with deterministic approaches.

3 - Constraint programming approaches, search strate-gies and bounds in problems of orthogonal packing.Marat Mesyagutov, Numerical Mathematics, DresdenUniversity of Technology, Zellescher Weg 12 - 14 (C 319),01069, Dresden, Germany, [email protected],Gleb Belov, Guntram Scheithauer

Constraint programming is an effective approach for discrete optimiza-tion problems. Since it allows the non-linear modeling, we considersome known higher dimensional approaches within that new paradigm.In the talk we discuss different know and new strategies, pruning rules

based on linear programming and some further amplification of theconstraint propagation, which are followed by the results of experi-mental study.

4 - A biased random key genetic algorithm for the 2Dand 3D bin packing problemJosé Fernando Gonçalves, Faculdade de Economia do Porto,Universidade do Porto, Rua Dr. Roberto Frias, s/n, 4200-464,Porto, Portugal, [email protected]

In this paper we propose a biased random key genetic algorithm for the2D and 3D bin packing problem. For each GA chromosome we con-struct a solution based on a maximal-space heuristic which packs theboxes according to the order supplied by chromosome. Two placementheuristics are considered and an improvement procedure is applied tothe solution. Computational tests carried on a set of 854 benchmarkinstances taken from the literature validate the quality of the solutionsand the approach. Supported by Fundação para a Ciência e Tecnologia(FCT) project PTDC/EGE GES/117692/2010.

� TA-10Tuesday, 8:30-10:00RB-Theta


Stream: Production Management & Supply ChainManagement (contributed)Contributed sessionChair: Farouk Yalaoui, Institut Charles Delaunay, ICD LOSI,University of Technology of Troyes, 12, Rue Marie Curie BP2060, 10000, Troyes, France, [email protected]: Christian Larsen, Economics, CORAL, Aarhus School ofBusiness, Aarhus University, Fuglesangs Alle 4, DK-8210, AarhusV, Denmark, [email protected]

1 - Dynamic supply chain inventory management: ageneric mathematical programming approachJoaquim Jorge Vicente, CEG - IST, Portugal,[email protected], Susana Relvas, Ana PaulaBarbósa-Póvoa

Inventory management plays an important role in supply chains. Thisis explored in the present work where a flexible Inventory Manage-ment (IM) policy is modeled through a mixed-integer linear pro-gramming model. This dynamically optimizes a multi-period, multi-warehouse/retailer and multi-product supply chain under deterministicand uncertain demands. Continuous and Periodic Review policies arealso modeled and tested over the same system for comparison. The re-sults show that the proposed IM policy outperforms the others leadingto lower operational costs.

2 - The relationship between purchasing strategy, man-ufacturing performance and financial performanceAhmed Attia, business management, damanhour university,elgomohoria st., 000, damanhour, elbehera, Egypt,[email protected]

Purchasing is one of the logistics activities playing a major role for get-ting a competitive advantage through reducing the total cost, reducingthe number of suppliers, building partnership with the suppliers, etc.The purchasing strategy will be the effective tool for achieving suchbenefits. The current study examines the direct effect of the purchas-ing strategy on the financial performance of the company, and also theindirect effect through manufacturing performance.

3 - Spare parts sharing with joint optimization of main-tenance and inventory policiesChristian Larsen, Economics, CORAL, Aarhus School ofBusiness, Aarhus University, Fuglesangs Alle 4, DK-8210,Aarhus V, Denmark, [email protected], Hartanto Wong, LarsRelund Nielsen

89


We consider a collaborative arrangement where a number of compa-nies are willing to share expensive spare parts, required for both fail-ure replacement and preventive maintenance purposes. We developa discrete-time Markov decision model for the joint optimization ofmaintenance and spare parts inventory policies for both a single- and amulti-company problem. The cost savings obtained as a result of shar-ing are evaluated. A number of cost allocation schemes, making all thecompanies involved being better off by sharing the spare parts ratherthan acting independently, are developed.

� TA-11Tuesday, 8:30-10:00RB-Iota

Transport Networks

Stream: Transportation and LogisticsInvited sessionChair: Anna Sciomachen, DIEM, University of Genova, ViaVivaldi 5, 16126, Genova, Italy, [email protected]

1 - Feeder bus network design problem: a new solvingprocedure and real size applicationsFrancesco Ciaffi, Dipartimento di Scienze dell’IngegneriaCivile, Università Roma Tre, Via Vito Volterra 62, 00146,Roma, Italy, [email protected], Ernesto Cipriani, MarcoPetrelli

The present research deals with a methodology for solving the feederbus network design problem in a real size large urban area. The solv-ing procedure is articulated in 2 phases: generation of two differentand complementary sets of feasible routes, using a heuristic algorithm,and the optimal network design using a GA, calculating simultaneouslyoptimal routes and frequencies, in order to provide a good balance be-tween maximization of the service coverage area and minimization ofthe overall travel time. The proposed procedure has been applied ontwo real-life networks, Winnipeg and Rome.

2 - Delivery splitting problem with multiple originsTomas Subrt, Dept. of Systems Engineering, CzechUniversity of Life Sciences, Kamycka 129, 165 21, Prague 6 -Suchdol, Czech Republic, [email protected]

This problem deals with logistic network design issues and is focusedon optimal splitting of deliveries from multiple sources to multiple des-tinations. It is supposed to have a logistic network where the summarysupply of all destinations can be satisfied by sources’ capacities. Ev-ery arc of such network is weighted by cost per item shipment func-tion. Due to the common effect of flow scale economies such func-tion is generally concave. This article demonstrates possible solutionmethods and results implementation using mathematical programmingtools.

3 - Freight terminal location in multimodal networksAnna Sciomachen, DIEM, University of Genova, Via Vivaldi5, 16126, Genova, Italy, [email protected],Daniela Ambrosino

In this work we present a heuristic method for evaluating possible lo-cations for freight intermodal terminals, that are dry ports. Aspectscoming from both classical simple plant location problems and shortestpath ones on multimodal graphs are combined in a two phase proce-dure. We first select the possible dry ports by analysing their position inthe network and their communication capabilities with the other nodesby using a multimodal connectivity criterion. Then we apply a heuris-tic algorithm for finding optimal multimodal O-D routes in network.Preliminary results are given

� TA-13Tuesday, 8:30-10:00RB-Tau

Semiconductor and PharmaceuticalsProductionStream: Manufacturing and WarehousingInvited sessionChair: Lars Moench, FernUniversität in Hagen, 58097, Hagen,Germany, [email protected]

1 - Stochastic Programming for New Product Introduc-tion in the Pharmaceutical IndustryKlaus Reinholdt Nyhuus Hansen, Section for Supply ChainManagement and Production, TU Munich, Arcisstrasse 21,80333, Munich, Germany, [email protected], MartinGrunow

Due to limited patent protection, pharmaceutical companies are focus-ing more on the new product introduction process and addressing therisks found prior to product launch. Where the widespread pipeline andcapacity planning models address R&D risks, we here focus on exoge-nous risks arising from the negotiations with the authorities such ase.g. delayed market authorization. We present how these risks can beaddressed via two-stage stochastic modeling for effective launch plan-ning and also how capacity can be improved to consider these risks.

2 - Enhancing standard mid-term planning in semicon-ductor manufacturingPhillip Kriett, Technical University Munich, Germany,[email protected], Martin Grunow

Enterprise-wide tactical planning determines a capacity-feasible prod-uct mix. State-of-the-art tactical planning frameworks aggregate re-source information and neglect the assignment of jobs to machines.Yet, certain semiconductor manufacturing steps such as testing canbe performed on flexible, parallel, and in particular non-identical re-sources. Here, this aggregation results in low capacity utilization andinfeasible production plans. We propose an exact solution approachwhich captures resources and is compatible with current software. Theapproach is tested on real-world data.

3 - Order release planning by iterative linear program-ming and simulation: An analysis of the underlyingcoordination mechanismHubert Missbauer, Information Systems, Production andLogistics Management, University of Innsbruck,Universitätsstrasse 15, 6020, Innsbruck, Austria,[email protected]

We deal with order release planning algorithms in manufacturing plan-ning and control that allow time-varying lead times by iterating be-tween a release models with fixed lead times and a simulation modelthat updates the lead times. Convergence is a critical issue here, anda theoretical foundation has not been provided up to now. We presentan analysis of the theory behind this iterative mechanism that shouldhelp to explain the convergence problems and to improve the methods.The analysis is performed using a single-stage order release model indiscrete and continuous time.

4 - A Scheduling Approach for Complex Job Shops withTransportationLars Moench, FernUniversität in Hagen, 58097, Hagen,Germany, [email protected], Rene Driessel

We consider a scheduling problem for complex job shops with auto-mated material handling systems (AMHS). Scheduling problems ofthis type are important in semiconductor manufacturing. The perfor-mance measure is the total weighted tardiness. We present a modi-fied shifting bottleneck heuristic for the integrated scheduling prob-lem. The heuristic is applied in a rolling horizon setting. We study theperformance using simulation. In addition, we change the number ofvehicles to see the impact of the capacity of the AMHS on the overallperformance.

90


� TA-14Tuesday, 8:30-10:00RB-Omega

Scheduling with variable parameters II

Stream: Scheduling under Resource ConstraintsInvited sessionChair: Yakov Shafransky, United Institute of InformaticsProblems, National Academy of Sciences of Belarus, Surganov str.6, 220012 Minsk, Belarus, 220012, Minsk, Belarus,[email protected]: Stanislaw Gawiejnowicz, Faculty of Mathematics andComputer Science, Adam Mickiewicz University, Umultowska 87,61-614, Poznan, Poland, [email protected]

1 - Minimizing Maximum Lateness for Single Machineunder Uncertain Due Dates and Precedence Con-straintsDzmitry Sledneu, Lund University, Sweden,[email protected], Yakov Shafransky

Problem of minimizing maximum lateness for one machine underprecedence constraints is considered. For each due date we have aninterval of its possible values and the due date may take any valuefrom this interval regardless of the will of the decision maker. A fea-sible schedule is called globally optimal schedule if it remains optimalunder any possible values of the due dates. We formulate necessaryand sufficient conditions for a schedule to be globally optimal and pro-pose an algorithm for constructing such a schedule. As a by-product,we formulate necessary and sufficient conditions for the optimality ofa schedule for the deterministic version of the problem.

2 - Scheduling precedence-constrained jobs with mixedprocessing times and maximum cost criterionMarek Debczynski, Faculty of Mathematics and ComputerScience, Adam Mickiewicz University, Umultowska 87Street, 61-614, Poznan, Wielkopolskie, Poland,[email protected], Stanislaw Gawiejnowicz

We consider single-machine scheduling problems with arbitrary prece-dence constraints and maximum cost criterion. We show how to solvethe problems in polynomial time when job processing times are vari-able and mixed, i.e. some of them are fixed, while the other ones arevariable and take into account the effects of learning, ageing or jobdeterioration.

3 - Scheduling jobs on unrelated parallel machines withgeneral positional deterioration to minimize the totalprocessing costYakov Shafransky, United Institute of Informatics Problems,National Academy of Sciences of Belarus, Surganov str. 6,220012 Minsk, Belarus, 220012, Minsk, Belarus,[email protected]

Set of independent jobs is to be processed on unrelated parallel ma-chines. The processing time of a job is the product of its basic pro-cessing time and a deterioration coefficient that depends on the job, onthe machine and on the position of the job in the job sequence on thismachine. Each of the coefficients is a non-decreasing function of theposition number. For each machine, we have a cost of processing a jobfor one time unit on the machine. The aim is to construct a schedulethat minimizes the total cost of processing all the jobs. We propose apolynomial algorithm to solve the problem

4 - Multicriteria optimisation of construction projectschedulesGrzegorz Ginda, Faculty of Materials and EnvironmentSciences, University of Bielsko-Biała, INSTITUTE OFTEXTILE ENGINEERING AND POLYMER MATERIALS,Pl.Fabryczny 5, 43-300, Bielsko-Biała, Poland,[email protected], Miroslaw Dytczak, TomaszWojtkiewicz

Complexity of optimal construction project scheduling results fromnumerous feasible orders of technological operations, availability ofdifferent operation realisation modes and possible concurrency among

operations due to limited availability of resources. Efficient approachis discussed which allows identification of the most suitable projectschedule instances. It applies simultaneous generation and evaluationof representative population of schedule instances. A criteria hierarchyis applied with this regard. The approach also facilitates identificationof desirable schedule features.

� TA-15Tuesday, 8:30-10:00RB-2101

Semi-infinite methods and applications

Stream: Semi-Infinite OptimizationInvited sessionChair: Ana I. Pereira, Department of Mathematics - ESTiG,Polytechnic Institute of Braganca, Campus de Sta Apolonia,Apartado 134, 5301-857, Braganca, Portugal, [email protected]

1 - Strong duality in robust linear semi-infinite program-mingMiguel Goberna, Estadística e Investigación Operativa,Universidad de Alicante, Ctra. San Vicente s/n, 03080, SanVicente del Raspeig, Alicante, Spain, [email protected] this talk, we propose a duality theory for semi-infinite linear pro-gramming problems under uncertainty in the constraint functions, theobjective function, or both, within the framework of robust optimiza-tion. We show that robust duality holds whenever a robust momentcone is closed and convex. We then show that robust moment coneis closed and convex if and only if robust duality holds for every lin-ear objective function of the program. We also give robust forms of theFarkas lemma for systems of uncertain semi-infinite linear inequalities.

2 - General semi-infinite programmingVladimir Shikhman, Dept. Mathematics C, RWTH AachenUniversity, Templergraben 55, 52056, Aachen, Germany,[email protected], Hubertus Th. JongenWe consider general semi-infinite programming (GSIP) from a topo-logical point of view. We focus on the description of the closure of thefeasible set, the Symmetric Mangasarian Fromovitz Constraint Quali-fication, the Nonsmooth Symmetric Reduction Ansatz, appropriate no-tion of Karush-Kuhn-Tucker points and critical point theory.

3 - Reduction method with multistart technique forsemi-infinite programming problemsAna I. Pereira, Department of Mathematics - ESTiG,Polytechnic Institute of Braganca, Campus de Sta Apolonia,Apartado 134, 5301-857, Braganca, Portugal,[email protected], Florbela P. Fernandes, M. Fernanda P. Costa,Edite M.G.P. FernandesSemi-infinite programming problems can be efficiently solved by re-duction type methods. In this work a new global reduction method forsemi-infinite programming is presented. The multilocal optimizationis carried out with a multistart technique and the reduced problem isapproximately solved by a primal-dual interior point method combinedwith a two-dimensional filter line search strategy. The filter strategy isused to promote the global convergence of the algorithm. Numericalexperiments with a set of well-known problems are shown and com-parisons with other methods are presented.

4 - Study of optimization problems with analytic con-straint functionsTatiana Tchemisova, Departmento of Mathematics, Universityof Aveiro, Campus Universitario de Santiago, 3810-193,Aveiro, Portugal, [email protected], Olga KostyukovaFor convex SIP problems with analytic constraint functions, we ap-ply our approach based on the notion of immobile indices and theirimmobility orders. Implicit Optimality Criterion proved for the gen-eral case of problems with 1-dimensional index set allows to replacetesting optimality in the convex SIP by testing in a special nonlinearproblem. We study some specific properties of this problem and provenew optimality conditions. Special attention is paid to the case whenthe constraints of the original SIP are presented by analytic functionsand do not satisfy the Slater condition.

91


� TA-16Tuesday, 8:30-10:00RB-2103

Asset management applied to energy andthe environmentStream: Uncertainty and Perturbations in Optimizationand in Environmental and Energy ModellingInvited sessionChair: Thomas Archibald, Business School, University ofEdinburgh, 29 Buccleuch Place, EH8 9JS, Edinburgh, UnitedKingdom, [email protected]

1 - Long-term asset maintenance optimization at Scot-tish WaterTravis Poole, Scottish Water, United Kingdom,[email protected], Thomas Archibald, RobertMurray

A Knowledge Transfer Partnership between Scottish Water and theUniversity of Edinburgh has produced an asset maintenance optimiza-tion tool which produces long-term (25-year) maintenance plans forScottish Water’s assets. The optimizer implements a number of opti-mization techniques, and addresses a set of different planning scenar-ios. By using a common currency for risks and costs, it is able to con-sider multiple asset types simultaneously. Scottish Water spends over£700m maintaining assets each year, so even a modest improvement inefficiency of plans will lead to significant savings.

2 - Practical Issues in Asset Management within UK Wa-ter IndustryJake Ansell, Business Studies, The University of Edinburgh,29 Buccleuch Place, EH4 1ET, Edinburgh, United Kingdom,[email protected]

Water companies are large complex organisations which deliver freshwater to households and industry, and take away waste water fromboth. The paper will review many aspects of managing within thiscontext including the nature of the data, intervention actions, the pos-sibilities of implementing optimal maintenance actions and impact ofdelays. It will explore issues around condition grade and time basedmodels.

3 - Modelling Asset Deterioration using Structured Ex-pert JudgementGraeme Blair, Scottish Water, United Kingdom,[email protected], Deborah Gee, MatthewRevie, Lesley Walls

This work is motivated by an industry problem; a utility seeks to modelasset deterioration to predict future failures. Model estimates informstrategic investment and are subject to scrutiny by the regulator. Amodel is developed for an asset class which has limited historical sta-tistical records. Modelling deterioration with respect to time or state isexplored. Using the theory of structured expert judgement, a processfor eliciting probabilities is designed. The model validation approachand the value of a subjective, state-based approach to asset deteriora-tion modelling are described.

� TA-17Tuesday, 8:30-10:00RB-2105

Covering Models

Stream: Applications of Location AnalysisInvited sessionChair: Ioannis Giannikos, Business Administration, University ofPatras, University of Patras, GR-26500, Patras, Greece,[email protected]

1 - Multi-criteria covering-based location of volunteerfire departmentsBrigitte Werners, Management and Economics,Ruhr-University Bochum, Universitätsstr, 44780, Bochum,Germany, [email protected], Dirk Degel

Timeliness is one of the most important objectives that reflect the qual-ity of emergency services such as firefighting systems. Sufficient fire-fighting facilities must be established and located to provide high qual-ity services over a wide area with spatially distributed demands. To op-timize their location, the objectives minimizing the average travel timeand minimizing the longest travel time are in conflict with minimizingcosts. Different multi-criteria methods are used to compare solutionsfor an urban area with several volunteer firefighting departments.

2 - Non-full coverage facility location problemsMaria Cortinhal, Quantitative Methods, ISCTE- CIO, Av. dasForças Armadas, 1649-026, Lisbon, Portugal, [email protected]

We present new discrete facility location problems. These new mod-els broad the generic setting for the single source facility capacitatedproblem and for the single source modular facility capacitated problemby allowing a non-full coverage of customer demands. Additionally, itis considered a penalty function that allows doing a cost-full coveragetrade-off. For each problem, two models are suggested. Computationalexperiments with CPLEX showed that reformulated models give betterresults. To enhance the quality of the results, some valid inequalitieswere developed and tested.

3 - Locating Undesirable FacilitiesH.a. Eiselt, University of New Brunswick, P.O. Box 4400,E3B 5A3, Fredericton, NB, Canada, [email protected],Vladimir Marianov

We locate undesirable facilities for nonhazardous materials, such aslandfills, given an objective that minimizes a combination of costs andpollution. In contrast to most models that consider the effects of pollu-tion only near the facilities, we assume that there are undesired effectson wider areas, as those caused by trucks transporting the waste. Ourmodel includes not only locations, but also facility sizes as parameters.The model is formulated, its properties are investigated, and some pre-liminary computational evidence is presented.

4 - Using Clustering Heuristics to Solve a Family of De-mand Covering ModelsIoannis Giannikos, Business Administration, University ofPatras, University of Patras, GR-26500, Patras, Greece,[email protected], Basilis Boutsinas, Antiopi Panteli

We discuss demand covering models to determine locations for serversthat optimally cover a given set of demand points. Our clusteringheuristic processes the input data to determine clusters that result insmaller problems. It is based on the concept of bi-clusters, which arespecial sub matrices of a given coverage matrix. The algorithm startsby identifying appropriate bi-clusters of the coverage matrix and thencombines them to define an aggregate solution, which is improved witha local search heuristic. The proposed algorithm is tested in a series ofwidely known test problems.

� TA-18Tuesday, 8:30-10:00RB-2107

Advances in Location Modeling

Stream: Network and Discrete LocationInvited sessionChair: Dmitry Krass, Rotman School of Mgmt, University ofToronto, 105 St. George st., M5S 3E6, Toronto, Ontario, Canada,[email protected]

1 - Prepositioning of Supplies in Preparation for a Hurri-cane with Forecast Information UpdatesRajan Batta, Industrial and Systems Engineering, Universityat Buffalo (SUNY), 208 F Davis Hall, 14260, Buffalo, NewYork, United States, [email protected], Gina Galindo

92


Given a foreseen hurricane, it is possible to use prediction models inorder to obtain periodic forecasts about its main characteristics such asits path, wind force and time of landfall. We present a dynamic modelfor pre-positioning supplies in such a setting. Our model has two dis-tinguishing features: the possible destruction of supply points duringthe disaster event and the use of forecast information updates in orderto strategically modify and improve the pre-positioned quantities andlocations.

2 - Organization of a public service through the solutionof a Districting ProblemCarmela Piccolo, Department of Business and ManagementEngineering, University of Naples Federico II, via CristoforoColombo 19, 80030, Castello di Cisterna, Italy,[email protected], Giuseppe BrunoA districting problem consists in subdividing a region into districts onthe basis of an objective function. Districting models can also be usedto describe problems when a service has to be organized in a region.The decision about the number of districts and the allocation of theusers can be performed considering the conflicting objectives of twoactors: the planner aiming at minimizing the total costs and the usersinterested to reduce a measure of accessibility. We propose a generalmathematical formulation and an application to a real case study. Theresults are analyzed and discussed.

3 - Inventory Location and Transshipment ProblemDmitry Krass, Rotman School of Mgmt, University ofToronto, 105 St. George st., M5S 3E6, Toronto, Ontario,Canada, [email protected], Oded Berman, AlexShlakhterWe analyze an inventory location problem where facilities are allowedto satisfy some of the demand by transshipping available inventory be-tween each other and customers may change the shopping patternsbased on stock availability at the facilities. These features make themodel much more difficult than either the inventory location or trans-shipment models in the literature. By applying Infinitesimal Pertur-bation Analysis (IPA) algorithm — a form of approximate dynamicprogramming we show how realistic-size versions can be solved andobtain some interesting managerial insights.

� TA-19Tuesday, 8:30-10:00RB-2111

Pension FundsStream: Actuarial Sciences and Stochastic CalculusInvited sessionChair: Elena Vigna, Dipartimento di Scienze Economico-Sociali eMatematico-Statistiche, Università di Torino and Collegio CarloAlberto, corso Unione Sovietica 218 bis, 10135, Torino, Italy,[email protected]

1 - A risk-based premium: What does it mean for DBplan sponsors?An Chen, Department of Economics, University of Bonn,Adenauerallee 24-42, 53113, Bonn, NRW, Germany,[email protected], Filip UzelacThis paper develops a risk-based valuation model for pension insur-ance provided to defined benefit (DB) plans by the Pension BenefitGuaranty Corporation (PBGC). It takes account of investment policiesof the pension fund and of the DB plan sponsor. In addition, our modelaccounts for the premature termination of the pension fund caused bythe underfunding of the plan sponsor, so-called distress termination.To obtain realistic estimates of the PBGC premium and to examine ourvaluation formula, we calculate risked-based premiums for 25 promi-nent DB plan sponsors.

2 - Mean-variance optimization in DC plan with stochas-tic interest rateElena Vigna, Dipartimento di Scienze Economico-Sociali eMatematico-Statistiche, Università di Torino and CollegioCarlo Alberto, corso Unione Sovietica 218 bis, 10135, Torino,Italy, [email protected], Francesco Menoncin

We solve a mean-variance optimization problem of a defined contribu-tion pension scheme in the accumulation phase. The financial marketconsists of: (i) the risk-free asset, (ii) a risky asset following a GBM,and (iii) a bond driven by a stochastic interest rate whose dynamicsfollows a mean-reverting process. We solve the associated HamiltonJacobi Bellman PDE and find a closed-form solution for the optimalinvestment strategy. Numerical applications show how the proportionsof bond and risky asset decrease when retirement approaches.

3 - Optimal consumption, investment and life insurancewith surrender option guaranteeMorten Tolver Kronborg, Department of MathematicalSciences, University of Copenhagen, Copenaghen, Denmark,[email protected]

We present the optimal consumption, investment and life insurancestrategy for an investor, endowed with labor income, searching to max-imize his utility over cumulative consumption, bequest upon death andterminal wealth when he is restricted to fulfill a binding capital con-straint of an American type. The optimal strategy combines a longposition in the optimal unrestricted allocation with an American putoption. In order to stay self-financing, the investor changes his bud-get constraint for the unrestricted allocation whenever the constraint isactive.

4 - Allowing for Tail Risk, and Aversion to Tail Risk, inOptimal Portfolios for Long-Horizon InvestorsIqbal Owadally, Faculty of Actuarial Science and Insurance,Cass Business School, City University, London, EC1Y 8TZ,London, United Kingdom, [email protected], ZinoviyLandsman

Asset return distributions are heavy-tailed and investors, particularlynearing retirement, are sensitive to extreme market events in the tailof these distributions. The tail mean-variance model proposed in in-surance risk management involves a criterion focusing on the risk ofrare but large losses. It does not satisfy the property of positive homo-geneity, and the optimal solution typically requires numerical methods.We use a convex optimization method and a mean-variance character-ization to find an explicit and easily implementable solution with andwithout a risk-free asset.

� TA-20Tuesday, 8:30-10:00RB-2113

Multicriteria Analysis for InvestmentDecision Making

Stream: Decision Making Modeling and Risk Assess-ment in the Financial SectorInvited sessionChair: Luiz F. Autran M. Gomes, Management, Ibmec/RJ, Av.Presidente Wilson, No. 118, Room 1110, 20030020, Rio deJaneiro, RJ, Brazil, [email protected]

1 - Frameworks of Strategic Decision Behavior on In-vestment and Risk ManagementMei-Chen Lo, National United University, 360, Miaoli,Taiwan, [email protected]

Effective investment and risk management, as the part of financial sys-tem, has strong impact on the performance of the enterprises. In thisstudy, the strategic decision behaviors are discussed, and two MCDMmethods were used: 1.Analytic Hierarchy Process method for settingthe weights of dimensions criteria; 2.Technique for Order Preferenceby Similarity to Ideal Solution for assessment the relative performanceviewed by different groups of experts. The procedures help to find theimprovement directions through the study of the strategic behaviors onthe investment and risk management.

93


2 - An Approach to Multicriteria Sorting within the Ver-bal Decision Analysis ParadigmEugenia Furems, Decision Making, Institute for SystemAnalysis, 9, Prospect of 60 let Octyabrya, 117312, Moscow,[email protected]

Multicriteria sorting (MS) is often used inter alia as a model for invest-ment alternatives assessment, where a class is the subset of alternativeswith equivalent feasibility. Verbal Decision Analysis (VDA) is one ofapproaches for MS problem solving. A number of VDA-based meth-ods use DM’s preferences as the only source of information and en-sure obtaining the completness and consistency of classification rules,while avoiding the exhaustive search. A new dialogue strategy for suchmethods is proposed to reduce a DM’s cognitive load and to enhancethe ways of rules consistency providing.

3 - Comparing decisional mapsValérie Brison, Mathematics and Operational Research,UMONS, Belgium, [email protected], Marc Pirlot

The purpose of this work is to provide models to compare maps rep-resenting the state of a region at different stages of its evolution. Weformulate three mathematical models all based on several axioms thatmake sense in some geographic contexts. We start with a simple one,which doesn’t take into account any geographic aspect. In the secondone, we introduce one geographic aspect and then we generalize thismodel to take into account any finite number of geographic aspects.

4 - Using the Choquet-extended TODIM method for amulticriteria analysis of improvements in a majorroad in the Southeastern Region of Brazil: theParaty-Cunha HighwayLuiz F. Autran M. Gomes, Management, Ibmec/RJ, Av.Presidente Wilson, No. 118, Room 1110, 20030020, Rio deJaneiro, RJ, Brazil, [email protected], Maria AugustaMachado

Once significantly improved, Paraty-Cunha Highway will be the majorroad linkage between Paraty and its hinterland of Brazil. A key moti-vation behind investing on the improvement of that highway has to dowith the fact that Paraty is dangerously close to Angra dos Reis. Weformulate the investment decision making problem in a multicriteriaframework. A behavioral multicriteria decision model is used. Theanalysis is based on the paradigm of cumulative prospect theory. Asynthesis of the analysis leads to recommendations to decision mak-ers.

� TA-21Tuesday, 8:30-10:00RB-2115


1 - Using the coverage data structure to solve a bi-criteria constructive model aiding to identify non-dominated portfoliosJavier Pereira, Escuela de Ingeniería Informática, UniversidadDiego Portales, Avenida Ejercito 441, Santiago, Chile,[email protected], Fernando Paredes

We previously proposed a bi-criteria constructive model to find thenon-dominated portfolios, a problem formulated in terms of satisfac-tion of a given set of technical requirements. We argued that this isapplicable in situations where early and detailed business informationis poor. Solution was based on a preference-programming model foundin literature. Here, we improve this approach considering the structureof the project-requirement matrix, which is not treated on traditionalportfolio modeling. Numerical examples show that this improves thepruning process of the original algorithm.

2 - An EPQ Model for Deterioration Items and Exponen-tial Demand Rate Taking into Account the Time Valueof MoneyVikram jeet Singh, MATHEMATICS, Lovely ProfessionalUniversity, Jalandhar-Delhi G.T Road(NH-1), 144402,Phagwara, Punjab, India, [email protected]

In this study we develop an inventory model with exponential demandwith Time Value of Money. Deterioration occurs as soon as the itemsare received into inventory and it follows two parameter Weibull distri-butions. There is no repair or replacement of deteriorating items duringthe replenishment cycle. Product transactions are followed by instan-taneous cash flow Shortages are allowed and partially backlogged. Thesystem operates for a prescribed period of a planning horizon. Produc-tion rate is demand rate dependent. The problem is discussed under theinflationary environment.

3 - Pricing fx forwards in OTC markets - new evidencefor the pricing mechanism when faced with counter-party riskStefan Stöckl, FIM Research Center Finance & InformationManagement, University of Augsburg, 86135, Augsburg,Bavaria, Germany, [email protected],Alexander Leonhardt, Andreas Rathgeber, Johannes Stadler

In line with existing empirical results, this paper demonstrates, usingdaily foreign exchange (fx) market data for five currencies (CHF, EUR,GBP, USD, YEN), that, since the beginning of the financial crisis, mar-ket pricing of fx forwards does not match the pricing formula derivedfrom the covered interest rate parity (CIP). Therefore, the CIP leadsto systematic mispricing in relation to future fx spot rates. Overall,four statistically significant factors for this mispricing were identified.The high significance of the counterparty risk demonstrates that pric-ing models should be reviewed.

4 - Qualitative multi-attribute model for evaluating com-panies’ interoperabilityBorka Jerman Blazic, Laboratory for Open Systems andNetworks, Jozef Stefan Institute, Jamova cesta 39, 1000,Ljubljana, Slovenia, [email protected], Vladislav Rajkovic,Arnas Bareisis, Ausra Kentrait

In the paper a model to support enterprises in improving interoperabil-ity will be presented. The methodology considers conceptual, organi-zational and technological maturity of an enterprise regarding businessstrategy, processes, services and data. For the evaluation we proposea qualitative hierarchical multi-attribute model. The model is imple-mented within a DEX framework and will be illustrated by examples.

� TA-22Tuesday, 8:30-10:00RB-2117

Institutional and Technological Changesin Market and Transitional EconomiesStream: Mathematical Models in Macro- and Microe-conomicsInvited sessionChair: Ludmilla Koshlai, Systems analysis and OR, Institute ofCybernetics, Gonchar str.,65-a,apt.20, 01054, Kiev, Ukraine,[email protected]: Jean-Francois Emmenegger, Quantitative Economics,University of Fribourg, Bd de Pérolles 90, F440, 1700, Fribourg,Fribourg, Switzerland, [email protected]

1 - DSS for technological planningOleksandr Pylypovskyi, Radiophysical, Taras ShevchenkoNational Univesity Of Kyiv, 252187, Kyiv, Ukraine,[email protected], Ludmilla Koshlai, Petro Stetsyuk

We consider a software for decision support for structural-technological changes based on intersectorial optimization models de-veloped by Prof. M. Mikhalevich. Optimization models are nonlinear

94


programming problems (NLP) with two objective functions for max-imization: final consumer incomes and multiplier "cost-value of pro-duction". Solutions for NLP are based on r-algorithm, programms arewritten on Octave language. MiSTC is developed using Qt library andprovide tools for analysis of input/output data for intersectorial models.The protocol of interaction with user is recorded in HTML.

2 - Modeling the Returns of Securities on the FinancialMarketsJean-Francois Emmenegger, Quantitative Economics,University of Fribourg, Bd de Pérolles 90, F440, 1700,Fribourg, Fribourg, Switzerland,[email protected]

A reason of the recent turbulences (since 2007) on the international fi-nancial markets is the under evaluation of risks and the over evaluationof expected returns. How can these return series be modeled? Bache-lier’s (1900) idea was to construct a random-walk model, for securityand commodity markets. The empirical evidence shows that Bache-lier’s idea does not hold, as Mandelbrot and Fama (1963) showed. Thispaper continues this discussion and proposes data-based methods tocalculate more efficient financial risk indicators.

3 - Forecasting Ukrainian cargo transport system indi-catorsElena Pervukhina, Dept. of Information Systems, SevastopolNational Technical University, Universitetskaya Str., 33,99053, Sevastopol, Crimea, Ukraine,[email protected], Jean-Francois Emmenegger,Victoria Golikova

The paper addresses forecasting of transport volumes based on empir-ical econometric models. The peculiarity of transitional economy istaken into account. Therefore the data generating processes of cargovolumes has to be modeled with special care. The present analysisconsiders the forecast performance of multivariate time series models,comprising different modes of the Ukrainian cargo transport system.Two forecast methods are proposed. Both results illustrate the lack ofcorrespondence of the transport industry with respect to the Europeanstandards.

4 - Peculiarities of Labour Market in a Transition Econ-omyLudmilla Koshlai, Systems analysis and OR, Institute ofCybernetics, Gonchar str.,65-a,apt.20, 01054, Kiev, Ukraine,[email protected]

The labour market of official sectors of economy in transition is con-sidered as the typical example of imperfect market. Therefore, we as-sume that monopsony and oligopsony are the main kinds of competi-tion, producing the impact on wages during the transition. We consideroptimization models for the case of monopsony and oligopsony. Ex-istence conditions formulated for optimal strategies of employers andproperties of these strategies are analyzed. Investigations demonstratethat dependence between wages and demand for employers’ produc-tion in oligopsony is similar to dependence in the case of monopsony.

� TA-23Tuesday, 8:30-10:00RB-Delta

Regularity in Nonsmooth Analysis

Stream: Generalized Differentation and OptimizationInvited sessionChair: Alexander Kruger, School of Science, InformationTechnology & Engineering, University of Ballarat, UniversityDrive, Mount Helen, P.O. Box 663, 3353, Ballarat, Victoria,Australia, [email protected]: Marco A. López-Cerdá, Statistics and Operations Research,Alicante University, Ctra. San Vicente de Raspeig s/n, 3071,Alicante, Spain, [email protected]

1 - An inverse function result and some applications tosensitivity of generalized equationsGabor Kassay, Mathematics, Babes-Bolyai University, 1 MKogalniceanu 1, 400084, Cluj, Romania,[email protected]

We extend an inverse function theorem to set-valued case in metricspaces. Defining a condition number and the concept of well condi-tioning for parametric generalized equations at a certain initial valueof the parameter, our inverse function theorem allows us to determinehow much the initial problem can be perturbed in order to preserveits well conditioning. An application to the framework of vector opti-mization problems is then given. Finally we identify a special class ofwell conditioned set-valued mappings.

2 - Extensions of Metric RegularityAlexander Kruger, School of Science, InformationTechnology & Engineering, University of Ballarat, UniversityDrive, Mount Helen, P.O. Box 663, 3353, Ballarat, Victoria,Australia, [email protected], Andrei Dmitruk

The presentation is devoted to some extensions of the metric regularityproperty for mappings between metric or Banach spaces. Several newconcepts are investigated in a unified manner: uniform metric regu-larity, metric multi-regularity for mappings into product spaces (wheneach component is perturbed independently), as well as their Lipschitz-like counterparts. The properties are characterized in terms of certainderivative-like constants. Regularity criteria are established based on aset-valued extension of a nonlocal version of the Lyusternik—Gravestheorem due to Milyutin.

3 - Frechet and proximal Regularity of perturbed dis-tance functions at points in the target set in BanachspacesMessaoud Bounkhel, Department of Mathematics, King SaudUniversity, College of Science, PO Box: 2455, 11451,Riyadh, Saudi Arabia, [email protected]

This talk is devoted to study the Frechet and proximal Regularity atpoints in the target set of perturbed distance functions dJ_S(·) deter-mined by a closed subset S and a Lipschitz function J(·). Also, weprovide some important results on the Clarke subdifferential of dJ_S(·)at those points in arbitrary Banach space.

4 - Some recent contributions to convex calculusMarco A. López-Cerdá, Statistics and Operations Research,Alicante University, Ctra. San Vicente de Raspeig s/n, 3071,Alicante, Spain, [email protected]

In this talk we present a general formula for the optimal set of a re-laxed minimization problem in terms of the approximate minima of agiven data function. This formula is used for deriving explicit charac-terizations of the subdifferential mapping of the supremum function ofan arbitrarily indexed family of functions, exclusively in terms of thedata functions. We also discuss on possible alternative approaches andapplications to subdifferential calculus.

� TA-24Tuesday, 8:30-10:00CC-A11

Demand Forecasting

Stream: ForecastingInvited sessionChair: Aris Syntetos, Centre for Operational Research andApplied Statistics - Salford Business School, University ofSalford, Maxwell Building, The Crescent, M5 4WT, SALFORD,United Kingdom, [email protected]

1 - Grouping seasonality and the formation of groupsAris Syntetos, Centre for Operational Research and AppliedStatistics - Salford Business School, University of Salford,Maxwell Building, The Crescent, M5 4WT, SALFORD,United Kingdom, [email protected], John Boylan,Mona Mohammadipour

95


Estimating seasonality based on individual seasonal indices (ISI) oftenleads to unsatisfactory results. An alternative is to use the knowledgeof demand at some aggregate/group level to improve forecasts at theindividual level. This approach is usually referred to as group sea-sonal indices (GSI); an important assumption in its application is thata grouping mechanism is available. Our work provides empirical evi-dence on the comparatives merits of ISI & GSI and the issue of choos-ing between them. The formation of seasonal groups is also explicitlyaddressed and linked to performance results.

2 - Forecasting in production planning by adaptive ex-ponential smoothing methodsFrank Herrmann, Innovation and Competence Centre forProduction Logistics and Factory Planning, University ofApplied Sciences Regensburg, PO box 120327, 93025,Regensburg, Germany, [email protected]

Demands of customers for products and of the production for parts arebeing forecasted quite often within the operational production plan-ning and control by IT Systems (PPS systems) like the SAP system viaexponential smoothing. There parameters need to be changed if thecharacteristics of the time series has changed. Since a manual changeis very time consuming, industrial practice is interested in adaptive ex-ponential smoothing methods. There state of the art, there limitationin terms of unstable forecasts, there performance in empirical studiesand there usage in PPS systems are presented.

3 - Forecasting Intermittent Demand by Hyperbolic-Exponential SmoothingArmagan Tarim, Management, Hacettepe University, Beytepe,06800, Ankara, Turkey, [email protected], StevenPrestwich, Roberto Rossi, Brahim Hnich

Croston’s method is superior to exponential smoothing when demandis intermittent, but is known to be biased. Variants of the method re-move the bias, but only the recent variant called TSB addresses theproblem of demand obsolescence. We describe a new forecastingmethod called Hyperbolic-Exponential Smoothing that handles obso-lescence by decaying hyperbolically when demand is zero. In a fullfactorial design experiment on various smoothing factors, demand dis-tributions and intermittencies, it has similar bias to TSB and even betterforecast accuracy.

4 - Stochastic models for time-dependent intermittentdemandClint Pennings, Rotterdam School of Management, ErasmusUniversity, Netherlands, [email protected], Jan van Dalen,Erwin van der Laan

Intermittent demand is characterized by several subsequent periodswithout demand. Croston developed a popular method for forecast-ing by assuming independence between inter-arrival time and quantitydemanded. However, this seems unnatural in many practical situa-tions (preventive or corrective maintenance) and has been empiricallyproven to be unrealistic. We developed a stochastic model to forecastdemand that explicitly takes into account the time dependency and as-sessed the potential for performance improvement in inventory man-agement relative to other popular frameworks.

� TA-25Tuesday, 8:30-10:00CC-A23

Modelling Markets with Imperfections

Stream: Energy, Environment and Climate 2Invited sessionChair: Christoph Weber, Universität Essen, Universitätsstr. 11,45117, Essen, Germany, [email protected]

1 - Cartelization in the natural gas industry, a model-based analysisAlbert Banal Estanol, Universitat Pompeu Fabra, 08002,Barcelona, Spain, [email protected], StevenGabriel, Olivier Massol

We investigate the stability of a cartel gathering the non-OECD natu-ral gas exporting countries. Our analysis is based on a deterministic,discrete-time, finite-horizon oligopoly model of investment and pro-duction equilibrium strategies. We implement a feedback informationstructure to solve the dynamic inconsistency of the open loop Stack-elberg dynamic game. The model is calibrated to represent the worldnatural gas market under a "business as usual" scenario (projections ofeconomic growth, demography, oil price) and used to formulate somepolicy recommendations for gas importing countries.

2 - Expectation Formation on Future Markets and Boom-and-Bust Cycles in Electricity Markets: Evidencefrom an Agent-Based SimulationDaniel Ziegler, Chair for Management Sciences and EnergyEconomics, University Duisburg-Essen, Universitaetsstrasse12, 45117, Essen, Germany, [email protected]

Agent-based simulations have the potential to provide insights howrisks and disequilibria affect investments and may induce investmentcycles. We present a modular agent-based model for the long termanalysis of interrelated wholesale electricity markets. A future markettranslates expectations and spot market outputs into noisy price sig-nals, which serves as inputs. First results indicate that the dynamicsof investment cycles strongly depend on the way market participantsform their expectations.

3 - Decision problems including limited liquidity in en-ergy marketsOliver Woll, Universitaet Duisburg-Essen, Universitaetsstraße12, 45117, Essen, Germany, [email protected],Christoph Weber

In most energy markets, the liquidity of products with long-term matu-rity is limited. Standard models for corporate hedging do not cope withthis situation. This paper provides an analytical approach for solvingthe problem by including a liquidity function into the standard mean-variance model going back to Markowitz. The paper then analysesoptimal hedging strategies for power generators. It is shown that theoptimal hedging strategy not only depends on price volatility and mar-ket liquidity but also on the size of the initial open position and on theremaining planning horizon.

4 - Planning of Energy Supply Chains based on Domes-tic Energy microgenerationGeorgios Kopanos, Department of Chemical Engineering,Centre for Process Systems Engineering, Imperial CollegeLondon, Roderic Hill Building, South Kensington Campus,SW7 2AZ, London, London, United Kingdom,[email protected], Michael Georgiadis, EfstratiosPistikopoulos

This work addresses an emerging non-classical Energy Supply Chain(ESC) based on decentralized energy production via microgeneration.In particular, several domestic microgenerators are connected so as toform an energy micro-grid that allows residents to interchange energyamong each other. Thermal and electrical energy demand of everyhousehold must be satisfied, while excessive electricity production canbe sold to the macro-grid. A new mathematical programming frame-work for the energy production planning of this new type of ESC isproposed, and a representative case study is presented.

� TA-26Tuesday, 8:30-10:00CC-A24

Optimization in Design of Water Systems

Stream: OR in Water ManagementInvited sessionChair: Elcin Kentel, Department of Civil Engineering, METU,Orta Dogu Teknik Universitesi, Universiteler Mahallesi,Dumlupinar Blv. No:1, 06800, Ankara, Turkey,[email protected]: Halil Önder, Civil Engineering, Middle East TechnicalUniversity, Middle East Technical University, Department of CivilEngineering, 06800, Ankara, Turkey, [email protected]

96


1 - An improved decomposition-based heuristic for thepressurized water irrigation network design problemMargarida Pato, ISEG (Technical University of Lisbon), CIO(University of Lisbon) and ISEG (Technical University ofLisbon), Rua do Quelhas, 6, 1200-781, Lisboa, Portugal,[email protected], Graça Gonçalves, Luis GouveiaDesigning a pressurized water distribution network for irrigation pur-poses (WDNP) is a highly complex problem that aims at minimizingcosts while obeying many technical constraints. A mixed binary non-linear programming model for WDNP will be presented along withan improved decomposition heuristic based on a linearization of themodel. The computational results obtained from the heuristic with in-stances simulating a real application will be shown and compared toresults from standard software. We conclude that the heuristic is asolid approach to the WDNP.

2 - Use of spreadsheet optimization for management ingroundwater engineeringHalil Önder, Civil Engineering, Middle East TechnicalUniversity, Middle East Technical University, Department ofCivil Engineering, 06800, Ankara, Turkey, [email protected] work presents a groundwater management model that combinesanalytical solutions of the groundwater flow equation with linear pro-gramming to determine the best pumping scheme and optimum config-uration for the dewatering system to drain groundwater in a construc-tion site. It is based on spreadsheet application and it is a relatively lowcost and rapidly applicable methodology. Using spreadsheets for opti-mization problem will be beneficial in closing the technology transfergap between the development of optimization technique and their ac-ceptance and practical use by practicing engineers.

3 - Estimation of aquifer parameters in groundwatermanagement using spreadsheet optimization as aninverse problemElcin Kentel, Department of Civil Engineering, METU, OrtaDogu Teknik Universitesi, Universiteler Mahallesi,Dumlupinar Blv. No:1, 06800, Ankara, Turkey,[email protected], Halil Önder, Cüneyt TaskanA spreadsheet optimization using a non-linear error function is utilizedfor estimation of aquifer parameter as an inverse problem based on ananalytical solution of groundwater flow. The values of transmissivityand storage coefficient are determined from measured drawdown val-ues at observation wells. Spreadsheet optimization provides the practi-tioners, educators, and researchers an efficient and convenient methodfor determining coefficients of aquifer, and a practical platform for rou-tine evaluations. The proposed development is a relatively low cost andrapidly applicable methodology.

4 - A hybrid model for forecasting the streamflow dataAlpaslan Yarar, Civil Engineering, Selcuk University, Turkey,[email protected], Mustafa Onucyildiz, Hülya YararIn this study, a hybrid model, Wavelet-Neuro Fuzzy (WNF), has beenused to forecast the streamflow data of 5 Flow Observation Station(FOS), which belong to Sakarya Basin in Turkey. In order to evaluatethe accuracy performance of the model, Linear Regression (LR) modelhas been used with the same data sets. The comparison has been madeby Root Mean Squared Errors (RMSE) of the models. Results showedthat hybrid WNF model forecasts the streamflow more accurately thanLR model.

� TA-27Tuesday, 8:30-10:00CC-A25

Inequality Averse Decisions

Stream: Decision ProcessesInvited sessionChair: Alec Morton, Management/ Operational Research, LondonSchool of Economics, Houghton St, London, wc2a2ae, London,England, United Kingdom, [email protected]: Nikolaos Argyris, Management, Operational ResearchGroup, London School of Economics and Political Science,Houghton Street, WC2A 2AE, London, United Kingdom,[email protected]

1 - Combining equity and utilitarianism in a mathemati-cal programming modelHilary Paul Williams, Operational Research, London Schoolof Economics, Houghton Street, WC2A2AE, LONDON,United Kingdom, [email protected], John N. HookerWe show how equity and utilitarianism may be combined in a math-ematical programming model, which applies particularly to the provi-sion of health care. When the disparity in the cost of care of a par-ticular group exceeds a threshold level the objective gradually movesfrom one of equity to one of utilitarianism. The resulting formula-tion, in its unmodified form, is non-convex and its hypograph is nonMIP representable. When reformulated the model is easily solved anddemonstrates interesting results on a dataset based on some US andUK data.

2 - Equitable preferences: representation and use inpractical problemsNikolaos Argyris, Management, Operational Research Group,London School of Economics and Political Science,Houghton Street, WC2A 2AE, London, United Kingdom,[email protected] consider the problem of centralized inequality-averse resource al-location. We first discuss a set of axioms that may characterize "equi-table’ preferences over allocation profiles. We show that, under mildconditions, an equitable preference ordering over a finite set of alloca-tions can always be represented by a symmetric, non-decreasing andconcave function. Further, we show that the set of all such represen-tations compatible with a set of revealed preferences has a succinctpolyhedral representation. Finally, we discuss how these results maybe used in choice and ranking problems.

3 - Incorporating preference information in multicriteriaproblems with equity concernsOzlem Karsu, Management, London School of Economics,Houghton Street, WC2A 2AE, LONDON, United Kingdom,[email protected], Alec MortonWe consider Multi Criteria Decision Making (MCDM) problems in-volving equity concerns which are highly encountered in real life, es-pecially in public sector. From a modelling point of view, preferencefor equity can be handled by imposing person anonymity and quasi-concavity conditions on the model of the DM’s preferences. One so-lution approach for MCDM problems is based on taking into accountvalue judgments which the Decision Maker (DM) has provided. Wesuggest such an approach for problems with equity concerns and intro-duce substantial new theory.

4 - Aversion to health inequalities in healthcare prioriti-sation: a multiobjective mathematical programmingperspectiveAlec Morton, Management/ Operational Research, LondonSchool of Economics, Houghton St, London, wc2a2ae,London, England, United Kingdom, [email protected] discuss the prioritisation of healthcare projects where there is aconcern about health inequalities. Our analysis begins with a standardwelfare economic model of healthcare resource allocation. We showhow the problem can be reformulated as one of finding a particular sub-set of the class of efficient solutions to an implied multiobjective op-timisation problem. Algorithms for finding such solutions are readilyavailable, and we demonstrate our approach through a worked exampleof treatment for clinical depression.

� TA-28Tuesday, 8:30-10:00CC-A27

Hierarchical Optimization

Stream: Structured Optimization and ApplicationsInvited sessionChair: Eleftherios Couzoudis, Economics, University of Zurich,Chair for Quantitative Business Administration, Moussonstrasse15, 8044, Zurich, Switzerland,[email protected]

97


1 - A poisson input queueing system for threshold pol-icy of machine repair problem with degraded failureRekha Choudhary, Mathematics Dept., Govt. Engg. College,110, Swarn Jayanti Nagar, 321001, Bharatpur, Rajasthan,India, [email protected]

This investigation deals with the study of Optimal threshold policy formachine repair problem with R heterogeneous removable repairmen.The first repairman turns on for repair only when K1 failed units areaccumulated and starts repair after a setup time, which is assumed tobe exponentially distributed. The life time and repair time of failedunits are assumed to be exponentially distributed. To obtain steadystate queue-size distribution, recursive technique is used. Various per-formance measures are established using steady-state probabilities. Acost function is also deduced.

2 - Expansion Planning of Storage Technologies in Re-newable Generation SystemsChristian Kraemer, Institute of Power Systems and PowerEconomics, RWTH Aachen University, Schinkelstr. 6, 52064,Aachen, Germany, [email protected], AndreasSchäfer, Manuel Jäkel, Albert Moser

Rising amounts of renewables and the age structure of European gen-eration systems require the planning of new generation units. Stor-age technologies are supposed to play a significant role for storing re-newable energy and providing flexibility. Expansion planning (EP) ofstorage units needs to consider time-dependencies in order to evaluatefuture benefits. It is shown that the long-term and large-scale EP prob-lem can be broken down to its most important constraints. The derivedmodel simulates the development of thermal and storage units underall relevant technical and economic constraints.

3 - The Optimality Antisymmetrical rank function inposetZahra Yahi, SEGC-LMD, Béjaia University, Faculty ofEconomic, Management and Trade Sciences, Algeria,[email protected], Sadek Bouroubi

In studying the optimality rank function, several results are given. Asufficient condition is given by Alexiv. The optimality of the rank func-tion depends also on the type of the poset; e.g., for the rank functionabout a distributive poset, a normal poset is optimal. Our objective arestudies of optimality of the Antisymmetrical rank function in a partic-ular poset. We give some sufficient condition of the Antisymmetricalrank function and some results in relation with the optimality of therank function in poset.

� TA-29Tuesday, 8:30-10:00CC-A29

MINLP in Airline and Air TrafficManagement applications

Stream: Mixed-Integer Non-Linear ProgrammingInvited sessionChair: Sonia Cafieri, Lab. MAIAA, Dept. de Mathematiques etInformatique, Ecole Nationale d’Aviation Civile, 7 Ave. EdouardBelin, 31055, Toulouse, France, [email protected]: Claudia D’Ambrosio, LIX, CNRS - Ecole Polytechnique,route de Saclay, 91128, Palaiseau, France,[email protected]

1 - On solving the collision avoidance problem for ATMby solving a Mixed 0-1 nonlinear optimization modelheuristicallyLaureano Fernando Escudero, Dept. de Estadística eInvestigación Operativa, Universidad Rey Juan Carlos,c/Tulipan, S/n, 28933, Mostoles (Madreid), Spain,[email protected], Antonio Alonso-Ayuso, F. JavierMartin-Campo

A mixed 0-1 nonlinear optimization model is presented to solve theCollision Avoidance for Air Traffic Management Problem. We give anew configuration for each aircraft such that every conflict situation isavoided. If some aircrafts violate the safety distances, a conflict sit-uation occurs. Due to the difficulty to solve the problem by MINLPengine software in short time, a heuristic way is developed for obtain-ing good solutions in short time. The main results will be reported bycomparing the exact solutions with the heuristic ones for short-scaleproblems.

2 - Conflict resolution by minor speed adjustmentsDavid Rey, LICIT, IFSTTAR, Bron, France,[email protected], Christophe Rapine, Rémy Fondacci,Nour-Eddin El Faouzi

We present a reformulation of a nonlinear optimization problem en-countered in Air Traffic Management (ATM). A current challenge inATM is to increase airspace capacity to deal with a growing traffic de-mand. One way to address the capacity problem is to smoothly regulatethe traffic flow by performing minor adjustments on aircraft speeds. Toguarantee the required safety standards in ATM, one needs to preventpotential conflicts from occurring. This speed regulation frameworkcan be modeled as a nonlinear optimization problem which we pro-pose to reformulate as a Mixed Integer Linear Program.

3 - Airline tail assignment optimization: the impact ofaircraft heterogeneityPatrícia Ribeiro, Technical University of Lisbon, Lisbon,Portugal, [email protected], Susana Relvas, NunoLeal

Operating with low margins in a competitive industry, airlines seekstrategies to minimize operating costs and improve operational effi-ciency. One possible strategy consists in considering heterogeneityamong aircraft tails when assigning them to the planned flights. Weaim to optimize the tail assignment process of the Airline TAP Portugalby taking into account the different characteristics and performances ofeach aircraft. The savings achieved by optimizing a mixed integer lin-ear programming model integrating the re-design of lines of flights andthe tail assignment process are presented.

4 - Aircraft conflict avoidance: a mixed-integer nonlin-ear optimization approachSonia Cafieri, Lab. MAIAA, Dept. de Mathematiques etInformatique, Ecole Nationale d’Aviation Civile, 7 Ave.Edouard Belin, 31055, Toulouse, France,[email protected]

Detecting and solving aircraft conflicts, which occur when aircraftsharing the same airspace are too close to each other according to theirpredicted trajectories, is a crucial problem in Air Traffic Management.We focus on mixed-integer optimization models based on speed regu-lation. We first solve the problem to global optimality by means of anexact solver. The problem being very difficult to solve, we also pro-pose a heuristic procedure where the problem is decomposed and it islocally exactly solved. Computational results show that the proposedapproach provides satisfactory results.

� TA-30Tuesday, 8:30-10:00CC-A31

Modeling Approaches and SolutionAlgorithms

Stream: Emerging Aspects of Production Planning inContinuous Process Industries: Theory, Optimization,and PracticeInvited sessionChair: Christoph Schwindt, Institute of Management andEconomics, Clausthal University of Technology, Julius-Albert-Str.2, 38678, Clausthal-Zellerfeld, Germany,[email protected]

98


1 - Heuristic decomposition and LP-based scheduling inmake-and-pack productionNorbert Trautmann, Department of Business Administration,University of Bern, Ordinariat für Quantitative Methoden derBWL, Schützenmattstrasse 14, 3012, Bern, BE, Switzerland,[email protected], Philipp BaumannWe deal with the short-term scheduling of make-and-pack productionprocesses. The planning problem consists in minimizing the produc-tion makespan while meeting a given demand subject to a large varietyof technological constraints. The problem size of real-world instancesrequires the usage of heuristics.

We propose to schedule groups of batches iteratively using acontinuous-time MILP formulation; the assignment of the batches tothe groups and the scheduling sequence of the groups are determinedusing a priority rule. We report on computational results for a real-world production process.

2 - Modeling Formulation and Solution Approach for aContinuous Process ProblemKrystsina Bakhrankova, Applied economics, SINTEF -Technology and society, Box 4760 Sluppen, S. P. Andersensveg 5, 7465, Trondheim, Norway,[email protected], Truls FlatbergProduction planning of continuous systems is less covered in the pro-cess industry literature relative to their batch counterparts. This workpresents a modeling formulation and a corresponding solution ap-proach for a real-world optimization problem, where bulk chemicalsare produced via a completely continuous non-intermittent process. Italso provides relevant analysis with respect to model properties andsolution times as well as potential economic benefits of applying theproposed methodology in practice.

3 - Planning of a continuous production process in theprinting industryPhilipp Baumann, Department of Business Administration,University of Bern, IFM, AP Quantitative Methoden,Schützenmattstrasse 14, 3012, Bern, Switzerland,[email protected], Norbert TrautmannOffset printing is a common method to produce large amounts of print-ing matter. We consider a real-world offset printing process that is usedto imprint customer-specific designs on napkin pouches. The planningproblem consists in allocating printing-plate slots to the designs suchthat a given customer demand is fulfilled, all technological require-ments are met and the total printing cost are minimized. We formulatethis planning problem as a mixed-integer linear program, and we reporton computational results for industrial problem instances.

4 - Comparing models for operational production plan-ning and scheduling in single-stage continuous pro-cesses: Process Systems Engineering vs. Opera-tions ResearchAna Paula Barbósa-Póvoa, Dept. De Engenharia e Gestao,IST, Av. Rovisco Pais, 1, 1049-001, Lisbon, Portugal,[email protected], Pedro Amorim, Bernardo Almada-Lobo,Tânia Pinto_VarelaSeveral researchers from the two communities, Operational Researchand Process Systems Engineering, have focused on mathematical for-mulations for the simultaneous lot-sizing and scheduling of single-stage continuous processes with complex set-up structures. This prob-lem arises in the glass, beer, and dairy production. Although recentresearch has been performed by both mentioned communities, there isno intense communication between these research efforts. This workattempts a systematic analysis of the recent formulation developmentsto narrow the gap between both communities on this problem.

� TA-31Tuesday, 8:30-10:00CC-A33

Data Mining: Web and Social-OrientedApplications

Stream: Data Mining, Knowledge Discovery and Artifi-

cial IntelligenceContributed sessionChair: Takashi Onoda, System Engineering Lab., CRIEPI, 2-11-1,Iwado Kita, Komae-shi, 201-8511, Tokyo, Japan,[email protected]: Anastasia Motrenko, Applied mathematics andmanagement, MIPT, Moscow, Russian Federation,[email protected]

1 - An evaluation of the Generalized Additive Neural Net-work spam filtering approachTiny Du Toit, School of Computer, Statistical andMathematical Sciences, North-West University, North-WestUniversity, 11 Hoffman Street, 2531, Potchefstroom,North-West, South Africa, [email protected], HennieKruger

The performance of a Generalized Additive Neural Network on a pub-licly available corpus is investigated in the context of statistical spamfiltering. The impact of different feature set sizes on an automated con-struction algorithm is studied using cost-sensitive measures. Experi-ments show high performance where legitimate e-mails are assignedthe same cost as spams. Applications where legitimate e-mails areallocated a much higher cost than spams should be avoided. This de-crease in performance is caused by the aggressive feature dimensionreduction required by the algorithm.

2 - Applying hybrid models to analyze the adoption in-tention of mobile applicationsYang-Chieh Chin, International Business, Asia University,500, Lioufeng Rd., Wufeng, 41354, Taichung, Taiwan,[email protected], Chiao-Chen Chang

Mobile applications are one of the most rapidly growing segments ofthe software market. The goal of this study is to gain insight into thefactors that affect user intention to adopt mobile applications. Firstly, adominance-based rough set approach (DRSA), a rule-based decision-making technique, is used to determine the adoption intentions associ-ated with decision rules in mobile applications. The second phase ofthe project uses formal concept analysis (FCA) to infer decision rulesand variables. Practical and research implications are also offered.

3 - Semantic Web services composition with queryrewriting under constraintsKhaled Sellami, LMA Laboratory, Bejaia University / EISTIFrance, 06000, Bejaia, Algeria, [email protected],Mohamed Ahmed-Nacer, Rachid Chelouah, Hubert Kadima,Nadia Halfoune

The concept of "Web services" is essentially an application availableon the Internet by a service provider, and accessible by customersthrough standard Internet protocols. Examples of services currentlyavailable are the booking travel online, banking or the implementationof the management of the supply chain. The purpose of this work is topresent, in one hand, a model of semantic annotations for describingthe Web services and the user request and in another hand, an algo-rithm which discovers and composes the Web services using the queryrewriting under constraints.

4 - Logical analysis of data for a study of proverbsJorge Santos, Matemaatica, Univ Evora, Rua RomãoRamalho, 59, 7000-671, Évora, Portugal,[email protected], Luís Cavique, Armando Mendes

A case study of proverbs is presented as a part of a wider project, basedon data collected by thousands of interviews and involving a set ofthousands Portuguese proverbs. We search the minimum informationneeded to identify the birthplace of an interviewee in a database withinconsistencies. An approach to data reduction is given by the Logi-cal Analysis of Data (LAD). LAD handicaps, like the inability to copewith the contradiction and the limited number of classification classes,will be overcome in this version of Logical Analysis of InconsistentData.

99


� TA-32Tuesday, 8:30-10:00CC-A34

Soft OR case studiesStream: Soft OR and Problem Structuring MethodsInvited sessionChair: Giles Hindle, Mangement Systems, Hull UniversityBusiness School, University of Hull, Cottingham Road, HU6 7RX,Hull, United Kingdom, [email protected]

1 - Extending WASAN — Case Study of a UK PoliceForceChris Smith, OIM Group, Aston University, Aston BusinessSchool, Aston Triangle, B4 7ET, Birmingham, WestMidlands, United Kingdom, [email protected], DuncanShaw

This paper extends the theoretical and practical application of Wasteand Source Matter Analysis (WASAN) methodology beyond its initialconception. Applying WASAN in a case outside of the environmentit was designed for begins to show that WASAN has generic proper-ties and can be used in a number of different problem situations. Tosuccessfully apply WASAN in the new case significant development ofthe methodology was required. The development of the methodologywill take an action research approach operationalised in a case studyenvironment. This paper will form part of the researchers PhD.

2 - The secret’s in the mix: using OR to inform learningand teaching developmentsJo Smedley, Centre for Excellence in Learning and Teaching,University of Wales, Newport, Lodge Road, Caerleon, NP183QT, Newport, Wales, United Kingdom,[email protected]

Problem structuring methods help to structure complex problems togain a common understanding among a group and progress on a com-mitted course of action. This paper illustrates how a combination ofsoft and hard OR approaches supported the design, development andimplementation of an organisational learning and teaching initiativeacross a UK University. OR as a collaborative learning "language"across disciplines and topics supported effective and efficient engage-ment helping to achieve successful outcomes and inform continuingdevelopments.

3 - A conceptual modelling framework for Public Healtheconomic modellingHazel Squires, ScHARR, University of Sheffield, RegentCourt, 30 Regent Street, S1 4DA, Sheffield, South Yorkshire,United Kingdom, [email protected], Jim Chilcott,Ron Akehurst, Jennifer Burr

Commonly used methods for health economic modelling are often di-rectly applied to Public Health (PH) interventions. However, an inad-equate understanding of these complex PH problems and the resultingimplications for model development means that inappropriate modelstructures may be used. We present a conceptual modelling frame-work (CMF) to facilitate model development within PH economicmodelling, developed from (i) a review of key challenges of PH eco-nomic modelling, (ii) a review of CMFs and (iii) qualitative research tounderstand modellers’ experiences developing PH economic models.

4 - Participatory Decision Processes in Rural Communi-ties in TanzaniaJoe Kakeneno, Management Systems, Tanzania PortsAuthority, P.O Box 9184, Dar es Salaam, N/A, Dar es Salaam,N/A, Tanzania, United Republic Of,[email protected], Cathal Brugha

We discuss the usability of the Structured Multi-Criteria Decision-Making (MCDM) methodology in participatory decision processesfor selecting suitable irrigation methods to be implemented by ruralcommunities in Tanzania. An MCDM framework which could be in-tegrated into existing community participatory decision making pro-cesses was developed and used to arrive at community preferences.The framework could be used to support distributed decision makingin various problem contexts and situations.

� TA-33Tuesday, 8:30-10:00CC-A37

OR and Transportation: Applications inDeveloping Countries

Stream: OR for Development and Developing Coun-triesInvited sessionChair: Youssef Masmoudi, University of Sfax, Hight School ofCommerce of Sfax, BP 954, 3018, Sfax, Tunisia,[email protected]: Habib Chabchoub, LOGIQ, Institut Supérieur de GestionIndustrielle, 3018, Sfax, Sfax, Tunisia,[email protected]

1 - Static multiclass truck assignment model: an appli-cation to NigeriaPrecious Ikem, Civil Engineering, imperial, SkemptonBuilding, sw7 2az, London, United Kingdom,[email protected], Michael BellThis paper reports on the development of a traffic model built to esti-mate the potential impact of a dry port solution on traffic congestion inLagos. It outlines a conceptual framework for assigning import freighttraffic to a road network while considering the characteristics of thefreight haulage industry in developing countries, particularly Nigeria.The use of a static multiclass user equilibrium assignment model withan inter-zone container flow survey data is proposed. The classes rep-resent containerized cargo categorized by the type of vehicle carryingthe container or its contents.

2 - New approach to the Single Track Railway Schedul-ing ProblemMaya Laskova, The Institute of Control Sciences V. A.Trapeznikov Academy of Sciences, Moscow, Russia, RussianFederation, [email protected], AlexanderLazarev, Elena MusatovaWe consider the single track railway scheduling problem with two sta-tions. Two subsets of trains are given, where trains from station one goto station two, and trains from station two go to station one. All trainshave the same speed. At most one train from station one or two can beon the track at any time. We consider the single track railway schedul-ing problem with two stations with optimization criteria of minimizingadditional completion time for each trains. We propose a heuristic ap-proach to The Single Track Railway Scheduling Problem.

3 - Humanitarian Operations Research in SouthernAfricaHildah Mashira, Applied Mathematics, National University ofScience and Technology, 14407 Inkhaka Circle, SelbournePark, Bulawayo, Zimbabwe, [email protected] HIV/AIDS pandemic has reached an unprecedented scale in SouthAfrica. The burden that this has placed on the country is already re-stricted health care resources place a greater emphasis on the valueof Anti Retroviral drug distribution. The aim of my project is to de-sign a distribution model for the department of health in the Limpopoprovince. This will seek to present major stages applicable to buildingeffective models and routes that are cost effective and ensure that drugsare delivered to almost every part of the province.

� TA-34Tuesday, 8:30-10:00CC-A39

Environment and SustainableDevelopment - Green Teachers

Stream: Optimization for Sustainable DevelopmentInvited sessionChair: Azizah Hanim Nasution, Graduate School of Natural

100


Resources and Environment Management, University of SumateraUtara, Indonesia, [email protected]

1 - A Markov-based model in managing coordination re-lationship of school communities to achieve environ-mental behaviourAzizah Hanim Nasution, Grad.School of Env. Manag &Natural Resources, University of Sumatera Utara, FMIPAUSU, 20155, Medan, North Sumatera Province, Indonesia,[email protected], Herman MawengkangThe world today concerns with environmental problems. Most of theproblems might be caused by human behavior.The most effective wayto enhance environmental behavior is through education. The commu-nities involved in the educational system of a school can be regardedas a multi agent system. In this paper we use Markov-based dynamicinfluence diagram to model the coordination relationships such thatagents are able to both represent and infer how their activities affectother agents’ activities in a way to achieve the environmental behaviorobjective.

2 - A decision analysis approach model to maximize stu-dents’ cognition of environmental problemsHidayati Hidayati, Environmental Protection Board,Government of North Sumatera Province, Jl. Diponegoro,Medan, Indonesia, [email protected], Azizah HanimNasutionA learner is a cognitive system that develops by his/her information andknowledge-processing activities. To maximise the learner’s cognitivedevelopment, knowledge-intensive environments are essential particu-larly for achieving environmental knowledge. This paper proposes amodel based on a decision analysis approach evaluating students’ cog-nitive development toward environmental problems.

3 - A Stochastic Programming model for SustainableProduction Planning of Crude Palm Oil IndustryHendaru Sadyadharma, Grad.School of Env. Manag &Natural Resources, University of Sumatera Utara, PSL USU,20155, Medan, North Sumatera Province, Indonesia,[email protected], Herman MawengkangThis paper addresses a multi-objective stochastic programming modelof the sustainable production planning of crude palm oil. The modeltakes into account conflicting goals such as return and financial riskand environmental costs. The uncertainty comes from the price ofcrude palm oil. Starting from it two single objective models are for-mulated: a maximum expected return model and a minimum financialrisk (pollution penalties) model. We transform the stochastic program-ming model into a deterministic multi-objective model using samplingaverage approach.

� TA-35Tuesday, 8:30-10:00CC-A41

Pricing and Supply Chain Management

Stream: Operations/Marketing InterfaceInvited sessionChair: Kathryn E. Stecke, University of Texas at Dallas, UnitedStates, [email protected]

1 - Impact of Returns Policies on Retail Assortment andPricing DecisionsAlex Grasas, Economics and Business, Universitat PompeuFabra, Ramon Trias Fargas 25-27, 08005, Barcelona, Spain,[email protected], Aydin AlptekinogluWe analyze a retailer’s assortment, pricing and return policy (fractionof price refunded upon return) decisions. The retailer must choose itsassortment from an arbitrary set of products, and determine the priceand refund amount for every product. We model the consumer choicebehavior using a nested-MNL-based choice model, and derive a sim-ple metric to rank products, which reveals the optimal assortment quiteefficiently. We compare the optimal policies for this integrated modelto the case when the return policy is exogenous, driven by category- orstore-wide considerations.

2 - Dynamic Pricing of Fashion Products: The Effects ofDemand Learning and Strategic Consumer BehaviorYossi Aviv, Olin Business School, Washington University inSaint Louis, Campus Box 1133, 1 Brookings Drive, 63130,Saint Louis, MO, United States, [email protected] propose a dynamic-pricing model of a retailer that sells a productto strategic consumers. The seller learns about the market size throughsales observations. This raises an interesting phenomenon: consumerscan influence the seller’s perception about the market size. We willdiscuss the latter point, and examine the effectiveness of price commit-ment strategies. We will also consider a model of a retailer that doesnot incorporate learning from early sales. This leads to an interest-ing question: is it better to become sophisticated (capable of learning)when facing strategic consumers?

3 - The study of a time-dependent Freight Transporta-tion ProblemXiang Song, School of Mathematics, University ofPortsmouth, Lion Gate Building, Portsmouth, UnitedKingdom, [email protected] transportation is a major component of the supply chain man-agement. We aim to study a "time-dependent’ freight transportationproblem, which takes varied time between two locations into consid-eration. The main methodology we adopt is a number of dynamic pro-gramming state space relaxation techniques and their applications tosolving the time-dependent Freight Transportation Problem. The sim-ulated data from a software-development company, which provides ve-hicle routing and scheduling solutions to UK-based customers, will beused for evaluating our result.

4 - An Integrated Approach to Green Product and Sup-ply Chain DesignCornelia Schoen, GISMA Business School, LeibnizUniversity Hannover, Germany,[email protected] many industries, product design increasingly requires to integrateeconomic objectives with environmental thinking - driven by raisingecological concerns, regulatory pressures, and the potential to create amarketing edge through sustainable operations. A major challenge ofgreen design is that it must not only embrace a product’s key featuresfrom a consumer perspective but also all underlying supply chain pro-cesses that determine a product’s ’greenness’. We present an integratedoptimization approach for green product and supply chain design alongwith applications.

� TA-36Tuesday, 8:30-10:00CC-A43


Stream: DEA and Performance MeasurementInvited sessionChair: Emmanuel Thanassoulis, Aston Business School, AstonUniversity, Operations and Information Management, B4 7ET,BIRMINGHAM, United Kingdom, [email protected]

1 - Prioritizing project activities via DEA methodologyYossi Hadad, Industrial Engineering and Management, SCE -Shamoon College of Engineering, Balik Bazel, 8 PeretzLasker St., Beer Sheva 84519, Israel, 84100, Beer Sheva,Israel, [email protected], Baruch Keren, Lea FriedmanRanking project activities by importance, and then clustering the activ-ities into subgroups that require additional attention is very important.Ranking project activities may assist project managers in identifyingthe most significant activities, and thus allocate resources effectively.A multi-criteria ranking based on several indexes may provide betterrankings than those based on a single index. This research proposes amethod for ranking project activities, where each activity is evaluatedby several indexes. The proposed method applies DEA methodologyand the Super Efficiency model.

101


2 - Cognitive Effects of DEA on Performance Assess-mentNadia Vazquez Novoa, Institute of Management Control andBusiness Accounting, TU Braunschweig, Pockelsstraße 14,38106, Braunschweig, Lower Saxony, Germany,[email protected], Heinz Ahn

Psychological evidence suggests that the use of DEA efficiency scoresmay lead to significant cognitive biases. An experimental study con-sidering the representativeness and anchoring effects of DEA scoreswas conducted. The findings not only confirm that DEA scores act asa strong performance marker when deciding on which DMUs shouldbe awarded for their non-financial performance, but also show thatthe scores may significantly influence a posterior financial assessment.These results have practical consequences for planning, reporting, andcontrolling processes that incorporate DEA scores.

3 - More New Features of Data Envelopment AnalysisOnline Software (DEAOS)Mohammad-Reza Alirezaee, School of Mathematics, IranUniversity of Science and Technology, Hengam St., ResalatSq., 16846, Tehran, Iran, Islamic Republic Of,[email protected], Ali Niknejad, Nassrin Alirezaee

DEAOS has been presented as a professional web application for dataenvelopment analysis (DEA) including several available models, mul-tiple time periods facilities, useful features, user friendly interface andreporting, importing data, exporting results and easy to use on inter-net. Here more new features, added to DEAOS, including categoricalselection, analytical hierarchy process (AHP) analysis, and many otheritems will be presented. DEAOS is available at www.DEAOS.com.

4 - Data Envelopment Analysis software for the ad-vanced usersAli Emrouznejad, Aston Business School, Aston University,B4 7ET, Birmingham, United Kingdom,[email protected], Emmanuel Thanassoulis

This paper presents software that takes its features closer to the lat-est developments in the DEA literature. The new software addresses avariety of issues such as: Assessments under a variety of possible as-sumptions of returns to scale including NIRS and NDRS; undesirableinput/output variables, slack-based measures, truly unlimited numberof assessment units (DMUs); analysis of groups of data by estimat-ing automatically separate boundaries by group; Malmquist Index andits decompositions; Super efficiency; Cross efficiency; Allocative effi-ciency and Bootstrapping.

� TA-37Tuesday, 8:30-10:00CC-Act

OR in Health & Life Sciences 1Stream: OR in Health & Life Sciences (contributed)Contributed sessionChair: Abraham Seidmann, University of Rochester, 14627,Rochester, NY, United States, [email protected]

1 - Utility of quantitative methods in strategy — makingprocess for blood donation system — logistics ap-proachSebastian Twaróg, Department of Business Logistics,University of Economics in Katowice, 1 Maja 50, 40-287,Katowice, Poland, [email protected], AnnaOjrzynska, Grazyna Trzpiot, Jacek Szołtysek

An effective blood donation system is centrepiece to assuring healthcare in every country. Improvement of such a system is an importanttask for logistician. This paper aims to point out the possibility of ap-plications of chosen quantitative methods in developing the blood do-nation system. An important part of our research was multivariate andspatial differentiation of discussed issues. Verified was the usefulnessof multivariate method data mining and spatial autocorrelations. Thoseresults are prerequisite for creating a blood donation system model.

2 - A simulation-based optimization tool for patient allo-cation to clinical studiesStephan Tiesler, University of Jena, Faculty of Business andEconomics, Chair of Business Informatics, Carl-Zeiss-Straße3, 03643, Jena, Thuringia, Germany,[email protected], Johannes Ruhland

In all sorts of clinical studies, patient enrolment and recruitment con-stitutes both a critical success factor and a scarce resource. In largeproject settings numerous studies with diverse intake requirements anddue dates are carried out simultaneously, creating competing demandfor a patient "supply’ that is stochastic and subject to numerous le-gal, structural and medical restrictions. We develop a simulation basedoptimization tool that will allow for optimal allocation of patients tostudies in real time and that can also be used to guide strategic issuesof study planning.

3 - Using decision analytic modeling for the optimal di-chotomization of diagnosticsGimon de Graaf, Epidemiology, University Medical CenterGroningen, Hanzeplein 1, P.O. Box 30 001, 9700 RB,Groningen, Netherlands, [email protected], DouwePostmus

The chosen cut-off on a diagnostic test for clinical decisions hasfar-reaching consequences on costs and health effects. Surprisingly,this cut-off is almost never chosen based on such downstream conse-quences, but rather on an arbitrarily chosen balance between false pos-itives and false negatives. We describe how an optimal cut-off basedon downstream consequences can be identified with help of multi-statemodeling of a patient’s disease progression. We illustrate this methodby means of a case study of a diabetes screening program.

4 - Custom contract and the role of Group PurchasingOrganizations (GPOs) as information intermediariesVera Tilson, University of Rochester, 14627, Rochester,United States, [email protected], AbrahamSeidmann, Rajib Saha

Many hospitals in the US join GPOs in order to get deeper group dis-count contracted by the GPOs. However, some members further nego-tiate directly with the same vendors. The common perception is thathospitals benefit from such directly-established "custom contracts’ asthey yield lower prices. Using a game-theoretic model, we find that theprovision for custom contracts, in fact, benefits vendors at the expenseof hospitals. We show, how with the provision for custom contracts,GPOs expectedly act as demand aggregators for small hospitals, andas information intermediaries for the rest.

� TA-38Tuesday, 8:30-10:00HH-Colombus

Interactive Multiobjective Optimization:Methods and Applications

Stream: Multiobjective OptimizationInvited sessionChair: Kaisa Miettinen, Dept. of Mathematical InformationTechnology, University of Jyvaskyla, P.O. Box 35 (Agora),FI-40014, University of Jyvaskyla, Finland, [email protected]: Jussi Hakanen, Dept. of Mathematical InformationTechnology, University of Jyvaskyla, P.O. Box 35 (Agora),FI-40014, University of Jyvaskyla, Finland, [email protected]

1 - Interactive Gradient Projection and Reference PointApproach for Multiobjective Performance PlanningDong-Ling Xu, Manchester Business School, The Universityof Manchester, F37 MBSE, Booth Street West, M15 6PB,Manchester, England, United Kingdom, [email protected],Jian-Bo Yang

In the paper, equivalence relationships are explored between data en-velopment analysis (DEA) and multiple objective linear programming

102


models, leading to a hybrid approach in which the two functions ofperformance assessment and planning can be integrated where DM’spreferences can be taken into account in an interactive fashion. It isproved that the input-oriented DEA dual models are identical to ref-erence point minimax models, both of which can be used for con-ventional efficiency analysis, whilst the latter can be used to conducttrade-off analysis for performance planning based on a gradient pro-jection approach. A case study for a supplier performance assessmentproblem is explored in detail.

2 - Decision making in a multi-objective paper mill de-sign problemIngrida Steponavice, Dept. of Mathematical InformationTechnology, University of Jyvaskyla, FI-40100, University ofJyvaskyla, Finland, [email protected], Sauli Ruuska,Kaisa Miettinen

Application of interactive multiobjective optimization to support deci-sion making in industry where modern production processes necessi-tate increasingly complex and computationally costly models imposeschallenges. A computationally expensive integrated control and designproblem concerning both flexibility in design and amount of capitalinvested in paper is considered. A computationally inexpensive surro-gate problem is prepared with the PAINT method while the interactiveIND-NIMBUS tool is used to direct a search in finding the most satis-factory solution according to the DM’s preferences.

3 - Consistent parametric representation of Pareto opti-mal solutions via geometric criteriaAlberto Lovison, Dipartimento di Metodi e ModelliMatematici, Università degli Studi di Padova, 35121, Padova,Italy, [email protected], Markus Hartikainen

In many practical multiobjective optimization problems, the set ofPareto optimal (PO) solutions exhibits non trivial manifold structuresboth in objective and decision spaces. In particular, for nonconvexobjective functions, different branches of the set of PO solutions canbe mapped one onto the other, causing serious problems to approx-imation algorithms. We propose an algorithm based on geometricalcriteria which can discriminate among different branches and produceconsistent parametric representations of the set of PO solutions, as isshown by examples.

4 - Parameterized achievement scalarizing functionsbased approach to solve multicriteria median loca-tion problemOlga Karelkina, Mathematics, University of Turku, TurunYliopisto, 20014, Turku, Finland, [email protected]

This research addresses interactive optimization techniques based onparameterized achievement scalarizing functions (ASFs). A decisionmaking process is simulated for multicriteria median location problemto illustrate the way how synchronous usage of ASFs may reduce thenumber of iterations in an interactive process and speed up its con-vergence to the most preferred solution. Numerical experiments areperformed for the case of three objectives.

� TA-39Tuesday, 8:30-10:00HH-Cousteau

MCDA: New Approaches and Applicationsin Financial MarketsStream: MCDA: New Approaches and ApplicationsInvited sessionChair: Gabriela Fernández Barberis, Quantitative Methods, SanPablo CEU University, 23, Julián Romea St., 28003, Madrid,Spain, Spain, [email protected]

1 - Multiple Criteria Decision Aid Methods with NewGeneralized Criteria: an application to the financialmarkets

Gabriela Fernández Barberis, Quantitative Methods, SanPablo CEU University, 23, Julián Romea St., 28003, Madrid,Spain, Spain, [email protected], María del Carmen EscribanoRódenas, Maria Del Carmen Garcia Centeno

We propose New Types of Generalized Criteria which gives the de-cision maker the chance to express his preferences more easily. Toillustrate the relevance of the preference functions introduced in theanalysis we applied the PROMETHEE Methods with New General-ized Criteria to order the preference between the daily financial returnsof some Stock Market Indexes of different countries. In the financialmarket, the volatility is an essential variable. We have used differenteconometric models to estimate the stochastic volatility.

2 - Towards a new classification of Spanish SavingBanks: application of multicriteria decision aid meth-odsMilagros Gutiérrez Fernández, Financial Economy andAcounting, University of Extremadura, Av, University s/n,10071, Cáceres, Spain, [email protected], GabrielaFernández Barberis, Ricardo Palomo Zurdo

The objective of this paper is to propose a classification of the SpanishSaving Banks under a criterion of integration that takes into accounteconomic and social variables. We have study 45 saving banks in theperiod 1999-2009. Multicriteria decisions methods belong to the fam-ily of Outranking Relations that have proved to be a tool which is veryimportant in this type of analysis.

� TA-40Tuesday, 8:30-10:00HH-Livingstone

Dynamic Pricing

Stream: Revenue ManagementInvited sessionChair: Mikhail Nediak, School of Business, Queen’s University,143 Union Street, K7L 3N6, Kingston, Ontario, Canada,[email protected]

1 - Revenue Management with Lifetime Value Consider-ations: Balancing Customer Acquisition and Reten-tion Spending for Firms with Limited CapacityAnton Ovchinnikov, University of Virginia, Darden School ofBusiness, 100 Darden Blvd, 22903, Charlottesville, VA,[email protected]

This paper discusses the interaction between revenue management(RM) and customer relationship management (CRM) for a firm thatoperates in a customer retention situation but faces limited capacity.We present a dynamic programming model for how the firm balancesinvestments in customer acquisition and retention, as well as retentionacross multiple customer types. We characterize the optimal policyand discuss how the policy changes depending on capacity limitations.We then contrast the modeling results with those of a behavioral ex-periment.

2 - Use of exact algorithms for Partially ObservableMarkov Decision Processes in solving a DynamicPricing problemShahin Abbaszadeh, School of Management, University ofSouthampton, University of Southampton, University Road,SO17 1BJ, Southampton, United Kingdom,[email protected]

We use Partially Observable Markov Decision Processes (POMDP) tofind optimal dynamic pricing policies for a retailer of short-life-cycleproducts with no replenishment during the season. The objective is tomaximize expected profit using predefined prices during the season.We show our problem has a Convex and Piecewise Linear value func-tion and employ three exact POMDP algorithms that make use of thatproperty. We analyse performance of these methods in comparison tocommon practices and strategies such as myopic pricing and discussmanagerial insights and potential future research.

103


3 - Name-Your-Own-Price sales channels: Can strategicconsumers improve their performance with informa-tion exchange or collaboration?Mikhail Nediak, School of Business, Queen’s University, 143Union Street, K7L 3N6, Kingston, Ontario, Canada,[email protected], Tatsiana Levina, Yuri Levin,Jeff McGillSupplier reluctance to openly advertise large discounts lead to the cre-ation of opaque name-your-own-price channels, where firms can selldiscounted products in generic categories without revealing their iden-tity until a purchase is completed. Prices in these channels are con-cealed and bid-sampling is restricted. Consumers responded by devel-oping social networking practices aimed at exploiting these services.In this work, we explore several robust bidding strategies and the valueof collaboration among consumers in the form of exchange of bid his-tory, coordinated bidding, and risks-pooling.

� TA-41Tuesday, 8:30-10:00HH-Heyerdahl

AHP/ANP 4Stream: Analytic Hierarchy Process, Analytic NetworkProcessInvited sessionChair: Y. Ilker Topcu, Industrial Engineering, Istanbul TechnicalUniversity, Istanbul Teknik Universitesi, Isletme Fakultesi, Macka,34357, Istanbul, Turkey, [email protected]

1 - Applying ANP to Select Strategic Alliances for Inter-national Tourist Hotels in TaiwanSu-Chuan Shih, Business Administration, ProvidenceUniversity, 200 Chung Chi Road, Business Administration,43301, Taichung City, Taiwan, Taiwan, [email protected] tourist hotels often adopt strategic alliance to increasetheir competition advantage. This study employs Analytic NetworkProcess to select strategic alliance types. The result shows the mostimportant key criterion is the performance goal of the alliance. Un-der the key criterion, the index of profit rate is weighted more thanthe others. The preference alliance types are ordered as: the complexactivities, the electronic channels, and the physical shops.

2 - An Integrated Multi-Criteria Approach for the Evalua-tion and Control of Strategic OptionsRalf Kaspar, Chair of Managerial Accounting, University ofOsnabrueck, 49069, Osnabrueck, Germany,[email protected] evaluation and control of strategic options within the strategicmanagement process is fundamentally complicated by multiple andmutually influencing objectives, internal and external factors, lackof transparency, complexity and conflicts of interest. To cope withthese difficulties a suitable multi-criteria decision support approachis required. Research has led to the development of an ANP basedframework. Supported by a case study it is a contribution to an im-proved evaluation of strategic options incorporating advanced SWOT-Analysis, BOCR modeling and group decision making.

3 - A Multi-Criteria Based Evaluation of "InnovationStrategy Selection’Ozge Surer, Industrial Engineering, Istanbul TechnicalUniversity, ITU Isletme Fakultesi, Macka, 34357, Istanbul,Turkey, [email protected], Sezi Cevik Onar, Y. Ilker TopcuInnovation is a crucial tool for firms to compete, thus it is necessary todefine the factors which affect this process. Containing interrelationsamong the related factors, a multi criteria approach utilizing networkmodels should be used. In accordance with the iterative steps of An-alytic Network Process, innovation strategy types and related factorsare identified by opinions of experts and the literature survey. Rela-tions among these factors are revealed and their relative importanceand the priorities of the strategies are assessed, based on the pairwisecomparison answers of the experts.

� TA-42Tuesday, 8:30-10:00BW-Amber

Simulation Models and Decision Analysis

Stream: Stochastic Modeling and Simulation in Engi-neering, Management and ScienceInvited sessionChair: Erik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39, 85577,Neubiberg, Germany, [email protected]: Silja Meyer-Nieberg, Department of Computer Science,Universität der Bundeswehr München, 85577, Neubiberg,Germany, [email protected]: Sonia De Cosmis, School of Science and Technology,Università di Camerino, via Madonna delle Carceri 9, 62032,Camerino (MC), ITALY, Italy, [email protected]

1 - Modelling a Distribution Operation Using DiscreteEvent-Based SimulationAmílcar Arantes, Civil Engineering-DECivil, IST/TechnicalUniversity of Lisbon, Av. Rovisco Pais, 1049-001, LISBOA,Portugal, [email protected]

This work considers the development of a simulation model and itsapplications to a case-study based on realistic inputs from a distribu-tion company. It considers the activities and resources involved in thedistribution process and the chain configuration. The model was de-veloped after a literature review and interviews, which were essentialfor problem characterization and to get realistic data. The results showthat event based simulation and the SIMUL8 can work as a valid designdistribution chain decision tool, especially because it can cope with thereal world inevitable randomness.

2 - Generation of EPC based Simulation ModelsChristian Mueller, Dep. of Management and Businesscomputing, TH Wildau, Bahnhofstrasse, D-15745, Wildau,Germany, [email protected]

In this paper, we present a approach, where business processes are for-mulated in EPC notation with some extensions about simulation rele-vant data. For EPC modeling Bflow, an Eclipse plugin is used. Fromthis EPC model we generate a DesmoJ simulation model. A DesmoJmodel contains a set of Java classes that use the DesmoJ simulationframework. Based on the Java programing language, the generatedmodels can extended in all directions. The generation mechanism sup-ports a technique, so that the modeling extensions are not lost by re-generation. Bflow and DesmoJ, are both open source projects.

3 - Rare-Event Estimation for Density Dependent Mod-elsAdam Grace, Mathematics, University of Queensland,University of Queensland, St Lucia, 4072, Brisbane, QLD,Australia, [email protected]

Many complex systems in chemistry, biology and engineering can bemodeled via density dependent models. Applications include chemicalreactions, population dynamics and traffic flow in telecommunicationnetworks. Rare-event estimation and simulation for such models canbe difficult. We present an alternative approach to this problem whichutilizes the limiting behavior of density dependent models. We demon-strate this method requires fewer samples than current methods, and isalso more robust to correlation issues.

4 - Adding adaptability to Discrete Event Simulationthrough parameterisationFrancesco Aggogeri, Department of Mechanical Engineering,University of Brescia, Via Branze 38, 25123, Brescia, Italy,[email protected], Gian Mauro Maneia, MarcoMazzola, Angelo Merlo

104


A novel approach is proposed aiming to facilitate simulation of man-ufacturing processes by fast modelling and evaluation. Architectureand validation of a DES-based parametric model are presented. Themodel consists on a flexible and adaptable solution enabling to simu-late either manufacturing cells and complex lines with same structure.So a user can tailor the model on his processes without specific knowl-edge of simulation software. The authors acknowledge the support ofthe European Commission through the project COPERNICO (CP-IP229025-2).

� TA-43Tuesday, 8:30-10:00BW-Granite

Stochastic programming in energy

Stream: Stochastic ProgrammingInvited sessionChair: Asgeir Tomasgard, Applied economics and operationsresearch, Sintef Technology and society, Trondheim, Norway,[email protected]

1 - The value of storage at household level: A smart gridperspectiveStein W. Wallace, Management Science, Lancaster UniversityManagement School, LA1 4YX, Lancaster, United Kingdom,[email protected]

As the percentage of wind power increases, storage of energy will beneeded. The purpose of this work is to understand the question of stor-age bottom-up. Is there any value in storage at residential level, and ifso, what is the value? We do not not assume behavioural changes, onlythat the building uses its storage capabilities optimally.

2 - Gas Network Topology Extension for Multiple Sce-nariosJonas Schweiger, Optimization, Zuse Institute Berlin (ZIB),Takustr. 7, 14195, Berlin, Germany, [email protected]

Due to the European regulation gas networks have to become moreflexible. Investments are needed to allow the network operator to ful-fill different demand scenarios. We developed a model for the networkextension problem for multiple demand scenarios. That is, we searchcost-optimal network extensions such that a variety of demand scenar-ios can be realized. The problem is described by a two-stage stochasticprogram. We propose a decomposition along the scenarios and solvethe problem by a Branch&Bound-algorithm. Non-convex MINLP areused to model the gas network and solved as subproblem.

3 - A Bilevel Programming Approach to Modeling ofTransmission Capacity PlanningPaolo Pisciella, Department of Information Technology andMathematical Methods, Unversity of Bergamo, Italy,[email protected], Marida Bertocchi, Maria TeresaVespucci

We introduce a model intended for the analysis of the upgrade of thenational transmission grid. The problem is modeled as a bilevel pro-gram with mixed integer structure in both, upper and lower level. Up-per level is defined by the transmission company problem which has todecide on how to upgrade the network so to avoid congestions. Lowerlevel models the reactions of generating companies, which take a de-cision on new facilities and power output, and market operator whichstrikes a new balance between demand and supply providing new zonalprices.

4 - Long-term expansion of the European Power SystemUnder Various Emission Mitigation ScenariosChristian Skar, Department of Electric Power Engineering,Norwegian University of Science and Technology, 7491,Trondheim, Norway, [email protected], AsgeirTomasgard, Gerard Doorman

This paper will present an analysis of optimal expansion of generationand transmission capacity for a country-level power system model ofEurope. Important driving factors for investments in new infrastruc-ture like annual demand, fuel prices and prices for CO2 emissions, aretaken directly from the Global Change Assessment Model (GCAM).As electricity demand in GCAM is only calculated for two regions,Western Europe and Eastern Europe, a disaggregation based on statis-tical data is used to get country-wise demand. The expansion problemis formulated as a two-stage stochastic linear program.

� TA-44Tuesday, 8:30-10:00BW-Marble

Computer vision (1)

Stream: Machine Learning and its ApplicationsInvited sessionChair: Teofilo deCampos, Center for Vision Speech and SignalProcessing, University of Surrey, GU2 7XH, Surrey, UnitedKingdom, [email protected]: Fei Yan, Center for Vision Speech and Signal Processing,University of Surrey, GU2 7XH, Surrey, United Kingdom,[email protected]: Ivan Reyer, Dorodnicyn Computing Centre of RAS, ul.Pererva d. 58, kv. 153, 109451, Moscow, Russian Federation,[email protected]

1 - Anomaly detection in machine perception systemsJosef Kittler, Centre for Vision, Speech and SignalProcessing, University of Surrey, GU2 7XH, Guildford,United Kingdom, [email protected]

Anomaly detection in engineering systems is cast as a problem of de-tecting outliers to the distribution of observations representing a stateof normality. We focus on anomaly detection in machine perception.We argue that in addition to outlier detection, anomaly detection in ma-chine perception systems requires other detection mechanisms. Theyinclude incongruence detection, data quality assessment, decision con-fidence gauging, and model drift detection. These mechanisms areelaborated and their application illustrated on a problem of anomalydetection in a sports video interpretation system.

2 - Learning Optimal Semantically Relevant Visual Dic-tionary using Evolutionary AlgorithmsAshish Gupta, CVSSP, University of Surrey, University ofSurrey, GU27XH, Guildford, Surrey, United Kingdom,[email protected]

We propose a novel optimally discriminative visual dictionary learn-ing method using evolutionary algorithms. Due to large intra-categoryappearance variation, semantically equivalent descriptors are scatteredover feature space. Current methods, using hard or soft assignment,are unable to cluster these equivalent descriptors, inter-leaved withother categories, together. We over partition feature space, generat-ing a huge population of tiny syntactically cohesive partitions, whichare optimally assigned, with locality constraints, so as to obtain a max-imally discriminative visual dictionary.

3 - Transductive Transfer Learning for Action Recogni-tionTeofilo deCampos, Center for Vision Speech and SignalProcessing, University of Surrey, GU2 7XH, Surrey, UnitedKingdom, [email protected], Nazli FarajiDavar

In video processing, it is often the case that one video may containseveral samples of actions that share a number of low level features(eg illumination, motion style, etc). Therefore, each video constitutesa domain. We show that, for action classification, a classifier that istrained with one video may lead to poor performance when appliedto another video, but if the unlabelled samples of this other video areavailable, it is possible to apply a transductive transfer learning tech-nique, leading to a significant improvement in performance.

105


� TA-45Tuesday, 8:30-10:00BW-Water

Stochastic Logistical Networks

Stream: Stochastic Modelling in Logistical NetworksInvited sessionChair: El-Houssaine Aghezzaf, Industrial Management, GhentUniversity, Technologiepark 903 -, Campus Ardoyen, 9052,Zwijnaarde, Belgium, [email protected]: Derek Verleye, Industrial Management, Ghent University,Technologiepark 903, 9052, Zwijnaarde, Belgium,[email protected]

1 - An Approximate Scenario-based Approach to DesignRobust Supply Chain Networks under UncertaintyWalid Klibi, Operations Management and InformationSystems Department, BEM / CIRRELT, 680 cours de laLibération, 33405, Bordeaux, France, [email protected],Alain Martel

Supply Chain Networks (SCN) must be designed to last for severalyears and cope with any plausible future, and thus involve a value max-imization problem under uncertainty. The original SCN design modelis intractable. We propose some complexity reduction mechanisms totackle the problem, and the design model is reduced to a bi-criterion,multi-period, two-stage stochastic program with recourse. This modelconsiders future scenarios including high-impact disruptive events andrecourse variables anticipating response policies in case of disruptions.

2 - On a strict solution algorithm for a stock replenish-ment policy with fill-rate constraints in the two-stepsupply chainTakashi Hasuike, Graduate School of Information Science andTechnology, Osaka University, 2-1 Yamadaoka, 565-0871,Suita, Osaka, Japan, [email protected], KojiOkuhara, Nobuyuki Ueno

This presentation proposes a strict solution algorithm to obtain a stockreplenishment policy with fill-rate constraints in the two-step supplychain. In previous researches, it is hard to solve this model strictly,because the model is formulated as a non-deterministic programmingproblem with several stochastic constraints. In order to overcomethe disadvantage and develop a strict and efficient solution approach,the deterministic equivalent transformations in mathematical program-ming are performed. Then, the strict optimal stock replenishment pol-icy is explicitly obtained.

3 - Stochastic and dynamic patient transportationJakob Puchinger, Mobility, AIT - ÖsterreichischesForschungs- und Prüfzentrum Arsenal Ges.m.b.H.,Giefinggasse 3, 1210, Wien, Austria,[email protected], Ulrike Ritzinger, Richard Hartl

We consider a patient transportation problem for an emergency med-ical service in Vienna. The organization has to perform a large num-ber of patient transports every day and, in addition, has to respond toarising emergency calls. In the considered application, about 60% oftransportation requests are known in advance, whereas other requestsare occurring throughout operations. Therefore, a stochastic and dy-namic variant of the dial-a-ride problem is considered. The probleminstances are big, consisting of a fleet of around 120 vehicles serving800 to 1000 requests a day.

4 - Supply chain coordination with capacity reservationand emergency optionGuillaume Amand, HEC-Ulg, University of Liège, boulevarddu rectorat, 7 - Bât. B31, 4000, Liège, Belgium,[email protected], Yasemin Arda

This study analyzes the contractual relation between a retailer and asupplier with the aim of determining possible deviations from the op-timal system performance. In order to face a random demand, the re-tailer submits a contract to the supplier based on the number of unitsdelivered and the capacity used to serve the retailer. Before the uncer-tainty is resolved, the supplier has to decide of the capacity that he re-serves. Once the demand is known, the carrier may request additional

capacity from an emergency option at a higher cost. An asymmetriccase is also discussed.

106

EURO 25 - Vilnius 2012 TB-03

Tuesday, 10:30-12h00

� TB-01Tuesday, 10:30-12h00RB-Alfa

Tutorial Lecture: Professor GuyDesaulniersStream: Keynote, Tutorial and Plenary LecturesTutorial sessionChair: Marc Sevaux, Lab-STICC - CNRS UMR 3192, Universitéde Bretagne Sud - UEB, Centre de Recherche - BP 92116, 2 rue deSaint Maude, 56321, Lorient, France, [email protected]

1 - 20 years of column generation for the vehicle routingproblem with time windowsGuy Desaulniers, École Polytechnique de Montréal andGERAD, H3C 3AT, Montréal, Canada,[email protected]

Column generation is a well-known mathematical programming tech-nique for solving large-scale optimization linear problems. In the mid-1980s, it was shown that it can be successfully embedded in a branch-and-bound framework (yielding a branch-and-price method) for solv-ing integer programs. Since then, it has become the most popular ex-act methododology for solving several classes of vehicle routing prob-lems. As a fertile testbed, the vehicle routing problem with time win-dows (VRPTW) played an important role in this success. It consistsof finding a set of least-cost vehicle routes to deliver goods to a set ofcustomers. Each customer must be visited exactly once within a giventime window and each route must respect vehicle capacity. In thistalk, we survey the evolution of the column generation methods devel-oped for the VRPTW, starting from the seminal work of Desrochers,Desrosiers, and Solomon (1992). In particular, we present various pathrelaxations proposed to avoid solving the NP-hard elementary short-est path subproblem, labeling algorithms used to generate their corre-sponding columns, and valid inequalities tightening the lower bounds,including those that are directly defined on the master problem vari-ables. We discuss certain acceleration strategies, report computationalresults, and highlight current and future challenges.

� TB-02Tuesday, 10:30-12h00RB-Beta

Scheduling and Logistics

Stream: SchedulingInvited sessionChair: Vitaly Strusevich, School of Computing and MathematicalSciences, University of Greenwich, Old Royal Naval College, ParkRow, Greenwich, SE10 9LS, London, United Kingdom,[email protected]

1 - Adding an Extra Machine to a Shop Parallel IdenticalMachinesVitaly Strusevich, School of Computing and MathematicalSciences, University of Greenwich, Old Royal Naval College,Park Row, Greenwich, SE10 9LS, London, United Kingdom,[email protected], Kabir Rustogi

We consider the problems of processing jobs on m parallel identicalmachines to minimize the makespan or the total flow time. We measurethe machine impact for m machines to the one with m+1 machines. Forthe makespan, in the preemptive case the impact is (m+1)/m, while inthe non-preemptive case the impact is 2. For minimizing the total flowtime, we prove that the impact is (m+1)/m. We present polynomial-time algorithms that determine an optimal number of machines, pro-vided that each machine incurs a cost and the objective function cap-tures a trade-off.

2 - Approximation Schemes for Scheduling on a Sin-gle Machine Subject to Cumulative Deterioration andMaintenanceKabir Rustogi, School of Computing and MathematicalScienceMathema, University of Greenwich, Old Royal NavalCollege, Park Row, Greenwich, SE10 9LS, London, UnitedKingdom, [email protected], Vitaly Strusevich, Hans Kellerer

We consider a scheduling problem on a single machine to minimizethe makespan. The processing conditions are subject to cumulative de-terioration, but can be restored by a single maintenance period (MP).The decision-maker is free to choose when to start the MP. We link theproblem to the Subset-sum problem (if the duration of MP is constant)and to the Half-product problem (if the duration of MP depends onits start time). For both versions of the problem, we adapt the exist-ing fully polynomial-time approximation schemes to our problems byhandling the additive constants, which are of the opposite sign with thevariable part of the objective.

3 - Supply Chain Scheduling in a Competitive Environ-mentMiguel Zamarripa, Departamento de Ingenieria Quimica,Universitat Politècnica de Catalunya, Spain,[email protected], Antonio Espuña

Someofthepublished works address Supply chain under uncertainty inthe available data, very few of them put the issue in a realistic opera-tional environment, taking into account that the SC of interest will haveto compete with other supply chains which, like to work as efficientlyas possible to cover the same market demands. This work proposes tointroduce into the model information about the expected performanceof the competing supply chains (including the SC of interest) to im-prove the operational decision making in a competitive/cooperative en-vironment using Game Theory optimization.

4 - Scheduling of coupled tasks with high multiplicityMichaël Gabay, Laboratoire G-SCOP, 46 avenue Félix Viallet,38031, Grenoble, France,[email protected], Gerd Finke, NadiaBrauner

The problem of scheduling tasks on a radar system can be modeled us-ing coupled-tasks. We consider the case where all tasks are identical.This case raises original complexity questions and belongs to so-calledhigh multiplicity scheduling problems. We present some complexityissues and results regarding the problem. We also describe a series ofcounter-examples showing that this problem is unlikely to be in NP.

� TB-03Tuesday, 10:30-12h00RB-L1


Stream: Logistics, Transportation, TrafficContributed sessionChair: Markus Bohlin, Swedish Institute of Computer Science,Box 1263, SE-16429, Kista, Sweden, [email protected]

1 - A heuristic solution approach for a real-size hierar-chical hub location problemJulia Sender, Institute of Transport Logistics, TU DortmundUniversity, Leonhard-Euler-Str. 2, 44227, Dortmund, NRW,Germany, [email protected], Uwe Clausen

In wagonload traffic, flows of single wagons with different origins anddestinations are consolidated on their routes through the railway net-work. The specific hierarchical hub-and-spoke-network is built up byseveral formations yards, which differ, for example, in their size andfunction. We present a specific hub location model for a strategic net-work design problem in wagonload traffic, which covers most charac-teristics of wagonload traffic in German railway logistics, and a newheuristic decomposition approach for solving the problem.

107

TB-04 EURO 25 - Vilnius 2012

2 - Auto-Carrier Distribution Problem with Highly Vari-able Daily Demand at DealersTsung-Sheng Chang, Graduate Institute of LogisticsManagement, National Dong Hwa University, 1, Sec.2,Da-Hsueh Rd., Shou-Feng, 974, Hualien, Taiwan,[email protected], Hao-Ping Jiang

Typically, auto-carriers must return to their original depots after com-pleting their daily duties. However, such an operations strategy is quiteineffective in the presence of highly variable daily demand at dealers.Hence, this research intends to tackle the auto-carrier distribution prob-lem that involves the main characteristics of multiple depots, multiplestaging areas, multiple trips, time window constraints, and no fixedhome depots. We first mathematically model the distribution problem,and then propose an optimization-based solution algorithm to solve theNP-hard distribution problem.

3 - Monte Carlo Simulation Techniques with ParallelHeuristics applied to solve the Capacitated VehicleRouting ProblemThiago Guimaraes, Centro Universitario Franciscano doParana, Rua Mateus Leme, 1244 - Apartamento 201 - CentroCívico, 80530000, Curitiba, Parana, Brazil,[email protected], Luis Gustavo Pereira, Wesley JoséNogueira Medeiros

It presents on this paper a new approach to solve the Capacitated Vehi-cle Routing Problem. The proposal is based on Cluster First-RouteSecond strategy, employing Monte Carlo Simulation Techniques tocluster the demand points and Classical Heuristics with permutation 2-opt to generation of the routes. Parallel computing techniques are usedto reducing the processing time. An extensive computational experi-ment with classical instances showed an average deviation of 1,40%compared with its exact results on the other hand a greatly reducedprocessing time.

4 - Shunting with car mixing using column generationMarkus Bohlin, Swedish Institute of Computer Science, Box1263, SE-16429, Kista, Sweden, [email protected],Florian Dahms, Holger Flier, Matus Mihalak

We consider a specific railway shunting problem that arose in theHallsberg hump yard, specified formally by Bohlin et al. who showedit to be NP hard. A natural MIP formulation give a weak linear re-laxation and consequently a large MIP gap. In our new work we usean extended formulation to model the problem’s specifications whilemaintaining a tight relaxation. To solve the linear relaxation of ourformulation we utilize column generation. The lower bounds providedby our relaxation are close (often exact) to the optimal objective andcan be calculated in reasonable time for large instances.


Approaches for Integer Programming

Stream: Combinatorial OptimizationInvited sessionChair: Juan José Salazar González, Estadística e InvestigaciónOperativa, Universidad de La Laguna (Tenerife), Av. AstrofísicoFrancisco Sánchez, s/n, Facultad de Matemáticas, 38271, LaLaguna, Tenerife, Spain, [email protected]: Martine Labbé, computer Science, Université Libre deBruxelles, CP210/01, Boulevard du Triomphe, 1050, Bruxelles,Belgium, [email protected]

1 - Optimal interval scheduling with a resource con-straintCarlo Filippi, Quantitative Methods, University of Brescia,Contrada S. Chiara 50, 25122, Brescia, BS, Italy,[email protected], Enrico Angelelli, Nicola Bianchessi

Consider a set of jobs and a set of parallel identical machines. Ev-ery job has: fixed start and finish times, a resource requirement, anda value. Every machine owns R units of a renewable resource andcan process more than one job at a time, provided the resource con-sumption is at most R. Two questions arise: Which is a job subsetof maximum total value that can be scheduled on a given number ofmachines? Which is the minimum number of machines required toschedule all jobs? For both problems we discuss complexity, heuristicand exact approaches, and preliminary computational experiments.

2 - Capacitated Network Design with Facility LocationIvana Ljubic, Department of Statistics and OperationsResearch, University of Vienna, Bruennerstr. 72, 1210,Vienna, Austria, [email protected], Bernard Gendron,Stefan Gollowitzer

We consider a new network design problem that arises in the design oflast mile telecommunication networks and combines the capacitatednetwork design problem with the single-source capacitated facility lo-cation. We compare several MIP formulations with respect to the qual-ity of their lower bounds and propose new sets of valid inequalities forthe problem. We also develop a branch-and-cut approach and providea computational study to assess its performance.

3 - A Branch-Cut-and-Price algorithm for the PiecewiseLinear Network Flow ProblemTue Christensen, Economics and Business, Aarhus University,Fuglesangs Allé 4, 8210, Aarhus V, Denmark, [email protected],Martine Labbé

In this paper we present an exact solution method for the network flowproblem with piecewise linear costs. This problem is fundamentalwithin supply chain management and extends the fixed-charge trans-portation problem in a straight-forward way and has numerous appli-cations. We propose a Dantzig-Wolfe reformulation of the problemand extend this to an exact method by branching and the addition ofvalid inequalities. On a number of randomly generated test instancesthe proposed method compares favorable to a standard mixed-integerprogramming solver with a significant reduction in runtime.


Problems on graphs I

Stream: Discrete Optimization, Geometry & Graphs(contributed)Contributed sessionChair: Nair Abreu, Production Engineering, Federal University ofRio de Janeiro, Rua Joao Lira 106/401, Leblon, 22430 210, Rio deJaneiro, Rio, Brazil, [email protected]

1 - On rational approximation of geometric graphVladimir Benediktovich, Department of CombinatorialModels and Algorithms, Institute of Mathematics of NAS ofBelarus, Surganova str., 11, 220072, Minsk, Belarus,[email protected]

A geometric graph is rational if all its edges have rational lengths. Kle-ber (2008) asked for what graph its vertices can be slightly perturbedin their e-neighborhoods so that the resulting graph becomes rational(e-approximation) and in addition vertices have rational coordinates(rational e-approximation). Any geometric cubic graph has a rationale-approximation for any e>0 (2008). In 2011 Dubickas assumed theexistence of up to 4 vertices of degree above 3. We prove that anyconnected geometric graph with maximum degree 4 and vertex w ofdeg(w)<4 has a rational e-approximation for any e>0.

2 - New spectral upper bounds on the size of k-regularinduced subgraphsDomingos Cardoso, Departamento de Matematica,Universidade de Aveiro, Campus Universitario de Santiago,3810-193, Aveiro, Portugal, [email protected], Sofia Pinheiro

108


In (Haemers’ Ph.D. thesis, 1980) an upper bound (based on the leastand largest adjacency eigenvalues) on the independence number of ar-bitrary graphs was introduced. Later, in (Lu et al, 2007), upper bounds(using the largest Laplacian eigenvalue) were deduced. For regulargraphs, in (Bussemaker et al, 1976) a spectral upper bound on the or-der of induced subgraphs with average degree d was obtained. Thisupper bound is herein extended to arbitrary graphs, introducing threenew upper bounds, based on adjacency, Laplacian and signless Lapla-cian eigenvalues, respectively.

3 - Graph spectra and combinatorial optimizationNair Abreu, Production Engineering, Federal University ofRio de Janeiro, Rua Joao Lira 106/401, Leblon, 22430 210,Rio de Janeiro, Rio, Brazil, [email protected],Claudia Justel, Domingos Cardoso

Wilf(1967), followed by Hoffman(1970), presented the first resultson graph spectra in combinatorial optimization by introducing grapheigenvalue bounds to the chromatic number. Later several articles ap-peared relating eigenvalues to classical invariants of graphs: stabilityand clique numbers, matching, connectivity and isoperimetric num-ber. Furthermore, some hard combinatorial optimization problemshave been approximated by using spectral techniques. We present asurvey of significant results about graphs spectra applied to combina-torial optimization.

� TB-06Tuesday, 10:30-12h00RB-Gamma

Performance evaluation and newmetaheuristicsStream: MetaheuristicsInvited sessionChair: Cristina C. Vieira, Department of Electronics andInformatics Engineering, University of Algarve, Faculty ofScience and Technologies, Campus de Gambelas, Faro,Localidade, Portugal, [email protected]

1 - Using a multicriteria function to evaluate metaheuris-tics performancePaulo Oswaldo Boaventura-Netto, Production Engineering,Federal University of Rio de Janeiro/COPPE, Rua PinheiroMachado 99, ap. 1508, Laranjeiras, 22231-090, Rio deJaneiro, RJ, Brazil, [email protected], Valdir Melo

We propose a multicriteria evaluation scheme based on Condorcetranking technique. It uses weights associated to the algorithms from aset being object of comparison. We used five criteria and a function onthe set of natural numbers to create a ranking. The comparison involvesthree well-known problems of combinatorial optimization, taking in-stances from public libraries. Each algorithm was used with essentiallythe same structure, the same local search was used and the initial so-lutions were similarly built. The metaheuristics GRASP, Tabu Search,ILS and VNS, participate from the comparison.

2 - Performance Assessment of Electromagnetism-likeAlgorithmAlkin Yurtkuran, Industrial Engineering Department, UludagUniversity, Uludag University, Industrial EngineeringDepartment, 16059, Bursa, Turkey, [email protected],Erdal Emel

Meta-heuristic algorithms are robust and efficient techniques for solv-ing complex real-world problems. Electromagnetism-like Algorithm(EMA) is one of the recently introduced population based meta-heuristic which simulate the behavior of charged particles on an elec-trical field. This work provides a comparison between EMA and pop-ular foraging and evolutionary algorithms using well-known CEC05benchmark functions based on the solution value and solution time.The results show that the performance of EMA is comparable to novelmeta-heuristic algorithms in terms of solution quality and time.

3 - From constructive search to local searchCristina C. Vieira, Department of Electronics and InformaticsEngineering, University of Algarve, Faculty of Science andTechnologies, Campus de Gambelas, Faro, Localidade,Portugal, [email protected], Carlos M. Fonseca

Constructive search may be reinterpreted as local search by consider-ing an appropriate solution space, defining a neighbourhood structurebased on the original construction rules and specifying how to evaluateincomplete solutions. The resulting problem formulation is compati-ble with both local and constructive optimisation methods, and allowsvery different metaheuristics to be studied on exactly the same (local)formulation of a given (global) optimisation problem. The proposedapproach is illustrated in the context of ACO and evolutionary algo-rithms on the TSP.

4 - Algorithm of inspired by virus and bacterium attack:A meta-heuristic approach as an optimization toolMahdi Bashiri, Shahed University, Iran, Islamic Republic Of,[email protected], Masoud Bagheri, ZeinabRasoolinezhad

In this paper, a new meta-heuristic algorithm has been proposed basedon virus and bacterium attack (VBA). Actually presented solution ap-proach is inspired by attack of virus and bacterium to human body.In the proposed algorithm, virus and bacteria attack to reach the bodycells death. Some numerical examples have been selected from the lit-erature to consider the proposed VBA’s efficiency. Also the algorithmhas been compared by the genetic algorithm (GA). The computationalresults verify its efficiency from computational time and solution qual-ity aspects as well.

� TB-07Tuesday, 10:30-12h00RB-Eta

Vehicle Routing and LogisticsOptimization

Stream: Vehicle Routing and Logistics OptimizationInvited sessionChair: Erdener Özçetin, Industrial Engineering, AnadoluUniversity, Iki Eylül Kamp., MMF END 107, 26000, Eskisehir,Turkey, [email protected]


We propose a new parallel search procedure for dynamic multi-objective traveling salesman problem. We design a multi-objectiveTSP in a stochastic dynamic environment. The proposed procedurefirst uses parallel processors to identify the extreme solutions of thesearch space for each of k objectives individually at the same time.These solutions are merged into a matrix E. The solutions in E are thensearched by parallel processors and evaluated for dominance relation-ship. The proposed procedure was implemented in two different ways:a master-worker architecture and a pipeline architecture.

2 - Solving the vehicle routing problem with time win-dows by an interior point branch-price-and-cutframeworkPedro Munari, ICMC, University of Sao Paulo, Sao Carlos,Sao Paulo, Brazil, [email protected], Jacek Gondzio

In this talk, we present a branch-price-and-cut framework for the ve-hicle routing problem with time windows which is based on a primal-dual interior point method for efficiently solving the linear relaxationsat each branch node. The interior point method is used to obtain well-centred, non-optimal solutions that are used to improve the genera-tion of columns and valid inequalities. Computational results usingthe Solomon’s instances show that the proposed approach has a bet-ter overall performance than the best branch-price-and-cut frameworkavailable in the literature.

109


3 - A Parallel Algorithm for Vehicle Routing Problem onGPUsErdener Özçetin, Industrial Engineering, Anadolu University,Iki Eylül Kamp., MMF END 107, 26000, Eskisehir, Turkey,[email protected], Gurkan Ozturk

Vehicle Routing Problem (VRP) is a common and difficult logisticsproblem. To reach a solution with exact methods is mostly hard, sev-eral meta-heuristics have been developed. Graphics processing units(GPUs) on enhanced graphics cards are used to solve hard optimiza-tion problems with implementing meta-heuristic algorithms by paral-lelizing. In this paper, a parallel algorithm has been proposed for VRP.The performance of the algorithm, especially in execution time hasbeen shown by solving literature test problems.

4 - Solving Vehicle Routing Problems with EquipmentAllocationPhilip Kilby, NICTA and The Australian National University,Locked Bag 8001, 2601, Canberra, ACT, Australia,[email protected], Fabien Tricoire

We look at a routing problem where some tasks require particularequipment (and may require more than one type of equipment). Theequipment allocated to a vehicle is a decision variable, and restricts thevehicles that can perform each task.

We look at two methods for solving the combined equipment alloca-tion and vehicle routing problem. One method uses a standard vehiclerouting solver, modified in a modular way to incorporate the additionalfunctionality. We also describe a method based on column generation.

� TB-08Tuesday, 10:30-12h00RB-Epsilon

VMI and Delivery

Stream: Supply Chain PlanningInvited sessionChair: Marcel Turkensteen, Business Studies, ASB, FuglesangsAlle 4, 8210, Aarhus V, Denmark, [email protected]

1 - Shipment consolidation by collaborative planningbetween industrial suppliers under VMI policyWaldemar Kaczmarczyk, Department of Operations Research& Information Technology, AGH University of Science &Technology, al. Mickiewicza 30, 30-059, Krakow, Poland,[email protected]

In industrial supply chains just-in-time policy forces frequent andsmall shipments. Transportation costs can be reduced by consolidationof shipments, but this is possible only if shipments can wait until theconsolidated shipment is accumulated. The Vendor Managed Inven-tory system opens another possibility, suppliers may coordinate theirshipments in advance. Proposed concept of horizontal collaborativeplanning enables more frequent consolidation without delivery delays.Presented results show potential benefits of such collaboration.

2 - An Improved Robust Particle Swarm Optimization Al-gorithm to the Uncertain Inventory Routing ProblemIsa Nakhai Kamalabadi, Department of IndustrialEngineering, School of Engineering, Tarbiat ModaresUniversity, Jalal Ale Ahmad Highway, P.O.Box: 14115-143,14115-143, Tehran, Tehran, Iran, Islamic Republic Of,[email protected], Ali Hossein Mirzaei

We address a multi-period multi-product uncertain inventory routingproblem where products are produced and delivered from one manu-facturer to a set of retailers through a fleet of heterogeneous capaci-tated vehicles. Here, split delivery is permissible but each vehicle cantravel just a tour per period. The uncertainty follows an unknown butbounded pattern. Due to the complexity of the problem, we developan improved robust particle swarm optimization algorithm to solve theproblem. The computational results show that the developed algorithmoutperforms the benchmark algorithms.

3 - A vendor managed inventory model with transporta-tion to geographically dispersed retailersMarcel Turkensteen, Business Studies, ASB, Fuglesangs Alle4, 8210, Aarhus V, Denmark, [email protected], Christian LarsenThis study considers a vendor that supplies a set of geographically dis-persed retailers and wishes to use a coordinated inventory and distri-bution policy that minimizes the total costs incurred. Routing consid-erations, an infinite time horizon, and demand uncertainty make theproblem hard to solve for existing approaches. Our solution is to es-timate transportation distances with continuous approximation. UsingMarkov chain modeling, we then derive the delivery zones and vehicleutilization policies in combination with optimal inventory policies atthe vendor and the retailers.

� TB-09Tuesday, 10:30-12h00RB-Zeta


Stream: Cutting and PackingInvited sessionChair: Andreas Bortfeldt, Dept. of Information Systems,University of Hagen, Profilstrasse 8, 58084 Hagen, Germany,58084, Hagen, FR Germany, Germany,[email protected]

1 - Optimal cutting of the raw materialsJaroslav Hanzel, Institute of Control and IndustrialInformatics, Slovak University of Technology, Ilkovicova 3,812 19, Bratislava, Slovakia, [email protected],Ladislav Jurisica, Anton Vitko, Marian Klucik, Peter PasztoThe contribution deals with optimal utilization of the materials in theindustry. The problem is to find and cut out the rectangular part fromthe metal plate with no defects and a maximal surface. Simple al-gorithm applicable in the embedded systems with restricted computa-tional resources is proposed. It is able to quickly find solutions closeto optimal solution with low computational cost. On the other hand theproblem was solved also by means of standard optimization method.The analysis and comparison of proposed and standard methodologyon the basis of computed data is given.

2 - Different criteria to sort the list of pieces in 2D-Orthogonal Strip Packing ProblemFernando Garcia_Perez, Matematicas Aplicadas, Universidadde Alcala de Henares, Edificio de Ciencias. CampusUniversitario, Carretera Madrid-Barcelona Km. 33,600,28805, Alcala de Henares, Madrid, Spain,[email protected], Joaquin Aranda Almansa, MiguelDelgado PinedaIn the two-dimensional strip packing problem, 2D-SPP, a set of n rect-angular items and a strip of width W and unlimited height are given.Each item has an height and a width. This problem consists of allo-cating orthogonally all the items, without overlapping, by minimizingthe overall height of the packing. It is usual to sort the small pieces inorder to improve the result. We study the influence on the outcome ofdifferent criteria to sort the list of pieces and how to select the optimalcriterion using around 1,000 instances from the literature.

3 - Generation of Two-dimensional Guillotine CuttingPatterns by Genetic Algorithms and Fit-HeuristicsLilian Caroline Xavier Candido, Federal University of Parana,Avenida Candido de Abreu, numero 304, apartamento 905,80530000, Curitiba, Parana, Brazil,[email protected], Thiago GuimaraesThis paper proposes a new strategy for generation of two-dimensionalguillotine cutting patterns, with non-stages and two stages with thepossibility of rotation of the items in both cases. The methodology isdivided into two phases: first, the items are selected and grouped intosubsets by genetic algorithms. In the second phase, heuristics are em-ployed to determine fitting geometric arrangement of the cutting pat-tern. Computational experiments performed with classical instancesfrom the literature indicate excellent results for some instances withvery low processing time for all of them.

110


4 - Algorithms Based on Particle Swarm Optimizationfor the Two-dimensional Knapsack ProblemDavid Álvarez, São Paulo State University, Brazil,[email protected], Ruben Romero

This study presents the (un)constrained (un)weighted k-staged fixedand rotated two-dimensional guillotineable single knapsack problem.An encoding based on slicing tree is presented. Four algorithms,based on the Particle Swarm Optimization (PSO) metaheuristic wereused: basic PSO, PSO with the mutation operator, a hybrid of VariableNeighborhood Search (VNS) with PSO and a hybrid of VNS and PSOincluding the mutation operator as a turbulence factor. The computa-tional results on large sets of test cases show that the methodology isvery efficient for the two-dimensional knapsack problems.

� TB-10Tuesday, 10:30-12h00RB-Theta


Stream: Production Management & Supply ChainManagement (contributed)Contributed sessionChair: Hicham Chehade, ICD- LOSI, University of Technology ofTroyes, 12 Rue Marie Curie, 10000, Troyes, France,[email protected]: Harry Groenevelt, Simon School of Business, University ofRochester, Wilson Blvd, 14627, Rochester, NY, United States,[email protected]

1 - An Inventory Model to Determine The Optimal Mix ofOwned and Rented ItemsLeonardo Epstein, School of Engineering, UniversidadAdolfo Ibáñez, Av. Diagonal Las Torres 2640, (of 209 B),Peñalolén, Santiago, Chile, [email protected], EduardoGonzález-Császár

Inventory models for rental items may be used to plan operations thatinvolve lending or renting tools, or providing service such as access totelephone lines. We address the situation where the service providerowns an inventory of items, which may not be sufficient to meet de-mand. Additional items may be obtained by renting them from anotherprovider. Thus, emerges the problem of determining the optimal num-ber of items the provider must own to maximize his expected profitwithin a time horizon. We consider the situation where the number ofusers is finite and future demand is uncertain.

2 - Analysis of Supply Contracts with Total MinimumQuantity Commitment and Fixed Order CostsFrank Y Chen, Department of Management Science, CityUniversity of Hong Kong, Hong Kong, China,[email protected]

We study a supply contract that specifies a minimum total (cumula-tive) order quantity over the planning horizon (e.g., one year). Eachreplenishment order incurs both fixed and variable costs. We show theoptimal policy to be of a modified (s, S) form.

3 - Heuristics for the (Q,R,S) inventory policyHarry Groenevelt, Simon School of Business, University ofRochester, Wilson Blvd, 14627, Rochester, NY, United States,[email protected]

We will compare the performance of a number of heuristics for the (Q,R, S) inventory policy. This is a general policy that includes periodicand continuous review base stock and order point-order quantity poli-cies as special cases. The heuristics trade off computational effort andaccuracy, and provide estimates for a number of different performancemetrics, such as inventory holding and backorder costs, unit fill rate,order fill rate, in-stock probability, etc..

� TB-11Tuesday, 10:30-12h00RB-Iota

Transportation and Logistics

Stream: Transportation and LogisticsInvited sessionChair: Sadegh Niroomand, Industrial Engineering, EastermMediterranean University, Turkey,[email protected]

1 - Developing a sustainable model for city transporta-tion networksNarges Shahraki, industrial engineering, koc university, kocuniversity, istanbul, Turkey, [email protected], MetinTurkay, Ali Fattahi

In this paper, we consider the design and upgrade of city transportationnetworks using robust and exact optimization methods. The transporta-tion network is represented with a directed graph G(N, A), where N isthe set of nodes and A is the set of edges connecting these nodes. Inthe design and upgrade of the network, it is important to assign de-mand for different sectors (such as housing, industry, commerce andservices) to certain locations in the metropolitan area while achievinga sustainable transportation system. We incorporate economical, socialand environmental aspects of triple bottom line of sustainability.

2 - Inventory and Transport Planning of Raw Materialsfor Steel EnterprisesHua-An Lu, Department of Shipping and TransportationManagement, National Taiwan Ocean University, #2,Pei-Ning Road, 202, Keelung, Taiwan,[email protected]

Steel enterprises required different kinds of raw materials for smeltingtheir products. For reducing transport costs, maritime carriers are used.The whole inventory and transport planning of raw material importsfor a steel company is a complicate problem. This study proposes twomodels: one for inventory scheduling of raw materials and voyage al-location of different ship types; one for ship assignment in consideringof deployment by own carriers or spot charter vessels. This planningprocedure was applied to a real case study for China Steel Corporationin Taiwan.

3 - Experimenting a Metaheuristics for a Flexible Trans-port ServicePasquale Carotenuto, Istituto per le Applicazioni del Calcolo"M. Picone", Consiglio Nazionale delle Ricerche, via deiTaurini 19, 00185, Roma, RM, Italy, [email protected],Daniele Monacelli, Marco Turco

A demand responsive transport system (DRTS) is a flexible system inwhich the stops are fixed and tours are variable. Client takes a reserva-tion about a trip which could be in the same day or in the next. DRTScan be analyzed by two kind of dynamisms. The first is about the wayin which requests arrives to the system. In this case DRTS could bestatic if the algorithm runs after received all the requests, or can be dy-namic on line if the algorithm runs while the requests are arriving tothe system. The second kind of dynamism is about when the requestsare served: on reservation or on service.

4 - Optimal Supermarket LayoutSadegh Niroomand, Industrial Engineering, EastermMediterranean University, Turkey,[email protected], Bela Vizvari

The optimal layout of a supermarket is determined. It is well-knownthat the amount of sales is an increasing function of the length of thecustomer’s path. To increase the sales, the objective is to increasethe travelled path in the shop. Based on real life observations, cus-tomers are classified into 16 classes. It is supposed that each customeruses the optimal solution of the relevant TSP when visits departments.The mathematical model maximizes the total weighted length of the16 TSPs’ optimal solutions. The model uses the dual of DFJ model.Numerical results are presented.

111


� TB-12Tuesday, 10:30-12h00RB-Omicron

Robustness in Public Transportation

Stream: Optimization in Public TransportInvited sessionChair: Marie Schmidt, Institut für Numerische und AngewandteMathematik, Georg-August-Universität Göttingen, Lotzestraße16-18, 37073, Göttingen, Germany,[email protected]

1 - Robust Lockmaster’s ProblemsChristina Büsing, Operations Research, RWTH AachenUniversity, Kackertstraße 7, 52072, Aachen, Berlin, Germany,[email protected], Elisabeth GüntherAn efficient solution of the lockmaster’s problem (LockP) reduces thetransportation time of goods along waterways and thus makes thismeans of transport a competitive and environmental friendly alterna-tive to trucks or railways. The LockP is to manage the lockage timessuch that ships arriving from different levels of a waterway get up ordown lifted minimizing the total waiting time. Due to delays a pre-defined schedule may turn infeasible. To deal with these changes, weintroduce robust models, obtain compact ILP-formulations and com-pare their performance on real data of the Kiel Canal.

2 - Robustness in Periodic and Aperiodic TimetablingMarc Goerigk, Georg-August Universitaet Goettingen, 37083,Goettingen, Germany, [email protected],Anita SchöbelIn public transport timetabling, arrival and departure times need to bescheduled for a set of vehicles, minimising travel times. This can bedone with or without periodically reoccurring events, resulting in theperiodic and aperiodic timetabling problem.

Finding timetables that also consider robustness against delays is a re-cently emerging field of research, with a wide variety of concepts howto define robustness.

In this talk an experimental evaluation of robust timetables on close-to real-world data is given. Both the aperiodic as well as the harderperiodic case are considered.

3 - On the improvement of robustness in railway stationareasThijs Dewilde, Centre for Industrial Management/Traffic &Infrastructure, University of Leuven, Celestijnenlaan 300A,bus 2422, B-3001, Leuven, Belgium,[email protected], Peter Sels, Dirk Cattrysse,Pieter VansteenwegenTo improve the robustness of a railway system in station areas, weintroduce a repetitive procedure to successively optimize the trainrouting problem and to reinforce this solution by applying timetablechanges in a tabu search environment. We introduce the spread of thetrains in the objective function for the route choice and timetablingmodules. The performance of our algorithms is evaluated based on thecase study for Brussels’ area, using a discrete event simulation. Thecomputational results indicate an improvement in robustness and a re-duction in knock-on delay.

4 - Robust bicriteria pathsMarie Schmidt, Institut für Numerische und AngewandteMathematik, Georg-August-Universität Göttingen,Lotzestraße 16-18, 37073, Göttingen, Germany,[email protected], Kenneth Kuhn, AndreaRaith, Anita SchöbelFinding robust solutions of optimization problems is important in prac-tice and well researched for single-objective problems. However, inmany applications one has to deal with several objective functions.Motivated by the application of routing airplanes through convectiveweather, where two conflicting objective functions are given by thelength and the risk of the chosen route, we extend robustness conceptsto the bicriteria case and develop new robustness concepts. The con-cepts are analyzed for different uncertainty sets and relations betweenthe different concepts are identified.

� TB-13Tuesday, 10:30-12h00RB-Tau

Geometric Clustering

Stream: Geometric ClusteringInvited sessionChair: Andreas Brieden, Universität der Bundeswehr München,Werner-Heisenberg-Weg 39, 85579, Neubiberg, Germany,[email protected]: Peter Gritzmann, Mathematics, TU München, Arcisstr. 21,D-80290, Munich, Germany, [email protected]

1 - On the Diameter of Partition PolytopesSteffen Borgwardt, Fakultät für Mathematik, TechnischeUniversität München, Boltzmannstr. 3, 85748, Garching,Bayern, Germany, [email protected]

We study the combinatorial diameter of so-called ’partition polytopes’,which are associated to partitioning a set of n items into clustersC_1,...,C_k of prescribed sizes t_1,...,t_k. We derive upper bounds ontheir diameter in the form of t_1+t_2, n-t_1 and n/2, which is a general-ization of the well-known diameter-2 result for the Birkhoff polytope.The constructive, graph-theoretical proof yields an effective algorithmto transfer such a partition into another one by applying a number of’cyclical exchanges’ in between the clusters that adheres to our prov-ably low diameter bounds.

2 - A New Algorithmic Approach in Mitigating SupplyChain RiskMichael Öllinger, Universität der Bundeswehr München,Germany, [email protected], Andreas Brieden,Peter Gritzmann

In our world of distributed production supply chain safety is an impor-tant issue. A new algorithmic model for supply chain safety is intro-duced based on geometric clustering. Each cluster represents a differ-ent supply chain component in the supply network, responsible for apart of the overall production. Aiming for a minimum probability ofa failure, the clustering constructs optimal supply chain components.The basic model is given in detail, complemented with simulations thatindicate the excellent behavior in terms of mitigating supply chain riskand constructing a robust supply chain.

3 - Forecasting risk by means of geometric clusteringAndreas Brieden, Universität der Bundeswehr München,Werner-Heisenberg-Weg 39, 85579, Neubiberg, Germany,[email protected], Peter Gritzmann

Applying the law of large numbers is a convenient way to predict riskfor a large and at the same time homogenous data set. Unfortunatelyin practice you have either just small but homogenous or many, but atthe same time inhomogenous, datasets. In the talk a new method ispresented that can deal with this problem.

4 - On optimal weighted balanced clusterings and powerdiagramsPeter Gritzmann, Mathematics, TU München, Arcisstr. 21,D-80290, Munich, Germany, [email protected], AndreasBrieden

We study weighted clustering problems under balancing constraintswith a view towards separation properties. We introduce gravity poly-topes and more general gravity bodies that encode all feasible cluster-ings. Then we show that their extreme points correspond to stronglyfeasible power diagrams, certain specific cell complexes, whose defin-ing polyhedra contain the clusters, respectively. Further, we character-ize strongly feasible centroidal power diagrams in terms of the localoptima of some ellipsoidal function over the gravity polytope.

112


� TB-14Tuesday, 10:30-12h00RB-Omega

New models in batch scheduling

Stream: Scheduling under Resource ConstraintsInvited sessionChair: Mikhail Y. Kovalyov, United Institute of InformaticsProblems, National Academy of Sciences of Belarus, Surganova 6,220012, Minsk, Belarus, [email protected]: Ammar Oulamara, Lorraine University, LORIA UMR7503, Parc de Saurupt, 54000, Nancy, France, [email protected]

1 - Scheduling a hybrid flowshop with a serial batch ma-chine: application to chemotherapy productionAmeur Soukhal, University of Tours, Computer ScienceLaboratory, 64, avenue Jean Portalis, 37200, Tours, France,[email protected], Jean-Charles Billaut, YannickKergosien, Christophe Lenté, Jean-François TournamilleThe Unity of Biopharmacy and Clinical Oncology of CHRU of Toursproduces cancer drugs and deliver them to nursing care services. Be-cause these services are dispersed within a vast hospital complex, aperson is dedicated to deliver the drugs. After being produced, thepreparations are controlled by an automaton that is in charge to checkthe correct dosage of the various compounds of each preparation. Thecontroller can be considered a serial batching machine. The objectiveof this study is to propose production methods to minimize the delays.

2 - Scheduling an unbounded batching machine withjob processing time compatibilitiesAdam Janiak, Institute of Computer Engineering, Control andRobotics, Wrocław University of Technology, Janiszewskiego11/17, 50372, Wroclaw, Poland, [email protected],Adrien Bellanger, Mikhail Y. Kovalyov, Ammar OulamaraThe problem of scheduling jobs on an unbounded batching machine isconsidered. The machine can process any number of jobs simultane-ously in a batch, subject to an additional constraint that, in the samebatch, the job processing times are compatible. There are processingtime intervals for the jobs. The job processing times are compatible ifthe corresponding processing time intervals intersect. The batch pro-cessing time is equal to the left endpoint of the intersection of the jobprocessing time intervals in this batch. The objective is to minimize aregular function of job completion times.

3 - An optimal algorithm to schedule jobs on two identi-cal parallel machines with an operatorDjamal Rebaine, Informatique et mathématique, Université duQuébec à Chicoutimi, 555, boul. de l’Université, G7H 2B1,Chicoutimi, Québec, Canada, [email protected], MohammedZouba, Pierre BaptisteWe consider the scheduling problem on two identical machines withan operator to minimize the makespan. The operator supervision isdone through a subset of working modes to divide up his interventionsover the machines. The processing times then become variable as theydepend on the working mode being utilized. Therefore, the process ofbuilding a schedule includes the search of the subset of working modesto utilize along with the order and the duration of their utilization. Apseudo-polynomial time algorithm is exhibited to produce an optimalsolution within a free changing mode.

� TB-15Tuesday, 10:30-12h00RB-2101

Algorithms for Semi-Infinite Optimization

Stream: Semi-Infinite OptimizationInvited sessionChair: Margarita Rodríguez Álvarez, Dpto. Estadística eInvestigación Operativa, Universidad de Alicante, Ctra.Alicante-San Vicente s/n, 03690, San Vicente del Raspeig,Alicante, Spain, [email protected]

1 - Numerical methods for solving semi-infinite pro-grams with infinitely many conic constraintsTakayuki Okuno, Graduate School of Informatics, KyotoUniversity, Yoshida-Honmachi, Sakyo-Ku, 606-8501, Kyoto,Japan, [email protected], Shunsuke Hayashi,Masao Fukushima

The semi-infinite program (SIP) is normally represented with infinitelymany inequality constraints, and has been studied extensively so far.However, there have been few studies on the SIP involving conic con-straints, even though it has important applications such as Chebyshev-like approximation, filter design, and so on. In this talking, we focuson the SIP with an infinite number of conic constraints. We providesome applications and algorithms for solving it.

2 - Zimmermann-cutting plane algorithm for solvingnon-symmetric fuzzy semi-infinite linear program-ming problemsSomayeh Khosravi, Mathematics, Shiraz University ofTecnology, Shiraz, Iran, Islamic Republic Of,[email protected], Alireza Fakharzadeh

This paper introduces a solution method for fuzzy semi-infinite linearprogramming problems, in which contains the crisp objective functionand the infinite number of fuzzy constraints. The solution method isthe kind of cutting plane algorithm in which its sub-problem are solvedby using the Zimmermann-method. The convergence of the algorithmis proved and some numerical examples are given.

3 - The alpha-BB based cutting plane method for semi-infinite program with multi-dimensional index setShunsuke Hayashi, Graduate School of Informatics, KyotoUniversity, Yoshida-Honmachi, Sakyo-Ku, 606-8501, Kyoto,Japan, [email protected], Kensuke Goumoto

Recently, Shiu and Wu proposed the algorithm for semi-infinite pro-gram (SIP) combining the alpha-BB method with the cutting planemethod. In this study, we extend their algorithm to the case where theindex set in the SIP is multi-dimensional. When the index set is one-dimensional, the "refinement step" in the algorithm is just to divide aclosed interval into the left and the right. However, such a division isnot unique in the multi-dimensional case. We also apply the algorithmto the three-dimensional design-centering problem, which can be castas an SIP with a two-dimensional index set.

4 - Weighted Voronoi cells via quadratic systemsMargarita Rodríguez Álvarez, Dpto. Estadística eInvestigación Operativa, Universidad de Alicante, Ctra.Alicante-San Vicente s/n, 03690, San Vicente del Raspeig,Alicante, Spain, [email protected], Miguel Goberna,Virginia N. Vera de Serio

Given a set T such that each point s has assigned a positive weightw(s), the multiplicatively weighted (MW) Voronoi cell of s is the set ofpoints closer to s than to any other point of T for the distance d(.,t)/w(t).The MW-Voronoi diagram of T is the family of MW-Voronoi cells ofall the elements of T. Recently, we applied the theory of linear semi-infinite systems to get information on the ordinary Voronoi cells anddiagrams (with the same weight for all the points of T). In this talk, weapply results on quadratic systems to MW-Voronoi cells and diagrams.

� TB-16Tuesday, 10:30-12h00RB-2103

Theory and Applications of ConstrainedEquilibria

Stream: Uncertainty and Perturbations in Optimizationand in Environmental and Energy ModellingInvited sessionChair: Jacek Krawczyk, School of Economics and Finance,Victoria University of Wellington, 6140, Wellington, NewZealand, [email protected]

113


1 - Global emission ceiling versus international cap andtrade: a comparison based on normalized Nash equi-libriaJacqueline Morgan, Department of Mathematics and Statistics& CSEF, University of Naples Federico II, Via Cinthia,80126, Napoli, Italy, [email protected], Fabien Prieur

We model climate negotiations as a two-stage game. In the first stage,players agree on a global emission cap (GEC). In the second one, theynon-cooperatively choose either their emission level or their emissionquota, under the cap that potentially binds them together. When theemission cap is binding, among the coupled constraints Nash equilib-ria, we select a normalized equilibrium. With and without emissiontrading, there exists a range of values for which a binding cap im-proves all players’ payoff. Moreover GEC outperforms the ITC systemin terms of emission reduction and payoff gains.

2 - Robust Nash Equilibrium in a Class of Multi-Leader-Follower GamesMasao Fukushima, Kyoto University, Japan,[email protected]

In this paper, we focus on a class of multi-leader-follower games un-der uncertainty with some special structure. By means of the robustoptimization technique, we first formulate the game as the robust Nashequilibrium problem, and then the generalized variational inequality(GVI) problem. We then establish some results on the existence anduniqueness of a robust leader-follower Nash equilibrium. We also ap-ply the forward-backward splitting method to solve the GVI formula-tion of the problem and present some numerical examples. [This is ajoint work with Ming Hu.]

3 - Regulation of pollution in a Cournot equilibriumLars Mathiesen, Economics, Norwegian School ofEconomics, Helleveien 30, 5045, Bergen, Norway,[email protected]

This paper is concerned with modeling polluters that form an oligopolyin their product market. Adding emission-constraints converts theCournot-Nash equilibrium into a generalized Nash equilibrium prob-lem (GNEP). When coupling constraints are common and correspond-ing shadow prices are equal to all players the resulting equilibrium is acomplementarity problem (CP). Conditions for convergence of a New-ton process to a unique equilibrium of the CP are shown to be similar tothose of the Cournot-Nash model. Computational results demonstrateits efficiency compared to other approaches.

4 - Economics of collective monitoring: a study of envi-ronmentally constrained electricity generatorsJacek Krawczyk, School of Economics and Finance, VictoriaUniversity of Wellington, 6140, Wellington, New Zealand,[email protected], Javier Contreras, James Zuccollo

We tackle welfare differences between individual and collective moni-toring of pollution. Given a model of emissions propagation the regu-lator can calculate the optimal emissions quota for each generator andenforce it through individual monitoring. That will maximise welfare,but may be expensive because of implementation costs. An alternativeis provied. We use a 3-node network, find that the welfare loss fromcollective monitoring can be small and that the imposition of trans-mission and environmental restrictions may benefit the “inefficient”generator and decrease consumer surplus.

� TB-17Tuesday, 10:30-12h00RB-2105

Hub Location IStream: Hub LocationInvited sessionChair: Ivan Contreras, Concordia University, H3G 1M8, Montreal,Canada, [email protected]

1 - P-HUB approach for the optimal park-and-ride facilitylocation problemVladimir Marianov, Electrical Engineering, PontificiaUniversidad Catolica de Chile, Vicuna Mackenna 4860,7820436 Macul, 7820436, Santiago, Chile,[email protected], Felipe Aros-Vera

We optimally locate Park and Ride (P&R) facilities, modeled as ca-pacitated hubs. P&R usage is maximized. Commuters can either usethe service or choose to travel by car only, and their behavior followsa gravitational model. We consider users’ generalized cost and attrac-tiveness of P&R facilities as input for the gravity rule. For small in-stances of the problem, we propose a novel linearization of the gravityrule. A modification of the Heuristic Concentration Integer procedureis applied to solve larger instances. Numerical results are shown, in-cluding a case in Queens, NY.

2 - Hub location under uncertaintyFrancisco Saldanha-da-Gama, CIO/DEIO, University ofLisbon, FCUL-DEIO, Bloco C6, Piso 4, 1749-016, Lisbon,Portugal, [email protected], Sibel A. Alumur, Stefan Nickel

In this work, we address several aspects concerning hub location prob-lems under uncertainty. Two sources of uncertainty are considered: theset-up costs for the hubs and the demands to be transported betweenthe nodes. Generic models are presented for single and multiple allo-cation versions of the problems. Using a set of computational tests weanalyze the changes in the solutions driven by the different sources ofuncertainty considered isolated and combined.

3 - Hub and Spoke Network Design with Single-assignment, Capacity Decisions and Balancing Re-quirementsStefan Nickel, Institute for Operations Research, KarlsruheInstitute of Technology (KIT), Geb 11.40, 76128, Karlsruhe,Baden-Wuertemberg, Germany, [email protected], IsabelCorreia, Francisco Saldanha-da-Gama

An extension of the capacitated single-allocation hub location problemis considered in which not only the capacity of the hubs is part of thedecision making process but also balancing requirements are imposedon the network. The decisions involve: the selection of the hubs, theallocation of the spoke nodes to the hubs, the flow distribution throughthe sub network defined by the hubs, and the capacity level at whicheach hub should operate. A set of computational experiments is per-formed and the results are reported.

4 - Supermodular Properties in Hub LocationIvan Contreras, Concordia University, H3G 1M8, Montreal,Canada, [email protected], Elena Fernandez

In this work we show how a generalized hub location problem, thatincludes as particular cases well-known hub node and hub arc loca-tion problems, can be stated as the maximization of a supermodularfunction. As a consequence, we obtain worst-case bounds on the per-formance of a greedy heuristic. Moreover, we present two integer pro-gramming formulations for the generalized hub location problem. Oneof these formulations is derived from the properties of supermodularfunctions, and involves variables with at most two indices. Computa-tional experiments confirm the efficiency of such formulation.

� TB-18Tuesday, 10:30-12h00RB-2107

Location Analysis: AdvancedOptimization Tools

Stream: Continuous LocationInvited sessionChair: Víctor Blanco, Quant. Methods for Economics & Bussines,Universidad de Granada, Facultad de Ciencias Economicas yEmpresariales, Campus Cartuja, 18011, Granada, Spain,[email protected]

114


1 - Expanding Search for a Hider on a NetworkSteve Alpern, Mathematics Department, London School ofEconomics, Houghton Street, WC2A 2AE, London, UnitedKingdom, [email protected], Thomas LidbetterA Hider H is hidden somewhere on a network Q. A Searcher choosescontinuously expanding sets, of total length. Starting at a given pointA, he captures the Hider at the first time T with H in set. This newsearch paradigm models searching by a team of agents. We solve theBayesian problem on a tree (where the distribution of H is known) andthe game where the Hider chooses H and T is the payoff, for certaintypes of networks. This extends the article S. Alpern, ’Find-and-Fetchon a Tree’, Operations Research 59 (2011).

2 - Solving the franchisor-franchisee bi-objective opti-mization problemPilar M. Ortigosa, Computer Architecture and Electronics,University of Almería, Ctra. Sacramento s/n, La Cañada deSan Urbano, 04120, Almería, Spain, [email protected], JuanaLópez-Redondo, Jose Fernandez, Aranzazu Gila Arrondo,Carmen G. Escamilla, Victoria Plaza LeivaThough the problem of obtaining the Pareto-front for this problem hasbeen previously tackled through exact general methods, they requirehigh computational effort. In this work, we propose a new multi-objective optimization heuristic algorithm, which deals with the prob-lem at hand in a fast and efficient way. It combines ideas from dif-ferent multi-objective and single-objective optimization evolutionaryalgorithms, although it also incorporates new devices which help to re-duce the computational requirements, and also to improve the qualityof the provided solutions.

3 - Optimal Timings of Project Control Points using Fa-cility Location Model and SimulationNarjes Sabeghi, Applied Mathematics, Ferdowsi University ofMashhad, Faculty of Mathematical Sciences,No10,ladan10,vakilabad BLV,Mashhad,Iran, Mashhad, Iran,Islamic Republic Of, [email protected], HamedReza Tareghian, Hassan Taheri, Mohammad AminiApproaches with static view to the project control suffer from the factthat the possibility of corrective actions is missing. In this paper, weconsider a dynamic view to the project control and use an adapted ver-sion of the facility location model to find optimal timings of the controlpoints. A simulation model is then used to predict the possible devia-tions from the base plan at each control point. Borrowing ideas fromdisturbance management, corrective actions are taken. This processcontinues until the timing of the last point of control is obtained.

4 - A hierarchy of SDP-relaxations for multifacility or-dered median problemsVíctor Blanco, Quant. Methods for Economics & Bussines,Universidad de Granada, Facultad de Ciencias Economicas yEmpresariales, Campus Cartuja, 18011, Granada, Spain,[email protected], Justo Puerto, Safae EL Haj Ben AliWe consider multifacility ordered median continuous location prob-lems with any norm, and in any dimension. We show that there is asequence of semi definite programming relaxations that converges tothe optimal solution of the problem by using a Lasserre’s moment ap-proach. Furthermore, in order to solve larger problems, we prove thatwe can exploit the sparsity and the symmetry of the original problemto reduce, considerably, the size of the SDP problems in the sequence.

� TB-19Tuesday, 10:30-12h00RB-2111

Quantitative Models for Risk Managementin InsuranceStream: Actuarial Sciences and Stochastic CalculusInvited sessionChair: Montserrat Guillen, Econometrics-UB, University ofBarcelona, Diagonal, 690, E-08034, Barcelona, Spain,[email protected]: Ana Maria Perez-Marin, Econometrics and Statistics,University of Barcelona, Av. Diagonal 690, Faculty of Economics,08034, Barcelona, Barcelona, Spain, [email protected]

1 - Selecting prospects for cross-selling financial prod-ucts using multivariate credibilityFredrik Thuring, Actuarial Science, Cass Business School,106 Bunhill Row, EC1Y 8TZ, London, United Kingdom,[email protected], Jens Perch Nielsen, Catalina Bolance

For many companies a possible way to expand its business is to sellmore products to preferred customers in its portfolio. Data on the cus-tomers’ past behaviour (e.g. past claiming history, for insurance prod-ucts, or past information on defaulting, for banking products) is storedin the company’s data base and can be used to assess whether or notmore products should be offered to a specific customer. This studyimplements a method for using historical data of each individual cus-tomer, and the portfolio as a whole, to select a target group of customerto whom more products would be offered.

2 - A logistic regression approach to estimating cus-tomer profit loss due to lapses in insuranceAna Maria Perez-Marin, Econometrics and Statistics,University of Barcelona, Av. Diagonal 690, Faculty ofEconomics, 08034, Barcelona, Barcelona, Spain,[email protected], Montserrat Guillen, Manuela Alcañiz

This article focuses on business risk management in the insurance in-dustry. A methodology for estimating the profit loss caused by eachcustomer in the portfolio due to policy cancellation is proposed. Us-ing data from a European insurance company, customer behaviour overtime is analyzed in order to estimate the probability of policy cance-lation and the resulting potential profit loss due to cancellation in twodifferent lines of business contracts. Implications for understandingcustomer cancellation behaviour as the core of business risk manage-ment are outlined.

3 - Developing customer fidelity programs and strate-gies within customer relationship management activ-itiesAras Keropyan, Economics and Business Organization,University of Barcelona, 08034, Barcelona, Spain,[email protected], Ana Maria Gil-Lafuente

Over the last decade, Customer Relationship Management (CRM) ac-tivities have become fundamental for companies. Every time it is beingmore costly to acquire new customers than retaining the existing busi-ness, therefore the big corporations are more disposed to invest moreresources for existing customers in order to assure their satisfaction andfidelity with the company. In this study our objective is to implementa fuzzy based Hungarian Method that will allow us to develop effi-cient customer fidelity programs and strategies for different customersegments within CRM activities.

� TB-20Tuesday, 10:30-12h00RB-2113

Financial Modeling and Optimization

Stream: Decision Making Modeling and Risk Assess-ment in the Financial SectorInvited sessionChair: Hüseyin Tatlidil, Statistics, Hacettepe University, HacettepeUniversitesi Beytepe Kampusu Istatistik Bolumu, Beytepe, 06532,Ankara, Turkey, [email protected]

1 - Portfolio optimization with a new quantile-based riskmeasureCristinca Fulga, Gheorghe Mihoc-Caius Iacob Institute ofMathematical Statistics and Applied Mathematics ofRomanian Academy, Calea 13 Septembrie No.13, Sector 5,050711, Bucharest, Romania, [email protected]

We develop a portfolio selection method which takes into considera-tion the recent positive evolution of the risky assets that are not com-prised in the portfolio available at the moment of the decision by usingthe Analytical Hierarchy Process. Our model relies on a new quan-tile based risk measure which is defined using the modified loss dis-tribution according to the decision maker’s loss aversion preferences.

115


We investigate the practical performance of the proposed Mean-Riskmodel on a portfolio composed of some of the most representative se-curities of the Bucharest Stock Exchange.

2 - Probability distribution models for european stockindex returnsChris Katris, Engineering Sciences, University of Patras,26504, Patras, Greece, [email protected], SophiaDaskalaki

We study the marginal distributions of daily returns for eight EuropeanStock Market Indices and attempt to fit the Generalized Hyperbolicand mixtures of Normal Distributions. We also create a model thatcombines the Generalized Pareto for the tails and other distributionsfor the rest of the data. The above marginal distributions are then con-sidered as static models for parametric estimation of the Value at RiskEvaluation of the models concludes this work.

3 - A Query for Statistical Arbitrage in Fixed Income Mar-kets for Developing CountriesSait Satiroglu, Center For Computational Finance, OzyeginUniversity, Kusbakisi Cad No:2, Altunizade Uskudar, 34662,Istanbul, Turkey, [email protected], Emrah Sener,Emrah Ahi

This paper studies the Nelson-Siegel framework driven fixed incomearbitrage strategy in three developing countries. We consider the richcheap analysis and buttery trades which historically present abnormalyields to maturity, taking as reference a robust zero-coupon yield curvefitted with bond prices. The main contribution of the paper is theconstruction-optimization, back-testing and comparison of our fittedyield curves and Z-score based strategies. We show that "intellectualcapital" required yield curve arbitrage strategies produce positivelyskewed returns also in emerging markets.

4 - The Performance of Value at Risk and ExpectedShortfall Under Subfamilies of Generalized Hyper-bolic Distributions: Evidence from ISE-30Ali Sabri Taylan, Market Oversight, Turkish DerivativesExchange, Vadeli Islem ve Opsiyon Borsasi A.S., AkdenizCd. Birsel Is Mrk. No:14 Kat:6, 35210, Izmir, Turkey,[email protected], Hüseyin Tatlidil

The approach based on normal distribution was used in the studies ofguessing the VaR as a standard risk measuring criterion. However, inmany researches it is understood that the distribution of return series offinancial variables has heavy tails and it is skewed and straight. In thisstudy, we fit subfamilies of generalized hyperbolic distributions andevaluate the performance of quantile-based risk measures such as VaRand Expected Shortfall with a dataset of daily prices of ISE-30 Indexand stocks. We also examine the differences between the success ratesof the calculated risk measures.

� TB-21Tuesday, 10:30-12h00RB-2115

Financial Mathematics and OR 6Stream: Financial Mathematics and ORInvited sessionChair: Tatiana Zolotova, Chair f Applied mathematics, StateUniversity of the Ministry of Finance of the Russian Federation,127006, Moscow, Nastasinsci lane, house 3, case 2, Researchfinancial institutions, 127006, Moscow, Russian Federation,[email protected]: Tansel Avkar, Department of Mathematics, Middle EastTechnical University, 06531, Ankara, Turkey,[email protected]

1 - The general approach to the modeling of the riskmanagement procedures and its specification for in-vestment problems

Tatiana Zolotova, Chair f Applied mathematics, StateUniversity of the Ministry of Finance of the RussianFederation, 127006, Moscow, Nastasinsci lane, house 3, case2, Research financial institutions, 127006, Moscow, RussianFederation, [email protected] work is devoted to the description of the general model of the riskmanagement, including two submodels: the model of an estimation ofsystem effectiveness and the model of an estimation of the risk of itsfunctioning. The general model can be applied to decision-making inconditions of risk for various complex systems and processes. Prob-lems of management are considered for stochastic processes at thestrategic and share investment, using various estimations of risk. Meth-ods of these problems decision in view of criteria of efficiency and riskare offered.

2 - Stochastic Hybrid Dynamical Financial Systems andTheir OptimizationGerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara,Turkey, [email protected], Büsra Temocin, Azar Karimov,Erdem Kilic, Yeliz Yolcu OkurWe represent dynamics in finance under uncertainty and covariance.An emerging representation tool for this is Stochastic DifferentialEquations. Is has become acknowledged that additionally an impulsivepart is needed, together amounting for Levy processes. Even these areoften composed piecewise with changes happening at thresholds. Wepresent several classes of these Hybrid Systems, discuss problems ofidentification, keeping them between barriers and optimizing portfo-lios with regard to them, discussing approaches of Hamilton-Jacobi-Bellman and Maximum Principle.

3 - Parameter Estimation for Stochastic DifferentialEquationsTansel Avkar, Department of Mathematics, Middle EastTechnical University, 06531, Ankara, Turkey,[email protected], Gerhard-Wilhelm WeberStochastic differential equations play a fundamental role in modelingdynamic phenomena in nature, technology and economy. Models de-fined via SDEs usually depend on finite or possibly infinite dimen-sional parameters. To obtain a useful model, it is critical to estimatethem. But these equations are hard to represent by a computer and toresolve. We express them in the manner of approximation by a dis-cretization and additive models with splines. Our estimation refers tothe linearly involved spline coefficients and the partially nonlinearlyinvolved probabilistic parameters.

4 - A Surrogate Constraint Approach for a Scenario-based Capital Budgeting ModelJosé Paixão, Dept. Statistics and Operations Research,Faculty of Sciences - University of Lisbon, Bloco C6 - CampoGrande, 1749-016, LISBOA, Portugal, [email protected],Anabela CostaContingent claims analysis can be used for project evaluation whenthe project develops stochastically over time and the decision to investcan be postponed. In that perspective, a model has been presented inthe literature. However, the corresponding linear integer program turnsout to be quite intractable even for small size instances. In this talk, wepresent a surrogate constraint relaxation for the original problem yield-ing the determination of lower and upper bounds for the optimal valueof the problem. Computational experience, allowing the comparisonwith other approaches, is reported.

� TB-22Tuesday, 10:30-12h00RB-2117

Mathematical Models in Macro- andMicroeconomics 1Stream: Mathematical Models in Macro- and Microe-conomicsInvited sessionChair: Marta Kostrzewska, Institute of Mathematics, University ofSilesia, ul. Bankowa 14, Katowice, 40-007, Katowice, Poland,[email protected]

116


1 - Efficiency-Indices for revealed preference tests:Complexity results and algorithmsBart Smeulders, Faculty of Business and Economics, KULeuven, Naamsestraat 69, 3000, Leuven, Belgium,[email protected], Frits SpieksmaIn this paper, we prove a number of complexity results regardingefficiency-indices for revealed preference tests. These indices are oftenused in micro-economics as a measure of consumer rationality. Whilea number of observations regarding the difficulty of calculating theseindices are made in the economics literature, few formal results exist.For one popular efficiency-index we provide exact polynomial timealgorithms, whilst for two other common efficiency-indices we proveNP-Hardness for four different axioms of revealed preference.

2 - Who’s Number One? Asymmetric Production Costsas a Determinant of Quality RankingsMihkel Tombak, Management, University of Toronto, 3359Mississauga Rd North, L5L 1C6, Mississauga, Ontario,Canada, [email protected], Sergio MezaWe examine how production cost asymmetries affect the way firmscompete in prices and quality in a vertical differentiation model andobtain an endogenously determined quality ranking. When cost differ-ences grow, quality differentials decrease, and when they grow largeenough, we have a unique duopoly equilibrium where the low-costfirm offers high quality. We illustrate the robustness of these results todifferent costs of quality formulations and find our results are strength-ened when adding costs of quality. We conduct a welfare analysis anddiscuss implications.

3 - Time-varying regression model of gross domesticproductDamir Atygayev, Control/Management and AppliedMathematics, Moscow Institute of Physics and Tehnology,Russian Federation, [email protected], Vadim MottlIn this paper, we review time-varying linear regression, which is theessence of recovering some hidden dependences (GDP’s components),which are time-varying in the general case. Feature of this model isthe assumption of zero presence of noise in the observations and hid-den components. This assumption imposes additional constraints onthe observations and the functional dependence of the hidden compo-nents. The method being proposed is simple in tuning and has linearcomputational complexity with respect to the signal length.

4 - Games in the network flowsMarta Kostrzewska, Institute of Mathematics, University ofSilesia, ul. Bankowa 14, Katowice, 40-007, Katowice, Poland,[email protected], Leslaw SochaThe aim of this presentation is to study the cooperative games relatedto the network cost flow problem. We consider the network in whichthe set of arcs is shared between K players. Every coalition is inter-ested in maximizing the value of the preference function together withminimizing the linear function representing the total cost of a flow.The coalition is wining if both criteria do not exceed defined extremevalues. The properties and the costs of stability of defined games arestudied.

� TB-23Tuesday, 10:30-12h00RB-Delta

Mathematical Programming andRegulatory - Ecological Systems

Stream: Mathematical ProgrammingInvited sessionChair: Erik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39, 85577,Neubiberg, Germany, [email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]: Joanna M. Makulska, Department of Cattle Breeding,University of Agriculture in Krakow, al. Mickiewicza 24/28,30-059, Kraków, Poland, [email protected]

1 - Robust Optimization in Spline Regression Models forTarget-Environment NetworksAyse Özmen, Scientific Computing, Institute of AppliedMathematics, Middle East Technical University,Kazimkarabekir Mah. 2048. Sok No:5/5, Etimesgut, 06790,Ankara, Turkey, [email protected], Erik Kropat,Gerhard-Wilhelm Weber

Solutions to optimization problems can represent significant sensitiv-ity to perturbations in the parameters of the problem. So, optimiza-tion affected by parameter uncertainty is a focus of the mathematicalprogramming and a need to tackle uncertain data when optimizationresults are combined within real-life applications. Since the regres-sion models of target-environment networks can be affected by noise,we introduce spline regression models using MARS, Conic MARS(CMARS) and, their robust versions under polyhedral uncertainty byusing robust optimization that deal with data uncertainty.

2 - Optimal parameters region for Pulse Width - PulseFrequency and Pseudo-Rate modulatorsM. Navabi, SMARTLAB, Shahid Beheshti University, GC,1234, NY, Iran, Islamic Republic Of,[email protected], Hamed Rangraz

Pulse Width - Pulse Frequency, the device that translates the continu-ous commanded torque signal to an on/off signal, is analyzed, the opti-mal range of its parameters is determined and compared with pseudo-rate modulator. Thrusters used as attitude control actuator operate inon-off mode, but controller create a continuous signal, therefore a unitis needed to translate continuous signal while providing lowest fuelconsumption and thruster firing for attitude control of spacecraft. Twomodulators are selected and optimized through system analysis.

3 - Economically optimal days open and days dry in highmilking cowsJoanna M. Makulska, Department of Cattle Breeding,University of Agriculture in Krakow, al. Mickiewicza 24/28,30-059, Kraków, Poland, [email protected], AndrzejWeglarz, Anna Stygar

Numerous investigations indicate a necessity to change standard rec-ommendations on the optimal length of days open and days dry in highyielding dairy herds. This study presents the solutions obtained fromthe multi-level hierarchic Markov model developed to support the eco-nomically optimal decisions on insemination, drying-off and replace-ment of high milking cows. The biological, technical and economicparameters of the model were derived based on the empirical data rep-resenting the conditions of dairy cattle production in Poland and on theresults of a literature review.

� TB-24Tuesday, 10:30-12h00CC-A11

Financial Forecasting and Risk Analysis

Stream: ForecastingInvited sessionChair: Patricia Xufre, CIO-FCUL and NovaSBE, Edificio C6,Campo Grande, 1749-016, Lisboa, Portugal, [email protected]

1 - Are the tails different from the body?Jose Faias, UCP, Catolica Lisbon School of Business andEconomics, Palma de Cima, VAT 501082522, 1649-023,Lisbon, Portugal, [email protected]

In recent years there has been an increasing interest in modeling depen-dence in heavy tail phenomena such as the latest turbulence episodesin financial markets. We measure extremal correlation in internationalequity markets. Our analysis offers a new perspective into the dynam-ics governing asset return comovements during periods of stress at thecross-country and cross-sectoral levels.

117


2 - Investment models with a VaR constraintPatricia Xufre, CIO-FCUL and NovaSBE, Edificio C6,Campo Grande, 1749-016, Lisboa, Portugal,[email protected], Antonio Rodrigues

We analyse how VaR (Value-at-Risk) can be used directly to designan investment strategy. We propose 2 methods to build an investmentmodel based on the profit maximization with VaR constrained: (i) thedistribution of the returns is stationary, (ii) a recursive way to estimateVaR allowing adaptation to new information. The estimation of thequantile is made through a new neurocomputational procedure. Bothmethods were tested for several price time series, as well as with arti-ficial data generated according to different stochastic models.

3 - Recursive quantile estimation vs. quantiles from re-cursive density estimationJose Luis Carmo, CIO and University of Algarve, Portugal,[email protected], Antonio Rodrigues

Many decision problems, namely in the areas of financial investing,inventory management or quality control, are approachable by the re-cursive estimation of quantiles, from (possibly) nonstationary time se-ries. In this work, we compare different recursive quantile estimationmethods, and end up concluding, on the basis of empirical evidence, infavour of an alternative approach: the computation of quantiles fromrecursively estimated probability densities, the latter being accom-plished through an original adaptive exponential smoothing method.

4 - How the stochastic problem drives the Brazilian elec-tric sectorReinaldo Souza, Departamento de Engenharia Elétrica,Pontifícia Universidade Católica do Rio de Janeiro, RuaMarquês de São Vicente, 22453-900, Rio de Janeiro, RJ,Brazil, [email protected], Pedro Ferreira, FernandoLuiz Cyrino Oliveira

This article aims, first of all, at presenting the reader a broad perspec-tive of the Brazilian Electric sector, identifying the process of trans-formation through which it underwent during the last decades, arrivingat the Brazilian Electric Sector’s (BES) New Model. With this NewModel, it is shown how a good modeling of the stochastic variableis determinant to the good functioning of the BES’ three fundamen-tal pillars, that is, planning, operation, and accounting and settlementprocess of the energy transactions in the short term market.

� TB-25Tuesday, 10:30-12h00CC-A23

Sustainable Supply Chains

Stream: Environmental Issues in Operations Manage-mentInvited sessionChair: Mehmet Soysal, Operations Research and Logistics,Wageningen University, Hollandseweg 1, 6706 KN, Wageningen,Netherlands, [email protected]

1 - Carbon-Optimal and Carbon-Neutral Supply ChainsCharles Corbett, UCLA Anderson School of Management,110 westwood plaza, box 951481, 90095-1481, Los Angeles,CA, United States, [email protected], FelipeCaro, Tarkan Tan, Rob Zuidwijk

We introduce a model of joint production of GHG emissions in gen-eral supply chains, decomposing total footprint into processes, each ofwhich can be influenced by any combination of firms. We examinewhen the supply chain can be carbon-optimal, ie. when all firms ex-ert first-best emissions reduction effort levels. We find that, to inducecarbon-optimal effort levels, emissions need to be over-allocated. Thismeans that the focus in the LCA and carbon footprinting literature onavoiding double-counting is, in the context of setting incentives, mis-guided.

2 - An application of Revenue Management in remanu-facturing: the case of a third party remanufacturerJoao Quariguasi, MACE, University of Manchester,Manchester, United Kingdom,[email protected], Andy Reade,Azadeh DindarianOptimally deciding as to whether recycle, refurbish, remanufacture orretrieve parts from a returned product is no trivial task, and is dictatedby a number of factors: the WTP for the remanufactured product, scrapprices, remanufacturer’s idle capacity. Moreover, in the case of largethird party remanufacturers, these decisions have to be made hundredsof thousands of times a year. In this paper, using real data obtainedfrom a large remanufacturer in the UK, and drawing on revenue man-agement techniques, we develop an automated system to support suchdecisions.

3 - Robust design of production networks for secondgeneration synthetic bio-fuelGrit Walther, Schumpeter School of Business and Economics,Chair of Production and Logistics, Bergische UniversitätWuppertal, Rainer-Gruenter-Str. 21, 42119, Wuppertal,Germany, [email protected], Thomas Spengler,Anne Schatka, Laura HombachSecond generation synthetic bio-fuels can make an important contribu-tion to sustainable mobility. However, there are currently high uncer-tainties with regard to development of production technology, futurepolitical requirements, and future demand. We develop a multi-periodMIP-model for integrated location, capacity and technology planningfor the design of production networks for second generation syntheticbio-fuels. Principle network configurations are developed, and the un-certainties are regarded using a scenario based approach.

4 - An integrated model for sustainable food logisticsmanagementMehmet Soysal, Operations Research and Logistics,Wageningen University, Hollandseweg 1, 6706 KN,Wageningen, Netherlands, [email protected], JacquelineBloemhof, Jack van der VorstFood supply chains (FSCs) require different quantitative models foraiding management decisions, mainly because of the perishability ofthe product. Addition of sustainability concerns makes models for theFSCs more challenging. In this study, we develop a generic MILPmodel for the logistics management of FSCs while considering the per-ishability of the products and the sustainability (GHG emissions) of thetransportation modes. The model minimizes various costs, while ful-filling demand requirements and can be used as a decision support toolfor improving the sustainability of FSCs.

� TB-27Tuesday, 10:30-12h00CC-A25

Decision Analysis in Practice

Stream: Decision ProcessesInvited sessionChair: Mara Airoldi, Dept of Management, London School ofEconomics and Political Science, Houghton Street, WC2A 2AE,London, United Kingdom, [email protected]

1 - Assessing the value for money of UK public servicesTom McBride, National Audit Office, SW1W 9SP, London,United Kingdom, [email protected] National Audit Office provides the UK Parliament with indepen-dent assurance on the use of public resources. One way we do this is bypublishing around 60 Value for Money (VfM) reports per year. Thesereports are on topics as diverse as dementia services and banking sup-port. OR plays an important role in our work. For example they can beused to construct a counterfactual to assess if VfM has been achieved;or to examine the integrity of Government forecasts. This presentationwill give an overview of our role, and present 3 case studies of how wehave used OR to analyse UK public services.

118


2 - Disinvestments in practiceMara Airoldi, Dept of Management, London School ofEconomics and Political Science, Houghton Street, WC2A2AE, London, United Kingdom, [email protected]

A challenge faced by many public sector organisations is how to re-duce costs. Disinvesting is more challenging than allocating additionalspend, because the staff who know the service best do not see honestpresentation of the options as being in their interest. We show howa decision conferencing approach was used to help a health authoritymake critical decisions about the reconfiguration of an eating disordersservice in order to secure cost savings, and discuss lessons learned.

3 - Tackling uncertainties in multi-criteria decision sup-port for sustainable water infrastructure planningLisa Scholten, System Analysis, Integrated Assessment andModelling, EAWAG:Swiss Federal Institute of AquaticScience and Technology, Ueberlandstr 133, P.O.Box 611,8600, Dubendorf, Zurich, Switzerland,[email protected], Judit Lienert

To support water infrastructure choices under uncertain scenarios,MAUT is applied to a case study in Switzerland. Different sources ofuncertainty from internal and external factors of MCDA will be con-sidered during evaluation. Due to the complexity of elicitation anduncertainty modeling, usually much simplified preference models areused (e.g. risk neutrality, additive aggregation). Knowing (from globalsensitivity analysis), which uncertainty really matters, will help to fo-cus on more adapted elicitation design. The overall decision analyticframework and first results will be presented.

� TB-28Tuesday, 10:30-12h00CC-A27

Optimization Modeling

Stream: Structured Optimization and ApplicationsInvited sessionChair: Marcel Sinske, Institute of Operations Research, KarlsruheInstitute of Technology, Kaiserstraße 12, 76131, Karlsruhe,Germany, [email protected]

1 - Taxonomy of concepts and terms in administration,OR and managementErwin Reizes, O.R., Fac.Ing./UdelaR,Uruguay, Luis A.deHerrera 1042,ap.1701 (home), ap.1701, 11300, Montevideo,Uruguay, [email protected]

A Taxonomy of concepts and terms seems necessary to scope with the"Babel" of synonyms and homonyms in administration, OR and man-agement theory and practice. Based on ideas of Logic, System Theory,Linguistics, Biology, OR and Informatics, a uniform correspondanceof concepts and terms is given.

2 - Measuring the Performance of TV Campaigns basedon DEA: an Empirical StudyKatsuaki Tanaka, Faculty of Business Administration,Setsunan University, 17-8 Ikedanakamachi, 572-8508,Neyagawa, Osaka, Japan, [email protected],Yoshinori Matano

Though much efforts have been done to have a understanding of theTV advertising effects, there is no definite standard. The frontier anal-ysis is useful clarifying whether the campaign is efficient or not. DEAis a frontier analysis based on the empirical data. By observing the dis-tance from the frontier, it provides the improvement of the inefficientDMU. In this paper, DEA is employed to measure the performance ofTV advertising campaigns. Each campaign is considered as a DMU -the inputs are the advertising expenditures whereas the outputs are theawareness and knowledge of the campaigns.

3 - Double and One Sided Moral Hazard in the Three-tierhierarchical Organization of the Venture Capital Fi-nanceAyi Gavirel Ayayi, Science de la gestion, Université duQuébec, 3351 Boulevard des Forges, G9A 5H7, TroisRivières, Québec, Canada, [email protected] paper studies double-sided moral hazard problem (MHP) betweenventure capitalists (VCs) and entrepreneurs and one-sided MHP be-tween the investors in the venture capital funds and VCs because ofthe direct impact of the moral hazard on the performance of the fundsas well as on the growth and the success of the investees. I show thatan optimal compensation contract can reconcile the different needs andconcerns of the three contracting parties; given the information distri-bution and the behavioral assumptions. Global optimization programis used to solve goal congruence optimal contract.

� TB-29Tuesday, 10:30-12h00CC-A29

MINLP and Nonlinear Network Flows inGas Transport I

Stream: Mixed-Integer Non-Linear ProgrammingInvited sessionChair: Benjamin Hiller, Optimization, Zuse Institute Berlin,Takustr. 7, 14195, Berlin, [email protected]

1 - Checking Feasibility in Stationary Models of GasTransportationClaudia Stangl, Mathematics, University of Duisburg-Essen,Butlerstrasse 10, 47058, duisburg, Germany,[email protected], Rüdiger SchultzChecking the feasibility of transportation requests belongs to the keytasks in gas pipeline operation. In its most basic form, the problem isto decide whether a certain quantity of gas can be sent through the net-work from prescribed entries to prescribed exit points. In the stationarycase, the physics of gas flow together with technological and commer-cial side conditions lead to a pretty big (nonlinear, mixed-integer, finitedimensional) inequality system. We present elimination and approxi-mation techniques so that the remaining system gets within the reachof standard NLP-solvers.

2 - The Gas Network Nomination Validation Problem - AnMIP ApproachBjoern Geissler, Mathematics, FAU Erlangen-Nürnberg,Discrete Optimization, Cauerstraße 11, 91058, Erlangen,Germany, [email protected], AlexanderMartin, Antonio Morsi, Lars ScheweNomination validation constitutes kind of a core problem for many op-timization tasks in gas economy. It is about to decide whether a gasnetwork can be operated such that given demands are satisfied subjectto legal, physical and technical constraints. The nomination validationproblem is most naturally modeled as nonconvex MINLP. However, forlarge networks these models are intractable by state-of-the-art MINLP-solvers. We present an alternative approach based on a new kind ofMIP-relaxations and provide computational evidence for its suitabilityeven for large real-life instances.

3 - Gas Network Topology OptimizationJesco Humpola, Optimization, Zuse Institute Berlin,Takusstrasse 7, 14159, Berlin, Berlin, Germany,[email protected], Thorsten Koch, Thomas Lehmann, JonasSchweigerWe aim for topology optimization of large-scale, real-world gas trans-mission networks. Given a nomination, i.e. a balanced allocation ofsource and sink flows, we look for a suitable setting of all networkcomponents and a cost optimal selection of new network elements,such that the requested amount of gas can be transmitted through thenetwork without violating physical or operational constraints. Theproblem can be stated as a non-convex MINLP, which we solve bya special tailored combination of the solvers SCIP and IPOPT and aheuristic that makes use of the dual values of the IPOPT calls.

119


4 - Modeling flow in gas transmission networks usingshape-constrained expectile regressionFabian Sobotka, Economics, Universität Goettingen, Platz derGöttinger Sieben 5, 37073, Göttingen, Niedersachsen,Germany, [email protected], RadoslavaMirkov, Werner Römisch, Thomas Kneib

The gas flow within a gas transmission network is studied with theaim to optimise such networks. A geoadditive model for describingthe dependence between the maximum daily gas flow and the tempera-ture on network exits is proposed. Semiparametric expectile regressionprovides the possibility to model the upper tail of the response distri-bution while accounting for the spatial correlation between differentexits. The effect of the temperature is modeled with shape constraintsto include knowledge about gas load profiles and to allow for a realisticprediction.

� TB-30Tuesday, 10:30-12h00CC-A31

Applications and Decision SupportSystems

Stream: Emerging Aspects of Production Planning inContinuous Process Industries: Theory, Optimization,and PracticeInvited sessionChair: Jim Everett, Centre for Exploration Targeting, University ofWestern Australia, 49 Goldsmith Rd, 6009, Nedlands, WA,Australia, [email protected]

1 - A decision-support tool for designing cost-effectivebiopharmaceutical manufacturing processesSofia Simaria, Biochemical Engineering, University CollegeLondon, Torrington Place, WC1E 7JE, London, UnitedKingdom, [email protected], Richard Allmendinger,Suzanne Farid

A decision-support tool using genetic algorithms to aid the design offlexible and cost-effective multi-product facilities that can cope withdiverse biopharmaceutical drug candidates is presented. Results froman industrially relevant case study show that the most cost-effectivepurification sequences and equipment sizing strategies that meet de-mand and purity targets for each product in the facility are identified.Emphasis is also placed on the use of graphical methods to allow thevisualisation of trade-offs in the set of optimal solutions so as to en-hance the decision-making process.

2 - Optimal Supply Chain Planning: Service Levels andProfitability ScoresAna Amaro, Applied Mathematics and Informatics, ISCAC,Accounting and Administration College of Coimbra, QuintaAgrícola, Bencanta, 3040-316, Coimbra, Portugal,[email protected], Ana Paula Barbósa-Póvoa

Globalization has led businesses to focus on the worldwide markets.To maintain competitive advantage, process industries and associatedsupply chains (SCs) have to address new challenges. The present con-tribution develops a Mixed Integer Linear Programming formulationto model different SC performance and planning criteria such as ser-vice levels and profitability. These represent efficiency goals and SCs’ability to manage customers’ expectations. The impact of each plan-ning strategy on the optimal plan is analysed and the applicability isillustrated through an industrial pharmaceutical SC.

3 - Joint lot sizing and routing problem of perishablegoodsMárcio Antônio Ferreira Belo Filho, Instituto de CiênciasMatemáticas e de Computação, Universidade de São Paulo,Avenida Trabalhador são-carlense, 400 - Centro, São Carlos,São Paulo, Brazil, [email protected], Pedro Amorim,Franklina Toledo, Christian Almeder, Bernardo Almada-Lobo

The joint production and distribution planning problem usually ad-dresses batching of orders disregarding lot sizing. We investigate thevalue of considering lot sizing versus batching for a multi-customer de-livery problem with routing and perishable products. Results show thatlot sizing is able to deliver better solutions than batching. Moreover,the flexibility due to lot sizing decisions can reduce production anddistribution costs. Finally, these savings are leveraged by customerswith tight time windows and production systems with non-triangularsetups.

� TB-31Tuesday, 10:30-12h00CC-A33

Data Analysis and Its Applications

Stream: Data Mining, Knowledge Discovery and Artifi-cial IntelligenceContributed sessionChair: Irina Yatskiv, Computer Science, Transport andTelecommunication institute, Lomonosova 1, LV-1019, Riga,Latvia, [email protected]

1 - Extending Learning Vector Quantization for mixed-type data classificationChung-Chian Hsu, Dept. of Information Management,National Yunlin University of Science and Technology, 123University Rd. Sec. 3, 640, Douliou, Yunlin, Taiwan,[email protected], Jiang-Shan Wang, Hung-Yi Tsai,JongChen Chen

Real-world data usually consist of categorical and numeric values.Learning Vector Quantization (LVQ), a useful classification technique,however handles only numeric data. Categorical values need to be con-verted to numeric ones prior to training. Nevertheless, careless conver-sion schemes can lose semantics inherent in categorical values and mayresult in reduced classification accuracy of LVQ. In this study, we pro-pose an extended LVQ which incorporates a distance scheme: distancehierarchy for taking into account the semantics. Experimental resultsdemonstrated effectiveness of ELVQ.

2 - Continuous predictor modality multinomial corre-lated interpretationAndrew Yatsko, ITMS Dept., University of Ballarat, Mt.Helen campus, 3350, Mount Helen, VIC, Australia,[email protected], Andrew Stranieri,Adil Bagirov, Herbert Jelinek

Compactness of data leads to improved performance of classificationalgorithms. Discretization of continuous features is also a way to uni-form representation of problems with nominal features. Certain reli-able, yet simple techniques exist. At the same time, interval optimiza-tion of multiple attributes poses a significant challenge. This articleexplores a suboptimal approach based on cluster analysis. The methodis being compared to others and appears to be competitive, althoughlimits exist to what can be achieved, imposed by noise, data specifics,and the bias of classification methods.

3 - Assessing Bayesian Network classifiers for softwaredefect predictionKarel Dejaeger, Faculty of Business and economics,Katholieke Universiteit Leuven, Naamsestraat 69, 3000,Leuven, Belgium, [email protected], ThomasVerbraken, Bart Baesens

Software testing is a crucial activity during software development anddefect prediction models assist practitioners herein by providing anupfront identification of faulty software. While Naive Bayes (NB) isoften applied in this regard, citing predictive performance and com-prehensibility as its major strengths, alternative Bayesian algorithmsboosting the promise of simpler networks remain underexposed. Weexplored 15 such classifiers, in combination with a Markov blanketfeature selection procedure, and found that simpler models can be con-structed using Bayesian classifiers other than NB.

120


4 - A decision support system for ERP implementationin small and medium sized enterpriseYing Xie, Dept of System Management and Strategy,University of Greenwich, Business School, Park Row, SE109LS, London, United Kingdom, [email protected]

ERP implementation is challenging for small and medium sized enter-prises (SMEs) as it is time consuming and risky. An integrated decisionsupport system for ERP implementation (DSS_ERP) is developed inte-grating analytical regression models, a simulation model and nonlinearprogramming models. The DSS_ERP offers an analytical model to ac-curately monitor the progress each critical success factor (CSF) makesalong the time horizon and monitor the cost consumed; it determinesthe priorities of CSFs and facilitates decision makings on resource al-locations.

� TB-32Tuesday, 10:30-12h00CC-A34

Developing Methodology in Soft OR

Stream: Soft OR and Problem Structuring MethodsInvited sessionChair: Amanda Gregory, Business School, University of Hull,HU6 7RX, Hull, United Kingdom, [email protected]

1 - Evaluating interventions using a lean systemsmethodologyGavin Betts, Hull University Business School, University ofHull, Cottingham Road, HU6 7RX, Hull, East Yorkshire,United Kingdom, [email protected]

This research seeks to evaluate the Vanguard Methodology. Thismethodology has been developed by Vanguard Consulting over thirtyyears and has been labelled a lean systems methodology in the litera-ture. It has been designed for the service sector and claims improvedservice at reduced costs. An evaluation framework developed to assesscases using this methodology will be presented.

2 - The use of the CAQDAS (Computer Aided QualitativeData Analysis Software) to structuring problems inOperations ResearchWilliam Vianna, Information Science, Federal University atSanta Catarina- Brazil, Rua Cônego Bernardo, 100. ap. 202,Trindade, 88036570, Florianópolis, Santa Catarina, Brazil,[email protected], Eduardo Hoji, Edilson Giffhorn

The aim of this study is to propose the use of CAQDAS (ComputerAided Qualitative Data Analysis Software) to support the analysis ofqualitative data in structuring problems in Operations Research. Theapplication is given in terms of data integration between postgraduateevaluation system and the Basic Education system evaluation in Brazil.The qualitative data repository favors recursivity and structured returnto data at any stage of development of the multicriteria model.

3 - Exploring the use of Collaboration Engineeringapproaches within the practice of Soft SystemsMethodologyAida Azadegan, Business School, Nottingham UniversityBusiness School, Apt. 1/3, 1 Rice Street, M3 4JL, Manchester,United Kingdom, [email protected], Giles Hindle

Soft OR contains a family of approaches that support groups in a rangeof tasks including tackling complex problem situations and makingstrategy. There has been increased interest in supporting the facilita-tion of Soft OR approaches in practice. This paper explores the useof Collaboration Engineering ideas within the practice of Soft SystemsMethodology. In particular, we examine the use of ThinkLets to scriptthe tools and steps of Soft Systems Methodology.

4 - It looks like a PSM and it does what a PSM does butis it a PSM?Amanda Gregory, Business School, University of Hull, HU67RX, Hull, United Kingdom, [email protected],Jonathan Atkins

The defining characteristics of a PSM and the set of approaches towhich the label can be applied are generally agreed. This paper fo-cuses on extending the label PSM to approaches that have the char-acteristics of a PSM but which have been developed outside of theOR/systems field and the potential (dis)advantages of this. The dis-cussion will be illustrated by reference to such an approach, the DP-SIR (Drivers — Pressures — State Changes — Impacts — Responses)framework, which is widely used within the particularly ’wicked’ con-text of ecosystem management.

� TB-33Tuesday, 10:30-12h00CC-A37

Logistic and Development

Stream: OR for Development and Developing Coun-triesInvited sessionChair: Youssef Masmoudi, University of Sfax, Hight School ofCommerce of Sfax, BP 954, 3018, Sfax, Tunisia,[email protected]: Habib Chabchoub, LOGIQ, Institut Supérieur de GestionIndustrielle, 3018, Sfax, Sfax, Tunisia,[email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - Time Dependant Intermodal Transportation in the Re-gion of MarmaraGiray Resat, Industrial Engineering, Koc University, Turkey,[email protected], Metin Turkay

This talk presents a multi-objective optimization model for combin-ing different transportation modes (such as Road, Rail and Sea ) in thedesign and operation of an intermodal transportation network in a re-gion. The problem is formulated as a mixed-integer optimization prob-lem that accounts for time and congestion dependent vehicle speeds.We represent the modeling approach, data analysis and outline the im-portant characteristics of the mathematical programming problem forminimization of the transportation cost and time simultaneously for theMarmara Region of Turkey.

2 - 3D Container Ship Loading Planning Problem withmultiple scenariosLuiz Salles Neto, Federal São Paulo State University,12231-280, São José dos Campos, São Paulo, Brazil,[email protected], Anibal Azevedo, Antônio Moretti,Antônio Chaves

The proposed paper presents a new 3D Container Ship Loading Plan-ning Problem with multiple scenarios. This problem is modeled asa bi-objective framework that minimizes two criteria: the number ofmovements and also the instability measure computed as the distancebetween gravity and geometric center for every different scenario. Todeal with this multiobjective and stochastic problem a new represen-tation that allows a compact and efficient encoding which enables thesolution of large-scale problems proper related with real-life scenarios.

3 - Graph Theoretic Properties and Algorithms with Ap-plication in Disaster ResponseReza Zanjirani Farahani, Kingston Business School, KingstonUniversity London, Room 330, Kingston Hill, KT2 7QH,Kingston Upon Thames, London, United Kingdom,[email protected], Hannaneh Rashidi Bajgan

121


This paper considers directing paths in transportation networks so thatall points could be reached. These problems are hard to solve optimallywhile sometimes even sub-optimal solutions are acceptable providedwe find them quickly; this application is evident in disasters. Thispaper presents two algorithms based on graph theoretic properties tosolve such problems; the first utilizes concept of faces in planar graphsand the second uses perfect matching rules. Such quick algorithms wedo not look for optimal solutions but these methods can also help speedup current optimal algorithms.

� TB-34Tuesday, 10:30-12h00CC-A39

Optimization for Sustainable Development1Stream: Optimization for Sustainable DevelopmentInvited sessionChair: Herman Mawengkang, Mathematics, The University ofSumatera Utara, FMIPA USU, KAMPUS USU, 20155, Medan,Indonesia, [email protected]

1 - A Chance-constrained Optimization Model for a Wa-ter Distribution Network ProblemAsrin Lubis, Mathematics, Unimed/Graduate School ofMathematics, University of Sumatera Utara, FMIPA USU,20155, Medan, Sumatra Utara, Indonesia,[email protected], Herman Mawengkang

The optimal design of a Water Distribution Network (WDN) consistsof the choice of a diameter for each pipe, the pipe line network design,and the pipe length. In this paper we propose a nonlinear stochas-tic optimization model for tackling a WDN under the considerationof reliability in water flows. The nonlinearities arise through pressuredrop equation. We transform the stochastic programming model into adeterministic multi-objective model using sampling average approach.Then we solve the result model using an interactive method.

2 - Land use management problems under uncertaintySiti Rusdiana, Mathematics, University of Syiah Kuala,FMIPA USU, 20155, Medan, North Sumatera Province,Indonesia, [email protected]

This paper is to find an optimal land portfolio composition throughtime under market uncertainty. We assume that an optimal land usemanagement program will involve a gradual sequence of conversiondecisions through time, evolving as each land allocation value becomesknown more accurately. In this paper we formulate stochastic pro-gramming model, which takes into account the uncertainty related tothe market value of revenues accruing from the land in different states.

3 - How to Determine Optimum Promotions through CLVMaximization?: A Case Study in a Turkish BankFusun Ulengin, Industrial Engineering, Dogus University,Dogus Universitesi, Muhendislik Fak. Endustri Muh. Bol,Acibadem, Kadikoy, 34722, Istanbul, Turkey,[email protected], Yeliz Ekinci, Nimet Uray

The purpose of this study is to develop a methodology to guide man-agers in determining the optimal promotion campaigns for differ-ent market segments in order to maximize the Customers’ LifetimeValue. It aims to find an optimal promotion plan for a service com-pany through stochastic dynamic programming. It also determines the’value’ states for the customers using classification and regression treeanalysis. An experiment is designed to observe the impacts of differentpromotion campaigns on customers. It is applied in a private Turkishbank and the improvement of profit is tested.

4 - A stochastic optimization model for multi-productfish production planning problem under uncertaintyin raw resources and demandIntan Syahrini, Mathematics, University of Syiah Kuala,FMIPA USU, 20155, Medan, North Sumatera Province,Indonesia, [email protected], Herman Mawengkang

A multi-product fish production planning produces simultaneouslymulti fish products from several classes of raw resources. The qual-ity of raw resources and market demand are uncertain during the plan-ning period. This paper proposes a multi-stage stochastic integer pro-gramming to model the production planning problem. A direct searchmethod is used to solve the model.

� TB-35Tuesday, 10:30-12h00CC-A41

Electric Mobility

Stream: Energy, Environment and Climate 2Invited sessionChair: Patrick Jochem, Chair of Energy Economics (IIP),Karlsruhe Institute of Technology (KIT), Hertzstr. 16, 76187,Karlsruhe, Germany, [email protected]

1 - A framework for modelling the electricity demand byplug-in electric vehiclesRashid Waraich, Institute for Transport Planning and Systems,ETH Zürich, Wolfgang-Pauli-Str. 15, 8903, Zürich, Zürich,Switzerland, [email protected]

Plug-in electric vehicles (PEVs) are envisioned by many as the cureto several current problems, especially environmental ones. What isoften neglected in this context is that a large scale penetration of suchvehicles may also pose new challenges, as charging these vehicles willresult in changes of current electricity demand patterns. The presentedwork introduces a framework, which models electricity demand byPEVs by extending a transport simulation called MATSim. The frame-work allows the evaluation of new charging policies and helps pinpointpossible bottlenecks in the electric grid.

2 - Technical and Economical Aspects of IntegratingPlug-in Electric Vehicles in Large Scale Power Sys-temsMiguel Carrión, Electrical Engineering, University of Castilla- La Mancha, Avda Carlos III, s/n, Campus Fábrica de Armas,45071, Toledo, Spain, [email protected], LuisSanchez-rodriguez

Plug in - Electric Vehicles (PEVs) are becoming an actual option toreplace traditional combustion-engine cars. In this situation, it is a pri-ority to analyze the impact of an extensive usage of electric vehiclesin actual power systems. In this work, we study the technical and eco-nomical aspects of the integration of PEVs in the Iberian Peninsulapower system (MIBEL). Three different approaches modelling the co-ordination between the system operator and PEVs users are consid-ered. Finally, a realistic case study is solved and sensitivity studies areperformed.

3 - Integrating Electric Vehicles into the Germany Elec-tricity Grid — an Interdisciplinary AnalysisPatrick Jochem, Chair of Energy Economics (IIP), KarlsruheInstitute of Technology (KIT), Hertzstr. 16, 76187, Karlsruhe,Germany, [email protected], Thomas Kaschub,Alexandra-Gwyn Paetz, Wolf Fichtner

This study analyses the integration of electric vehicles (EV) into theGerman power grid including different demand respond (DR) ap-proaches from a technical, economical and user perspective. For thisan overview of the future German electricity market with the focus onEV integration is given. It is shown that the shortage in the regionaland local electricity grid is increased. DR can help to tackle this issueby controlled unidirectional or bidirectional (V2G) charging of EV.From the current perspective especially the first alternative seems to beattractive in all considered disciplines.

122


� TB-36Tuesday, 10:30-12h00CC-A43


Stream: DEA and Performance MeasurementInvited sessionChair: Zilla Sinuany-Stern, Industrial Engineering andManagement, Ben Gurion University, Beer-Sheva, 84105,Beer-Sheva, Israel, [email protected]

1 - Analysing the Research and Teaching QualityAchievement FrontierDavid Mayston, Economics and Related Studies, Universityof York, University of York, Y010 5DD, York, UnitedKingdom, [email protected]

The paper analyses the achievement possibility frontier between re-search and teaching quality in higher education. It finds sev-eral important reasons why the associated feasible set is likely tobe non-convex, making use of the standard performance analy-sis technique of Data Envelopment Analysis less valid. The pa-per therefore investigates the alternative Free Disposal Hull tech-nique, and compares the results of deploying these techniquesto the performance evaluation of UK Departments of Economics.See: http://york.ac.uk/media/economics/documents/ discussionpa-pers/2011/1125.pdf

2 - Benchmarking the implementation path to efficiency:The case of small regional airportsEkaterina Yazhemsky, Business School, Hebrew University ofJerusalem, Mount Scopus, 91905, Jerusalem, Israel,[email protected], Nicole Adler, Tolga Ülkü

Small and regional airports suffer from limited traffic, large fixed facil-ities and insufficient revenues to cover their costs. The question is howthese airports should be structured, managed and financially supportedin order to survive. Efficient operations contribute to decreasing thefinancial dependency of airports on subsidies. According to a DEA-based, bound adjusted measure, the efficiency levels of 89 Europeanairports are determined. Due to heterogeneity across the sample, mul-tiple ideal points are applied to identify implementation paths and tohighlight managerial best practices.

3 - Performance assessment of wind farmsClara Vaz, School of Technology and Management,Polytechnic Institute of Bragança, Campus de Santa Apolónia,Apartado 134, 5301 - 857, Bragança, Portugal, [email protected]

This study develops a methodology to provide insights regarding theperformance of wind farms from a European player in energy sec-tor. Firstly, the DEA is used to measure the performance of windfarms in generating electricity from the resources available and non-discretionary variables such as wind (speed and availability). Thisanalysis enables to identify the benchmarks whose best practices canbe emulated by inefficient units and provides useful information forregulators in the sector. Secondly, bootstrap procedures are applied toobtain statistical inference on the efficiency estimates.

4 - Hospitals efficiency in Israel: 1999-2009Zilla Sinuany-Stern, Industrial Engineering and Management,Ben Gurion University, Beer-Sheva, 84105, Beer-Sheva,Israel, [email protected], Dov Chernichovsky, Lea Friedman

Data Envelopment Analysis (DEA) is used to measure the efficiencyof major general hospitals in Israel, during 1999-2009. The inputs arenumber of standard beds; the outputs are number of hospitalizationdays and number of discharges. In the second phase, we used regres-sion analysis, to examine the effect of size, and type of ownership onhospitals efficiency. We found that independent hospitals are the mostefficient and have decreasing returns to scale, followed by the gov-ernmental and public hospitals which have increasing returns to scale.Also Malmquist analysis was used over time.

� TB-37Tuesday, 10:30-12h00CC-Act

OR in Health & Life Sciences 2Stream: OR in Health & Life Sciences (contributed)Contributed sessionChair: Leanne Smith, School of Mathematics, Cardiff University,Senghennydd Road, CF24 4AG, Cardiff, United Kingdom,[email protected]: Paul Harper, School of Mathematics, Cardiff University,CF24 4AG, Cardiff, United Kingdom, [email protected]

1 - A Takt time based resource optimization techniqueto minimize patient wait time in hospitalsJyoti R. Munavalli, Computer Science & Engineering, PESIT,India,Ph.D Research Scholar with Maastricht University,Netherlands, Ring Road„ BSK III stage, 580083, bangalore,karnataka, India, [email protected], Frits vanMerode, Srinivas A

We propose a technique to address the resource optimization, patientflow balancing and wait time minimization in a hospital set-up. Wepresent a mathematical model for resource optimization which is basedon Takt time concept (average time a patient spends in a hospital fortreatment) for each department in all the timeslots and considers the de-pendency constraints. The Linear programming solution provides theresource allocation pattern for each department during every time slotwithin the resource availability to balance the patient flow and achievehigher patient satisfaction levels.

2 - Optimizing the emergency distribution network in thecase of epidemic outbreaks responseTom Dasaklis, Department of Industrial Management,University of Piraeus, 80 Karaoli & Dimitriou str., 18534,Piraeus, Attica, Greece, [email protected], NikolaosRachaniotis, Costas Pappis

Effective control of an epidemic’s outbreak calls not only for a rapidresponse but also for the establishment and management of an emer-gency supply chain. During the control effort huge amounts of medicalsupplies should be distributed from central warehouses to local areas.In this paper an epidemic diffusion model based on differential equa-tions is coupled with an emergency logistics distribution model. Anintegrated solution approach is proposed combining epidemics controlrules and logistics network design. A numerical example is finally pre-sented illustrating the methodology proposed.

3 - Modelling ambulance location and deployment inWalesLeanne Smith, School of Mathematics, Cardiff University,Senghennydd Road, CF24 4AG, Cardiff, United Kingdom,[email protected], Paul Harper, Vincent Knight, IsraelVieira, Janet Williams

Response time targets for the Welsh Ambulance Service Trust (WAST)are not currently being met, particularly for high priority emergenciesin rural areas. A discrete event simulation makes use of initial vehicleallocations given by a developed location model, in order to identifythe affect demand, number of available vehicles and handover timeshave on regional response. The aim of these mathematical models is tohelp WAST make better decisions on locations, capacities and deploy-ments, so that they may provide a more efficient and effective serviceand achieve the Government set targets.

4 - MetSim: a simulation support tool using meteorolog-ical information to improve the planning and man-agement of hospital servicesPaul Harper, School of Mathematics, Cardiff University,CF24 4AG, Cardiff, United Kingdom, [email protected],R John Minty

Improved short-term predictions of hospital admissions and bed oc-cupancy offer the potential to plan resource needs more accuratelyand effectively. The MetSim project explores the relationship betweenweather and health, building novel Bayesian models that are more sen-sitive to the weather. Short-term forecasts of the numbers of admis-sions categorised. In turn, MetSim uses hazard ratios embedded within

123


a simulation framework to provide forecasts of short-term bed needs.The web-based tool has been piloted in UK hospitals and is supportedby the Met Office.

� TB-38Tuesday, 10:30-12h00HH-Colombus

Hybrid Evolutionary MultiobjectiveOptimization Algorithms

Stream: Multiobjective OptimizationInvited sessionChair: Karthik Sindhya, of Mathematical Information Technology,University of Jyväskylä, Finland, [email protected]

1 - A Preference-based Evolutionary Algorithm for Multi-objective Optimization: The Weighting AchievementScalarizing Function Genetic AlgorithmAna Belen Ruiz, Applied Economics (Mathematics),University of Malaga, C/ Ejido, 6, 29071, Malaga, Spain,[email protected], Rubén Saborido Infantes, Mariano Luque

We suggest a preference-based EMO algorithm which asks the DMfor a reference point (RP) and which is based on the EMO algorithmNSGA-II and on achievement scalarizing functions (ASFs). In prac-tice, an ASF finds the closest efficient solution to the RP, and thissolution also depends on a weight vector used in the ASF. The pro-posed approach modifies the dominance criterion of NSGA-II in orderto highlight solutions closer to the RP. The classification of the pointsinto the nondominated fronts is done according to the values that eachsolution takes on an ASF, using different weight vectors.

2 - Preference-based Evolutionary Algorithm for Multi-objective Bilevel OptimizationPekka Malo, Information and Service Economy, AaltoUniversity School of Economics, Runeberginkatu 22-24,00100, Helsinki, Finland, [email protected]

We present an evolutionary framework for solving multiobjectivebilevel problems, where the upper-level has multiple objectives andlower-level has a single objective. The implementation of the algo-rithm is based on a new technique for approximating the optimal lower-level solution mapping by a class of smooth and continuous single-valued functions. The algorithm is well-motivated by the recent resultsobtained for general parametric optimization problems and set-valuedanalysis. The performance of the algorithm is demonstrated with re-spect to a collection of scalable test-problems.

3 - A Hybrid Evolutionary Multi-objective OptimizationAlgorithm for Enhanced Convergence and DiversityKarthik Sindhya, of Mathematical Information Technology,University of Jyväskylä, Finland, [email protected],Kaisa Miettinen, Kalyanmoy Deb

Evolutionary multi-objective optimization (EMO) algorithms are usedto find approximate Pareto optimal solutions, though they have slowconvergence speed. We present a modular hybrid EMO algorithm in-cluding a local search module (involving an achievement scalarizingfunction) to increase the convergence speed, a diversity enhancementmodule, etc. It can be connected with various EMO methods likeMOEA/D. Numerical tests on test problems indicate the efficacy ofour hybrid EMO algorithm, which can also be used to handle dynamicprocess simulation based multi-objective optimization problems.

� TB-39Tuesday, 10:30-12h00HH-Cousteau

MCDA: New Approaches and Applications1Stream: MCDA: New Approaches and ApplicationsInvited sessionChair: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]: Yves De Smet, SMG - CODE, Université Libre deBruxelles, Boulevard du Triomphe CP 210-01, 1050, Bruxelles,Belgium, [email protected]

1 - Monotonicity and minimax biproportional apportion-mentsPaolo Serafini, Dept. of Mathematics and Computer Science,University of Udine, Via delle Scienze 206, 33100, Udine,Italy, [email protected]

We deal with biproportional apportionment problems (BAP). On onehand we may approach BAP via the axiomatic approach where someaxioms are stated and the assigned seats must satisfy the axioms. An-other approach requires the definition of quotas and the seats must min-imize an error measure wrt the quotas. It is therefore important to un-derstand to what extent error minimization methods satisfy the mostcommon axioms. In this talk we investigate the axiom of monotonic-ity and show that minimization of the maximum absolute error withrespect to regional quotas satisfies monotonicity.

2 - Multi attribute regional market location problembased on the clustering approachMohsen Yahyayi, Shahed University, Iran, Islamic RepublicOf, [email protected], Mahdi Bashiri

Regional market location is considered in this paper; a hybrid approachis presented including 3 phases of clustering, elimination and decisionmaking. In the first phase cities are clustered with maximum homo-geneity. Then low potential points are eliminated. Finally cities areranked to be a regional market according to (1) Distance from the clus-ter median point optimized by SFLP to have less transportation cost,(2) City population, (3) Distance from major transportation corridors,(4) Tourism attractions. Our approach was applied in Iran regionalmarket using 10-means clustering.

3 - A Careful Look at Criterion Importance and WeightsPekka Korhonen, Aalto University School of Economics,00100, Helsinki, Finland, [email protected], JyrkiWallenius, Anssi Oorni

We investigate the connection between announced importance of crite-ria and weights. Our considerations are based on a simple experimentalsetting, where the subjects were asked to pairwise compare alternativeswith two criteria. We used the subjects’ responses to choices to esti-mate the weights and compared them to "direct’ (given) weights. Thecomparison was based on prediction power of a linear function. Ourresults point out that there is a reason to question the common beliefthat the values of the weights have the same rank order as criteria im-portance.

4 - Rank reversal in the PROMETHEE I and II rankings: asummary of recent investigationsYves De Smet, SMG - CODE, Université Libre de Bruxelles,Boulevard du Triomphe CP 210-01, 1050, Bruxelles,Belgium, [email protected], Céline Verly, JulienRoland

The multicriteria methods based on pairwise comparisons suffer frompossible rank reversal occurrences when the set of alternatives is modi-fied. We study this distinctive feature in the scope of the PROMETHEEI and II methods. First, empirical tests are conducted on the basis ofartificial data sets in order to quantify the likelihood of rank reversalinstances. Then conditions to avoid this phenomenon are provided.Finally, we investigate manipulation risks on a simplified version ofPROMETHEE II.

124


� TB-40Tuesday, 10:30-12h00HH-Livingstone

Preference Learning 1

Stream: Preference LearningInvited sessionChair: Krzysztof Dembczynski, Institute of Computing Science,Poznan University of Technology, Piotrowo 2, 60-965, Poznan,Poland, [email protected]: Willem Waegeman, Department of Applied Mathematics,Biometrics and Process Control, Ghent University, Coupure links653, 9000, Ghent, Belgium, [email protected]

1 - Active learning for preference learningNir Ailon, Computer Science, Technion Israel Institute ofTechnology, Taub Building, Technion City, 32000, Haifa,Israel, [email protected]

Analyzing preferences lies in the intersection of many fields, most no-tably econometrics (discrete choice theory), combinatorial optimiza-tion and machine learning theory and practice. In a typical setting, theultimate goal is to be able to predict the order of a given set of alter-natives given a set of observed, noisy pairwise comparisons. In thiswork I will discuss the adversarial noise case, and present a new ac-tive learning ERM (Empirical Risk Minimization) algorithm with analmost optimal preference query complexity and provable guarantees.

2 - Optimal recommendation setsPaolo Viappiani, Computer Science Dept, Aalborg University,selma lagerfos vej 300, 9000, Aalborg, Denmark,[email protected]

Utility elicitation is an important component of many applications,such as decision support systems. Such systems query users about theirpreferences and offer recommendations based on the system’s beliefabout the user’s utility function. We analyze the connection betweenthe problem of generating optimal recommendation sets and the prob-lem of generating optimal choice queries, considering both Bayesianand regret-based elicitation. Our results show that, somewhat surpris-ingly, under very general circumstances, the optimal recommendationset coincides with the optimal query.

3 - Large-scale training of linear ranking support vectormachinesAntti Airola, University of Turku, Finland, [email protected]

We introduce a large-scale training algorithm for the linear rankingsupport vector machine, which is one of the most widely applied meth-ods for inferring ranking models using machine learning. The methodgeneralizes the most efficient previously known method, which is ap-plicable to ordinal regression problems with a small constant numberof classes, to tasks with an arbitrary number of different levels of util-ity. This is achieved by combining a cutting plane optimization tech-nique with the use of self-balancing binary search trees. The methodeasily scales to millions of data points.

� TB-41Tuesday, 10:30-12h00HH-Heyerdahl

AHP/ANP 5Stream: Analytic Hierarchy Process, Analytic NetworkProcessInvited sessionChair: Eizo Kinoshita, Urban Science Department, MeijoUniversity, 4-3-3 Nijigaoka, 509-0261, Kani, Gifu, Japan,[email protected]

1 - Super Pairwise Comparison Matrix in MDAHP

Takao Ohya, School of Science and Engineering, KokushikanUniversity, 4-28-1 Setagaya, 154-8515, Setagaya-ku, Tokyo,Japan, [email protected], Eizo KinoshitaWe have proposed a Super Pairwise Comparison Matrix (SPCM) toexpress all pairwise comparisons in the evaluation process of the dom-inant analytic hierarchy process or MDAHP as a single pairwise com-parison matrix. This paper shows, by means of a numerical counterex-ample, that in MDAHP an evaluation value resulting from the applica-tion of LLSM to a SPCM does not necessarily coincide with that of theevaluation value resulting from the application of the GMMDAHP tothe evaluation value obtained from each pairwise comparison matrixby using the geometric mean method.

2 - Research on the comparison of electronic monetaryvalue in the information societyShunei Norikumo, General Management Department, OsakaUniversity of Commerce, 4-1-10, Mikuriyasakae-machi,577-8505, Higashiosaka, Osaka, Japan,[email protected] study considers the social influence that computerization of themoney to advance rapidly gives to an existing value of money andcompares the value of money using dominant AHP. In late years themoney changed from a form printed on paper to electronic money andvirtual money. Outwardly the money that the property was totally dif-ferent from the conventional substantial money and currency in eventhe same money came to flow on information system.

3 - An Application of Energy Alternative Selection: AFuzzy AHP-ELECTRE MethodologyAbit Balin, Industrial Engineering, Yildiz TechnicalUniversity, Yildiz Technical University, Department ofIndustrial Engineering, Besiktas, 34000, Istanbul, Turkey,[email protected], Özge Nalan Alp Bilisik, HayriBaraçlıEnergy planning problem is a very complex issue and has a lot of cri-teria that must be taken into account to solve this problem. The assess-ment criteria that will be used in this study are technical, economic, en-vironmental and social. All of the criteria must be determined and eval-uated carefully because these criteria are so critical for governments,human beings and environment. In this study, to make a decision aboutenergy planning problems we use fuzzy AHP to determine the criteriaweights and ELECTRE to determine the best energy resource.

4 - Dominant AHP as Measuring Method of ServiceValueEizo Kinoshita, Urban Science Department, Meijo University,4-3-3 Nijigaoka, 509-0261, Kani, Gifu, Japan,[email protected] author believes is surpassed by AHP, a measuring method of valueof goods in a service economy. This is because a utility function whichrepresents value is expressed based on an absolute scale (best suitedfor an industrialized society) in utility theory, whereas in AHP, it is ex-pressed based on a relative scale (best suited for a service economy).However, the author mathematically identifies a fundamental flaw ofSaaty’s AHP in this paper, and proves that Dominant AHP, proposedby Kinoshita and Nakanishi, is superior to Saaty’s AHP as a measuringmethod of value of services.

� TB-42Tuesday, 10:30-12h00BW-Amber

Networks under Uncertainty

Stream: Stochastic Modeling and Simulation in Engi-neering, Management and ScienceInvited sessionChair: Erik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39, 85577,Neubiberg, Germany, [email protected]: Sandra Rauscher, Management Science, LancasterUniversity, United Kingdom, [email protected]: Selis Onel, Chemical Engineering, Hacettepe University,Department of Chemical Engineering, Beytepe Campus, 06800,Beytepe, Ankara, Turkey, [email protected]

125


1 - Research of the methods of stochastic approxima-tion and genetic algorithms for the transport systemmicroscopic simulation models calibrationElena Yurshevich, Mathematical Methods and ModelingDepartment, Transport and Telecommunication Institute,Lomonosova str. 1, 1019, Riga, Latvia, [email protected] is considered the calibration procedure of traffic flow’s microscopicmodels as an optimization problem of the vector of the model numer-ical parameters. The problem is complicated by the absence of theanalytical model, capable to describe the relationship between the vec-tor of numerical parameters and objective function. It was consideredthe capability of the stochastic approximation and heuristic methodsapplication for calibration. The investigation was implemented usingthe transport system simulation model constructed in VISSIM packageand an aggregated data of traffic flow.

2 - Simulation Metamodeling using Bayesian NetworksKai Virtanen, Mathematics and Systems Analysis, AaltoUniversity, School of Science, Finland, [email protected] introduce a novel approach towards simulation metamodeling us-ing Bayesian networks (BNs). A BN represents the joint probabilitydistribution of a set of random variables and enables the efficient calcu-lation of their marginal and conditional distributions. A BN metamodelgives a non-parametric description for the joint probability distributionof simulation inputs and outputs as well as provides various what-ifanalyses by combining simulation data with expert knowledge. Theconstruction and utilization of BN metamodels are illustrated with asimulation example involving a queueing model.

3 - On the Peter Principle: An Agent Based Investigationinto the Consequential Effects of Social Networksand Behavioural FactorsAngelico Fetta, Maths, Cardiff University, United Kingdom,[email protected], Paul Harper, Vincent Knight,Israel Vieira, Janet WilliamsThe Peter Principle (PP) is a theory used to describe job incompetencein a hierarchical organisation. It argues that should staff be compe-tent at a given level, their competence may not be implicit at a higherlevel. Through the use of Agent Based Simulation coupled with socialnetworks and organisational thought, the PP is assessed under previ-ously unexplored conditions. Results show that the PP may not be asdetrimental to organisational efficiency as previously thought, whilepromotion on merit offered the best efficiency trade off in the absenceof clear evidence of the PP paradigm.

4 - Hybrid lateral transshipments in multi-item inventorynetworksSandra Rauscher, Management Science, Lancaster University,United Kingdom, [email protected], KevinGlazebrook, Colin Paterson, Thomas ArchibaldWe present a new approach for addressing shortages in multi-locationinventory systems facing multivariate, non-homogeneous Poisson de-mand. A hybrid lateral transshipment policy allows rebalancing of in-ventory levels along with reacting to stock-outs. We derive this byapplying a dynamic programming policy improvement step to a no-transshipment policy. Shipment costs are modelled with a knapsack-like structure to accommodate different item types. We carry out an ex-tensive simulation study showing the benefit of modelling multi-itemtransshipments against using single item models in parallel.

� TB-43Tuesday, 10:30-12h00BW-Granite

Recent Applications of ProbabilisticProgramming

Stream: Stochastic ProgrammingInvited sessionChair: Milos Kopa, Department of Probability and MathematicalStatistics, Charles University in Prague, Faculty of Mathematicsand Physics, Sokolovská 83, CZ 186 75, Prague, Czech Republic,[email protected]

1 - Robustness in stochastic programs with the first or-der stochastic dominance and/or probabilistic con-straintsMilos Kopa, Department of Probability and MathematicalStatistics, Charles University in Prague, Faculty ofMathematics and Physics, Sokolovská 83, CZ 186 75, Prague,Czech Republic, [email protected], Jitka DupacovaThe paper presents robustness results for stochastic programs with thefirst order stochastic dominance and probabilistic constraints. Due totheir frequently observed lack of convexity and/or smoothness, theseprograms are rather demanding both from the computational and ro-bustness point of view. Under suitable conditions on the structure ofthe problem and for discrete distributions we suggest a new robust port-folio efficiency test with respect to the first order stochastic dominance.We exploit the contamination technique to analyze the resistance withrespect to additional scenarios.

2 - Chance constrained problems: reformulation usingpenalty functions and sample approximation tech-niqueMartin Branda, Department of probability and mathematicalstatistics, Charles University in Prague, Ke Karlovu 3, 12116,Prague 2, Czech Republic, [email protected] compare basic formulations of stochastic programming problemswith chance constraints, generalized integrated chance constraints andpenalties in the objective function. It is shown that the problemsare asymptotically equivalent. We discuss solving the problems us-ing sample approximation technique. Rates of convergence and esti-mates on the sample size are derived for problems with mixed-integerbounded sets and/or several constraints. We present applications to aninvestment problem with real features and to a vehicle routing problemwith time windows, random travel times and demand.

3 - Robust Pricing of Monopolistic Cloud ComputingServices with Service Level AgreementsVladimir Roitch, Department of Computing, Imperial CollegeLondon, 180 Queen’s Gate, SW7 2AZ, London, UnitedKingdom, [email protected], Daniel Kuhn, Yike GuoWe address the optimal pricing of cloud computing services from theperspective of a monopolistic service provider that needs to managedemand responsiveness and uncertainty. We formulate the pricingproblem for on-demand based services as a multi-stage stochastic pro-gram and model service level agreements via chance constraints. Un-der weak assumptions about the demand uncertainty we show that theresulting model can be reduced to an equivalent two-stage stochasticprogram. We state various new results. Finally, we investigate thepossibility of selling unused capacity on a spot market.

4 - Tractable Dynamic Model of Price and Volume GivenUninformed Market MakerMartin Smid, Econometrics, UTIA, Czech Academy ofSciences, Pod Vodarenskou vezi 4, CZ 182 08, Praha 8, CzechRepublic, [email protected] construct a simple yet realistic model of a market with a marketmaker, maximizing the expected consumption while keeping the prob-ability of bankruptcy small. The market maker estimates the fair priceby means of a proxy. The order sizes are random. We show our modelto conform several stylized facts. Further, we describe a dynamics ofthe price and the traded volume and we show that the price volatilitycan be decomposed into the part caused by the movements of the fairprice, the part due to the order arrivals and the part due to the uncer-tainty.

� TB-44Tuesday, 10:30-12h00BW-Marble

Computer Vision (2) and MachineLearning

Stream: Machine Learning and its ApplicationsInvited sessionChair: Teofilo deCampos, Center for Vision Speech and SignalProcessing, University of Surrey, GU2 7XH, Surrey, UnitedKingdom, [email protected]

126


Chair: Fei Yan, Center for Vision Speech and Signal Processing,University of Surrey, GU2 7XH, Surrey, United Kingdom,[email protected]: Ivan Reyer, Dorodnicyn Computing Centre of RAS, ul.Pererva d. 58, kv. 153, 109451, Moscow, Russian Federation,[email protected]

1 - Efficient Large Scale Scene UnderstandingRamin Zabih, Cornell, Computer Science, 4130 Upson Hall,14853, Ithaca, New York, United States, [email protected],Philip Torr

Image understanding can be thoughts of a discrete labelling problemin which the goal is to decide a per pixel label. Labels might be suchthings as depth or object labels. This problem is very large scale dis-crete optimization, with millions of labels needing to be determined inthe course of a video. In this talks I will discuss some tricks for makingthis process efficient.

2 - Skeleton Base and Scalable Boundary-SkeletalShape ModelIvan Reyer, Dorodnicyn Computing Centre of RAS, ul.Pererva d. 58, kv. 153, 109451, Moscow, Russian Federation,[email protected], Ksenia Zhukova

A scalable shape model of an image is presented. It uses a polygonalfigure approximating a raster object and its skeleton base. The skele-ton base is a subset of the figure’s skeleton. It defines a closed planardomain (skeleton core) describing possible position of a skeleton’s es-sential part for any region with Hausdorff deviation from the figure notexceeding the given value. The skeleton base changes steadily and con-tinuously as the admissible deviation increases. The family of skeletonbases defines a scalable boundary-skeletal model reflecting shape fea-tures at various levels of detail.

3 - Structured Output Learning for Automatic Analysisof Court GamesFei Yan, Center for Vision Speech and Signal Processing,University of Surrey, GU2 7XH, Surrey, United Kingdom,[email protected], Josef Kittler

We investigate the application of structured output learning (SOL) inautomatic analysis of court games. We formulate the problem as one oflearning a mapping from observations to structured labels, where thelabels are event sequences, and employ max-margin Markov networksto achieve a max-margin solution in the joint feature space. We com-pare closely the more popular generative approach based on a hiddenMarkov model (HMM) with our discriminative approach, and demon-strate that our approach leads to an improved performance in real worldtennis video analysis.

� TB-45Tuesday, 10:30-12h00BW-Water

Logistics and stochastics 1

Stream: Stochastic Modelling in Logistical NetworksInvited sessionChair: Birger Raa, Mgt Information Science & Operations Mgt,Ghent University, Tweekerkenstraat 2, 9000, Gent, Belgium,[email protected]: Wout Dullaert, Faculty of Economics and BusinessAdministration, VU University Amsterdam, De Boelelaan 1105,1081 HV, Amsterdam, Netherlands, [email protected]

1 - The effect of capacitated replenishments on periodicreview inventory systemsWout Dullaert, Faculty of Economics and BusinessAdministration, VU University Amsterdam, De Boelelaan1105, 1081 HV, Amsterdam, Netherlands,[email protected], Birger Raa

We study periodic review inventory systems with lost sales and ca-pacitated replenishments. This capacity restriction implies that the in-ventory position may not reach the order-up-to level when placing anorder, such that additional safety stock is needed to reach the same ser-vice level. To determine the required level of safety stock, a genericiterative procedure based on Markov chain analysis is proposed. Com-putational experiments illustrate the impact of a limited replenishmentcapacity on the required safety stock level, and the effectiveness of theproposed iterative calculation method.

2 - A stochastic model for controlling the deteriorationof a two machines production system: Extensions toremanufacturingPierre Dejax, IRCCyN, Ecole des Mines de Nantes, LaChantrerie, 4, rue Alfred Kastler, BP 20722, 44307, Nantescedex 3, France, [email protected], Annie FrancieKouedeu, Jean-Pierre Kenne, Victor Songmene

A deteriorating production system consisting of two parallel machinesis investigated. The goal is to find the production rates of machinesso as to minimize a long term average expected cost. The failure rateof one machine depends on its production rate. The paper proposesa stochastic programming formulation of the problem and derives theoptimal production policies numerically. Control policy parameters areobtained by simulation. Sensitivity analyses of system parameters areconducted to validate the proposed methodology. We finally presentan extension to reverse logistics.

3 - A stochastic formulation of the disassembly line bal-ancing problemMohand Lounes Bentaha, Industrial Engineering andInformatics, Ecole des Mines de Saint-Etienne, 158 CoursFauriel, 42023, Saint-Etienne, France, [email protected], OlgaBattaïa, Alexandre Dolgui, Frédéric Grimaud

The disassembly line balancing problem is studied under uncertainty.Disassembly task times are assumed random variables with knownprobability distributions. An AND/OR graph is used to model theprecedence relations among tasks. The goal is to assign the disassem-bly tasks to workstations while respecting precedence and cycle timeconstraints. The objective is to minimize the total line cost includingthe incompletion cost arising from tasks not completed within the cy-cle time. A stochastic linear mixed integer programming formulationis developed.

4 - Quantification of value of flexibility in a hybrid MTOand MTS systemBeyazıt Ocaktan, Industrial Engineering, Sakarya University,Balikesir University,Engineering Faculty, IndustrialEngineering Department M 5211, 10100, Balikesir, Turkey,[email protected], Ufuk Kula

We consider a production system which produces products for make-to-order and for make-to-stock. Our objective is to quantify the valueof flexible in this system. To achieve this, a production environmentconsisting of 2 parallel machines is considered. Machine 1 producesproduct 1 only. The other machine produces product 1 and product 2.Product 2 is produced to order. We formulate this problem as an MDPand use approx. D.P. to determine approximately optimal control pol-icy of such system and quantify the value of flexibility, and provideinsight on when such flexibility is most beneficial.

127

TC-01 EURO 25 - Vilnius 2012

Tuesday, 12:30-14:00

� TC-01Tuesday, 12:30-14:00RB-Alfa

Keynote Lecture: Professor BjornJorgensen

Stream: Keynote, Tutorial and Plenary LecturesKeynote sessionChair: Peter Letmathe, Faculty of Business and Economics,RWTH Aachen University, Chair of Management Accounting,Templergraben 64, 52062, Aachen, Germany,[email protected]

1 - AccountingBjorn Jorgensen, Accounting and Business Law, University ofColorado at Boulder, UCB 419, 995 Regent Drive, 80309,Boulder, CO, United States, [email protected]

This presentation uses optimization and stochastic processes to pro-vide a formal introduction to disclosure, financial accounting, and val-uation. Financial accounting produces balance sheets, income state-ments, and cash flow statements relying on two basic principles: Dou-ble entry bookkeeping and the accrual basis of accounting. Disclo-sure of financial accounting statements is today mandatory for publiclytraded firms. A single-period benchmark setting is initially used to es-tablish what disclosures would maximize firm value in the absence ofmandatory disclosure. Second, in a multi-period setting, the so-calledclean surplus relation ties together consecutive balance sheets throughcomprehensive income from the income statement. Third, under theaccrual basis of accounting, income decomposes into cash flows andaccruals. Since accruals arise from the non-cash component of incomesuch as depreciation, accruals must reverse over time in the sense thatthey add up to zero over the complete life cycle of the firm. I proposeto model accruals as a moving average time-series process with a unitroot. This modeling approach has implications for how market valuesrelate to earnings, cash flows, and balance sheet book values.

� TC-02Tuesday, 12:30-14:00RB-Beta

Scheduling in Robotics andManufacturing

Stream: SchedulingInvited sessionChair: Seda Hezer, Industrial Engineering, Selcuk University,Department of Industrial Engineering, 420031, Konya, Turkey,[email protected]

1 - The Job Shop with Blocking and a Rail-Bound Trans-fer SystemReinhard Bürgy, Dept of Informatics, University of Fribourg,Bd de Pérolles 90, 1700, Fribourg, Switzerland,[email protected], Heinz Gröflin

Considered is the following Job Shop with no buffers. Jobs are pro-cessed on machines and transferred from one machine to the next byrobots (or cranes) moving on a single rail. The robots do not passeach other, but can "move out of the way’. The objective is to findthe starting times for all operations, the assigned robots for all trans-fer operations and the trajectory of each robot, in order to minimizemakespan. We give a disjunctive graph formulation and use the job in-sertion (sub-) problem as a mechanism for local search. Computationalresults support the value of the approach.

2 - A special case of the minimization of tool switchesproblemHoracio Yanasse, LAC, INPE, Av. dos Astronautas 1758, CP515 - INPE/CTE, 12227-010, São José dos Campos, SP,Brazil, [email protected]

Consider the case where the MOSP graph corresponding to a mini-mization of tool switches problem (MTSP) has some cutting vertex.If we remove this vertex and all edges incident to it, an optimal solu-tion for the original problem can be built from optimal solutions of theconnected components of the resulting graph. We can use this resultto develop polynomial algorithms for special cases of the MTSP whenthe corresponding MOSP graph has particular topologies.

3 - Two sided disassembly line balancing problemSeda Hezer, Industrial Engineering, Selcuk University,Department of Industrial Engineering, 420031, Konya,Turkey, [email protected], Yakup Kara

The disassembly line balancing problem (DLBP) can be stated as theassignment of disassembly tasks to workstations such that all the prece-dence relations are satisfied and some measure of effectiveness is op-timized. In this study a mathematical model is proposed for two sideddiasssembly line balancing problem (TSDLBP) in which product flowsalong a production line with workstations on both sides of the line. Tothe best of our knowledge, the TSDLBP has not been applied to theDLBP and so this is the first study about TSDLBP. The model’s per-formance is evaluated using test problems.

4 - Scheduling activities in a research centre to mini-mize the total energy consumptionMaciej Lichtenstein, Institute of Computer Engineering,Control and Robotics, Wroclaw University of Technology,Janiszewskiego 11/17, 50-372, Wroclaw, Poland,[email protected], Adam Janiak, TomaszKrysiak, Mateusz Gorczyca

We study the problem of scheduling activities in a research centre withthe objective of the total electrical power consumption minimization .The considered research center is powered by some number of "greenenergy’ power plants and the activities conducted in that center dealswith the research on those power plants. There are given a time char-acteristics of the power plants efficiency, and the energy demands ofall activities. We present the precise model of the considered researchcenter in terms of scheduling activities, which minimizes the total en-ergy consumption over time.

� TC-03Tuesday, 12:30-14:00RB-L1

Sustainable Network Design andRailway/Sea Transport

Stream: Sustainable Transport PlanningInvited sessionChair: Joachim R. Daduna, Hochschule für Wirtschaft und RechtBerlin Berlin, Badensche Str. 52, 10825, Berlin, Germany,[email protected]

1 - Road-Rail-Sea hub location-allocation with sustain-ability considerationsAli Fattahi, Industrial Engineering, Koc University,Rumelifeneri, Sariyer, 34450, Istanbul, Istanbul, Turkey,[email protected], Metin Turkay

Integration of different transportation modes in geographical regionsoffers a number of advantages with respect to cost, environmentalimpact and some social considerations. Important decisions in thisintegration include hub location and flow allocation. Although thisproblem is very important, it has not been examined with sustainabil-ity considerations previously. In this paper, the design of intermodaltransportation is expressed as a MILP problem including sustainabilityconsiderations. The accuracy and efficiency of the model is tested ona number of instances derived from real data.

128

EURO 25 - Vilnius 2012 TC-05

2 - Green supplier network design based on customersegmentationLeyla Ozgur, Industry Engineering, Pamukkale University,Pamukkale Üniversitesi Kampüsü Rektörlük Binası BilgiIslem Daire Baskanlıgı, 20070, Denizli, Turkey, Turkey,[email protected], Semih Coskun, Askiner Gungor, OlcayPolat

Customer awareness on environmental issues has increased signifi-cantly for the last couple of decades. Despite good intentions, thereis a huge variation of demand for green products among customers inthe market. For this reason, there is a segmentation requirement forcustomers’ green purchasing trends. Green companies therefore needto redesign their green supplier networks (GSN) according to customersegments. In this study, a mathematical model is proposed to designa GSN based on customer segmentation by including the importantcriteria of production and marketing in green supply chain.

3 - Designing efficient transportation processes for sus-tainable city logisticsJoachim R. Daduna, Hochschule für Wirtschaft und RechtBerlin Berlin, Badensche Str. 52, 10825, Berlin, Germany,[email protected]

Discussions about City Logistics are mostly concerned with freighttransport in urban areas. However, passenger traffic is not taken intoaccount, even though it is essential that customers are able to reach re-tail locations to make purchases. The objective is therefore to presenta concept of a Passenger Traffic and Logistics Service Center thatstresses the accessibility of inner-city shopping areas by customers.Here the underlying structures include the reduction of traffic in ur-ban areas by reducing motorized individual traffic as well as deliveryprocesses in the commercial sector.

� TC-04Tuesday, 12:30-14:00RB-L2

Optimization problems on graphs andnetworksStream: Combinatorial OptimizationInvited sessionChair: Ivana Ljubic, Department of Statistics and OperationsResearch, University of Vienna, Bruennerstr. 72, 1210, Vienna,Austria, [email protected]

1 - Minimum Spanning Trees with Generalized DegreeConstraintsPedro Moura, DEIO, Faculty of Sciences University ofLisbon, Cidade Universitária, Campo Grande, 1749-016,Lisbon, Lisbon, Portugal, [email protected], Luis Gouveia,Amaro de Sousa

We describe a MST problem which arises in the design of wirelessnetworks. In a wireless link, the signal strength on the receiver sidedecreases with the distance to the transmitter side. To guarantee thatthe interference on the receiver side is under a given threshold, weimpose a maximum amount of interference caused by the frequencychannels used by the other links on both end nodes of the link. Dif-ferent types of links may be installed between two nodes dependingon the distance between them and their degrees. We introduce threemodels and compare the corresponding LP relaxations.

2 - Layered Graph Approaches for the Hop ConstrainedSteiner Tree Problem with multiple Root NodesMarkus Leitner, Institute of Computer Graphics andAlgorithms, Vienna University of Technology, Favoritenstraße9-11, 1040, Vienna, Austria, [email protected], LuisGouveia, Ivana Ljubic

We consider the following generalization of the hop constrainedSteiner tree problem: Given an edge-weighted undirected graph whosenodes are partitioned into root nodes, terminals and Steiner nodes, finda minimum-weight subtree that spans all the roots and terminals so that

the number of hops between each terminal and an arbitrary root doesnot exceed a given hop limit H. In this work, we show how to model theproblem over layered graphs and introduce a branch-and-cut approachfor solving this model. Furthermore, comparisons to previously pro-posed flow- and path-based models are given.

3 - On a time-dependent formulation for the travellingsalesman problemLuis Gouveia, DEIO, University of Lisbon, Campo Grande,Bloco C6, 1749-016, Lisbon, Portugal, [email protected],Maria Teresa Godinho, Pierre Pesneau

We present an updated classification of formulations for the asymmet-ric travelling salesman problem (ATSP) where we contextualize a newtime-dependent formulation presented in Godinho et al (2010). Thenew formulation is an improved version of the Picard and Queyranne(PQ) formulation. The PQ formulation uses, as a subproblem, an exactdescription of the n-circuit problem. The new formulation uses morevariables and is based on using, for each node, a stronger subproblem,namely a n-circuit subproblem with the additional constraint that thecorresponding node is not repeated in the circuit.

4 - The Recoverable Robust Two-Level Network DesignProblemEduardo Álvarez-Miranda, Dipartimento di Elettronica,Informatica e Sistemistica, Università di Bologna, Viale delRisorgimento 2, 40136, Bologna, BO, Italy,[email protected], Ivana Ljubic, S. Raghavan, Paolo Toth

We consider a problem modeled as the two level network design prob-lem under uncertainty. One of two technologies can be installed oneach edge and all nodes of the network need to be served by at leastthe secondary one. There is uncertainty regarding the set of primarycustomers; only a discrete set of realizations of them is available. Thenetwork is built in two stages. We model this problem using recover-able robustness, we provide mixed integer programming models anduse a branch & cut algorithm to solve it.

� TC-05Tuesday, 12:30-14:00RB-L3

Problems on graphs II

Stream: Discrete Optimization, Geometry & Graphs(contributed)Contributed sessionChair: Victor Lepin, Institute of Mathematics, National Academyof Sciences of Belarus, 11 Surganova str, 220072, Minsk, Belarus,[email protected]

1 - Extremal graphs for the randic index when minimum,maximum degree and order of graphs are oddLjiljana Pavlovic, Department of Mathematics, Faculty ofNatural Sciences and Mathematics, Radoja Domanovica 12,34000, Kragujevac, Serbia, [email protected], TomicaDivnic

Let G(k,m,n) be the set of connected simple n vertex graphs with min-imum vertex degree k and maximum vertex degree m. The Randicindex of G is the sum of d(u)d(v) raised to the power of -1/2, whered(u) is the degree of vertex u and the summation extends over all edgesuv of G. We find for k greater or equal to n/2, and k, m, n are odd,extremal graphs for which this index attains its minimum value. Weshow that the extremal graphs have only vertices of degree k, m andm-1, the number of vertices of degree m-1 is one and the number ofvertices of degree k is as close to n/2 as possible.

2 - Disjoint Path Problem in Mixed Acyclic GraphsCong Zhang, Graduate School of Informatics, KyotoUniversity, sakyoku, higashitakeyamachi, 606-8393, kyoto,kyoto, Japan, [email protected], Hiroshi Nagamochi

129


The disjoint path problem asks to find a set of vertex-disjoint paths be-tween k pairs of source and sink vertices. It is known that the problemis NP-hard in a general undirected/directed graph, and the problemwith a fixed k admits a polynomial time algorithm for an undirectedgraph or a DAG (directed acyclic graph). In this talk, we show a com-mon generalization of these two results; i.e., the problem with a fixed kcan be solved in polynomial time when a mixed graph becomes a DAGby contracting each component consisting of undirected edges into asingle vertex.

3 - Full cycle extendability of K(1,4)-restricted locallyconnected graphsPavel Irzhavski, Department of Discrete Mathematics andAlgorithmics, Faculty of Applied Mathematics and ComputerScience, Belarusian State University, Nezalezhnastsi ave., 4,220030, Minsk, Belarus, [email protected], Yury OrlovichA graph G is K(1,p)-restricted if, for any subgraph H of G which is iso-morphic to the complete bipartite graph K(1,p), the number of edges ofthe subgraph induced in G by all the vertices of H is at least p+(p-2).We prove that a connected locally connected K(1,4)-restricted graphon at least three vertices is either fully cycle extendable or isomorphicto one of five exceptional (non-Hamiltonian) graphs. This result gen-eralizes several previous results on the existence of Hamilton cycles ingraphs.

4 - Biclique partition for graphs with special blocksVictor Lepin, Institute of Mathematics, National Academy ofSciences of Belarus, 11 Surganova str, 220072, Minsk,Belarus, [email protected], Oleg DuginovThe biclique partition problem is to find a minimum number of edge-disjoint bicliques (complete bipartite subgraphs) that cover all edgesof a given graph. We present a linear-time algorithm for the bicliquepartition problem in graphs whose blocks are complete graphs, cycles,complete bipartite graphs, special bipartite graphs or wheels.

� TC-06Tuesday, 12:30-14:00RB-Gamma

Rich vehicle routing problems

Stream: MetaheuristicsInvited sessionChair: Jose Brandao, Management, University of Minho, Largo doPaço, 4704 -553, Braga, Portugal, [email protected]

1 - The Pickup and Delivery Problem with Cross-Docking OpportunitiesHanne Petersen, Department of Transport, TechnicalUniversity of Denmark, Denmark, [email protected],Stefan RopkeWe present a pickup and delivery problem, where a number of requestswith time windows must be served. Each request consists of a pickupand/or a delivery, and the aim is to construct a set of feasible vehicleroutes at the lowest possible cost. Several cross-docks are availableto help carry out the transportation operations, allowing for transfer ofgoods between vehicles and intermediate storage, and thus one requestcan be served by one or multiple vehicles. The problem is solved us-ing a Parallel Adaptive Large Neighbourhood Search, on real-life andgenerated data.

2 - Metaheuristics for the vehicle routing problem withbackhauls and soft time windowsJose Brandao, Management, University of Minho, Largo doPaço, 4704 -553, Braga, Portugal, [email protected] vehicle routing problem with backhauls and soft time windows(VRPBSTW) contains two distinct sets of customers: those that re-ceive goods from the depot and those that send goods to the depot. Toeach customer is associated an interval of time (time window), duringwhich each one should be served. This time window is called soft be-cause it can be violated, but this represents an additional cost. We solvethe VRPBSTW using iterated local search and iterated tabu search.The performance of these metaheuristics is tested using a large set ofbenchmark problems from the literature.

3 - A Multi-Objective Algorithm for Open Vehicle Rout-ing Problems. A Real Case Study for a Spanish Com-pany with time constraintsAna Dolores López Sánchez, Economía, MétodosCuantitativos e Historia Económica, Universidad Pablo deOlavide, Carretera de Utrera, km 1, 41013, Sevilla, Spain,[email protected], Alfredo G. Hernandez-Diaz, Julian Molina,Rafael Caballero, Macarena Tejada

Vehicle Routing Problems (VRP) arise in transportation problems op-timizing one or several objectives in conflict. We will focus on OpenVehicle Routing Problems with a homogeneous fleet and three objec-tives (total distance, maximum travel time and number of routes). Wepropose a multiobjective local search based algorithm capable of ob-taining efficient solutions in reasonable times. A real case study for aSpanish Company concerning the collection of workers at their homesis also considered. Results demonstrate the validity of the proposedalgorithm compared to current routes.

4 - Tabu Search and Genetic Algorithms for the solu-tion of the Capacitated Vehicle Routing Problem withTime WindowsArthur Gomez, University of Vale do Rio dos Sinos, AVunisinos 950, São Leopoldo, Brazil, Av. Joao Satt 25 apt802A, 91390364, porto alegre, rs, Brazil,[email protected], Cristiano Galafassi, LeonardoChiwiacowsky

This paper addresses the Vehicle Routing Problem with constraints re-lated to the vehicle capacity and time windows for customer service.The problem is solved by a Hybrid Algorithm based on Tabu Searchand Genetic Algorithms metaheuristics. The algorithm performanceis evaluated assuming different values for the metaheuristics parame-ters. Finally, the results are compared with the best heuristic and exactmethods results found in the literature. The Hybrid Algorithm hereproposed shows efficiency and robustness, with several optimal solu-tions achieved.

� TC-07Tuesday, 12:30-14:00RB-Eta

Warehouse Operations Optimization

Stream: Vehicle Routing and Logistics OptimizationInvited sessionChair: Hasan Akyer, Industrial Engineering, PamukkaleUniversity, Pamukkale University Engineering Faculty, IndustrialEngineering Department, 20070, Denizli, Turkey,[email protected]

1 - Development of a systematic method to evaluatepackaging areasJudith Weiblen, Institute for Material Handling and Logistics,KIT, 76131, Karlsruhe, Germany, [email protected], DominikBerbig, Kai Furmans

Consolidation & packaging areas cause 26% of total costs in a distri-bution center. Due to a big rate of manual jobs, it is the most importantprocess after storage & picking. But there is hardly any literature con-cerning packaging areas. Thus we present both a systematic methodto structure and evaluate packaging areas including a KPI-System andbest-practices to design such areas. Above all, a market study of tech-nical realization is shown. These results help to make tendering sim-pler,tenders can be compared better and monitoring and evaluation ofpacking areas gets possible.

130


2 - Optimal number of vehicles and paths of movementin GridFlow systems with AGVsMelanie Schwab, Material Handling and Logistics, KIT,Karlsruhe, Germany, [email protected]

The new concept GridFlow is an automatic materialflow system tobuffer, sort and store pallets. The pallets are arrangned in a grid with atleast one empty location and each can be moved in four directions, if noother pallet is blocking it. Pallets are moved by AGVs which drive un-derneath the pallets. Sequential movement of pallets allows to retrievepallets that can not be accessed directly. As investment for vehicles isthe main cost factor in the system, the optimal number of vehicles andpath of movement is derived from a mixed integer optimization modelwhich shall be presented.

3 - Optimization of the order-picking processes in thewarehouseMichal Jakubiak, Logistics, Wroclaw University ofEconomics, Komandorska 118/120, 50-345, Wrocław,Poland, [email protected]

The order picking problems in a warehouse are an important aspect inimproving the productivity of modern logistic hubs. According to vari-ous estimates, the costs associated with order picking in the warehouseare from 55 - 65% of the total material handling costs. The author, us-ing simulation tools, examines what factors influence the selection ofmoving methods during the process of order picking in the warehouse.

4 - A novel heuristic approach for the joint order batch-ing and picker routing problems in warehousesHasan Akyer, Industrial Engineering, Pamukkale University,Pamukkale University Engineering Faculty, IndustrialEngineering Department, 20070, Denizli, Turkey,[email protected], Osman Kulak, Mustafa Egemen Taner

A novel heuristic approach is proposed to solve order batching andpicker routing problems jointly for multiple cross aisles warehousesystems. The clustering algorithm integrated with GAs is developedto provide fast and effective solutions for the order batching problem.We adjust the routing problem of pickers as a TSP and propose efficientTSP heuristics. Numerical experiments are carried out to evaluate theperformances of the proposed methods. GANN comes out to be themost attractive method in terms of solution quality and computationalefficiency.

� TC-08Tuesday, 12:30-14:00RB-Epsilon

Applications

Stream: Supply Chain PlanningInvited sessionChair: Mikael Rönnqvist, Département de génie mécanique,Université Laval, G1V 0A6, Québec, Canada,[email protected]

1 - Mathematical programming models for master plan-ning in ceramic tile supply chains. Evaluation andcomparison of distributed and centralised scenarios.David Pérez Perales, Organización de Empresas, (CIGIP),Universidad Politécnica de Valencia, Camino de Vera, 46022,Valencia, Spain, [email protected], Mareva Alemany

This paper deals with the replenishment, production, and distributionproblem in a ceramic tile supply chain. A distributed decision-makingapproach made up of two decisions centers is initially considered in amulti-item, multi-facility and multi-level context. For each one, a de-terministic MIP model for the multi-period master planning aiming tomaximize their own local profits is proposed. Then, an ideal central-ized approach is also evaluated as a benchmark. An example based onmodifications of real-world industrial problems is presented to evaluateand compare both scenarios.

2 - A Mixed Integer Programming Model for Shelf Life In-tegrated Planning, Scheduling and Distribution in Yo-ghurt ProductionÇagrı Sel, Department of Industrial Engineering, Dokuz EylulUniversity, Dokuz Eylul University, Department of IndustrialEngineering, Buca, 35160 Izmir, Turkey, 35160, Izmir,Turkey, [email protected], Bilge Bilgen

In this study, we deal with a supply chain problem on yoghurt pro-duction to make production and distribution decisions in an integratedway and focus not only on the packaging stage, but also the fermen-tation/incubation stage. We introduce a new multi-echelon, multi-period and multi-objective mixed integer linear programming model.Sequence-dependent times and costs, perishability constraints, shelflife, labor and working time restrictions are explicitly taken into ac-count. As a result of the study, optimal solutions are obtained within areasonable computational time.

3 - A production-inventory problem in pharmaceuticalsupply chain — mixed integer approachPawel Hanczar, Department of Logistics, University ofEconomics, ul. Komandorska 118/122, 54-608, Wroclaw,Poland, [email protected]

The paper considers the production inventory problem in pharmaceu-tical supply chains. First the structure of the supply chain being an-alyzed is presented and the production process is briefly described.Three optimization models were taken into account in the main part ofthe research. The first two models were used to ensure (only if possi-ble) the feasibility of production and to minimize setup times for oneperiod. The third model takes into account all periods on the planninghorizon. The paper finishes with the presentation of results from a reallife application of the presented models.

4 - Use of Lagrangian decomposition in coordination ofrefinery production and sales planningJiehong Kong, Department of Finance and ManagementScience, NHH Norwegian School of Economics, Helleveien30, 5045, Bergen, Norway, [email protected], MikaelRönnqvist

The integrated planning model in refinery operation is a highly nonlin-ear problem. The standard decoupled approach is to divide the probleminto two subproblems representing refinery production and sales plan-ning. Information such as approximate component and product pricesare used to coordinate the solutions. To converge to a solution, re-strictions on the volumes are also needed. In this paper we proposea decomposition scheme based on LaGrange decomposition. It gener-ates solutions that are more stable than the existing approaches withoutrequiring additional volume constraints.

� TC-09Tuesday, 12:30-14:00RB-Zeta

Colourings, Independence andAssignment

Stream: Graphs and NetworksInvited sessionChair: Reinhardt Euler, Informatique, Université de Brest, 20 av.Le Gorgeu, B P 817, 29285, Brest, France,[email protected]

1 - Lessons for finding Mutually Orthogonal LatinSquares (MOLS) for n=10 from new and old proofsfor infeasibility of n=6Gautam Appa, Operational Research, London School ofEconomics, Houghton Street, WC2A 2AE, London, UnitedKingdom, [email protected], Reinhardt Euler, AnastasiaKouvela, Dimitris Magos, Yiannis Mourtos

131


A Latin Square of size n is an nxn matrix with numbers 1 to n appearingonce in each row & column. Two Latin Squares are mutually orthog-onal if each of the n square pairs of numbers (s,t) where s,t range in1,...,n, occur exactly once in the n square cells. It is known that theredoes not exist an orthogonal pair for n=6. We review the proofs ofthis by Tarry(1901), Fisher&Yates (1934), Yamomoto (1954), Stinson(1984), Dougherty (1994), Appa&Magos&Mourtos (2004) and extendthe work of Mann (1944). The purpose is to see what can be useful indeciding whether there exist 3 MOLS of size n=10.

2 - On the recognition of 2-row orthogonal Latin rectan-gle circuitsAnastasia Kouvela, Management Science Group, LondonSchool of Economics and Political Science, 29 Abercorn Pl,Flat 171, NW89DU, London, United Kingdom,[email protected], Gautam Appa, Reinhardt Euler,Dimitris Magos, Yiannis Mourtos

This study focuses on the completabilty of incomplete pairs of orthog-onal Latin squares to identify all reasons that forbid it. We present thisproblem as a 3-index assignment problem and as a first step give thecomplete clutter of 2-row orthogonal Latin rectangle circuits. Theseare minimal structures that forbid completability. To identify them inany incomplete pair of orthogonal Latin squares we present a polyno-mial time recognition algorithm.

3 - Polyhedral aspects of cardinality constraintsYiannis Mourtos, Management Science & Technology,Athens University of Economics & Business, Greece,[email protected]

A cardinality constraint states that every value of a discrete set must bereceived by a minimum and a maximum number of variables involvedin that constraint. Adopting an IP perspective, we study the polytopedefined as the convex hull of vectors satisfying multiple cardinalityconstraints. We provide families of facets and efficient separation al-gorithms and establish a convex hull relaxation for the case of 2 con-straints. We discuss the relationship of multiple cardinality constraintsto some generalizations of graph colouring.

4 - Approximating the multi-level bottleneck assignmentproblemFrits Spieksma, Operations Research and Business Statistics,Katholieke Universiteit Leuven, Naamsestraat 69, B-3000,Leuven, Belgium, [email protected],Trivikram Dokka, Anastasia Kouvela

We consider the multi-level bottleneck assignment problem (MBA).This problem is described in the recent book "Assignment Problems"by Burkard et al. (2009) on pages 188 - 189. One of the applicationsdescribed there concerns bus driver scheduling. We view the problemas a special case of a bottleneck m-dimensional assignment problem.We give approximation algorithms and inapproximability results, de-pending upon the completeness of the underlying graph.

� TC-10Tuesday, 12:30-14:00RB-Theta

Pricing and Applications

Stream: Production Management & Supply ChainManagement (contributed)Contributed sessionChair: M. Güray Güler, Industrial Engineering, BogaziciUniversity, Turkey, [email protected]

1 - A pricing problem in a two-ends model under uncer-taintyAlberto A. Álvarez-López, Quantitative Applied EconomicsII, UNED (Spanish National University of DistanceEducation), Paseo Senda del Rey, 11, 28040, Madrid, Spain,[email protected], Inmaculada Rodríguez-Puerta

An intermediary firm has to buy some quantity of an output directlyfrom the producer. But the producer has two possible ends to allo-cate its production between them, one end being certain and the otherone being uncertain. The problem of the intermediary firm is a pric-ing problem: that of determining the price in the certain end in orderto get exactly the desired amount of output from the producer. In thispaper, we model a situation of this kind, and study some aspects of thebehavior both of the producer and of the intermediary firm.

2 - Integrated Pricing and Scheduling for Order Out-sourcing in a Supply NetworkWoldemar Wedel, Business Information Systems andOperations Research, University of Kaiserslautern, 8050,Zurich, Switzerland, [email protected], Yike Hu, OliverWendt

Our work analyzes various pricing strategies and multiattribute dy-namic pricing mechanisms for the outsourcing of manufacturing andlogistics service orders in a supply network. Via multi-agent sys-tem simulation we investigate how prices should depend on resourceavailability, the negotiable contract execution time, the cost structureand willingness to pay of the network participants and their marketstructure. We compare the performance of cost-, competition-, andmarket/demand-oriented pricing strategies for determining an adequatelower price bound as a basis for interactive pricing.

3 - On Coordination under Random Yield and RandomDemandM. Güray Güler, Industrial Engineering, Bogazici University,Turkey, [email protected], Taner Bilgic

The aim of this study is to analyze coordination in a supply chain withrandom yield and random demand. We study five well known contractsfrom the literature and show that all but the wholesale price contractare able to coordinate the chain under voluntary compliance. We alsoprovide a sensitivity analysis on the optimal policy parameters.

4 - Layout Planning and Shelf Space Allocation in RetailStore: A Case Study for Chain Markets in TurkeyYelda Cansu Adali, Industrial Engineering, Istanbul KulturUniversity, Istanbul Kultur University, Atakoy Kampus,Istambul, Turkey, [email protected], Rifat GürcanÖzdemir, Kayhan Karaman

The study addresses layout planning and shelf space allocation prob-lems simultaneously in a retail store. In the layout planning, placementof category areas is determined on the store floor such that areas do notoverlap. In the shelf space allocation problem, space area and locationof each unit on the shelf need to be determined. A two stage mathemat-ical model will be developed for solving the above mentioned problemfor maximizing total sales revenue in the period. The developed modelwill be implemented in a real life retailer’s store layout and shelf spaceallocation problem.

� TC-11Tuesday, 12:30-14:00RB-Iota

Exact Methods for Arc Routing Problems

Stream: Arc RoutingInvited sessionChair: Claudia Bode, Chair of Logistics Management, GutenbergSchool of Management and Economics, Johannes GutenbergUniversity Mainz, 55099, Mainz, Germany,[email protected]

1 - Mixed Capacitated General Routing Problem: Exactand Heuristic MethodsFrancesca Vocaturo, Dipartimento di Economia e Statistica,Università della Calabria, Via Pietro Bucci - Cubo 0C, 87036,Arcavacata di Rende (CS), Italy, [email protected], AdamoBosco, Demetrio Laganà, Roberto Musmanno

132


The general routing problem (GRP) arises in contexts where it isneeded not only to service some arcs or edges of a graph, but also someisolated vertices. It is common in many real-life applications of wastecollection and mail delivery. Despite the importance, relatively fewstudies have been published on this problem. Moreover, most worksdeal with the uncapacitated case. We present exact and heuristic ap-proaches for the capacitated GRP defined on a mixed graph. Compu-tational results are also reported.

2 - Modeling the Rural Postman Problem with Time Win-dowsIngrid Marcela Monroy Licht, Mathematics and IndustrialEngineering, École Polytechnique de Montréal, 2900, boul.Édouard-Montpetit. Chemin de Polytechnique., H3T 1J4,Montréal, Québec, Canada, [email protected],Ciro Alberto Amaya, André Langevin

During the winter season it is necessary to monitor the formation of iceon roads. This is the case in Quebec, Canada, where a patrol vehiclemust check the status of roads during certain time intervals, dependingon weather conditions. To design efficient monitoring, we have mod-eled this problem as a Rural Postman Problem with Time Windows.We explored three models to formulate the problem: a mixed integermodel in arcs, an equivalent problem in nodes and a constraint pro-gramming problem. Some results are presented comparing the modelson directed and non-directed graphs.

3 - Analysis of Pricing Problem Relaxations for the Ca-pacitatad Arc Routing ProblemClaudia Bode, Chair of Logistics Management, GutenbergSchool of Management and Economics, Johannes GutenbergUniversity Mainz, 55099, Mainz, Germany,[email protected], Stefan Irnich

We present an in-depth analysis of the pricing problem and possiblerelaxations in the branch-and-price context for the CARP. Any pricingproblem relaxation must be able to handle two sets of tasks for mod-eling elementary routes and for branching. This complicates the pric-ing relaxations. We expose the following relaxations of CARP pricingproblem: partially elementary tours, NG-route relaxation (including adynamic version), and k-loop-free pricing. The requirement to con-struct task-2-loop free tours is rather intricate for k-loop elimination.Finally, acceleration techniques are addressed.

� TC-12Tuesday, 12:30-14:00RB-Omicron

Capacity and Infrastructure

Stream: Optimization in Public TransportInvited sessionChair: Rob Goverde, Transport and Planning, Delft University ofTechnology, Stevinweg 1, 2628 CN, Delft, Netherlands,[email protected]

1 - OR in shunt planningJoel van ’t Wout, Netherlands Railways, Netherlands,[email protected]

The final planning phase for a railway operator is the shunt planning.This involves the processes within the local station area, mainly choos-ing parking tracks and finding routes between the station and the shunt-ing yard. OR models for shunt planning can be of great help to theplanners of Netherlands Railways, because of limited parking and rout-ing capacity in the Dutch railway network. Moreover, NS would liketo make the shunt plans as late as possible, generating time pressurefor the planners. We will present a model that solves the problem ofmatching, parking and routing rolling stock.

2 - A model for balancing price, capacity and crowded-ness in public transportPaul Bouman, Decision and Information Sciences, RotterdamSchool of management, Erasmus University, BurgemeesterOudlaan 50, 3062 PA, Rotterdam, Netherlands,[email protected]

In a public transport setting without seat reservations, the typical waypeak demand can be balanced against operational and inventory costsby balancing price strategy against rolling stock allocation. Since pas-sengers have to share capacity, the satisfaction of a passenger is influ-enced by the allocation of other passengers. We will discuss a modelwhere we take individual tolerance levels for crowdedness into ac-count. We aim to optimize a social optimum for this model and thencompare the results to a simulation where the passengers try to maxi-mize their own individual satisfaction.

3 - Improving railway infrastructure utilization throughpricing mechanisms - Modelling train flowsMichal Kaut, SINTEF Technology and Society, PO Box 4760Sluppen, 7465, Trondheim, Norway, [email protected],Adrian Werner, Arnt-Gunnar Lium, Truls Flatberg, TeodorGabriel Crainic, Teodor Gabriel Crainic

The goal of the project is to improve the utilization of the Norwe-gian railway network by creating incentives for train operators to movesome of their load to off-peak times, by assigning penalties to the us-age of specific sections at given times. This leads to a bilevel problem:in the upper level, we set the prices and in the lower level, the trainoperators create their routes.

We focus on the lower-level optimization problem, in which we haveto approximate the way the train operators react to the usage prices,while ensuring that the proposed schedule is feasible in the network.

4 - Computing capacity consumption of railway linesand networksRob Goverde, Transport and Planning, Delft University ofTechnology, Stevinweg 1, 2628 CN, Delft, Netherlands,[email protected]

This paper describes a mathematical model to compute the capacityconsumption of a railway line and extends this concept to railway net-works. Capacity consumption is measured as the occupation time plusbuffer time relative to the scheduled time interval. The occupationtime is obtained by compressing the timetable while maintaining fea-sibility. This is modelled as a critical path problem over precedencegraphs. The line capacity problem is based on a microscopic modelwhich gives input to the macroscopic model of the network capacityproblem that also includes connections at stations.

� TC-13Tuesday, 12:30-14:00RB-Tau

Assembly systems and automotiveproduction

Stream: Manufacturing and WarehousingInvited sessionChair: Rüdiger Berndt, Computer Science 7, UniErlangen-Nürnberg, 91058, Erlangen, Germany,[email protected]

1 - Job rotation in assembly lines with disabled workersAlysson M. Costa, University of São Paulo, Brazil,[email protected], Mayron César Oliveira Moreira

We study the problem of planning job rotation schedules in assemblylines with heterogeneous workers. Practical motivation for this studycomes from the context of balancing assembly lines in sheltered workcenters for the disabled. We propose a new problem in which the tasksare kept fixed in the stations for the whole planning period while eachworker rotates within a subset of the stations. A mathematical modeland some preliminary results are discussed.

133


2 - Improving ergonomics by OR methods applied in theplanning process: The example of the automobile in-dustryAlena Otto, Management Science, Friedrich-SchillerUniversity Jena, Carl-Zeiß-Straße 3, 07743, Jena, Germany,[email protected]

Poor workplace ergonomics results in occupational diseases and inhigh costs for firms, e.g., from absenteeism and errors. It is possible toimprove ergonomics at low costs, especially at the earlier steps of plan-ning. Then, mitigation of ergonomic risk factors also goes with raisesin productivity and a higher quality. Currently, firms have difficultiesto incorporate ergonomic aspects into the planning process. We showexamples of how to integrate ergonomic aspects into the planning deci-sions. We discuss important planning steps, formulate problem settingsand point out OR models.

3 - Precompiled Configuration Spaces for the Verifica-tion of Consistency in Production Planning Pro-cessesRüdiger Berndt, Computer Science 7, UniErlangen-Nürnberg, 91058, Erlangen, Germany,[email protected], Peter Bazan,Kai-Steffen Hielscher

Highly customizable products and mass customization - as increasingtrends of the last years - are mainly responsible for an immense growthof complexity within the digital representations of knowledge. Thisalso applies to the scope of car manufacturing and to the productionplanning processes and the involved data. We develop a method todetect and analyze inconsistencies by employing a Multi-Valued Deci-sion Diagram (MDD) which encodes the set of all valid product con-figurations. On this basis, we state a number of rules of consistencythat are checked by a set-based verification scheme.

4 - Queue Optimization in Deterministic Linear Manufac-turing SystemsZuzana Nemcová, Department of Information Technologies,Faculty of Informatics and Management, University of HradecKrálové, Rokitanského 62, 500 02, Hradec Králové, CzechRepublic, Czech Republic, [email protected], MartinGavalec

The contribution describes cost optimization in deterministic linearqueuing systems. A manufacturing system is linear if it consists of aseries of servers. Knowledge of the development of the queue lengthsduring the run of the system is useful for controlling the total costs.The method of finding the optimal setup of service capacities of theservers in each time period in which the service times stay unchangedis suggested. It helps the manager of the system in deciding when andhow the service times of the servers should be changed. This results insignificant cost savings.

� TC-14Tuesday, 12:30-14:00RB-Omega

Nurses Scheduling

Stream: Timetabling and RosteringInvited sessionChair: Geetha Baskaran, Computer Science, The University ofNottingham Malaysia Campus, Faculty of Science, Jalan Broga,43500, Semenyih, Kajang, Malaysia,[email protected]

1 - Agent-based Cooperative Meta-heuristic search forFairness in Nurse RosteringSimon Martin, Mathematics, University of Portsmouth, LionBuilding, Lion Gate, PO1 3HE, Portsmouth, Hampshire,United Kingdom, [email protected], Pieter Smet,Djamila Ouelhadj, Ender Özcan, Greet Vanden Berghe

Traditional approaches to solving the nurse rostering problem havemainly focused on the minimisation of the number of constraint vi-olations across all the nurses. However, these rosters generally failto ensure that individual nurses are treated fairly and the workload isshared evenly amongst nurses. In this paper, we introduce new fair-ness measures and propose an agent-based framework for coopera-tive search that combines the strength of several heuristic and meta-heuristic agents to optimise these measures in order to generate fairnurse rosters.

2 - Multi-level Decision Support in Rostering andStaffingPatrick De Causmaecker, Computer Science/CODeS,Katholieke Universiteit Leuven, Campus Kortrijk, EtienneSabbelaan 53, BE-8500, Kortrijk, Flanders, Belgium,[email protected], StefaanHaspeslaghThe connection between rostering software and inter-ward decisionmaking is studied. The assignment of nurses to shifts in a ward issubject to a large numbers of constraints and must allow flexibility inmeeting unforeseen circumstances. Often software is available to sup-port the assignment task. We model the impact of the rostering strategyand the ward configuration on its operational platform in terms of de-mand and occupation. By studying the impact of algorithm behavioron the operational efficiency, a better informed management of loadand resource distribution becomes possible.

3 - The Nurse Scheduling Problem: Modelling theNurses’ Preferences in a Hospital in TurkeyAlejandra Duenas, Management, IESEG School ofManagement, Socle de la Grande Arche, 1 Parvis de laDéfense, 92044, Paris, France, [email protected], G.yazgiTÜtÜncÜThe nurse scheduling problem consists of generating a work schedulethat considers working days and days off for nursing staff in a Turk-ish hospital. Schedules meet all hard constraints and consider nurses’preferences modelled by fuzzy sets and aggregated to determine anoverall preference cost function. The schedules are generated by a hy-brid approach based on an interactive method which is combined witha multi-objective genetic algorithm(GA). The GA’s fitness function isbased on goal and aspiration levels. The results reveal that the pro-posed approach generates good quality solutions.

4 - From Simplified to Detailed Solutions To The NurseRostering ProblemGeetha Baskaran, Computer Science, The University ofNottingham Malaysia Campus, Faculty of Science, JalanBroga, 43500, Semenyih, Kajang, Malaysia,[email protected], Andrzej Bargiela,Rong QuThis paper presents a novel approach to solving the nurse rosteringthrough information granulation. Compared to the established schedul-ing methods,our approach involves judicious grouping (granulation) ofshift types that lead to a reformulation of the original problem in a sig-nificantly reduced solution space. The schedules are then translatedback into the original problem space by considering the constraints thatcould not be represented in the reduced space. The proposed methodhas been evaluated on the standard benchmark problems and has shownto be capable of finding high quality schedules.

� TC-15Tuesday, 12:30-14:00RB-2101

Vector and Set-Valued Optimization I

Stream: Vector and Set-Valued OptimizationInvited sessionChair: Vicente Novo, Matematica Aplicada, Universidad Nacionalde Educacion a Distancia, Juan del Rosal no. 12, 28040, Madrid,Spain, [email protected]: Beatriz Hernández-Jiménez, Economics, University Pablode Olavide, Edificio No 3, José Moñino - 2a planta-despacho26,Ctra. de Utrera, Km. 1- 41013 Sevilla, 41013, SEVILLA, Spain,[email protected]

134


1 - Variational-like inequality problems and vectorial op-timization problems in Banach spacesGabriel Ruiz-Garzón, Estadistica e I.O., University of Cadiz,Campus de Jerez de la Frontera, Avda. de la Universidad s/n,11405, Jerez de la Frontera, Cádiz, Spain,[email protected], Lucelina Batista dos Santos, AntonioRufián-Lizana, Rafaela Osuna-Gómez

This work is devoted to the study of relationships between solutions ofStampacchia and Minty vector variational-like inequalities and Paretoand weak Pareto solutions of vector optimization problems in Ba-nach spaces under pseudo invexity and pseudo monotonicity hypothe-ses. We have extended the results given by various authors to Banachspaces and the relationships obtained to weak efficient points in San-tos et al. are completed and enabled to relate Pareto points and solu-tions of the weak and strong Minty, Stampacchia and perturbed vectorvariational-like inequalities.

2 - Efficiency in multiobjective variational problem un-der generalized convexityManuel Arana-Jiménez, Estadistica e InvesitigacionOperativa, University of Cadiz, Facultad CCSS y de lacomunicacion, Av. Universidad s/n, 11405, Jerez de laFrontera, Cadiz, Spain, [email protected], AntonioRufián-Lizana, Rafaela Osuna-Gómez, Gracia M.Nieto-Gallardo

We study effciency in multiobjective variational problems. For thispurpose, we introduce new classes of functions, called V-KT(FJ)-pseudoinvex. These functions are characterized in order that all criti-cal points (Kuhn-Tucker/Fritz John) are efficient solutions. That is, wepropose the weakest properties on the functions involved in the mul-tiobjective variational problem, such that we can ensure that a criticalpoint is an efficient solution. These new classes of functions and re-sults are a generalization of others recently appeared in multiobjectivemathematical programming problems.

3 - The exact l1 penalty function method for vector opti-mization problemsTadeusz Antczak, Faculty of Mathematics and ComputerScience, University of Lodz, Banacha 22, 90-238, Lodz,lodzkie, Poland, [email protected]

In our considerations, we use the vector exact l1 penalty functionmethod for solving convex and nonconvex vector optimization prob-lems. The most important property of this method, that is, exactness ofthe penalization, is examined. We show that, there exists a lower boundof the penalty parameter, above which this value, the equivalence be-tween the set of (weak) Pareto optimal solutions in multiobjective pro-gramming problem and the set unconstrained (weak) Pareto optimalsolutions in its associated vector penalized optimization problem withthe vector exact l1 penalty function holds.

4 - Characterization of weakly efficient solutions fornon-regular vector optimization problems with conicconstraintsBeatriz Hernández-Jiménez, Economics, University Pablo deOlavide, Edificio No 3, José Moñino - 2a planta-despacho26,Ctra. de Utrera, Km. 1- 41013 Sevilla, 41013, SEVILLA,Spain, [email protected], Rafaela Osuna-Gómez, Marko A.Rojas-Medar

In non-regular problems the classical optimality conditions are totallyinapplicable. Meaningful results were obtained for scalar problemswith conic constraints by Izmailov and Solodov. In this paper we givea necessary optimality condition for a non-regular vector optimiza-tion problems with conic constraints defined using abstract spaces, andappropriate generalized convexity notions to show that the optimal-ity condition given is both necessary and sufficient to characterize theweakly efficient solutions set. The results that exist in the literature upto now, even for the regular case, are particular instances of the onespresented here.

� TC-16Tuesday, 12:30-14:00RB-2103

Perturbation and sensitivity in Markovcontrol problems and games

Stream: Uncertainty and Perturbations in Optimizationand in Environmental and Energy ModellingInvited sessionChair: Konstantin Avrachenkov, MAESTRO team, INRIA SophiaAntipolis, 2004 Route des Lucioles B.P. 93, 06902, SophiaAntipolis, France, [email protected]: Yi Zhang, Mathematical Sciences, University of Liverpool,Department of Mathematical Sciences, University of Liverpool,L69 7ZL, Liverpool, Merseyside, United Kingdom,[email protected]

1 - Markov decision processes with unknown transitionmatricesMasayuki Horiguchi, Department of Mathematics andPhysics, Faculty of Science, Kanagawa University, 2946Tsuchiya„ 259-1293, Hiratsuka, Kanagawa, Japan,[email protected]

We are concerned with Markov decision processes (MDPs) with un-known transition matrices. We introduce a Bayesian approach of in-tervals of prior measures to this uncertainty model, by which we canestimate unknown transition matrices to be posterior interval of closedconvex set in each component and formulate interval estimated MDPs.We characterize this model as a controlled Markov set-chain modelwith intervals of transition matrices. Some convergence properties foruncertain MDPs of Bayesian approach are discussed and also numeri-cal experiments are shown.

2 - Detecting non-Hamiltonicity in cubic graphs usinglinear feasibility problemsJerzy Filar, School of Computer Science, Engineering andMathematics, Flinders University, Sturt Road, 5042, BedfordPark, SA, Australia, [email protected], MichaelHaythorpe, Serguei Rossomakhine

We present a series of linear feasibility problems containing apolynomially-bounded number of constraints and variables. We showthat the feasibility problems are guaranteed to be feasible when gener-ated for Hamiltonian cubic graphs, but may not be feasible otherwise.We prove that cubic graph graphs always generate an infeasible setof constraints. We provide strong empirical evidence that almost allnon-Hamiltonian cubic graphs can be identified as such using this ap-proach.

3 - Transformation method for discounted continuous-time Markov decision processes with unboundedratesYi Zhang, Mathematical Sciences, University of Liverpool,Department of Mathematical Sciences, University ofLiverpool, L69 7ZL, Liverpool, Merseyside, United Kingdom,[email protected], Alexei Piunovskiy

The standard uniformization technique for reducing a continuous-timeMarkov decision process (CTMDP) to a discrete-time Markov deci-sion process has been well known for long. This technique requiresthe boundedness on the transition rates. In this talk, for a discountedCTMDP in Borel spaces, we present a transformation method that gen-eralizes the standard uniformization technique to the case of arbitrarilyunbounded transition rates. We also obtain a necessary and sufficientcondition for the finiteness of the underlying value function, whichleads to a non-explosion condition for a CTMDP.

4 - The expected total cost criterion for Markov decisionprocesses under constraints: a convex analytic ap-proachFrancois Dufour, Institut de Mathematiques de Bordeaux,Universite Bordeaux, 351 Cours de la Liberation, 33400,Talence, France, [email protected], MasayukiHoriguchi, Alexei Piunovskiy

135


This paper deals with discrete-time Markov Decision Processes underconstraints where all the objectives have the same form of an expectedtotal cost over the infinite time horizon. The existence of an optimalcontrol policy is discussed by using the convex analytic approach. Wework under the assumptions that the state and action spaces are gen-eral Borel spaces and the model is non-negative, semi-continuous andthere exists an admissible solution with finite cost for the associatedlinear program. Our result ensures the existence of an optimal solutionto the linear program given by an occupation measure of the processgenerated by a randomized stationary policy.

� TC-17Tuesday, 12:30-14:00RB-2105

Hub Location IIStream: Hub LocationInvited sessionChair: Barbaros Tansel, Industrial Engineering, Bilkent University,Bilkent, 6800, Ankara, Turkey, [email protected]

1 - A new approach to economies of scales in hub loca-tion problemsMartin Baumung, Deutsche Post Endowed Chair ofOptimization of Distribution Networks, RWTH AachenUniversity, Kackertstrasse 7, 52072, Aachen, Germany,[email protected]

Despite economies of scales being one of the main reasons for hub &spoke networks, little attention has been paid to their adequate consid-eration in hub location models. In most of nowadays models unit trans-portation costs are assumed to be independent of the actual flow andare discounted by a constant factor on interhub links to take account ofeconomies of scale. We present a new approach where transportationcosts are not incurred by flows but by vehicle tours. This way consol-idation can be achieved on any link of the network and the number ofhubs no longer needs to be constrained.

2 - The ring star hub location problem: A Benders de-composition approachRicardo Camargo, Engenharia de Produção, UniversidadeFederal de Minas Gerais, 30161-010, Belo Horizonte, MinasGerais, Brazil, [email protected], Gilberto Miranda,Henrique Luna, Elisangela Martins de Sá

The ring star hub location problem is the design of a two level hier-archical network: At the top, a ring network linking hubs, which areselected from a node candidates set, while observing scale economieson the links; and, at the bottom, a star network, single allocating ex-changing flow customers to a hub. The problem is an extension of hublocation problems to rapid transit and telecommunication applications.A new formulation is proposed and tackled by a Benders decomposi-tion algorithm, which efficiently solves large scale instances to opti-mality, due to a new cut selection scheme.

3 - The p-Hub Median Problem on Incomplete Trans-portation NetworksBarbaros Tansel, Industrial Engineering, Bilkent University,Bilkent, 6800, Ankara, Turkey, [email protected],Ibrahim Akgün

The p-hub median problem has been studied so far under the assump-tion of a fully connected (complete) network where there is an undi-rected arc present for each pair of nodes. Transportation networks inthe physical world are generally incomplete. That is, many of the arcsmay be missing. While an incomplete network problem can alwaysbe handled by converting it first to a complete network problem, westrongly deviate from this tradition and give a new formulation thatrepresents the problem directly on the given incomplete network. Wepresent the new formulation and certain modeling advantages relatedto it.

� TC-18Tuesday, 12:30-14:00RB-2107

Competitive Location.

Stream: Continuous LocationInvited sessionChair: Blas Pelegrin, Statistics and Operations Research,University of Murcia, Spain, [email protected]

1 - Solving a Huff-type location model on networksRafael Blanquero, Estadística e Investigación Operativa,Universidad de Sevilla, Faculty of Mathematics, Tarfia S.N.41012-Seville (Spain), 41012, Seville, Spain,[email protected], Emilio Carrizosa, Amaya Nogales, FrankPlastria

In this talk we address the problem of locating a single facility on anetwork according to the Huff competitive location model, where thenodes of the network represent users asking for a certain service. Theobjective function of this problem is shown to be multimodal and canbe written as the difference of two convex functions. This fact is ex-ploited in a branch and bound procedure which allows us to solve thisproblem on networks of reasonable size. Computational results arepresented.

2 - Location models and GIS tools for franchise distribu-tionRafael Suarez-Vega, Metodos Cuantitativos en Economia yGestion, Universidad de Las Palmas de Gran Canaria, EdificioDepartamental de Ciencias Economicas y Empresariales,Campus Universitario de Tafira, 35017, Las Palmas de GranCanaria, Las Palmas, Spain, [email protected], DoloresR. Santos-Peñate, Pablo Dorta-González

We consider a competitive location model and GIS tools to make de-cisions in a franchise distribution system when a single facility has tobe located in a planar space. Two normally in conflict objectives areinvolved, the maximization of the market share of the firm and the min-imization of the cannibalization. Maps representing the market share,the cannibalization effect, or a combination of them, are obtained. GIStools allow the incorporation of forbidden regions and other restric-tions, and the visualization of the effects produced by the opening of anew facility into the market.

3 - On a biobjective location model for firm expansionwith binary customer behaviourAlgirdas Lancinskas, Institute of Mathematics andInformatics, Vilnius University, Akademijos st. 4, LT-08663,Vilnius, Lithuania, [email protected], JuliusZilinskas, Blas Pelegrin, Pascual Fernandez

We study a bi-objective location model on a network for firm expan-sion in competition with other firms. The aim of the expanding firm isto maximize its market share and to minimize the cannibalism effectcaused by new facilities. Demand in each market is fixed and cus-tomers select the most attractive facility while attraction depends onfacility quality and distance to the customers. Under general condi-tion, efficient locations are proved to be in the set of nodes. A heuristicalgorithm is proposed to find efficient solutions for a large set of mar-kets. Computational results are discussed.

� TC-19Tuesday, 12:30-14:00RB-2111

Risk Modeling and Risk Measures

Stream: Actuarial Sciences and Stochastic CalculusInvited sessionChair: David Wozabal, TUM School of Management, TechnischeUniversität München, Arcisstraße 21, DE-80333, München,Germany, [email protected]

136


1 - Modeling the ATM Cash Delivery Problem as a Vehi-cle Dispatch Problem and Solving It via a New HybridHeuristic AlgorithmKübra Fenerci, Industrial engineering, Bogaziçi University,Maresal Çakmak Cd. Ince Su Sk. Abacı Apt 8/1Sirinevler/IST., Istanbul, Turkey,[email protected], Ilhan Or, Ekrem DumanThe Savings’ and Sweep algorithms are two classical, but widely usedapproaches to find good near optimal solutions for vehicle routingproblems. A varient heuristic integrating the basic philosophies ofthese algorithms is developed and applied to the vehicle routing prob-lem arising in the route optimization of ATM cash delivery vehicles.An "angle criterion" (reflecting the spatial tendencies of the sweep ap-proach) is embedded into the "savings criterion" to generate a hybridalgorithm. Relative weights of the "sweep" and "savings" criteria areexperimentally determined.

2 - "Jumping Assets", "Timing Effect" and "SystemicRisk" in Portfolio Selection ProblemsSergio Ortobelli, MSIA, University of Bergamo, Via deiCaniana, 2, 24127, Bergamo, Italy, [email protected], EnricoAngelelliIn this paper we propose a comparison among the ex-post sample pathsof the wealth obtained optimizing some large scale dynamic portfoliostrategies. We use the stock of the main US market of the last 20 years.This empirical portfolio selection analysis has permitted: 1) To iden-tify some assets, called "jumping assets", that present returns with bigoscillations in a very short time. 2) To use portfolio strategies that max-imize the time the portfolio reaches a loss and minimize the time theportfolio reaches an earn (Timing effect). 3) To forecast the periods ofsystemic risk.

3 - Robustifying Convex Risk Measures: A Non-Parametric ApproachDavid Wozabal, TUM School of Management, TechnischeUniversität München, Arcisstraße 21, DE-80333, München,Germany, [email protected] are robustifying convex, version independent risk measures. Therobustified risk measures are defined as the worst case portfolio riskover a Kantorovich neighborhood of a reference distribution. Wedemonstrate that the optimization problem of finding the worst caserisk can be solved analytically and consequently closed form expres-sions for the robust risk measures are obtained. We use these results toderive robustified versions for several examples of risk measures anddemonstrate the favorable properties of the resulting robust measuresin rolling-window out-of-sample evaluations.

� TC-20Tuesday, 12:30-14:00RB-2113

Risk Assessment in the Financial SectorStream: Decision Making Modeling and Risk Assess-ment in the Financial SectorInvited sessionChair: Tara Keshar Nanda Baidya, Industrial Engineering,PUC-Rio, Rua Marquês de São Vicente 225 — Gávea, 22451-041,Rio de Janeiro, RJ, Brazil, [email protected]

1 - A Study on Quantitative Evaluation of the Model forthe Piano FingeringKeisuke Hotta, Faculty of Information and Communication,Bunkyo University, 1100 Namegaya, 253-8550, Chigasaki,Kanagawa, Japan, [email protected] fingering problem is to decide the natural fingering for the musicscore. In the previous research, I studied the fingering of the musicalpassage with single-note, 3rd or 6th-chord and the chord for both rightand left hand respectively, by optimizing them as IP. The passage withsingle-note is the most difficult because of the flexibility. Then, unnat-ural fingering was occurred in several passages. In order to improvethe model and the cost function, I study the effects of them by measur-ing the existing qualitative results of the score quantitatively, and seekthe better fingering.

2 - On distributional robust probability functions andtheir computationsMan Hong Wong, Systems Engineering and EngineeringManagement, The Chinese University of Hong Kong, Room609, William M. W. Mong Engineering Building, TheChinese University of Hong Kong, Shatin, N.T., Hong Kong,Hong Kong, [email protected], Shuzhong Zhang

Consider a random vector, and assume that a set of its moments infor-mation is known. Among all possible distributions obeying the givenmoments constraints, the envelope of the probability distribution func-tions is introduced in this paper as distributional robust probabilityfunction. We show that such a function is computable in the bi-variatecase under some conditions. Connections to the existing results in theliterature, as well as the applications in risk management, will be dis-cussed.

3 - Recent Features of UK Income Protection Claim Du-rationsIsabel Cordeiro, Escola de Economia e Gestão, Universidadedo Minho/CEMAPRE, ISEG, Escola de Economia e Gestão,Universidade do Minho, Campus de Gualtar, 2710-057,Braga, Portugal, [email protected]

In this paper we calculate average claim durations conditioned on re-covery, on death and not conditioned on any particular mode of claimtermination, using the most recent graduations of the transition inten-sities defined for a multiple state model for Income Protection (IP).These average claim durations are compared with similar durationspresented in Cordeiro (2002, 2007), calculated with graduations foran earlier period. We also analyse other features of the distribution ofthe duration of an IP claim. All this information can be very useful forinsurance companies selling IP policies.

4 - Valuation of employee Stock Option with stochasticexercise price for a Brazilian FirmTara Keshar Nanda Baidya, Industrial Engineering, PUC-Rio,Rua Marquês de São Vicente 225 — Gávea, 22451-041, Riode Janeiro, RJ, Brazil, [email protected], Fernando Visser

Employee Stock Option is being used more and more in Brazil re-cently. The objective of our research is to propose a model to pricethis option in the Brazilian context. For this purpose, we have chosena Brazilian Airline firm called TAM S.A. The exercise price of thisoption is stochastic, it depends on the General Price Index of Brazilcalled IGPM. It has another particularity, the stock option can be exer-cised only after a certain period of time. In our case, there are optionsissued with a waiting period of 2, 3 and 4 years.

� TC-21Tuesday, 12:30-14:00RB-2115

Asset Pricing

Stream: Long Term Financial DecisionsInvited sessionChair: Andreas Loeffler, Banking and Finance, Freie UniversitätBerlin, Boltzmannstr 20, 14195, Berlin, Germany, [email protected]

1 - Optimal Stopping for Portfolio ManagementAlexandra Spachis, Business School, Imperial College,Exhibition road, SW7 2AZ, London, United Kingdom,[email protected], EleniHadjiconstantinou, Nicos Christofides

Optimal Stopping problems (OPM) offer a substantial improvement interminating a procedure at an earlier stage thus in real-life problemssaving costs of additional observations. The focus of this paper is onthe application of an OPM to the area of Portfolio Management. Morespecifically, the Portfolio Management Problem (PMP) aims at max-imizing the expected return of an investor investing in a portfolio. ADynamic Programming approach is proposed to solve the PMP and anillustrative example is presented to explain the proposed mathematicalformulation and its computational implementation.

137


2 - Misperceptions of Long-Term Investment Perfor-mance: Insights From An ExperimentMichael Stutzer, Finance, University of Colorado, 419 UCB,80309, Boulder, CO, United States,[email protected], Susan Jung Grant

The expected cumulative return is a reasonable long-term result in-dicator for repeated fixed dollar gambles. But long-term investmentsare better modeled by cumulating percentage returns, which are highlypositively skewed, making the median a far better measure of central-ity. Our experimental findings show that many subjects, including oneswho have authored papers in the area, would make long-term invest-ments in highly volatile assets with high expected return, but regretdoing so once they are shown the distribution of long-term cumulativereturns resulting from the investment.

3 - A Note On Estimating CAPM-BetasAndreas Loeffler, Banking and Finance, Freie UniversitätBerlin, Boltzmannstr 20, 14195, Berlin, Germany,[email protected], Thomas Burkhardt

Any CAPM-test uses an ex post formulation of the capital market line.In our note we show that the distribution of the noise has to be endo-genised.To this end we consider an model where assets follow a multi-dimensional Brownian motion. Then we posses a model with multipleprice path that enables us to simulate the beta evaluation. Such, weare able to determine the distribution of betas. Simulations using avery simple Brownian setup already show that the resulting beta dis-tributions have an enormous range. Even with largeR2 empiricallyestimatedCAPM-returns should be carefully interpreted.

� TC-22Tuesday, 12:30-14:00RB-2117

Mathematical Models in Macro- andMicroeconomics 2Stream: Mathematical Models in Macro- and Microe-conomicsInvited sessionChair: Ludmilla Koshlai, Systems analysis and OR, Institute ofCybernetics, Gonchar str.,65-a,apt.20, 01054, Kiev, Ukraine,[email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]: Jean-Francois Emmenegger, Quantitative Economics,University of Fribourg, Bd de Pérolles 90, F440, 1700, Fribourg,Fribourg, Switzerland, [email protected]

1 - An empirical assessment of customer lifetime valuemodelsAbdulkadir Hiziroglu, Management Information Systems,Sakarya University, Esentepe Campus M-5 Building, Sakarya,54187, Sakarya, Turkey, [email protected], SerkanSengul

Customer lifetime value plays a key role in customer relationship man-agement and has been carried out many sectors for decades. Re-searchers have proposed specific models for calculating customer life-time value. Literature lacks of comparative research on assessing theexisting models. We evaluate two customer lifetime value models fromsegmentation perspective. RFM model and basic structural model werecompared and an evaluation from segmentation perspective was done.The results indicated that basic structural model was superior to itspeer using the same databases and variables.

2 - On one nonlinear optimization problem for LeontiefmodelsPetro Stetsyuk, Nonsmooth optimization methods, Institute ofCybernetics, 40 Prospect Glushkova, 03187, Kiev, Ukraine,[email protected]

We consider the nonlinear programming problem for Leontief mod-els (NLPL) with a bilinear objective function and linear constraintsboth for primal and dual Leontief models and two quadratic constraintswhich normalize final demand and value added. We show that for pro-ductive Leontief matrix the set of solutions of NLPL is determined bythe eigenvectors corresponding to the maximum eigenvalues of a cer-tain symmetric matrices. For productive and indecomposable Leontiefmatrices NLPL has a unique solution. Numerical examples for the ag-gregated 15 sectors Leontief model of Ukraine are given.

3 - Upper bounds for optimization problem of inter-sectoral planning of structural and technologicalchangesTamara Bardadym, Dept. of Nonsmooth OptimizationMethods, V.M.Glushkov Institute of Cybernetics, ProspektAkademika Glushkova, 40, 03187, Kyiv, Ukraine,[email protected], Oleg BerezovskyiTwo optimisation problems of interbranch planning of structural andtechnological changes formulated as quadratic type problems are con-sidered: to maximize the total income of consumers, and to maximizeKeynesian multiplicator "income growth - production growth". Pos-sible ways for finding upper bounds by Shor’s technique with the useof superfluous constrains are described. Numerical experiments for7-branch model were accomplished.

� TC-23Tuesday, 12:30-14:00RB-Delta

Complementarity Problems, VariationalInequalities and Equilibrium

Stream: Mathematical ProgrammingInvited sessionChair: Sandor Zoltan Nemeth, School of Mathematics, TheUniversity of Birmingham, The Watson Building, Edgbaston, B152TT, Birmingham, United Kingdom,[email protected]: Song Wang, Mathematics & Statistics, The University ofWestern Australia, 35 Stirling Highway, Crawley, 6009, Perth,Western Australia, Australia, [email protected]

1 - The Influence of Technical, Market and LegislativeFactors on E-Waste FlowsThomas Nowak, Transport and Logistics Management,Vienna University of Economics and Business,Nordbergstrasse 15, 1090, Wien, Austria,[email protected], Fuminori Toyasaki, TinaWakolbinger, Anna NagurneyIn this paper, we analyze how technical, market, and legislative factorsinfluence the total amount and prices of e-waste that is collected, recy-cled and (legally and illegally) disposed of. We formulate the e-wastenetwork flow model as a variational inequality problem with four tiers:the model covers the behavior of sources of e-waste, collectors, pro-cessors, and demand markets for precious and hazardous materials. Inthe numerical simulation of our model we can quantify the change ofprices and quantities when penalities for illegal dumping for exampleis varied.

2 - Power penalty methods for nonlinear complementar-ity problemsSong Wang, Mathematics & Statistics, The University ofWestern Australia, 35 Stirling Highway, Crawley, 6009, Perth,Western Australia, Australia, [email protected] real-world problems in classical and financial engineering aregoverned by nonlinear optimization problems with complex derivativeconstraints called obstacles. Usually, these problems can be formu-lated as Nonlinear Complementarity Problem (NCP). We present re-cent advances in the power penalty approach to NCP, Mixed NCP andbounded NCP and prove converges results at an exponential rate. Nu-merical results will be presented to demonstrate the theoretical expo-nential convergence rates.

138


3 - Necessary conditions for variational relation prob-lemsAnulekha Dhara, Mathematics, Indian Institute of TechnologyGandhinagar, Vishwakarma Government Engineering CollegeComplex„ Chandkheda, Visat-Gandhinagar Highway„382424, Ahmedabad, Gujarat, India,[email protected] model of variational relation problem was introduced recentlyin 2008 by Luc. It englobes several models of fixed points, opti-mization, variational and quasivariational inequalities, equilibrium andmany others. In literature it is well known that if a point is a solutionof variational inequality then it is a solution of a projection problem.But such a condition need not hold for a variational relation problem.In this paper we aim to obtain necessary conditions for a point to solvethe variational relation problem.

� TC-24Tuesday, 12:30-14:00CC-A11

Judgmental Forecasting

Stream: ForecastingInvited sessionChair: Zoe Theocharis, Cognitive, Perceptual and Brain Sciences,UCL, 26 Bedford Way, WC1H 0AP, London, United Kingdom,[email protected]

1 - Why do people damp trends when using their judg-ment to make forecasts from time series?Nigel Harvey, Cognitive, Perceptual and Brain Sciences,University College London, Gower Street, WC1E 6BT,London, London, United Kingdom, [email protected],Stian ReimersPeople’s forecasts underestimate the steepness of trends in time se-ries. This trend damping may occur because 1) people anchor on thelast data point and make insufficient adjustment to take the trend intoaccount, 2) they adjust towards the average of the trends they haveencountered within an experiment, or 3) they are adapted to dampedtrends in the environment. To eliminate 1), we show anti-damping(the opposite of damping) for series that are negatively accelerated orhave shallow slopes. To eliminate 2), we show both damping and anti-damping when people forecast from a single series.

2 - Judgmental interval forecasts: The effects of assess-ment order and trend directionMustafa Sinan Gonul, Business Administration, Middle EastTechnical University, Universiteler Mah. Dumlupinar Blv.No:1, Cankaya, 06800, Ankara, Turkey,[email protected], Mary Thomson, Dilek Önkal-AtayTo generate judgmental forecasts of point, 90% and 50% intervals, fivedistinct values need to be assessed: i) 90% upper bound, ii) 90% lowerbound, iii) 50% upper bound, iv) 50% lower bound and v) point fore-cast. The interval bounds can either be assessed in an ordinary man-ner (i&ii first, followed by iii & iv) or concentric manner (i&iii first,followed by ii&iv) with points generated before or after the intervals.This research aims to explore the effects of assessment order and trenddirection on the performance of resultant forecasts. Findings are dis-cussed and implications are given.

3 - An exploration of judgmental forecasting biases us-ing dynamic task environmentsZoe Theocharis, Cognitive, Perceptual and Brain Sciences,UCL, 26 Bedford Way, WC1H 0AP, London, UnitedKingdom, [email protected], Nigel HarveyJudgmental forecasting tasks, which have mainly employed static dis-plays of graphs or tables, have revealed several robust biases in fore-casters’ performance. These biases include trend damping, positiveautocorrelation illusion and noise introduction. We investigated howpeople’s forecasts are influenced by presenting the data series as a dy-namic display of successive bar charts. We found that both forecasteraccuracy and the underlying information processing strategies were af-fected by this. As a result, biases differed substantially from thosepreviously reported.

4 - Change detection training improves time series fore-castingMatt Twyman, Cognitive, Perceptual and Brain Sciences,University College London, United Kingdom,[email protected], Nigel HarveyParticipants were asked to judge whether changes had occurred in timeseries, express confidence in those judgments, and forecast eight datapoints at the end of each series. Different participants performed thechange detection or forecasting tasks (on the same set of time series)in different orders. There were no differences in change detection per-formance or confidence between the two groups, but participants whowere asked to forecast first gave elevated and less accurate forecasts,indicating that change detection training improves forecasting.

� TC-25Tuesday, 12:30-14:00CC-A23

Measurement of Sustainability

Stream: Environmental Issues in Operations Manage-mentInvited sessionChair: Jutta Geldermann, Chair of Production and Logistics,Universität Göttingen, Platz der Göttinger Sieben 3, D-37073,Göttingen, Germany, [email protected]

1 - Developing the data envelopment analysis modelwith incorporating the carbon footprint to measurethe business operational efficiencyLi-Ting Yeh, Business Administration, National CentralUniversity, I1-209,NO.300 Jung-da Rd.,Jung-liCity,Taoyuan,Taiwan 320,R.O.C, Jhongli, Taiwan,[email protected], Dong Shang Chang, Wenrong LiuThe three scopes of carbon footprint have become most suitable meth-ods to measure the greenhouse gases from the business operating activ-ities. This environmental perspective is considered by previous stud-ies to measure the business operational efficiency. With incorporatingthe carbon footprint, the slack-based measure model is developed toevaluate environmental performance of business operations. Researchsample is adopted from Carbon Disclosure Project with Fortune 500Top Companies. The finding is useful to the government for amendingthe industry and energy policy with sustainability.

2 - Strategic and operational implications of climatechange for supply chain managementCostas Pappis, Department of Industrial Management,University of Piraeus, 80 Karaoli & Dimitriou str., 18534,Piraeus, Attica, Greece, [email protected], Cristobal Miralles,Tom Dasaklis, Nikos KaracapilidisEarth’s climate is rapidly changing as a result of increases in emissionsof greenhouse gases caused by human activities. Escalation in regula-tions, market forces and shareholders’ value are paving the way forthe decarbonization of supply chains. In this paper the implications ofclimate change for supply chain management in terms of strategic andoperational planning are identified and insights in the means, by whichsupply chains could be better positioned to cope with climate changehazards in a low carbon economy, are provided.

3 - Resource efficiency in supply chains: a review of net-work design and operational planningGábor Herczeg, Department of Management Engineering,Technical University of Denmark, Produktionstorvet bygn.424, 2800, Lyngby, Denmark, [email protected] sustainable development in supply chain management byengaging in green operations is receiving increasing attention. Greensupply chain operations can be enhanced by increasing resource effi-ciency, e.g., through sustainable use of finite raw materials, etc.. Wereview the state-of-the-art on resource efficiency in supply chains fromthe aspect of supply chain design and operational planning. Further-more, we develop a conceptual framework that identifies future direc-tions and tools for enhancing theory and practice to improve greenoperations.

139


4 - The varied contexts of environmental multicriteriadecision support processesJutta Geldermann, Chair of Production and Logistics,Universität Göttingen, Platz der Göttinger Sieben 3, D-37073,Göttingen, Germany, [email protected],Meike Schmehl, Katharina AmannVarious tools for analysing the impact of the environmental burden likeLife Cycle Assessment have been developed, and the need to apply de-cision analysis is widely recognised. This paper draws conclusions onthe process of multicriteria decision support in various contexts: (1)the development and validation of evaluation criteria for sustainablebioenergy concepts, like a bioenergy village; (2) evolving requirementsof electricity supply networks, which arise from projected fluctuatingregenerative power generation, from a managerial viewpoint; (3) in-dustrial risk and emergency management.

� TC-26Tuesday, 12:30-14:00CC-A24

OR in Forestry II

Stream: OR in Agriculture, Forestry and FisheriesInvited sessionChair: Patrick Hirsch, Institute of Production and Logistics,University of Natural Resources and Life Sciences, Vienna,Feistmantelstrasse 4, 1180, Wien, Austria,[email protected]

1 - On the optimal land allocationJussi Uusivuori, Finnish Forest Research Institute Metla,Unioninkatu 40 A, FIN-00170, Helsinki, Finland,[email protected] allocation between productive land and conservation is studied.Private decision-making is described in two cases: when access to con-served land is closed, and when there is an open access to conservedland. In the latter case the land size designated to conservation willbe smaller than in the former. Policy rules are derived for a social op-timum. When a closed-access regime is opened, optimal policies be-come dependent on wealth differentials between land-owning and non-landowning members of society. Utility of both groups is increased ifwillingness-to-pay for conservation is utilized.

2 - Scheduling sawmill operations using robust opti-mizationSergio Maturana, Ingenieria Industrial y de Sistemas, P.Universidad Catolica de Chile, Casilla 306 Correo 22,Santiago, Chile, [email protected], Mauricio Varas, JorgeVeraScheduling sawmills operations is difficult due to the uncertainty oflog supply and product demand. In this work robust optimization ap-proach is used to take into account uncertainties in product demandand raw material availability when scheduling sawmill operations. Theperformance of the robust solutions was tested using Montecarlo’smethod. Numerical evidence shows an interesting relationship be-tween the product’s holding/shortage costs and the level of conser-vatism, and also provides some insights about the impact of the budgetsof uncertainty in the performance of the robust solutions.

3 - Vehicle routing for transporting forest fuel and com-bined mobile chipper schedulingJan Zazgornik, Institute of Production and Logistics,University of Natural Resources and Life Sciences, Vienna,Feistmantelstrasse 4, 1180, Vienna, Austria,[email protected], Manfred GronaltThis work presents a vehicle routing problem for transporting forestfuel including combined mobile chipper scheduling. A mathematicalmodel was formulated as a multi-depot vehicle routing problem withtime windows. Additional constraints were used to model the queu-ing of trucks at a mobile chipper that chips wood directly into a truck.The objective is to minimize total working times of the drivers, whichincludes to keep waiting times of trucks low at the mobile chipper.Besides the model formulation a dynamic tabu search with alternatingstrategy was used to solve larger problem instances.

4 - Environmental impacts versus driving times - Therouting of log-trucks with a new near-exact solutionapproachMarco Oberscheider, Institute for Production and Logistics,University of Natural Resources and Life Sciences, Vienna,Feistmantelstraße 4, 1180, Vienna, Austria,[email protected], Jan Zazgornik, ManfredGronalt, Patrick Hirsch

The routing of log-trucks is compared for two objectives: minimizingtotal driving times against minimizing the greenhouse gas emissionsof timber transport. A new near-exact solution approach (NE) is in-troduced for this multi depot vehicle routing problem with pick-up anddelivery and time windows. As benchmark a Tabu Search strategy witha dynamically alternating neighborhood is used. The extensive numeri-cal studies with real life data - concerning the supply of biomass powerplants - reveal the potential of the NE and show how efficient planningcan contribute to climate change mitigation.

� TC-27Tuesday, 12:30-14:00CC-A25

Policy Analytics

Stream: Decision ProcessesInvited sessionChair: Alexis Tsoukiàs, CNRS - LAMSADE, Université ParisDauphine, 75775, Paris Cedex 16, France,[email protected]: Vivien Kana Zeumo, MIDO, University Paris Dauphine,LAMSADE, University Paris Dauphine, 75775 Paris Cedex 16,75775, Paris, Paris, France, [email protected]

1 - Policy analytics: what it is and how to distinguish itValerie Belton, Dept. Management Science, University ofStrathclyde, 40 George Street, G1 1QE, Glasgow, UnitedKingdom, [email protected], Alexis Tsoukiàs

In this talk we discuss the term "policy analytics" as a distinguishedfield of scientific investigation as well as of decision support practice.We present a policy cycle analysis and we show why decision analysisand analytics need to be specifically conceived in order to take care ofthe new challenges in this exciting area.

2 - An ANP-based approach for addressing sustainabil-ity in urban transformation projectsValentina Ferretti, Urban and Regional Studies and Planning,Politecnico di Torino, Viale Mattioli 39, 10125, Torino, Italy,[email protected], Marta Bottero, Andrea DiMaggio, Silvia Pomarico, Marco Valle

The paper addresses the problem of sustainability assessment of a newtransport infrastructure through the use of the Analytic Network Pro-cess. The model takes into consideration the different aspects of thedecision-making process, namely, economic, environmental, social,transport and urban planning aspects that have been organized accord-ing the categories of Benefits, Costs, Opportunities and Risks. Theresults of the work are a ranking of alternative solutions and a list ofthe most important elements of the decision problem that have beenrepresented by means of specific thematic maps.

3 - Poverty measurement: a decision aiding perspectiveVivien Kana Zeumo, MIDO, University Paris Dauphine,LAMSADE, University Paris Dauphine, 75775 Paris Cedex16, 75775, Paris, Paris, France, [email protected], AlexisTsoukiàs

In this paper we present a new methodology aiming at aiding the de-sign, implementation and monitoring of poverty reduction policies.The new methodology extends capabilities theory in a practical wayallowing to assess how clusters of similarly "poor" can improve theircondition.

140


4 - InViTo: an integrate approach for supporting theevaluation of transport scenariosIsabella Lami, Polytechnic of Turin, 10125, Turin, Italy,[email protected], Francesca Abastante, Stefano Pensa,Marta Bottero, Elena Masala

This study presents an innovative approach for integrating the AnalyticNetwork Process (ANP) and the visualisation tools. By the use of In-teractive Visualization Tool - InViTo, the work shows the contributionof the proposed approach to the decision-making process in handlingthe data collected, using specific 2D and 3D maps created by the sys-tem. The case study is an application of the ANP and InViTo to asection of the Project CODE24 showing the potentialities of the con-joint use of the two techniques in helping selecting a suitable scenariofor transport improvement.

� TC-28Tuesday, 12:30-14:00CC-A27

Boundaries and InterfacesStream: Models of Embodied CognitionInvited sessionChair: Susie Vrobel, The Institute for Fractal Research,Ernst-Ludwig-Ring 2, 61231, Bad Nauheim, Hessen, Germany,[email protected]

1 - Cold-blooded loneliness: Social exclusion leads tolower skin temperaturesHans IJzerman, Social Psychology, Tilburg University,Warandelaan 2, 5037AB, Tilburg, Netherlands,[email protected]

Recent findings have stressed links between physical warmth and af-fection (see IJzerman & Koole, 2011). We promote the role of skintemperature, arguing that simulations of temperature are used to inter-pret communal relationships. The results support that complex socialmetaphors are scaffolded onto bodily temperature changes: Social ex-clusion leads to lower finger temperatures (Study 1), while negativeaffect typically so experienced is alleviated after holding a warm cup(Study 2). We discuss interactions between body and social relationsspecifically and core cognitive systems in general.

2 - Are Free Association Clusters Ad-Hoc HumanoidAgents ?Antal Borbely, N.Y. Psychoanalytic Institute, 675 West EndAve, #1A, 10025, New York, NY, United States,[email protected]

During free association in psychoanalysis association clusters areformed. They represent themes relevant at that moment but they canalso be compared to ad-hoc humanoid agents. Their functioning andthe relationships formed with each other as well as with the analysandand analyst are explored. Agents relate to each other on social, human,and intrapsychic levels. Every agent is constituted by other agents aspart of an inexhaustible hierarchy. The spatial and temporal structureof this hierarchy is maintained by metaphoric processes involved inmemory, anticipation, and creativity.

3 - Toward a ’rationality’ tuned with human lifeMarina Alfano, Conservatorio di Musica GirolamoFrescobaldi, Ferrara, Ministero Pubblica Istruzione, Italy, ViaPreviati 22, 44121, Ferrara, Italy,[email protected], Rosolino Buccheri

In order to include interaction in physics and to account for the philo-sophical consequences of Gödel theorems, we have to accept the an-thropological datum that, at the dawning of the civilization, humancognition was based on a process of co-oscillatory resonance estab-lished by man with his environment as the effect of a hearing ampli-fied by his body and transmitted to his consciousness. By admittingthat such ’subjective’ knowledge is nowadays disregarded but not lost,we look to a more effective rationality by integrating the ’scientific’one into a wider, more comprehensive wisdom

4 - Dimensions of trust involved in human actionsagainst climate changeAnnette Hohenberger, Middle East Technical University(METU), Informatics Institute, Cognitive Science Program,Cankaya Ilcesi Universiteler Mah, Dumlupinar Bulvari No:1,06800, Ankara, Turkey, [email protected]

This paper explores the spatio-temporal, cognitive, social and emo-tional dimensions of trust involved in people’s actions against climatechange. Due to its vast spatio-temporal scale, climate change is a majorchallenge that necessitates humans to extend their interfaces with theenvironment on the above-mentioned dimensions. Trust is an essen-tial precondition in this process. To the extent that we can trust in thevarious environments that embed us and into the available resources,we can reduce and possibly handle the higher complexity with whichclimate change challenges us.

� TC-29Tuesday, 12:30-14:00CC-A29

MINLP and Nonlinear Network Flows inGas Transport II

Stream: Mixed-Integer Non-Linear ProgrammingInvited sessionChair: Jesco Humpola, Optimization, Zuse Institute Berlin,Takusstrasse 7, 14159, Berlin, Berlin, Germany, [email protected]

1 - An automated method for the booking validationproblemBenjamin Hiller, Optimization, Zuse Institute Berlin, Takustr.7, 14195, Berlin, [email protected]

We propose an automated approach to solve the booking validationproblem faced by gas network operators: Given a set of transmissioncapacity contracts (a booking), check whether all balanced gas flowsthat may result from those contracts are technically realizable. Ourapproach is based on MINLP methods for checking gas flow realiz-ability complemented by a method for generating a representative sub-set of the gas flows that may arise. This generation method combinesa stochastic model for gas offtakes and a deterministic model for thecontractual limitations of both injections and offtakes.

2 - Heuristic Long-term Planning of Natural Gas Trans-portation NetworksDaniel Heuberger, Institute of Power Systems and PowerEconomics, Schinkelstr. 6, 52056, Aachen, Germany,[email protected], Albert Moser

New challenges for natural gas TSO arise, due to the liberalization ofthe energy markets and the started regulation of natural gas networks inEurope. Long-term planning is an acknowledged method to quantifythe impact of different uncertainties in order to calculate cost efficientnetwork structures. This paper introduces a heuristic method for cal-culating cost-efficient natural gas transportation networks. In this con-text requirements for the optimization are presented and the developedmethod is introduced. Finally, the practical applicability is proved bymeans of exemplary results.

3 - Design of Hydrogen Transmission Pipe Networks forFranceDaniel De Wolf, Economie Gestion, Université du Littoral,189B avenue Maurice Schumann, B.P. 5526, 59379,Dunkerque Cedex 1, France, [email protected]

We develop an economic model for the deployment of hydrogen trans-port network. First, we present a state of the art of the existing methodsto design pipe networks. Having established the physical constraintsand the specific costs for the transport of hydrogen, we present themathematical program in order to determine the optimal topology andsizing of the network. A theoretical important result concerning thetopology of the networks is demonstrated. Then we present the heuris-tics used to solve the model. Finally, the model is applied to the devel-opment of a new hydrogen network in France.

141


4 - Stochastic nonconvex MINLP models and global op-timization for natural gas production network designunder uncertaintyXiang Li, Chemical Engineering, Queen’s University, Dupuis403, 19 Division Street, K7L 3N6, Kingston, Ontario, Canada,[email protected], Asgeir Tomasgard, Paul I. BartonScenario-based stochastic nonconvex MINLP models are developedto facilitate the design of natural gas production networks under un-certainty, where the nonconvexity comes from bilinear, quadratic andpower functions involved. As a gas network involves large invest-ments, a small performance gain made in the design can translate intosignificant increase in profits, it is desirable to solve the problem toglobal optimality. A novel global optimization method, called non-convex generalized Benders decomposition (NGBD), is developed toachieve this goal efficiently.

� TC-30Tuesday, 12:30-14:00CC-A31

Data confidentiality 1

Stream: Data ConfidentialityInvited sessionChair: Robert Garfinkel, OPIM, University of Connecticut, BoxU41IM, 06029, Storrs, CT, United States,[email protected]

1 - Solving L1-CTA in 3D tables by an interior-pointmethod for block-angular problemsJordi Cuesta, Chemical Engineering, University Rovira iVirgili, Av. Països Catalans 26, Sant Pere i Sant Pau, 43007,Tarragona, Catalunya, Spain, [email protected], JordiCastroGiven a table with sensitive information, CTA looks for the closest safetable. We focus on CTA for 3D tables using the L1 distance betweenthe original and protected tables. The resulting linear programmingproblems are solved by a specialized interior-point algorithm whichsolves the normal equations by a combination of Cholesky factoriza-tion and preconditioned conjugate gradients (PCG). Quadratic regular-izations are used to improve the performance of PCG. Computationalresults are reported for a set of linear optimization problems of up to50 million variables and 25 million constraints.

2 - Optimal Data-Independent Noise for Differential Pri-vacyJosep Domingo-Ferrer, Dept. of Computer Engineering andMaths, Universitat Rovira i Virgili, Av. Països Catalans 26,43007, Tarragona, Catalonia, Spain, [email protected],Jordi Soria-ComasDifferential privacy is a notion offering general and clear privacy guar-antees for the individuals in a database. The effect of the presenceor absence of any single individual on the response to a query is lim-ited. The response mechanism can be viewed as an output perturbationwhere the real response is computed and noise-added before release.The usual choice is Laplace noise adjusted to the global sensitivity ofthe query function. We show that Laplace noise is not optimal, and weconstruct optimal noise distributions.

3 - Solving L2-CTA by perspective reformulationsJordi Castro, Statistics and Operations Research, UPC, 08034,Barcelona, Catalonia, Spain, [email protected], ClaudioGentile, Antonio FrangioniControlled tabular adjustment (CTA) is an emerging technology forthe protection of tabular data. CTA minimizes the distance betweenthe (unsafe) original and the protected table. CTA results in challeng-ing MILP or MIQP problems, if, respectively, L1 or L2 distances areconsidered. We provide a novel analysis of perspective reformulationsfor the particular MIQP-CTA formulation, with pairs of linked semi-continuous variables. Computational results will be provided, togetherwith a comparison of the utility of the protected tables with L1-CTAand L2-CTA.

4 - On Generating a Secure, Representative Samplefrom a DatabaseRobert Garfinkel, OPIM, University of Connecticut, BoxU41IM, 06029, Storrs, CT, United States,[email protected]

A database manager (DBA) has access to a number of records pertain-ing to entities. The fields of the database can be partitioned into bins.Each entity can allow its record to be included in a sample, based on aprivacy constraint that specifies the minimum number of other partici-pants that satisfy "similarity’ conditions. Two measures are given, onelocal and one global, for the "representativeness’ of a sample based onthe population of the database. Then it is determined which samplesizes are privacy-feasible, and how to maximize representativeness forthese sample sizes.

� TC-31Tuesday, 12:30-14:00CC-A33

Applications of the Neural Networks

Stream: Data Mining, Knowledge Discovery and Artifi-cial IntelligenceContributed sessionChair: Yoshifumi Kusunoki, Osaka University, Japan,[email protected]: Anastasia Motrenko, Applied mathematics andmanagement, MIPT, Moscow, Russian Federation,[email protected]

1 - Supplier’s performance measurement using intelli-gent clusteringMeltem Gulserin, Industrial Engineering, Istanbul TechnicalUniversity, Macka, 34367, Istanbul, Turkey,[email protected], Mine Isik, Gulgun Kayakutlu

This research aims to propose a supplier performance measurementmodel including the supplier relations. Factors affecting the perfor-mance for procurement are accumulated through the literature surveybut improved with the expert support.The proposed model clusters thesuppliers according to the performance using Self Organized Mapsbased on Kohonen networks.This model benefits the historical data tosupport the future plans in a changing business environment. A casestudy on automotive suppliers is presented. This research will be valu-able for designing the roadmap of supplier management.

2 - Modelling the behaviour of masonry walls under outof plan loading by ANNM. Sami Donduren, Selcuk University, Turkey,[email protected]

In this study, out-of plane failures of two masonry walls having differ-ent mortar properties were investigated and the results were modeledwith Artificial Neural Networks (ANN).First masonry wall is knittedand plastered by conventional mortar and the second test specimen isknitted and plastered by improved mortar which has Sikalatex as anadditive. As a result, the failure strength of the wall increased by 25%,which had improved binding property with additive material. And ithas been seemed that the ANN model can produce acceptable results.

3 - Estimating the behaviors of reinforced concretecolumns under axial load effect by ANNM.Tolga Cogurcu, Selcuk University, Turkey,[email protected]

The aim of this study is to estimate the load capacity and stress-strainrelationship of columns which have the same cross-sectional areas un-der the effect of axial load by ANN. The results obtained in this studyhave shown that ANN is a sufficient method to estimate the behaviourof the columns. Results are analyzed with variance analysis on SPSSprogram and with security edge controls.

142


4 - Modelling the flexural behaviour of beams by ANNMehmet Kamanli, Selcuk University, Turkey,[email protected]

In this study artificial neural network method was used to predict thedeflection values of beams and compared with the experimental re-sults of a testing series. Fort that aim, 6 reinforced concrete beamswith constant rectangular cross-section were prepared and tested underpure bending. The neural network procedure was applied to determineor predict the deflection values of 1/1 scaled model beams. The analyt-ical results were compared with the test results and further predictionsincluding different mix designs can be possible at the end of the study.

� TC-32Tuesday, 12:30-14:00CC-A34

Risk and Uncertainty in Energy Models

Stream: Emissions Trading and EnergyInvited sessionChair: Dominik Möst, Chair of Energy Economics, TechnischeUniversität Dresden, Hertzstraße 16, 01062, Dresden, Germany,[email protected]

1 - An Electricity Market Model with Generation Capac-ity Investment under UncertaintyAndreas Schröder, Energy, Transportation, Environment,DIW Berlin, Mohrenstr. 58, 10117, Berlin, Preussen,Germany, [email protected]

This article presents an electricity dispatch model with endogenouselectricity generation capacity expansion for Germany over the hori-zon 2035. The target is to quantify how fuel price uncertainty impactsinvestment incentives of thermal power plants. Results point to twofindings which are in line with general theory: Accounting for stochas-ticity increases investment levels overall and the investment portfoliotends to be more diverse.

2 - Wind power in Europe: Modeling of wind feed-in timeseries with an hourly resolutionAlexander von Selasinsky, TU Dresden, Germany,[email protected], Michael Zipf

The research to be presented addresses the modeling of wind feed-in for EU27-countries. Based on six hour time series of wind speedthat are publically available, it is shown how time series of hourlywind feed-in can be constructed. In this context, an algorithm is pre-sented that enables the transition from six hour to one hour time series.Whereas modeling wind feed-in on an hourly resolution is especiallyrelevant for energy system analyses, the research to be presented evalu-ates calculated data by discussing the smoothing of wind power outputthrough the spatial distribution of generation.

3 - Power Transmission Grid Expansion using BendersDecompositionDavid Gunkel, Chair of Energy Economics, TU Dresden,Münchner Platz 3, 01069, Dresden, Germany,[email protected], Friedrich Kunz

Power generation and demand of electric power spatially diverge as aresult of the proceeding energy transition. Consequently, the networkhas to be extended to cope with the new requirements. This paper ad-dresses a way to solve large-scaled mixed-integer problems of networkexpansion planning applying benders decomposition. The results ofthe algorithm are exemplarily shown in a grid model.

4 - Valuation of Pumped Hydro Storages with respect tothe Electricity Spot and Reserve Power MarketsOliver Woll, Universitaet Duisburg-Essen, Universitaetsstraße12, 45117, Essen, Germany, [email protected], BastianFelix, Christoph Weber

To cope with increasing fluctuating renewable infeed, new pump stor-age capacity needs to be installed to guarantee the stability of the powergrid. Whereas recent scientific publications focus on the electricityspot market for the valuation of pump storages, we propose an ap-proach which combines the spot- with a reserve market optimization.We derive the optimal operating strategy theoretically and solve theoptimization problem by applying dynamic programming on a (nu-merical constructed) recombining price tree. In addition, we calculatethe value of a representative pump storage.

� TC-33Tuesday, 12:30-14:00CC-A37

Understanding, Modeling andImplementation of OR for SustainableDevelopment

Stream: OR for Development and Developing Coun-triesInvited sessionChair: Alexander Makarenko, Institute for Applied SystemAnalysis, National Technical University of Ukraine "KPI",Prospect Pobedy 37, 03056, Kiev, Ukraine, [email protected]

1 - Foreign Direct Investment Trends and ComparativeAnalysis for Chosen Transition CountriesSnjezana Pivac, Faculty of Economics, University of Split,Department of Quantitative Methods in Economics, Maticehrvatske 31, 21000, Split, Croatia, [email protected]

Foreign direct investments (FDI) have high positive impact on eco-nomic impetus especially in transition countries. So in the focus ofthis paper are the structure and forms of FDI and their influence oncrucial macroeconomic variables. In the FDI inflow analysis the coun-try risk rating estimation has been essential. FDI efficiency analysisis enriched with correlation between Incremental Capital Output Ratio(ICOR) and GDP increase. Multivariate classification of ten transitioncountries has been carried out according to all regarding indicators.

2 - Panel data analysis of institutional variables in at-tracting foreign direct investment in countries of cen-tral and southeastern EuropeElza Jurun, Faculty of Economics, University of Split,Department of Quantitative Methods in Economics, Maticehrvatske 31, 21000, Split, Croatia, [email protected], MarioPecaric, Tino Kusanovic

The aim of the paper is to find out the relevance of institutional fac-tors vis-à-vis traditional variables indicating macroeconomic stabil-ity in attracting foreign direct investment as a means of further eco-nomic restructuring and international competitiveness improvement.The Arellano-Bond dynamic panel based on multivariate factor analy-sis of related variables has been carried out for 11 countries from Cen-tral and South-Eastern Europe. The results show that after macroeco-nomic stability is realised institutional reform dynamism gains on theimportance

3 - General Sustainable Development Problems and ORAlexander Makarenko, Institute for Applied System Analysis,National Technical University of Ukraine "KPI", ProspectPobedy 37, 03056, Kiev, Ukraine, [email protected]

The sustainable development (SD) concept is one of general recentproblems. Many definitions of SD exist: the definition by Bruntlandis most recognized and usable, but this definition is verbal. So forcomputer modeling or quantitative decision—making on SD, furtherdevelopment and formalization of the concept are necessary. So wepropose the new formal definition of sustainable development. Alsowe describe the role which strong anticipation plays in sustainable de-velopment. Finally the new prospects for proposed ideas on SD aredescribed, including OR and education problems.

143


4 - A two-stage approach for distribution network designproblem in cosmetic & cleaning industryRifat Gürcan Özdemir, Industrial Engineering Department,Istanbul Kültür University, Atakoy Campus,Atakoy-Bakirkoy, 34156, Istanbul, Turkey,[email protected], Zeki Ayag, Ufuk Kula

The paper proposes a two-stage approach which combines a linearmathematical model and a simulation model to solve the problem ofdistribution network design in one of the leading cosmetic and clean-ing companies. The mathematical model determines number of ware-houses, amount of material flows and selects third party logistic com-panies based on their vehicle fleet size and types of vehicles inside thefleet. The simulation model involving uncertain nature of delivery isused to modify parameters of the mathematical model according to theperformance of network system based on service time.

� TC-34Tuesday, 12:30-14:00CC-A39

Optimization for Sustainable Development2Stream: Optimization for Sustainable DevelopmentInvited sessionChair: Herman Mawengkang, Mathematics, The University ofSumatera Utara, FMIPA USU, KAMPUS USU, 20155, Medan,Indonesia, [email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - Multi objective stochastic programming model forwaste management in crude palm oil industryMeslin Silalahi, Mathematics, University ofSisingamangaraja-Tapanuli/Grad School of MathematicsUSU, FMIPA USU, 20155, Medan, North Sumatera Province,Indonesia, [email protected], HendaruSadyadharma, Herman Mawengkang

The crude palm oil process does not need any chemicals as a pro-cessing aid. However, there are a number of environmental problems,such as high water consumption, the generation of a large amount ofwastewater with a high organic content, and the generation of a largequantity of solid wastes and air pollution. We include constraints in-volving random parameters of waste, which are required to be satisfiedwith a pre-specified probability threshold. In this paper we propose amulti-objective stochastic optimization model for tackling such envi-ronmental risk production planning problem.

2 - Optimization approach for measuring the ecologicalsuitability of product packagingZoran Rakicevic, Faculty of Organizational Scineces,University of Belgrade, Jove Ilica 154, 11000, Belgrade,Serbia, [email protected], Jovana Kojic

Limitation of available resources in the world and the general need ofreducing operating costs create a need to review the entire process ofcreating each element of the product, such as product packaging. Thispaper presents a new and different approach to the analysis of the eco-logical suitability of product packaging, using an optimization modelthat is focused on its dimensions. General method is further appliedto some of the most commonly used consumer products. The studyshows that suboptimal packaging leads to higher material consump-tion and increase of prices of the final products.

3 - Modeling the spread of infectious disease based ondynamic social networkFirmansyah Firmansyah, Mathematics, University MuslimNusantara, FMIPA USU, 20155, Medan, North SumateraProvince, Indonesia, [email protected], HermanMawengkang

The dynamic of human contact can be considered as a potential pointfor predicting the spread of infectiuous disease. In this paper we con-sider the SIRS model. We include the dynamic of human social inter-action and social distance in the model such that it would be possibleto control the spread of the disease.

4 - Decomposition-based optimization of a large-scalemesh-structured drinking water supply networkDerek Verleye, Industrial Management, Ghent University,Technologiepark 903, 9052, Zwijnaarde, Belgium,[email protected], El-Houssaine Aghezzaf

Optimization of production and distribution of drinking water in largenetworks leads to significant economic savings. However, the result-ing optimization model belongs to the class of mixed integer nonlinearproblems (MINLP). In addition to this inherent theoretical complex-ity some hydraulic restrictions are nonconvex. In this presentation wediscuss some possible decomposition approaches to solve practical in-stances of this challenging optimization problem. These instances aregenerated from an existing drinking water network in Flanders, Bel-gium.

� TC-35Tuesday, 12:30-14:00CC-A41

Innovation in Logistics

Stream: OR and Real ImplementationsInvited sessionChair: Belarmino Adenso-Diaz, Engineering School at Gijon,Universidad de Oviedo, Campus de Viesques, 33204, Gijon,Spain, [email protected]: Ben Lev, Decision Sciences, Drexel University, LeBowCollege of Business, 101 N. 33rd st., 19104, Philadelphia, Pa,United States, [email protected]

1 - Company-wide Production Planning Using a MultipleTechnology DEA approachSebastián Lozano, Dept. of Industrial Management,University of Seville, Escuela Superior de Ingenieros, Caminode los Descubrimientos, s/n, 41092, Seville, Spain,[email protected]

The proposed centralized-DEA approach aims at minimizing the in-put costs incurred to attain specified total, company-wide output lev-els. Specific features of the proposed approach are the considerationof multiple technologies, lower and upper bounds on the productionlevels of each plant and the possibility of shutting down a plant, if nec-essary. The consideration of multiple technologies, in particular, takesinto account the existence of heterogeneity within the company’s facil-ities. In that scenario, grouping all plants within a single technologygives misleading results.

2 - Bicriteria model for designing a container depots lo-gistic networkBelarmino Adenso-Diaz, Engineering School at Gijon,Universidad de Oviedo, Campus de Viesques, 33204, Gijon,Spain, [email protected], Antonio Palacio, SalvadorFurió, Sebastián Lozano

The container is a basic tool for all maritime logistic operations. Onceunloaded by the importers, containers must be stored in the terminalport or in a depot, waiting for an exporter needing them. The locationof these container depots has an impact on the logistic costs, as wellas in the environment of the area, given the heavy traffic load inherentto these operations. In this research we propose a bicriteria model fordesigning the depots network in a hinterland. Results are applied to thecase of the Port of Valencia.

3 - A mathematical model to design an inland network ofintermodal terminalsCarlos Andres, Management Department, PolytechnicUniversity of Valencia, Camino de Vera S/N, 46022, Valencia,Spain, [email protected], Salvador Furió, Julien Maheut

144


This paper shows how a mathematical model based on the typical hublocation problem has been used to develop a first analysis for the de-sign of a network of intermodal inland terminals giving support to mainSpanish container seaports and their traffics. The main objective of themodel is to reduce total transport costs through the optimal location ofa set of inland terminals connected to seaports by railway. The prob-lem has been successfully solved using Gusek software, allowing aninteresting analysis about inland terminals location and their influencearea.

� TC-36Tuesday, 12:30-14:00CC-A43

QMPD Session1Stream: Quantitative Models for Performance and De-pendabilityInvited sessionChair: Markus Siegle, Computer Science, Universitaet derBundeswehr Muenchen, Universitaet der Bundeswehr Muenchen,85577, Neubiberg, Germany, [email protected]

1 - Memory-efficient bounding algorithm for the two-terminal reliability problemMinh Lê, I10, TU Munich, Boltzmannstraße 3, 85748,Garching b. München, Bayern, Germany, [email protected] the two-terminal reliability of arbitrary redundancy sys-tems modeled as combinatorial graphs is known to be NP-hard. Thus,bounding algorithms are used to cope with large graph sizes. Still,they have huge demands on memory. We propose a memory-efficientimplementation of an extension of the Gobien-Dotson bounding al-gorithm. Without reducing runtime, the compression of relevant datastructures allows to use low-bandwidth high-capacity storage. Thisway, available hard disk space becomes the limiting factor. We canhandle graphs with sizes reaching around 100 components.

2 - An enhanced two-phase MAP fitting methodAndrás Mészáros, Department of Telecommunications,Budapest University of Technology and Economics, MagyarTudósok Körútja 2., 1117, Budapest, Hungary,[email protected], Miklós TelekMarkov arrival processes (MAPs) are used extensively in traffic mod-eling. A wide variety of fitting procedures have been developed. Mostof these are computationally demanding or not general enough. To re-solve this problem, two-step procedures of a specific type have beenmade, which fit a phase-type distribution (PH) to static parameters inthe first step, and extend it to a MAP in the second while fitting dy-namic parameters. We present a method that aims at providing a goodstarting point for the second step by optimizing the representation ofthe PH that was produced by the first step.

3 - Time-parallel simulation for stochastic automata net-worksJean-Michel Fourneau, PRiSM, UVSQ, Universite deVersailles, 45 avenue des Etats Unis, 78000, Versailles,France, [email protected], Franck Quessette, Thu Ha DaoThiTime Parallel Simulation (TPS) is the construction of the time-slices ofa sample-path on a set of parallel processors (see Fujimoto2000, chap.6 and references therein). TPS has a potential to massive parallelism asthe number of logical processes is only limited by the number of timesintervals which is a direct consequence of the time granularity and thesimulation length. Stochastic Automata Networks allow the construc-tion of extremely large Markov chains. Here we show how we can useTPS to solve efficiently some SAN models. The approach is based onsome notions of monotone models.

4 - M/G/c/c state dependent travel time models andpropertiesJ. MacGregor Smith, Mechanical and Industrial Eng. Dep.,University of Massachusetts at Amherst, 114 Marston Hall,University of Massachusetts, 1003, Amherst, MA, UnitedStates, [email protected], F. R. B. Cruz

We show that state dependent M/G/c/c queuing models provide a quan-titative foundation for three-phase traffic flow theory. An importantproperty shown with the M/G/c/c model is that the travel time functionis not convex, but S-shaped (i.e. logistics curve). Extensive analyticaland simulation experiments are shown to verify the S-shaped nature ofthe travel time function and the use of the M/G/c/c model’s method ofestimation of travel time over vehicular traffic links as compared withtraditional approaches.

� TC-37Tuesday, 12:30-14:00CC-Act

OR in Health & Life Sciences 3Stream: OR in Health & Life Sciences (contributed)Contributed sessionChair: Yw Chen, DYU, Taiwan, [email protected]

1 - Planning in a regional blood distribution networkJose luis Andrade, Industrial Management, School ofEngineering, University of Seville, Camino de losDescubrimientos s/n, 41092, Seville, Spain,[email protected], Pedro L. Gonzalez-R

We present an optimization model to plan the supply of blood productsto hospitals in a regional blood distribution network. Blood units arefixed shelf life resources to be used for the demanded surgeries, con-sidering urgency and blood substitution rules. The model outcomesweekly decisions on the inventory levels of blood products and theamount of units transferred from one location to another, to performlow wastage and shortage levels. Our plan offers a single strategyproviding an average good performance with uncertain demands andoutcomes the needed level of donation promotion.

2 - A cloud service to allocate emergent patients byaffinity setYuh-Wen Chen, Institute of IE and MOT, Da-Yeh University,112 Shan-Jeau Rd., Da-Tsuen„ Chang-Hwa 51505, Taiwan.,51505, Chang-Hwa, Taiwan, [email protected]

A large disaster always generates numerous trauma patients in a veryshort time. This study develops a cloud service to allocate these pa-tients to appropriate hospitals in order to save their lives and guaranteethe service level of emergency medicine. First, the web-platform ofcapturing vital signs of emergent patients is established. Second, a de-cision model of multi-attribute decision making based on affinity setis developed. Finally, the decision model is practically launched inthe web-platform to aid the transport decision for allocating emergentpatients to hospitals.

3 - A business model for synchronized and patient-centered Tele-health serviceJiun-Yu Yu, Business Administration, National TaiwanUniversity, No. 85, Sec. 4, Roosevelt Road„ 106, Taipei,Taiwan, [email protected], Kwei-Long Huang

Tele-health is a promising solution to take care of patients with car-diovascular disease. In this study, an integrated approach is appliedto construct a self-sustainable business model for a newly establishedTele-Health Center (THC) to dynamically measure and continuouslyimprove its performance. With process analysis, VISOR framework,and balanced scorecard, a business model in the form of key perfor-mance indicator (KPI) matrix is constructed. The causal structure ofKPIs is investigated so that the growth-stimulating loops are identified,helping THC prioritize its tasks to move forward.

145


� TC-38Tuesday, 12:30-14:00HH-Colombus

Theory of Set-valued Optimization andApplications

Stream: Multiobjective OptimizationInvited sessionChair: Tamaki Tanaka, Mathematics, Niigata University, Ikarashi2, 8050, 950-2181, Niigata, Niigata, Japan,[email protected]: Gue Myung Lee, Department of Applied Mathematics,Pukyong National University, 599-1 Daeyeon 3-dong Nam-gu,608–737, Busan, Korea, Republic Of, [email protected]

1 - On Nonsmooth Optimality Theorems for Robust Mul-tiobjective Optimization ProblemsGue Myung Lee, Department of Applied Mathematics,Pukyong National University, 599-1 Daeyeon 3-dongNam-gu, 608–737, Busan, Korea, Republic Of,[email protected]

In this talk, we present optimality theorems for the robust counterparts(the worst-case counterparts) of nonsmooth multiobjective optimiza-tion problems with uncertainty data, which are called robust multiob-jective optimization problems. The involved functions in the problemsare locally Lipschitz. Moreover, we give examples showing that theconvexity of the uncertain sets and the concavity of the constraint func-tions are essential in the optimality theorems.

2 - On DC set optimization problemsDaishi Kuroiwa, Shimane University, Japan,[email protected]

Set optimization is a minimization problem whose objective functionis set-valued and criteria of solution are based on set-relations. In thispresentation, we propose a notion of DC set-valued functions and westudy optimization problems with an objective DC set-valued function.

3 - Unified scalarization for sets and Ekeland’s varia-tional principle for set-valued mapsTamaki Tanaka, Mathematics, Niigata University, Ikarashi 2,8050, 950-2181, Niigata, Niigata, Japan,[email protected]

In this paper, we introduce a certain mathematical methodology on thecomparison between sets and Ekeland’s variational principle for set-valued maps via several scalarizing functions. The original idea on thesublinear scalarization for vectors was dealt by Krasnosel’skij in 1962and by Rubinov in 1977, and then it was applied to vector optimizationwith its concrete definition by Tammer (Gerstewitz) in 1983. Recently,Kuwano, Tanaka, and Yamada define twelve types of unified scalariz-ing functions for sets based on six kinds of set-relations proposed byKuroiwa, Tanaka, and Ha in 1997.

� TC-39Tuesday, 12:30-14:00HH-Cousteau

Latest advances on MCDA softwareStream: Innovative Software Tools for MCDAInvited sessionChair: Bertrand Mareschal, Solvay Brussels School of Economicsand Management, Université Libre de Bruxelles, Boulevard duTriomphe CP 210/01, 1050, Brussels, Belgium,[email protected]

1 - On Multi-Criteria ClusteringAlexandru-Liviu Olteanu, CSC, University of Luxembourg,Luxembourg, [email protected], Patrick Meyer,Raymond Bisdorff

A closely related type of problem in MCDA to that of sorting is clus-tering. As in Data Mining, in MCDA too, no unanimous definition ofclustering is given. We address this by formally defining several typesof clustering in a MCDA context. We base our definitions on any out-ranking relation from which different preferential situations can be ex-tracted between the alternatives. We define the ideal result with respectto these definitions and develop a method to find them. We present theimplementation of this approach in the diviz software and illustrate itusing a well-known problem in MCDA.

2 - A new web service for the stable sorting of a set ofalternativesThomas Veneziano, CSC, University of Luxembourg,Universite du Luxembourg, FSTC, 6, rue RichardCoudenhove-Kalergi, 1359, Luxembourg, Luxembourg,[email protected], Patrick Meyer, Raymond Bisdorff

We present a new web service which computes the stable sorting ofsome alternatives into predefined categories, bounded with some givenprofiles, according to the principles of the outranking methods. It takesinto account the notion of stability, which informs on the dependencyof the outrankings to the preorder of the weights, and defines for eachalternative an interval of categories on which it is affected in a sta-ble manner: knowing the preorder of the weights only, not a precisesetting, we ensure that the alternative will be sorted in one of the cate-gories of the interval.

3 - Visual PROMETHEE - A New Multicriteria DecisionAid SoftwareBertrand Mareschal, Solvay Brussels School of Economicsand Management, Université Libre de Bruxelles, Boulevarddu Triomphe CP 210/01, 1050, Brussels, Belgium,[email protected]

Visual PROMETHEE is the latest software implementation of thePROMETHEE and GAIA multi-criteria decision aid methods. It in-cludes all the standard features of the methods: PROMETHEE I and IIrankings, GAIA descriptive analysis, weight sensitivity analysis anddata management tools. New features are introduced: visual rep-resentations of the partial and complete PROMETHEE rankings, 3-dimensional GAIA analysis, visual weight stability intervals, GAIA-web action profile display, input-output analysis, GIS (Google Maps)integration and an enhanced multi-scenarios model for GDSS capabil-ities.

� TC-40Tuesday, 12:30-14:00HH-Livingstone


Stream: Preference LearningInvited sessionChair: Krzysztof Dembczynski, Institute of Computing Science,Poznan University of Technology, Piotrowo 2, 60-965, Poznan,Poland, [email protected]

1 - Efficient algorithms for robust ordinal regressionTommi Tervonen, Econometric Institute, Erasmus UniversityRotterdam, PO Box 1738, 3033HA, Rotterdam, Netherlands,[email protected], Remy Spliet

UTAGMS is an ordinal regression method that evaluates necessary andpossible preference relations caused by holistic preference statementson a set of decision alternatives evaluated in terms of multiple crite-ria. Traditionally the relations are computed by defining appropriateLPs and using a solver. In our presentation we show that the relations’computation can be formulated differently and asymptotically fasteralgorithms used. We also present and evaluate an MCMC samplingtechnique for uniform sampling of the general value functions used instochastic extensions of UTAGMS.

146


2 - Multicriteria sorting functions using valued assign-ment examplesOlivier Cailloux, Laboratoire Génie Industriel, École CentraleParis, Grande Voie des Vignes, 92295, Châtenay-Malabry,France, [email protected], Brice Mayag, VincentMousseau, Luis C. DiasWe propose to extend a classical elicitation technique for learningMCDA sorting functions. Such a function assigns each alternativefrom a set of alternatives, evaluated on multiple criteria, to a categoryin a set of preference ordered categories. We ask the decision makerfor a set of assignment examples where each example contains severalpossible destination categories for an alternative, each associated withcredibility. We propose procedures to deduce from this informationseveral sorting functions, giving more or less precise results dependingon the chosen credibility level.

3 - On the estimation of the parameters of Electre Trimodel in multi criteria ordinal sorting problem: a pro-posal of new approach in two phasesValentina Minnetti, Scienze Statistiche, Università LaSapienza, Italy, [email protected] this paper we are interested in the Multi Criteria Sorting Problem; tosolve it we use the Electre Tri method, based on the outranking relationcreated through the Decision Maker’s preference information in termsof the Assignment Examples (AEs). This method requires the elici-tation of preferential parameters (weights, thresholds, profiles, cuttinglevel) in order to construct a preference model which best matches theAEs. We propose a new methodology in two phases, taking into ac-count that the core of the analysis is the profiles’ estimation made bylinear programming problem.

4 - A ranking approach based on outranking relationsusing mathematical programmingSelin Ozpeynirci, Industrial Systems Engineering, IzmirUniversity of Economics, Sakarya Cad. No156, 35330, Izmir,Turkey, [email protected], Ozgur Ozpeynirci, YigitKazancogluWe consider the problem of ranking alternatives evaluated on multiplecriteria, which may arise in many real life applications such as rankingcountries, projects, universities and students. Several outranking meth-ods in the literature require the explicit definition of all parameters inadvance, which can be difficult for the decision maker. In this study,we develop an outranking based ranking method that uses mathemat-ical modeling approach and requires only a subset of the parametersdefined by the decision maker. A real life example is used to illustratethe method.

� TC-41Tuesday, 12:30-14:00HH-Heyerdahl

AHP/ANP 6Stream: Analytic Hierarchy Process, Analytic NetworkProcessInvited sessionChair: Pekka Korhonen, Economics and Management Science,Helsinki School of Economics, Runeberginkatu 22-24, 100,Helsinki, Finland, [email protected]

1 - R&D Personnel Selection with Fuzzy Analytic Hierar-chyErhan Berk, Defence Science, Hava Duragi Lojmanlari 4.BlokDaire:1, 06790, Ankara, Turkey, [email protected] the 21st century where the competitive environment becomes verytough for companies in parallel with the drastic development in tech-nology, only innovative companies became prominent. The R&D per-sonnel in particular have played the major role in these emerging inno-vations. In this study a Fuzzy Analytical Hierarchy Process (AHP)method for the selection of R&D personnel has been proposed. Acase study has been conducted by using the weights obtained from theFuzzy AHP method and the results have been compared.

2 - Determining effective criteria of Customers’ satisfac-tion in kitchen worktops by using AHPMajid Azizi, Department of Wood and Paper, Faculty ofNatural Resources,University of Tehran, Daneshkadeh St.,31585-4314, Karaj, Tehran, Iran, Islamic Republic Of,[email protected], Vahid Nabavi, Mehdi Faezipour

Determination of effective criteria in customer satisfaction of work-tops is very important for any company that produces this product. Toidentify effective criteria for customers’ satisfaction five major groupswhich are selling condition and price, delivery condition, guaranty, rawmaterial quality and technological specification were selected. Thesemajor groups have 19 sub-criteria. The priority rate of these criteriaand sub-criteria was obtained by AHP. Sub-criteria of price, paymentcondition, top veneer quality, MDF quality and fast delivery had thehighest priorities, respectively.

3 - Analysing numerically three different methods to es-timate a priority vector from inconsistent pairwisecomparison matrices in AHPAttila Poesz, Corvinus University of Budapest, Fovam ter 8,1093, Budapest, Hungary, [email protected],Pekka Korhonen

In our paper we present a simulation framework in which the perfor-mance of the estimating methods used to determine priority vectorcould be analysed in various inconsistency levels. Furthermore, westudy whether the implementation of the "Condition of Order Preser-vation" (COP) defined by Bana e Costa and Vasnick (2008) could pro-vide a better solution for the decision makers. The importance of ouranalysis is based on the author’s claim that the COP has to be satis-fied by any priority methods, otherwise the method has a fundamentalweakness.

� TC-42Tuesday, 12:30-14:00BW-Amber

Queueing Systems I

Stream: Stochastic Modeling and Simulation in Engi-neering, Management and ScienceInvited sessionChair: Erik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39, 85577,Neubiberg, Germany, [email protected]: Büsra Temocin, Middle East Technical University, 06531,Ankara, Turkey, [email protected]: Devin Sezer, Middle East Technical University, Institute ofApplied Mathematics, Eskisehir Yolu, 06531, Ankara,[email protected]

1 - The Availability and Separability of Unreliable Net-worksYat-wah Wan, Institute of Logistics Management, NationalDong Hwa University, 1 Sec. 2, Da-Hsueh Road, Shou-Feng,974, Hualien, Taiwan, [email protected]

Many networks of reliable stations have their steady-state distributionsin product form. Similar separable results exist for networks with unre-liable stations, where arrivals finding stations unavailable are blocked,and the ups and downs of stations are reversible processes independentof numbers of customers at stations. We show that the steady-state dis-tribution of an unreliable network can still be separable when ups anddowns of stations are dependent in some strong sense. We also lookfor general separable results for state-dependent arrivals and ups anddowns of stations.

2 - Inequality of Stability in the M2/M2/1 Queue with Pre-emptive PriorityHamadouche Naima, Recherche Operational, LAMOS,University of Bejaia, Algerie, Bejaia, Algeria,[email protected]

147


This paper studies the M2/M2/1 queue with preemptive priority. Weuse the strong stability method to approximate the characteristics ofthe M2/M2/1 queue with preemptive priority by those of the M/M/1queue. When the arrival intensity of the priority request is sufficientlysmall. First we give conditions of this approximation, second we givethe bound of approximation, in the last, we includes a numerical ex-ample and simulation results.

3 - Transient queues with correlated arrivals. Algorith-mic approachClaus Gwiggner, Information Systems, FU Berlin, Garystr.21, 14195, Berlin, Germany, [email protected],Sakae NagaokaWe look into a queueing system where customers are pre-scheduled butarrive with a certain delay. This is an old problem with few analyticalresults. In particular, the arrival counting process contains serial cor-relation. We identify this correlation structure for some special cases.Based on this we obtain transient solutions of the queue length. Asan application we derive decision rules for aircraft delay managementunder uncertainty.

4 - Optimal control by unreliable queueing systems withbroadcasting serviceAlexander Dudin, Applied Mathematics and ComputerScience, Belarusian State University, 4, Independence Ave.,Minsk-30, 220030, Minsk, Belarus, [email protected], Bin SunWe consider an unreliable MAP/PH/N type queueing system whereeach customer may be served in parallel by several available servers.We show such advantages of this service discipline in comparison tothe classical service discipline as smaller average sojourn time andhigher probability of the correct service. Problem of the optimal choiceof the limiting number of servers, which can be involved into the ser-vice of an arbitrary customer, is numerically solved. Numerical exam-ples illustrating effect of correlation in arrival process are presented.

� TC-43Tuesday, 12:30-14:00BW-Granite

Stochastic programming in industry II

Stream: Stochastic ProgrammingInvited sessionChair: Csaba I. Fabian, Operations Research, Eotvos LorandUniversity, Pazmany Peter setany 1/C, 1117, Buapest, Hungary,[email protected]: Achim Koberstein, DS&OR Lab, University of Paderborn,33098, Paderborn, Germany, [email protected]

1 - On integrating financial hedging decisions into amodel for global production network designAchim Koberstein, Business Administration,Goethe-University of Frankfurt, Grueneburgplatz 1, 60323,Frankfurt am Main, Germany,[email protected], Elmar Lukas, MarcNaumannIn this talk we present a multi-stage stochastic programming modelfor integrated planning of strategic production network design and fi-nancial hedging under uncertain demands and exchange rates. Finan-cial portfolio planning decisions on two types of financial instruments,namely forward-contracts and options, are explicitly represented bymulti-period decision variables and a multi-stage scenario tree. Weshow computational results on an illustrative case study, which indi-cate that the integrated model can result in better planning decisionscompared to the traditional separated approach.

2 - A Stochastic Second Order Cone Model for aStochastic Capacitated Traveling Salesmen LocationProblem with RecourseFrancesca Maggioni, Dept. of Mathematics, Statistics,Computer Science and Applications, University of Bergamo,Via dei Caniana n. 2, 24127, Bergamo, Italy, Italy,[email protected]

In this talk we present a Stochastic Capacitated Traveling SalesmenLocation Problem in which a single facility has to be located in a givenarea. This facility is used to serve a set of customers with unknownposition by traveling salesmen. The aim is to determine the servicezone that minimizes the expected cost of the traveled routes. We pro-pose a two-stage stochastic second order cone program where scenar-ios are represented by different sets of customers. Optimal solutionsare compared with the Capacitated Traveling Salesmen Location Prob-lem heuristic proposed by Simchi-Levi (1991).

3 - Optimal timing and capacity choice for pumped hy-dropower storageAne Marte Heggedal, ENOVA, 7491, Trondheim, Norway,Norway, [email protected], Emily Fertig, GerardDoorman, Jay Apt

Pumped hydropower storage can smooth output from intermittent re-newable electricity generators, facilitating their large-scale use in en-ergy systems. Pumped storage ramps quickly enough to smooth windpower and could profit from arbitrage on the short-term price fluctua-tions wind power strengthens. We consider five capacity alternativesfor a pumped storage facility in Norway that practices arbitrage in theGerman spot market. Real options theory is used to value the invest-ment opportunity, since unlike net present value, it accounts for uncer-tainty and intertemporal choice.

4 - Mathematical Optimization of an Industrial-Scale En-gineering SystemSelis Onel, Chemical Engineering, Hacettepe University,Department of Chemical Engineering, Beytepe Campus,06800, Beytepe, Ankara, Turkey, [email protected]

Performance of the drying section of a paper machine is analyzedbased on desired production rate while keeping drying parameterswithin the limits. A steady state analytical model is developed by em-ploying material and energy balances on the system and is solved usingdata collected for several paper grades to determine the drying parame-ters most effective on energy consumption and the drying efficiency. Alinear programming method is used for the optimization of the param-eters minimizing the total cost for electricity, natural gas, and steamwhile maximizing the efficiency.

� TC-44Tuesday, 12:30-14:00BW-Marble

Ensemble Learning and Artificial NeuralNetworksStream: Machine Learning and its ApplicationsInvited sessionChair: Sureyya Ozogur-Akyuz, Department of Mathematics andComputer Science, Bahcesehir University, Bahcesehir University,Dept of Mathematics and Computer Science, Cıragan cad.Besiktas, 34353, Istanbul, Turkey,[email protected]: Terry Windeatt, Electronic Engineering, Surrey University,Faculty of Engineering & Physical Sciences, University of Surrey,GU2 7XH, Guildford, Surrey, United Kingdom,[email protected]

1 - New Considerations on Bias-variance Decomposi-tion in Ensemble Classifer FusionDavid Windridge, FEPS, University of Surrey, UnitedKingdom, [email protected]

Much progress has been made in the area of decision fusion thoughthe separation of the process into its effects on the bias and varianceof the ensemble. We investigate a previously unconsidered aspect ofthis process when feature spaces within the constituent processes arenon-overlapping, drawing an analogy between the processes of classi-fier decision fusion and tomographic reconstruction. We demonstratethat standard classifier fusion introduces a morphological bias at thesame time as reducing variance. By taking this into account we cansubstantially improve classification performance.

148


2 - Ensemble Pruning via DC ProgrammingSureyya Ozogur-Akyuz, Department of Mathematics andComputer Science, Bahcesehir University, BahcesehirUniversity, Dept of Mathematics and Computer Science,Cıragan cad. Besiktas, 34353, Istanbul, Turkey,[email protected], Terry Windeatt, RaymondSmithEnsemble learning is a method of combining learners, however theensemble sizes are sometimes unnecessarily large which causes extramemory usage and decrease in effectiveness. Error Correcting OutputCode (ECOC) is one of the well known ensemble techniques for multi-class classification which combines the outputs of binary base learnersto predict the classes for multiclass data. We formulate ECOC for en-semble selection problem by using difference of convex functions (dc)programming and zero norm approximation to cardinality constraint.Experiments show that it outperforms the standard ECOC.

3 - Short-term forecasting of light rail passenger de-mand by artificial neural networksGökhan Sürmeli, Management Engineering, IstanbulTechnical University, ITU Isletme Fakültesi Oda No: A406,Macka Kampusu-Besiktas, 34357, Istanbul, Turkey,[email protected], Dilay Çelebi, Demet BayraktarAccurate demand information is a key factor for the scheduling of pub-lic transportation systems. In this study, ANN is applied to passengerdemand forecasting problem for the light rail service in Istanbul. Dueto its success of solving forecasting problems and its simple architec-ture, an MLP model is preferred. Daily demand data is split into fourtime slots and neural networks are developed for each of them. Theperformance of the models is compared to the best fitting ARIMA orexponential smoothing models in terms of root mean squared error andmean absolute percentage error.

4 - Optimizing the Backpropagation Algorithm for Train-ing Artificial Neural NetworksGeraldo Miguez, COPPE / PESC, Universidade Federal doRio de Janeiro, Brazil, R Mariz e Barros, 652/502, 20270-002,Rio de Janeiro, RJ, Brazil, [email protected],Nelson Maculan, Adilson Elias XavierThis paper uses a new activation function, the Bi-Hyperbolic func-tion, which offers a higher flexibility, and a faster evaluation time.It improves the performance of the Backpropagation algorithm actingstraight at the topology of the network. Computational experimentswith traditional problems of the literature show greater generalizationcapacity, faster convergence, higher computational speed and smallernumber of neurons network architecture. The network configuration isalso simplified, making this process faster and easier to be completed.

� TC-45Tuesday, 12:30-14:00BW-Water

Logistics and stochastics 2

Stream: Stochastic Modelling in Logistical NetworksInvited sessionChair: Dieter Fiems, Ghent University, St-Pietersnieuwstraat 41,9000, Gent, -, Belgium, [email protected]: Eline De Cuypere, TELIN, University Ghent,St-Pietersnieuwstraat 41, 9000, Ghent, Belgium,[email protected]

1 - Estimating Repair Effects in Failing SystemsErnie Love, Segal Graduate School of Business, Simon FraserUniversity, 500 Granville Street, Vancouver, Canada, V6C1W6, Vancouver, British Columbia, Canada, [email protected] The purpose of this research is to estimate the repair effectsa repairable system undergoes repeated failures and repairs. Data wascollected on a cement kiln incurring150 failures and 5 shutdowns foroverhaul. Causes of failure were identified as operational, mechanicalor electrical. The unit was restarted but with an (expected) improvedfailure rate by virtue of the repair. Repair improvements are modeled isa Kijima Type-II virtual age framework. Covariate data was also usedto better estimate the failure/repair process.

2 - Confidence-based Optimization for the NewsvendorProblemRoberto Rossi, Business School, University of Edinburgh, 29Buccleuch place, EH8 9JS, Edinburgh, United Kingdom,[email protected], Steven Prestwich, Armagan Tarim,Brahim Hnich

We consider a Newsvendor problem in which the decision maker isgiven a set of past demand samples. We employ confidence intervalanalysis in order to identify a range of candidate order quantities that,with prescribed confidence probability, includes the real optimal or-der quantity for the underling stochastic demand process with knowndistribution and unknown parameter(s). For these order quantities, wealso produce cost upper and lower bounds. We apply our approach toBinomial, Poisson, and Exponential demand. For two of these distri-butions we discuss the case of unobserved lost sales.

3 - Economic order quantity of an inventory control sys-tem with order backlogEline De Cuypere, TELIN, University Ghent,St-Pietersnieuwstraat 41, 9000, Ghent, Belgium,[email protected], Koen De Turck, DieterFiems

We present a mathematical model which evaluates the performance ofinventory control systems with order backlog. In terms of cost, we ac-count for holding, set-up and backlog costs. Production of units startswhen the inventory level drops below the threshold value and stopswhen stock capacity is reached. Moreover, orders are backlogged andcoupled to the completion of the units. The inventory problem is stud-ied in a Markovian setting. Set-up times are phase-type distributedwhile order and product arrivals are modelled by a Markovian arrivalprocess.

4 - Two Modified Preventive Maintenance Models in a Fi-nite Time Span with Failure Rate ReductionChun-Yuan Cheng, Dept. of Industrial Engineering &Management, Chaoyang Univerisity of Technology, 168Gifong E. Rd., Wufeng„ 413, Taichung County, Taiwan,[email protected], Min Wang, Mei-Ling Liu

For a preventive maintenance (PM) model over a finite time span,shorter PM interval can result in a better expected total maintenancecost (TC). However, the original optimal PM policies shown in liter-ature are obtained by searching the optimal PM interval over a con-strained range. We propose two modified failure-rate-reduction opti-mal PM policies: one is developed by releasing the constraint of thePM interval searching range; another is constructed by postponing thestarting time of the PM program based on the fact that equipment isnot easily fail in the early stage of its useful life.

149

TD-01 EURO 25 - Vilnius 2012

Tuesday, 14:30-16:00

� TD-01Tuesday, 14:30-16:00RB-Alfa

Tutorial Lecture: Professor JitkaDupacova

Stream: Keynote, Tutorial and Plenary LecturesTutorial sessionChair: Stein W. Wallace, Management Science, LancasterUniversity Management School, LA1 4YX, Lancaster, UnitedKingdom, [email protected]

1 - Stochastic programming - a flexible tool for decisionmaking under uncertaintyJitka Dupacova, Probability and Math. Statistics, CharlesUniv, Faculty of Math. and Physics, Sokolovska 83,CZ-18675, Prague, Czech Republic,[email protected]

The development of stochastic programming started six decades agowhen Beale (1955), Dantzig (1955), Charnes and Cooper (1959) andothers observed that for many linear programs to be solved, the valuesof the coefficients were not available. To solve these problems, the un-known values of coefficients were mostly replaced by their observedpast values. The obtained solutions were then feasible and optimalwith respect to one, maybe very rare set of observed coefficients. Thesuggestion was to replace the deterministic view by a stochastic oneand to assume that the unknown coefficients or parameters, say u, arerandom and that their probability distribution P is known and indepen-dent of the decision variables. A decision has to be taken before therealization of the random coefficients is known, using the knowledgeof probability distribution P.

The prototype stochastic program (we focus on in the sequel) aims atthe selection of the “best possible" decision x belonging to a decisionspace X subject to certain constraints. To build a decision model meansto specify the set of feasible decisions and to define a real objectivefunction which generates, independently of individual realizations ofrandom factors, a preference relation on the set of feasible decisions."Hard" constraints are usually reflected in definition of X. The outcomeof the decision x which satisfies them depends on realization of randomparameter u and is quantified as f(x,u). For different realizations oneobtains different optimal values and different optimal solutions and thequestion is which of these decisions should be accepted. The commonway is to use the expectation Ef(x,u) of the random objective functionf(x,u) as the overall objective function, which is then independent ofrealizations of random parameters and exploits solely their probabilitydistribution P. Numerous other criteria of optimality with such prop-erty can be constructed. The risk-adjusted expected return objectivefunction of the Markowitz model, Markowitz (1952), is a well knownearly example. Of course, the choice of the criterion should reflect thepurpose of the solved real-life problem.

When the random factors appear also in constraints, one may requirethat the constraints are satisfied for all possible realizations, a requestcommon in robust optimization. In this case, the set of feasible solu-tion could be unnecessarily small or even empty whereas, dependingon the solved problem, some violation of constraints might be toler-able. Hence, we have to specify which decisions will be accepted asfeasible. Again several approaches can be utilized:

* To use the expected values of the random constraint functions;

* To require that the constraints are satisfied with prescribed probabil-ities for each constraint separately – individual probabilistic or chanceconstraints, the idea of Charnes and Cooper (1958), or jointly for sev-eral constraints – joint probabilistic or chance constraints;

* To penalize constraints violations by a penalty q(x, u) assigned todecision x when it violates the constraints for the realization u and toincorporate the penalty term into the final objective function.

For stochastic linear programs a special type of penalty was designed:q(x, u) equals the optimal value of the second stage linear programwhich evaluates the effort for the cheapest possible compensation ofdiscrepancies caused by the decision x when the realization u occurs.This approach is known as stochastic linear programming with re-course. The simplest choice of the second stage program, called now

the simple recourse problem, appeared independently in Beale (1955)and Dantzig (1955); they suggested to distinguish and penalize twotypes of violations only.

The significant progress in the seventies happened thanks to Prékopa(1971) who opened applications of models with joint probabilistic con-straints. Another important extension, to multistage stochastic pro-grams, has aimed at a more realistic treatment of the dynamic or se-quential structure of real-life decision problems: When the informa-tion becomes known sequentially period-by-period, recourse decisionsresponding to the new information can be made; see e.g. Dupacová(1995) for a brief introduction.

The theoretical results for various types of stochastic programmingmodels were influenced and supported by achievements in optimiza-tion, probability and statistics and in computer technologies. In the1980’s, a special care to software development was initiated and hasresulted in the IIASA volume edited by Ermoliev and Wets (1988).Areas of further prospective applications have been delineated.

At present there are standardized types of stochastic programmingmodels (e.g., two-stage and multistage stochastic programs with re-course, models with individual and joint probabilistic constraints, inte-ger stochastic programs) with links to statistics and probability, to para-metric and multiobjective programming, to stochastic dynamic pro-gramming and stochastic control with relevant software systems avail-able or in progress. We refer to recent monographs and textbooks Birgeand Louveaux (1997, 2011), Kall and Mayer (2005, 2011), Shapiro,Dentcheva, and Ruszczyski (2009), handbook edited by Shapiro andRuszczyski (2003), collection edited by Infanger (2011) and to thestochastic programming bibliography by van der Vlerk (1996–2007).

The first applications appeared already in the fifties of the last century,e.g. Ferguson and Dantzig (1956), Charnes, Cooper, and Symonds(1958). They were based on simple types of stochastic programmingmodels, such as stochastic linear programs with simple recourse andmodels with individual probabilistic constraints. Moreover, special as-sumptions about the probability distribution P were exploited.

By now, significant applications of stochastic programming have beenmade in many areas, such as transportation, production planning andallocation of resources, management of hydropower and water re-sources, energy production and transmission, optimization of tech-nological processes, telecommunications, financial modeling and riskmanagement.

There are many excellent recent papers on successful large-scale real-life applications of stochastic programming. We refer to collectionsZiemba and Mulvey (1999) for a selection of financial problems,Wallace and Ziemba (2005) devoted to non-financial applications, toConejo, Carrión, and Morales (2010) focused on energy markets andcollection edited by Bertocchi, Consigli, and Dempster (2011) on ap-plications in finance and energy.

Stochastic programming is an example of a field which connects var-ious areas of optimization, probability and statistics, numerical math-ematics and computer science with the broad territory of operationsresearch. The success of applications is conditioned by a close collab-oration with users and it benefits from the team work. Even if imple-mentation of complex stochastic programming models frequently re-quires simplifications and/or approximations, their outcome may pointat new decision strategies.

We shall focus on the modeling issues. From the modeling point ofview, stochastic vehicle routing, stochastic networks and stochastic fa-cility location problems have been mostly treated as a natural extensionof the stochastic transportation problem with simple recourse, whereasindividual probabilistic constraints have appeared in the context of thestochastic nutrition model and in water resources management mod-els. The impressive early collection of case studies related to real-lifeapplications of joint probabilistic constraints was edited by Prékopa(1978, 1985); it reflects the prevailing interest in applications to waterresources problems at that time.

To apply stochastic programming means primarily to be familiar withthe real-life problem in question, its purpose, constraints, and with theavailable level of information concerning the influential random fac-tors. It is important to exploit the problem structure and to be aware ofalternative models. One can consider different objective functions andto model the same problem using probabilistic constraints or penaliz-ing the constraints violation or to combine both. Very often there is noconvincing reason why to prefer one approach to another. The choiceof model is tied to the data structure and it is frequently influencedby the decision maker’s individual attitude and the software available.To use the penalty or recourse approach means to quantify the lossescaused by the considered discrepancies, which may be a difficult taskfor the decision maker, who often prefers to assign probabilities ad hoc

150

EURO 25 - Vilnius 2012 TD-02

and to solve a stochastic program with individual probabilistic con-straints. This model, however, does not distinguish the magnitudes ofconstraints violations.

We shall illustrate the modeling issues and the interplay between theproblem setting and the stochastic programming model building by se-lected examples related mainly to water resources management andplanning; see e.g. Prékopa (1978, 1985), Dupacová (1979), Dupacováet al. (1991), Escudero (2000), Edirishinghe et al. (2000) and refer-ences therein.

ACKNOWLEDGEMENT. This work was partly supported by theCzech Science Foundation grant P402/11/0150.

MAIN REFERENCES

Beale, E. (1955), On minimizing a convex function subject to linearinequalities, Journal of the Royal Statistical Society B 17:173-184.

Bertocchi, M., Consigli, G., Dempster, M. A. H., eds. (2011), Stochas-tic Optimization Methods in Finance and Energy, Springer, New York.

Birge, J. R., Louveaux, F. (2011), Introduction to Stochastic Program-ming, 2nd Ed., Springer, New York.

Charnes, A., Cooper, W. W. (1959), Chance-constrained programming,Management Science 5:73-79.

Charnes, A., Cooper, W. W., Symonds, G. H. (1958), Cost horizonsand certainty equivalents: An approach to stochastic programming ofheating oil, Management Science 4:235-263.

Conejo, A. J., Carrion, M., Morales, J. M. (2010), Decision MakingUnder Uncertainty in Electricity Markets, Springer, New York.

Dantzig, G. B. (1955), Linear programming under uncertainty, Man-agement Science 1:197-206.

Dupacová, J.(1979), Water resources system modelling using stochas-tic programming with recourse, in: Kall, P., Prékopa, A., eds., RecentResults in Stochastic Programming, LNEMS 179, Springer, Berlin, pp.121-133.

Dupacová, J. (1995), Multistage stochastic programs: The-state-of-the-art and selected bibliography, Kybernetika 31:151–174.

Dupacová, J., Gaivoronski, A. A., Kos, Z., Szántai, T. (1991), Stochas-tic programming in water management: A case study and a comparisonof solution techniques, European J. Oper. Res. 52:28-44.

Edirisinghe, N. C. P., Patterson, E. I., Saadouli, N. (2000), Capacityplanning model for multipurpose water reservoir with target priorityoperation, Ann. Oper. Res. 100:273-303.

Ermoliev, Yu., Wets, R. J.-B., eds. (1988), Numerical Techniques forStochastic Optimization Problems, Springer, Berlin.

Escudero, L. F. (2000), WARSYP, a robust modelling approach forwater resources system planning under uncertainty, Ann. Oper. Res.95:313–339.

Ferguson, A., Dantzig, G. B. (1956), The allocation of aircraft toroutes: An example of linear programming under uncertain demand,Management Science 3:45-73.

Infanger, G., ed. (2011), Stochastic Programming, Springer, NewYork.

Kall, P., Mayer, J. (2011), Stochastic Linear Programming; Models,Theory, and Computation, 2nd Ed., Springer.

Markowitz, H. M. (1952), Portfolio Selection, J. of Finance 6:77–91.

Prékopa, A. (1971), Logarithmic concave measures with application tostochastic programming, Acta Scientiarium Mathematicarum (Szeged)32:301-316.

Prékopa, A., ed. (1978), Studies in Applied Stochastic Programming,MTA SzTAKI, Budapest, 80/1978 and 167/1985.

Shapiro, A., Dentcheva, D., Ruszczynski, A. (2009), Lectures inStochastic Programming. Modeling nd Theory, MPS-SIAM.

Shapiro, A., Ruszczyski, A., eds. (2003), Stochastic Programming,Vol. 10 of Handbooks in Operations Research and Management Sci-ence, Elsevier.

van der Vlerk, M., Stochastic Programming Bibliography. World WideWeb, http://www.eco.rug.nl/mally/spbib.html, 1996-2007.

Wallace, S. W., Ziemba, W. T., eds. (2005), Applications of StochasticProgramming, MPS-SIAM.

Ziemba, W. T., Mulvey, J., eds. (1999), World Wide Asset and Liabil-ity Modeling, Cambridge Univ. Press.

� TD-02Tuesday, 14:30-16:00RB-Beta

Scheduling in Transport

Stream: SchedulingInvited sessionChair: Jose M. Framinan, Industrial Management, School ofEngineering, University of Seville, Avda. de los Descubrimientoss/n, Seville, Spain, [email protected]

1 - The problem of minimization maximum weightedlateness of orders for two railway stations/Dmitry Arkhipov, ORSOT, Institute of Control Sciences,Pervomayskaya str., 28a - 53, 141707, Dolgoprudny, Moscow,Russian Federation, [email protected], AlexanderLazarev

The following scheduling problem is considered. There are two rail-way stations with double track. In each station there exists a varietyof orders. Each order is a wagon with it’s release date, value(weight)and due date. Wagons are delivered by train. All trains consist of thesame number of wagons and overcome the distance between stationsin the same amount of time. Dynamical programming algorithm con-structing the optimal schedule in compliance with criterion of minimalmaximum weighted lateness in O(n6) operations.

2 - Public transport system route reliability estimationusing macromodellingIrina Pticina, Transport and Telecommunication institute,Lomonosova 1, LV-1039, Riga, N/A, Latvia, [email protected],Irina Yatskiv

The main purpose of this research is to investigate the possibility ofestimation the Arrived Time (ATR) and Travel Time Reliability (TTR)for a single route on the macromodel of urban transport system. Theexperiments were conducted at the macromodel (VISUM) of Rigatransport system. The actual travel trip time between stops and be-tween route’s end stops and the vehicle arrived time at the end stopwere obtained as the results of the experiments. According to obtaineddata have been calculated ATR and TTR for the single bus route andfor its each stops.

3 - Scheduling interfering jobs in a permutation flow-shopPaz Perez Gonzalez, Industrial Management, University ofSeville, Camino de los descubrimientos s/n, 41092, Sevilla,Spain, [email protected], Jose M. Framinan

Most contributions regarding scheduling interfering jobs refer to one-machine setting. This paper studies the problem of scheduling twosets of jobs on permutation flowshop with the objectives of total com-pletion times for both sets. Since one machine case is known to beNP-hard and there are some approximate algorithms, we analyze theiradaptations to the flowshop setting. The methods are compared in anextensive test bed and computational results are provided.

4 - A Variable Neighbourhood Search (VNS) algorithmfor solving Read Mix Concrete Production and De-livery Scheduling Problem (RMCPDSP)Anna M. Coves, Institute of Industrial and ControlEngineering, UPC, Diagonal 647, 08028, Barcelona, Spain,[email protected], M. Antonia de los Santos

In this communication we present a VNS algorithm for solving RM-CPDSP. Integration of production and delivery is critical in the case ofReady Mix Concrete, as it expires approximately an hour and a halfafter it is produced. We propose three rules of VNS based on codingthat includes the following aspects: plant order assignment; deliverypriority; and time of first delivery for each order. We then combinedthis VNS algorithm with an MS algorithm that we had previously de-signed. We have found that using the two algorithms together providesbetter results by running only the VNS algorithm.

151


� TD-03Tuesday, 14:30-16:00RB-L1

Sustainable Road Transport and VehicleRouting

Stream: Sustainable Transport PlanningInvited sessionChair: U.Mahir Yildirim, Industrial Engineering, SabanciUniversity, Sabanci University, Faculty of Engineering and NaturalSciences, Tuzla, 34956, Istanbul, Turkey,[email protected]

1 - Minimum cost VRP with time-varying speed data andcongestion chargeRichard Eglese, The Management School, LancasterUniversity, Department of Management Science, LA1 4YX,Lancaster, Lancashire, United Kingdom,[email protected], Liang WenA heuristic algorithm is introduced for vehicle routing and schedulingproblems to minimize the total travel cost, including fuel, driver andcongestion charge costs. The fuel cost depends on the speed for a ve-hicle to travel along any road in the network and varies according tothe time of travel. If a vehicle enters the congestion charge zone dur-ing the daytime, a fixed charge is applied. The approach is tested ona constructed data set and also using real traffic data for the Londonarea. The results are also evaluated in terms of the CO2 emissions.

2 - Analysis of optimal vehicle speed: from supersonicwave devices to GPS transceiversHiroyuki Kawano, Department of Systems Design andEngineering, Nanzan University, Seirei-cho 27, Seto,4890863, Seto, Aichi, Japan, [email protected] Hanshin Expressway Company, vehicle traffic is monitored by su-personic wave detection devices along 500 meter intervals, but it isnot possible to observe accurate traffic conditions. Recently, trafficlog of highway patrol vehicles equipped GPS transceivers are storedinto a database system. We integrate the log data, including vehiclespeed, latitude and longitude, and weather data. Based on optimal ve-hicle speed model, we calculate continuous complement positions ofvehicles. We analyze optimal traffic volume and flows depending ondifferent lines and time zones.

3 - A bi-objective Pollution-Routing Problem: trade-offsbetween fuel consumption and driving timeEmrah Demir, School of Management, University ofSouthampton, 67 Kitchener Road, Portswood, SO17 3SF,SOUTHAMPTON, Hampshire, United Kingdom,[email protected], Tolga Bektas, Gilbert LaporteThis paper studies a bi-objective Pollution-Routing Problem, an exten-sion of the Vehicle Routing Problem with Time Windows, where oneobjective is to minimize fuel consumption, and the other total driv-ing time. A solution method is described where an Adaptive LargeNeighborhood Search algorithm, combined with a specialized speedoptimization algorithm, is used as a search engine for the applicationof four a posteriori methods to generate Pareto-optimal solutions to theproblem. Results of extensive computational experimentation show theefficiency of the proposed approaches.

4 - Greenest paths in time-dependent transportationnetworksU.Mahir Yildirim, Industrial Engineering, Sabanci University,Sabanci University, Faculty of Engineering and NaturalSciences, Tuzla, 34956, Istanbul, Turkey,[email protected], Bülent ÇatayThe literature on the shortest path algorithms (SPA) on time-dependentnetworks is mainly focused on finding the fastest (minimum time) path.Yet the current approaches are not applicable to determine the green-est (minimum greenhouse gas emitting/fuel consuming) path on a realtransportation network. We discuss new approaches to find the green-est path given the speed profile of the network and an appropriate emis-sion function of speed. Computational test using random data show thepotential savings and sustainability benefits of the new approaches overthe existing time-dependent SPA.

� TD-04Tuesday, 14:30-16:00RB-L2

Routing problems

Stream: Combinatorial OptimizationInvited sessionChair: Eleni Hadjiconstantinou, Business School, ImperialCollege London, South Kensington Campus, SW7 2AZ, London,United Kingdom, [email protected]: Vladimir Ejov, School of Computer Science, Engineeringand Mathematics, Flinders University, Sturt Road, 5042, BedfordPark, SA, Australia, [email protected]

1 - An Adaptive Memetic Algorithm using Large Neigh-bourhood Search for the Multi-Depot Pickup and De-livery ProblemPairoj Chaichiratikul, Business School, Imperial CollegeLondon, 18, Elvaston Place, SW7 5QF, London, UnitedKingdom, [email protected], EleniHadjiconstantinou

The objective of this NP-hard problem is to minimise the routing costof the fleet of vehicles serving transportation requests in a large ge-ographical coverage, subject to customer requirements. We proposea new heuristic approach that involves hybridisation between Adap-tive Large Neighbourhood Search (ALNS) and Adaptive Memetic Al-gorithm (AMA). The proposed heuristic is computationally tested onstandard benchmark instances from the literature. The preliminarycomputational results are promising; the proposed heuristic is capableof obtaining improved feasible solutions for some instances.

2 - The capacitated minimum spanning tree problemwith time windowsManolis Kritikos, Department of Management Science andTechnology, Athens University of Economics and Business,76, Patission street, 10434, Athens, Athens, Greece,[email protected], George Ioannou

We address the Capacitated Minimum Spanning Tree Problem withTime Windows (CMSTPTW). The CMSTPTW is an extension of theCapacitated Minimum Spanning Tree Problem (CMSTP). In the CM-STPTW, it is required to form the capacitated minimum-cost tree net-work such that the time windows constraints on the demand nodes aresatisfied. For the solution of the CMSTPTW, we propose a greedymethod. Our proposed method outperforms the greedy approach ofSolomon for MSTPTW. This is an indication of the quality of ourgreedy method for the CMSTPTW.

3 - "Snakes and Ladders" heuristic algorithm (SLH) forthe Hamiltonian cycles problem.Vladimir Ejov, School of Computer Science, Engineering andMathematics, Flinders University, Sturt Road, 5042, BedfordPark, SA, Australia, [email protected], SergueiRossomakhine

We place the vertices of an undirected graph in some order on a circle.Arcs between adjacent vertices on the circle are called snakes, whileother arcs represented by chords are called ladders. The SLH algorithmuses operations inspired by the Lin—Kernighan heuristic to reorder thevertices on the circle so as to transform some ladders into snakes. Theuse of a suitable stopping criterion ensures that the heuristic terminatesin polynomial time. We have observed that SLH is successful even incases where such cycles are extremely rare. Online demonstration willaccompany the presentation.

152


� TD-05Tuesday, 14:30-16:00RB-L3

Problems on graphs III

Stream: Discrete Optimization, Geometry & Graphs(contributed)Contributed sessionChair: Peter Recht, OR und Wirtschaftsinformatik, TU Dortmund,Vogelpothsweg 87, 44227, Dortmund, Germany,[email protected]

1 - A Coloring Algorithm for Triangle-Free GraphsMohammad Jamall, Mathematics and Statisics, ZayedUniversity, UAE, United Arab Emirates,[email protected]

We give a randomized algorithm that properly colors the vertices ofa graph G on n vertices with girth greater than 3 using O(D / log D)colors, where D is the maximum degree of G. The algorithm succeedswith high probability, provided D is large enough. The number of col-ors is best possible up to a constant factor for triangle-free graphs. Theexistence of such a proper coloring was established by Kim and Jo-hansson respectively.

A complete version of our paper can be found at arXiv:1101.5721.

2 - An Efficient Algorithm for Stopping on a Sink in a Di-rected Acyclic GraphGrzegorz Kubicki, Mathematics, University of Louisville,Natural Sciences 227, 40292, Louisville, Kentucky, UnitedStates, [email protected], Wayne Goddard, EwaKubicka

We consider an unordered version of the classical secretary problem.Vertices of an unknown directed acyclic graph of order n are revealedone by one in some random permutation. At each point, we know thesubgraph induced by the revealed vertices. Our goal is to stop on a sink,a vertex with no out-neighbors. We show that this can be achieved withprobability inversely proportional to the square root of n. This is bestpossible because the optinal stopping time on a sink in a directed pathhas the probability of success of that order of magnitude.

3 - On the Cubic Dimension of certain Classes of Bal-anced Binary Trees into HypercubeKamal Kabyl, Laboratory of Modeling and Optimization ofSystems LAMOS, Commercial Sciences Department,University of Bejaia, Algeria, 06000, Bejaia, Algeria,[email protected], Abdelhafid Berrachedi

The problem of embedding trees into the hypercube is much studiedin graph theory. In fact many efforts have been devoted to find suf-ficient conditions for which a tree T is a subgraph of the hypercube.In this paper, we introduce some new classes of balanced binary treesby giving their cubical dimensions. Furthermore, these classes of treessatisfy the conjecture of Havel [1984], which states that every balancedbinary tree with 2 to the power of n vertices can be embedded in an n-dimensional hypercube.

4 - A "min-max-theorem" for the cycle packing problemin Euler graphsPeter Recht, OR und Wirtschaftsinformatik, TU Dortmund,Vogelpothsweg 87, 44227, Dortmund, Germany,[email protected], Eva-Maria Sprengel

We look for a set of edge-disjoint cycles of maximum cardinality in agraph G=(V,E). The problem is tackled by considering local traces T(v)of G . Local Traces are not uniquely defined but their packing numbersare. We prove that if G is is Eulerian every maximum edge-disjointcycle packing of G induces maximum local traces T(v) or every v in V.In the opposite, if the total size of the local traces is minimal then thepacking number of G must be maximum. The determination of suchtraces leads to multi-commodity flow-problem with quadratic objectivefunction.

� TD-06Tuesday, 14:30-16:00RB-Gamma

Relaxation Adaptive MemoryProgramming

Stream: MetaheuristicsInvited sessionChair: Dorabela Gamboa, CIICESI-GECAD, ESTGF, InstitutoPolitécnico do Porto, Casa do Curral, Rua do Curral, Apt. 205,4610-156, Felgueiras, Portugal, [email protected]

1 - A RAMP Algorithm for Generalized AssignmentsCesar Rego, University of Mississippi, University, 38677,Oxford, MS, United States, [email protected], LutfuSagbansuaWe introduce a primal-dual RAMP algorithm for the solution of themulti-resource generalized assignment problems. The algorithm com-bines cross-parametric relaxation with the scatter search evolution-ary procedure as a way to create compound memory structures thateffectvely integrate information from the primal and dual solutionspaces. We present extensive computation results that clearly supportthe merit of the proposed algorithm when compared with the best al-ternative approaches.

2 - RAMP for the Capacitated Facility Location ProblemTelmo Matos, CIICESI, ESTGF, Instituto Politécnico doPorto, Casa do Curral, Rua do Curral, Apt. 205, 4610-156,Felgueiras, Portugal, [email protected], DorabelaGamboa, Cesar RegoWe present a relaxation adaptive memory programming (RAMP) ap-proach for the capacitated facility location problem (CFLP). The al-gorithm uses dual ascent with tabu search to explore primal-dual rela-tionships in a RAMP framework. A comparative analysis with the bestperforming algorithms of the literature is presented and discussed.

3 - A PD-RAMP Algorithm for the UFLPDorabela Gamboa, CIICESI-GECAD, ESTGF, InstitutoPolitécnico do Porto, Casa do Curral, Rua do Curral, Apt.205, 4610-156, Felgueiras, Portugal, [email protected],Telmo Matos, Fábio Maia, Cesar RegoWe address the well-studied uncapacitated facility location problem(UFLP) aimed at determining the set of possible locations where facil-ities should be established in order to serve a set of customers with min-imum cost. A relaxation adaptive memory programming (RAMP) ap-proach is developed by combining dual relaxation with scatter search.The results obtained on a standard testbed show the advantages of thisapproach over the current state-of-the art algorithms for the UFLP.

4 - A Parallel RAMP Algorithm for the Set Covering Prob-lemTabitha James, Business Information Technology, VirginiaTech, 1007 Pamplin Hall, 24061, Blacksburg, VA, UnitedStates, [email protected], Cesar Rego, José HumbertoAblanedo RosasThe relaxation adaptive memory programming (RAMP) framework al-lows for the combination of (primal) heuristic search methods with ex-act solution methodologies that solve a (dual) relaxation of the originalproblem in order to create an algorithm that leverages both approachesfor difficult combinatorial optimization problems. In this study, wepropose a parallel RAMP algorithm for the solution of large scale setcovering problems.

� TD-07Tuesday, 14:30-16:00RB-Eta

Manufacturing Issues

Stream: Supply Chain PlanningInvited sessionChair: Ignacio Eguia, School of Engineering. Dpt. of

153


Management Science, University of Seville, Camino de losDescubrimientos s/n, 41092, Sevilla, Spain, [email protected]

1 - Transportation scheduling in assembly lines understorage constraints: A case study of the automobileindustryMasood Fathi, Organziación Industrial, TECNUNUniversidad de Navarra, School of Engineering (Tecnun)„ PManuel Lardizabal 13„ 20018, San Sebastian, Guipuzcoa,Spain, [email protected], María Jesús Alvarez,Victoria Rodríguez

Study attempts to introduce an integrated model for production, in-ventory and transportation in assembly lines under storage constraints.An integrated optimization model which is a mixed integer linear pro-gramming formulation is presented with multiple objectives such asminimizing the number of tours, inventory level. To validate the pre-sented model, it has been tested in a real case study in VolkswagenNavarra plant, Spain. Given that different priorities for objectives de-fined in the discussed real case study exist, it has been tried to solvethe model through a goal programming framework.

2 - Modelling machine loading problem in reconfig-urable manufacturing systemsIgnacio Eguia, School of Engineering. Dpt. of ManagementScience, University of Seville, Camino de losDescubrimientos s/n, 41092, Sevilla, Spain, [email protected],Jesus Racero, Fernando Guerrero

A Reconfigurable Manufacturing System (RMS) is designed to pro-duce a family of parts that share some similarities using modular Re-configurable Machines. When modules are changed in a RM, differentoperations can be performed on the new machine configuration. Thiswork is concerned with the loading problem in a RMS. This problemdeals with the allocation of the operations and the associated mod-ules of a set of selected parts’ families among the RM, subject to thetechnological and capacity constraints of the RMS. A MILP model isdeveloped with an objective function of balancing the workload.

3 - Manufacturing process flexibility and capacity ex-pansion using Robust Optimization features inAIMMSOvidiu Listes, Paragon Decision Technology, ParagonDecision Technology, Schipholweg 1, 2034 LS, Haarlem,Netherlands, [email protected], Frédéric Babonneau,Christian Van Delft, Jean-Philippe Vial

We address the problem of manufacturing process flexibility and ca-pacity expansion under uncertain demands using Robust Optimization.The model involves binary product-to-plant assignments and decisionsfor expanding capacity. Affine decision rules are used for capacityexpansion and the amounts of processed products. We illustrate howAIMMS can accommodate the model formulation under uncertaintyand generate the robust counterpart automatically. We also show howthe intuitive modeling concepts in AIMMS allow for flexible experi-ments and comparison of results based on various uncertainty sets.

4 - Multi-objective synchronized planning in dynamiccollaborative manufacturing networksSenay Sadıç, University of Porto, Porto, Portugal,[email protected], Jorge Pinho de Sousa, José Crispim

In manufacturing networks, collaboration and information sharing maystrongly contribute to the global system performance. Dynamic net-work planning by synchronizing operations and allocating activities isa very hard problem due to its complex nature and increasing productcomplexity. In this study we develop a multi-objective MILP modeltackling this problem to minimize the overall operational costs of thenetwork and provide some flexibility to the units. The main drivers ofthe model are cost minimization and the concept of flexibility, as a wayto deal with different types of uncertainty.

� TD-08Tuesday, 14:30-16:00RB-Epsilon

Graph cut algorithms in Computer Vision

Stream: Boolean and Pseudo-Boolean OptimizationInvited sessionChair: Ramin Zabih, Cornell, Computer Science, 4130 UpsonHall, 14853, Ithaca, New York, United States, [email protected]

1 - Graph Cuts for Scene UnderstandingPhilip Torr, computing, oxford brookes university,Department of Computing, School of Technology, Design andEnvironment Oxford Brookes, OX33 1HX., oxford, oxon,United Kingdom, [email protected]

Within this talk I will describe how graph cuts can be used as the ba-sic optimization algorithm for scene understanding. The problem ofscene understanding is a basic one in computer vision which involvesrecognizing and recovering the shape of objects within a still image orvideo.

2 - Proposal selection in higher-order graph cutsHiroshi Ishikawa, Department of Computer Science andEngineering, Waseda University, Okubo 3-4-1, Shinjuku,169-8555, Tokyo, Japan, [email protected]

Recently, optimization of higher-order MRF with more than two labelsbecame practical using higher-order graph cuts: the combination of i)the fusion move algorithm, ii) the reduction of higher-order binary en-ergy minimization to first-order, and iii) the roof dual algorithm. Wediscuss the choice of proposal labeling for fusion move to improve theefficiency of the higher-order graph cuts and present a simple tech-nique for generating proposal labelings that speeds up the algorithmdramatically.

3 - Submodular Relaxations for Pseudo-Boolean Opti-mizationFredrik Kahl, Centre for Mathematical Sciences, LundUniversity, 22100, Lund, Sweden, [email protected],Petter Strandmark

We show how to use submodular relaxations for solving large-scaleboolean optimization problems. The maximum lower-bounding relax-ation is defined and analyzed. A method for constructing such relax-ations using generators of the cone of submodular functions is given,leading to a polynomial-time solution procedure via graph cuts. Theapproach is applied to several challenging vision problems, e.g. stereoand denoising. Experimental results confirm that our relaxations per-form better than state-of-the-art, both in terms of tighter lower boundsand increased number of assigned variables.

� TD-09Tuesday, 14:30-16:00RB-Zeta

EURO/ROADEF Challenge Session 1

Stream: EURO/ROADEF ChallengeInvited sessionChair: Christian Artigues, LAAS, CNRS, 7 avenue du ColonelRoche, 31077, Toulouse Cedex 4, France, [email protected]: H. Murat Afsar, Industrial Systems, University ofTechnology of Troyes, 12 rue Marie Curie, BP 2060, 10010,Troyes, France, [email protected]: Ender Özcan, Computer Science, University ofNottingham, Jubilee Campus, Wollaton Road, NG8 1BB,Nottingham, United Kingdom, [email protected]: Emmanuel Guere, Google, 38 Avenue de l’Opera, 75002,Paris, France, [email protected]: Eric Bourreau, COCONUT, LIRMM, 161 Rue Ada, 34000,Montpellier, France, [email protected]

154


Chair: Kedad-Sidhoum Safia, LIP6 - UPMC, 4 Place Jussieu,75005, Paris, France, [email protected]: Marc Sevaux, Lab-STICC - CNRS UMR 3192, Universitéde Bretagne Sud - UEB, Centre de Recherche - BP 92116, 2 rue deSaint Maude, 56321, Lorient, France, [email protected]

1 - ROADEF/EURO Challenge 2012 : Machine Reassign-mentEric Bourreau, COCONUT, LIRMM, 161 Rue Ada, 34000,Montpellier, France, [email protected], H. Murat Afsar,Christian Artigues, Emmanuel Guere, Ender Özcan,Kedad-Sidhoum Safia

We present the ROADEF/EURO 2102 challenge, an international op-timization contest propose jointly by EURO, the French OR society(ROADEF) and an industrial partner (GOOGLE). The subject is to im-prove the usage of a set of machines. A machine has several resourcesas for example RAM and CPU, and runs processes. You can move itfrom one machine to another. More than 80 teams from 33 differentcountries were initially registered. 30 teams are still competing for thefinal round and will present their work during all the dedicated Stream.Results will be announced during the last session.

2 - Attacking Google’s machine reassignment problemusing mathematical programming, metaheuristicsand heuristics.Christos Gogos, Finance and Auditing, TechnologicalEducational Institute of Epirus, Psathaki, Nea Ktiria TEI,48100, Preveza, – Please select one –, Greece,[email protected], Christos Valouxis, Panayiotis Alefragis,Efthymios Housos

Machine Reassignment is the problem of finding the best possible as-signment of processes to machines. Each process move incurs cost thathas to be won back by better assignments of processes to machines.Through the use of mathematical programming and metaheuristics ourapproach reaches good results. It consists of an Integer ProgrammingModule (IPM) and a Metaheuristic Module (MM). The IPM is capableof solving partitions of the problem involving subsets of processes andmachines. The MM exploits a set of moves under the Late Acceptancemetaheuristic. The two modules are activated in turn.

3 - Hybrid Method for Machine Reassignment ProblemSaïd Hanafi, ISTV2, LAMIH-SIADE, University ofValenciennes, Le Mont Houy, 59313, Valenciennes, France,[email protected], Hideki Hashimoto, KojiNonobe, Michel Vasquez, Yannick Vimont, MutsunoriYagiura

Our algorithm combines exact and heuristic approaches for solvingthe machine reassignment problem (ROADEF/EURO challenge 2012).We employ ILS, tabu search, relaxation and projection technique. Lo-cal search uses two neighborhoods: shift and swap neighborhood, andincorporates an automatic control mechanism of the weights in theevaluation function. Exploration of the search space avoids the trapof local optimality by using the reverse elimination method to providean exact tabu list management (Glover 1990). This method allows ourteam to be qualified for the final phase with rank 5.

� TD-10Tuesday, 14:30-16:00RB-Theta


Stream: Cutting and PackingInvited sessionChair: Silvio de Araujo, Departamento de Ciências daComputação e Estatística-DCCE, Universidade EstadualPaulista-UNESP, R. Cristovao Colombo, 2265 - Jd Nazareth,15054-000, São José do Rio Preto, São Paulo, Brazil,[email protected]

1 - How to deal with alternative cutting lengths in a one-dimensional cutting stock problemMarkus Siepermann, Business Information Management, TUDortmund, 44221, Dortmund, Germany,[email protected], Richard Lackes,Torsten NollThe input material of the tubes production process consists of steel barsthat have to be cut into pieces before they are heated and rolled. In or-der to avoid waste of input material the cuts of the input material haveto be optimized which is a classical one-dimensional cutting problem.A special characteristic of the tubes production is that the same outputcan be reached with different cutting lengths. Thus, the minimizationof input material gets more difficult. This paper presents a solutionhow the optimization and the known algorithms have to be adapted.

2 - A pseudo-polynomial time algorithm for a subclassof the Variable-Sized BPP with ConflictsMohamed Maiza, Applied Mathematics Laboratory, MilitaryPolytechnic School of Bordj El Bahri, EMP (Ex-ENITA), BP17, Bordj El Bahri, 16111, Algiers, Algeria,[email protected], Mohammed Said Radjef, SaisLakhdarWe propose a pseudo-polynomial time algorithm for the Variable SizedBPP with Conflicts in the case where the size of the items is at leastequal to one third of the largest bin capacity. After assigning eachthree compatible items with size equal to one third of the largest bin tothe same largest bin, the algorithm solves the problem as a minimum-weight matching problem in the weighted compatibility graph wherenodes are the items and adjacent items are mutually compatible. Theweight of a given edge is the minimum cost for packing the corre-sponding two items in the same or in separate bins.

3 - Multiobjective genetic algorithms to the one-dimensional cutting stock problemSilvio de Araujo, Departamento de Ciências da Computação eEstatística-DCCE, Universidade Estadual Paulista-UNESP, R.Cristovao Colombo, 2265 - Jd Nazareth, 15054-000, São Josédo Rio Preto, São Paulo, Brazil, [email protected],Helenice Florentino, Kelly PoldiThis paper deals with the one-dimensional integer cutting stock prob-lem. The case in which there is just one type of object in stock avail-able in unlimited quantity is studied. A heuristic method based onthe multi objective genetic algorithms concept is proposed to solve theproblem considering two conflicting objective functions: minimizationof the number of objects and the number of different cutting patterns.This heuristic is empirically analyzed by solving randomly generatedinstances and the results are compared with other methods from theliterature.

4 - Optimization of the outbound boxes dimensions ofmobile phones in a factory located in the IndustrialPark of Manaus in Amazon State, Northern of Brazil.Fabricio Costa, Nokia Institute of Technology - INdT,+559232365418, +559281065960, Manaus, Amazon, Brazil,[email protected] mobile phone is a device with high value added. Those productsare transported by airplane, volume and weight of boxes influences thefreight costs. Therefore the dimensions of outbound packages are veryimportant to that business. We considered three different types of con-straints, such as dimensions, weight of boxes and operations. The re-sults of new dimensions were: reduction of empty spaces inside boxesand freight costs, reduction of the product damages in operation, trans-port and collaborate to reduce the environment impacts. This industryis located in Manaus, Amazon state, Brazil.

� TD-11Tuesday, 14:30-16:00RB-Iota

Heuristic Methods for Arc RoutingProblemsStream: Arc RoutingInvited sessionChair: Ana Catarina Nunes, Dep. Métodos Quantitativos, ISCTE -

155


IUL / Centro IO, Portugal, Av. das Forças Armadas, 1649-026,Lisbon, Portugal, [email protected]

1 - Metaheuristics for household refuse collectionAna Catarina Nunes, Dep. Métodos Quantitativos, ISCTE -IUL / Centro IO, Portugal, Av. das Forças Armadas,1649-026, Lisbon, Portugal, [email protected], MariaCortinhal, Cândida Mourão

The household refuse collection in large urban areas may be modeledby the sectoring-arc routing problem (SARP). The SARP groups twofamilies of problems: sectoring (or districting) problems and capaci-tated arc routing problems (CARP). The SARP is defined over a mixedgraph. Its aim is to identify a given number of similar sectors (sub-graphs) and to build a set of collecting trips in each sector such thatthe total duration of the trips is minimized. Metaheuristics are pre-sented, and computational results over a set of benchmark instancesare reported.

2 - Solving a location-arc routing problem using a TS-VNS approachRui Borges Lopes, Dep. of Economics, Management andIndustrial Engineering, CIO / University of Aveiro, CampusUniversitário de Santiago, 3810-143, Aveiro, Portugal,[email protected], Carlos Ferreira, Beatriz Sousa Santos

Within location-routing it is easy to consider scenarios where the de-mand is on the edges of a network, rather than on the nodes (referredin the literature as location-arc routing problems — LARP). Examplesinclude locating facilities for postal delivery, garbage collection, roadmaintenance and street sweeping. For the LARP we present an itera-tive approach (TS-VNS) composed of a tabu search (TS) and a vari-able neighbourhood search (VNS), respectively, for the location and(arc) routing phases. Results are obtained for a set of test instancesand compared with other approaches.

� TD-12Tuesday, 14:30-16:00RB-Omicron

Rescheduling

Stream: Optimization in Public TransportInvited sessionChair: Leo Kroon, Rotterdam School of Management, ErasmusUniversity Rotterdam, P.O. Box 1738, 3000 DR, Rotterdam,Netherlands, [email protected]: Anita Schöbel, Institute for Numerical and AppliedMathematics, Georg-August Universiy Goettingen, Lotzestrasse16-18, 37083, Göttingen, Germany,[email protected]

1 - Recovery from disruptions with recovery length con-trolledLuis Cadarso, Matemática Aplicada y Estadística,Universidad Politécnica de Madrid, Pz. Cardenal Cisneros, 3,28040, Madrid, Spain, [email protected], Ángel Marín,Gabor Maroti

This paper studies the disruption management problem of rapid tran-sit rail networks. The task is to simultaneously adjust the timetableand the rolling stock schedules, also dealing with the dynamic natureof demand. The objective is to minimize passenger inconvenience,operational costs and the length of the recovery period (after whichoperations are back to the undisrupted schedules). We apply a discretechoice model for computing passenger demand, and we propose an in-tegrated MIP model for timetable and rolling stock recovery. We studyproblem instances of the Spanish operator RENFE.

2 - Robustness in crew re-schedulingDennis Huisman, Econometric Institute, Erasmus University,Rotterdam, Netherlands, [email protected], LucasVeelenturf, Daniel Potthoff, Leo Kroon, Gabor Maroti, AlbertWagelmans

We present a framework that shows how the re-scheduling problemcan be solved in a robust way. This means that the new schedule isto some extent insensitive to further changes in the environment. Wedemonstrate our framework with a partical application in disruptionmanagement, namely the railway crew rescheduling problem. Testson several instances of Netherlands Railways demonstrate that our ap-proach can be useful: if the crew schedule has to be modified morethan once, then the later rescheduling costs are lower.

3 - A geometric approach for rescheduling railwaytimetables by considering passenger transfers in thesystemFrancisco A. Ortega Riejos, Applied Mathematics I,University of Seville, 41012, Seville, Spain, [email protected],Juan A. Mesa, Miguel Angel Pozo, Justo Puerto

This paper considers a transit line where trains circulate and stop atthe stations according to a predetermined schedule which is knownby the users. At any instant, passengers arrive to different stations inorder to board these vehicles according to a model of distribution ofpassenger arrivals that is assumed deterministic. In this scenario, aservice rescheduling forced by an incidence is determined in order tominimize the loss of passengers who require transfers between differ-ent line runs at the interchange stations (supported by Spanish/FEDERprojects MTM2009-14243 and P09-TEP-5022).

4 - Railway crew rescheduling on ultra-short termPieter Fioole, PI, Netherlands Railways, stalpaertstraat 56,3067 XS, Rotterdam, Netherlands, [email protected]

When extreme weather is predicted, Netherlands Railways switchesto a different timetable in order to be more robust. The correspond-ing rolling stock and crew duties must be changed then as well, whichleads to changing over 1,000 crew duties several hours before execut-ing the plan. We will discuss the process of how this is done, as wellas the OR techniques that are used to achieve this. We will also discussthe benefits of a changed timetable compared to the original one on aday with extreme weather, as well as the benefits of OR techniques forrescheduling compared to manual dispatching.

� TD-13Tuesday, 14:30-16:00RB-Tau

Logistics and Manufacturing &Warehousing

Stream: Inventory ManagementInvited sessionChair: Luca Bertazzi, Dept. of Quantitative Methods, Universityof Brescia, C.da Santa Chiara, 50, 25122, Brescia, Italy,[email protected]

1 - Stochastic Dynamic Programming Algorithms for anIntegrated Logistic System with Outsourced Trans-portationDemetrio Laganà, DEIS, University of Calabria, Ponte PietroBucci, 41/C, 87036, Rende, Italy, [email protected],Adamo Bosco, Luca Bertazzi

We propose an exact dynamic programming and a rollout algorithmfor a logistic system with one product, one supplier, a set of retailersand outsourced deliveries over a time horizon. Each retailer has aninitial inventory level, a fixed storage capacity and a random demand.An order-up-to level policy is applied. An inventory cost is charged ifthe inventory level is positive, a penalty otherwise. The supplier has aninitial inventory level and a production at each time. The problem is tominimize the expected total cost over the horizon. Experiments werecarried out on benchmark instances.

156


2 - A modelisation for multimodal network design: ap-plication to the hinterland of a portArnaud Knippel, Laboratoire de Math, INSA Rouen, INSARouen - LMI, Avenue de l’Universit, 76801, Saint-Etienne duRouvray, France, [email protected], SophieMichel, Ibrahima Diarrassouba, Cédric Joncour

We are interested on the evacuation of the containers towards the hin-terland of a port. The transport modes tend to evolve by developingthe alternative modes with the road: river or railway. The way inwhich these modes must be organized put the question of the designof a multimodal network. In addition to choosing the mean of trans-port, one must decide where install the sites of transshipment and theirdimensioning. To model the strategic problem, we propose a gadgetto transform a multiple non oriented graph to a simple oriented graph.Preliminary numerical results are shown.

3 - Approximated neighborhood evaluation for complexlogistics support design problemsEmanuele Manni, Department of Engineering, University ofSalento, via per Monteroni, 73100, Lecce, Italy,[email protected], Abdallah Alalawin,Gianpaolo Ghiani, Chefi Triki

This paper deals with the design of the logistics support of complexsystems, based on a Level of Repair Analysis (LORA). We model amulti-echelon and multi-indenture LORA variant with the simultane-ous definition of the spare parts at stock and the maintenance resourcemanagement policy. The goal is to minimize the life-cycle cost, sub-ject to constraints on the expected availability. The solution approachuses an optimization via simulation method to explore the search spaceefficiently through an Approximated Neighborhood Evaluation model,relying on the estimation of a number of parameters.

4 - Combinatorial optimization techniques for the de-sign of reconfigurable machinesOlga Battaïa, IE & Computer Science, Ecole des Mines deSaint Etienne, Saint Etienne, France, [email protected],Alexandre Dolgui, Nikolai Guschinsky, Genrikh Levin

The problem is to design machine modules, such as working positions,turrets and spindle heads, suitable for the given production process.This problem is formulated as a combinatorial optimization problemwhere a known set of machining operations is partitioned into subsetswhich are assigned to the pieces of equipment under precedence, in-clusion, exclusion constraints among operations as well as productivityand location constraints. The objective is to minimize the total equip-ment cost. The solution method is based on searching the shortest pathin a specific graph.

� TD-14Tuesday, 14:30-16:00RB-Omega

Workforce Scheduling I

Stream: Timetabling and RosteringInvited sessionChair: Jan Lange, TU Dresden, Germany,[email protected]

1 - Ants for the Integrated Vehicle and Crew SchedulingProblemDavid Pash, University of Primorska, PINT, 6000, Koper,Slovenia, [email protected], Andrej Brodnik

We present an ACO (ant colony optimization) based meta-heuristicfor the integrated vehicle and crew scheduling problem. The quali-ties of the produced solutions are compared to the optimal solutionsof an IP based approach. We show on real-world instances that themeta-heuristic finds solutions of better quality than the sequential IPapproach. Moreover, our meta-heuristic method can be applied to theintegrated vehicle-crew scheduling and rostering problem.

2 - A Model for the Vessel Crew Scheduling ProblemAlexander Leggate, Management Science, University ofStrathclyde, Graham Hills Building, 40 George Street, G11QE, Glasgow, United Kingdom,[email protected], Tibor Illés, Robert Van derMeer, Kerem Akartunali

Crew scheduling problems have been widely studied in various trans-portation sectors — airlines, railways, urban buses, etc. However todate there appear to be no applications of these problems in a mar-itime setting. Here, we explore key differences between the maritimeproblem and problems in other transport settings, and present formu-lations for both the crew scheduling and re-scheduling problems for acompany operating a fleet of vessels on a global scale. Computationalresults are given, along with proposals for implementing this model asa tool within the company’s scheduling process.

3 - Heuristic Solution Approaches for Check-in Coun-ters’ Workforce Scheduling with HeterogeneousSkillsEmilio Zamorano de Acha, Chair of Production Management,Univerisity of Mannheim, Schloss SO 230, University ofMannheim, 68161, Mannheim, Baden-Württemberg,Germany, [email protected], Raik Stolletz

Ground-handling agencies deal with the allocation of check-in coun-ters’ employees to shifts and day-offs in order to meet airlines’ de-mand. Since airlines use different computer systems they schedule dif-ferently qualified agents for the check-in process, thus ground-handlersneed to assign multiskilled staff. To this problem we propose a MILPmodel and develop several heuristic solution approaches. Using real-world demand scenarios from a ground-handler at a German airport,we analyze the influence of different qualification sets among employ-ees in the workforce scheduling process.

4 - Scheduling Preventive Maintenances for Semicon-ductor Manufacturing Using CPJan Lange, TU Dresden, Germany,[email protected]

Scheduling preventive maintenance tasks is complex when also regard-ing human resource limits and effects to the manufacturing process. Inthis research a CP-based model for advanced maintenance schedul-ing is presented. Besides machines it comprises partly ordered main-tenances which are restricted to individual time frames and are per-formed by several engineers having specific qualifications and avail-abilities. The objectives are the balancing of engineer activities, main-tenance minimization and best fit to the expected WIP. The proposedapproach is examined based on real manufacturing data.

� TD-15Tuesday, 14:30-16:00RB-2101

Vector and Set-Valued Optimization II

Stream: Vector and Set-Valued OptimizationInvited sessionChair: Vicente Novo, Matematica Aplicada, Universidad Nacionalde Educacion a Distancia, Juan del Rosal no. 12, 28040, Madrid,Spain, [email protected]: Enrico Miglierina, Dipartimento di DisciplineMatematiche, Finanza Matematica ed Econometria, UniversitàCattolica del Sacro Cuore, via Necchi 9, 20123, Milano, Italy,[email protected]

1 - Lower and upper Ginchev derivatives of vector func-tions and their applications to multiobjective opti-mizationMarcin Studniarski, Faculty of Mathematics and ComputerScience, University of Lodz, ul. S. Banacha 22, 90-238, Lodz,Poland, [email protected], El-Desouky Rahmo

157


Based on the definitions of lower and upper limits of vector functionsintroduced in our earlier paper, we extend the lower and upper Ginchevdirectional derivatives to functions with values in finite-dimensionalspaces where partial order is introduced by a polyhedral cone. Thisallows us to obtain some new higher-order optimality conditions for acone-constrained multiobjective optimization problem.

2 - Saddle point results for approximate proper effi-ciency in vector optimization problemsLidia Huerga, Applied Mathematics, UNED, C/Juan delRosal, 12, 28040, Madrid, Spain, [email protected], CésarGutiérrez, Vicente Novo

We consider a constrained vector optimization problem on a locallyconvex Hausdorff topological linear space with generalized cone con-vexity assumptions. We introduce a new concept of approximateproper saddle point based on a new set-valued Lagrangian function andwe give necessary and sufficient conditions for a type of approximateBenson proper efficiency through approximate proper saddle point re-sults, where the complementary slack condition is bounded.

3 - Global optimization in Rn with box constrains andapplicationsMiguel Delgado Pineda, Matematicas Fundamentales, UNED,Edificio de Ciencias. UNED, C/ Senda del Rey, 9, 28040,Madrid, Spain, [email protected]

An algorithm is presented for global optimization of nonconvex nons-mooth multiobjective optimization programs, min fs(x) with box con-straints and all functions fs are Lipschitz. A Maple code is attachedfor this monotonic set contraction algorithm. It converges onto theentire exact Pareto set and yields its approximation with given preci-sion. Approximations for the ideal point and for the function valuesover Pareto set are obtained. The code does not create ill-conditionedsituations and it is ready for engineering and economic applications.Results of examples are presented, with graphs.

4 - A notion of condition number in multiobjective opti-mizationEnrico Miglierina, Dipartimento di Discipline Matematiche,Finanza Matematica ed Econometria, Università Cattolica delSacro Cuore, via Necchi 9, 20123, Milano, Italy,[email protected], Monica Bianchi, Elena Molho,Rita Pini

We introduce a condition numbers for a multiobjective optimizationproblem. Since, in multiobjective optimization, the solution set almostalways is not a singleton, we develop an approach where the solutionset is considered as a whole. A pseudodistance between function isdefined such that the condition number provides an upper bound onhow far from a well-conditioned function a perturbed function can bechoosen in order that it is well-conditioned also. An extension of clas-sical Eckart-Young distance theorem is proved, even if taking in ac-count only a special class of perturbations.

� TD-16Tuesday, 14:30-16:00RB-2103

Statistical, Games and OptimizationAnalysis in Climate Meta-Models

Stream: Uncertainty and Perturbations in Optimizationand in Environmental and Energy ModellingInvited sessionChair: Alain Haurie, ORDECSYS, Place de l’étrier 4, 1224,Chêne Bougeries, Switzerland, [email protected]: Frédéric Babonneau, Ordecsys and EPFL, CH-1224,Geneva, Switzerland, [email protected]

1 - Computational Analysis of a Stochastic ClimateGame ModelAlain Haurie, ORDECSYS, Place de l’étrier 4, 1224, ChêneBougeries, Switzerland, [email protected], Olivier Bahn,Roland Malhamé, Julien Thénié

This paper is an exercise in computational economics, applied to astochastic game model of climate policy. We aim to demonstrate thenumerical implementation of an approximate dynamic programmingapproach to compute Nash equilibrium solutions in a climate modelwith two groups of non-cooperating nations, via piecewise determinis-tic differential games. We characterize equilibrium strategies throughthe solution of a family of interconnected deterministic equivalent dif-ferential games, propose an approximation method and describe its im-plementation and analyze 2 case studies.

2 - Modeling Climate Negotiations as a Game DesignProblemGuillaume Jean Tarel, Ordecsys, Place de l’Etrier 4, 1224,Chêne-Bougeries, Switzerland, [email protected],Alain Haurie, Frédéric Babonneau, Marc Vielle

An international emissions trading scheme is implemented and a globalsafety emission budget is shared according to an equity rule which fa-vors developing or emergent countries with large populations. Theyhave the possibility to use strategically their emission rights to maxi-mize the economic benefit from the supply of the emission rights mar-ket. We propose a meta-model built from statistical emulation of acomputable general equilibrium model. The optimal design of thisgame is formulated and solved. The solution is interpreted in terms ofstable international agreement.

3 - Meta-modelling of coupled climate-economic dynam-icsFrédéric Babonneau, EPFL-ORDECSYS, Switzerland,[email protected], Neil Edwards, Phil Holden, AmitKanudia, Maryse Labriet

This paper applies the oracle-based method ACCPM for the cou-pling of the climate simulation model GENIE with the world energymodel TIAM. The objective is to compute a solution that maximizesthe TIAM economic utility function subject to a constraint on globalwarming. In this paradigm, a meta-model controls the global iterationbetween the models which are treated as oracles. Here we use emu-lations in place of the simulation models to permit direct control overthe smoothness of data exchanged in the process, while acceleratingthe overall process (emulated models are very fast).

4 - Climate Statistics for Energy MeteorologyJohn Boland, School of Mathematics and Statistics, Universityof South Australia, Mawson Lakes Blvd., 5095, MawsonLakes, South Australia, Australia, [email protected]

This will be a survey of the statistical tools I and my team have em-ployed in evaluating the performance of energy systems subject tothe influence of climate variables. It will include the following top-ics:Probabilistic forecasting of solar radiation time series at daily andhourly time scales; Estimation and forecasting of wind farm outputvolatility; Construction of a model for prediction of diffuse solar ra-diation; Interpolation of time series using Fourier Series. Keywords:Time Series, Dynamical Systems, Volatility, Diffuse Solar Radiation,Interpolation Methods

� TD-17Tuesday, 14:30-16:00RB-2105

Global Optimization 1

Stream: Global OptimizationInvited sessionChair: Herman Mawengkang, Mathematics, The University ofSumatera Utara, FMIPA USU, KAMPUS USU, 20155, Medan,Indonesia, [email protected]

1 - On solving non-convex mixed-integer nonlinear pro-gramming problems using active constraint strategyHardi Tambunan, Mathematics, Quality University, FMIPAUSU, 20155, Medan, North Sumatera Province, Indonesia,[email protected], Herman Mawengkang

158


The nonlinear mathematical programming problem addressed in thispaper has a structure characterized by a subset of variables restricted toassume discrete values, which are linear and separable from the con-tinuous variables. The strategy of releasing nonbasic variables fromtheir bounds, combined with the "active constraint" method, has beendeveloped. This strategy is used to force the appropriate non-integerbasic variables to move to their neighbourhood integer points. Suc-cessful implementation of these algorithms was achieved on varioustest problems.

2 - Developing an integrated model for river water qual-ity to estimate wastewater removal efficienciesSyafari Syafari, Mathematics, University Negeri Medan,FMIPA USU, 20155, Medan, North Sumatera Province,Indonesia, [email protected], Herman Mawengkang

Rivers and streams are an important component of the natural environ-ment, and need to be protected from all sources of pollution. Rivers,however, are increasingly under threat from different pollutants, whichinclude conventional pollutants and hazardous substances. In this pa-per, dynamic integrated modelling of basic water quality and organiccontaminant fate in rivers are explored. In this framework, a basicriver water quality model and organic contaminant submodel were de-veloped and then linked in order to estimate the wastewater removalefficiencies at each wastewater discharge site.

3 - Modeling the dynamic interaction of social networksMaya Silvi Lydia, Computer Science, University of SumateraUtara, FMIPA USU, 20155, Medan, North SumateraProvince, Indonesia, [email protected], HermanMawengkang

In terms of graph, social interactions can be modeled as networks. Thecharacteristic of the interactions is dynamic, changing continuouslyover time. Most research are on static social networks. This pape pro-poses a mathematical model to analyze the dynamic interactions in thesocial networks.

4 - Comparison of Novel Optimization Algorithms on In-telligent Well Production PerformanceMorteza Hassanababdi, Mathematics, Amirkabir University ofTechnology, 424 Hafez Ave, Tehran, Iran„ 15875-441,Tehran, Iran, Islamic Republic Of, [email protected],Ali MirHassani, Mahdi Nadri Pari, Seyyed Mahdia Motahhari

A main challenges of oil and gas reservoir management is produc-tion optimization. We employed 3 methods of Response Surface,Taguchi and Neural Network to model reservoir behavior. Optimiza-tion algorithms were applied over the output of the foregoing methods:Quadratic Programming, NLP as a Gradient optimization algorithmimplemented in Response Surface and Taguchi methods, and ParticleSwarm Optimization implemented in NN. We present results and acomparison, and we rcommend on applicability of our algorithms forother type of reservoirs.

� TD-18Tuesday, 14:30-16:00RB-2107

Location Analysis

Stream: Continuous LocationInvited sessionChair: Burak Boyacı, ENAC, EPFL, EPFL ENAC INTER LUTS,GC C2 406 (Bâtiment GC) Station 18, 1015, Lausanne, Vaud,Switzerland, [email protected]

1 - Operational Research connecting Statistics and Car-tographyJosé L. Pino, Estadística e Investigación Operativa,Universidad de Sevilla, Avenida Reina Mercedes s/n, 41012,Sevilla, Spain, [email protected], Ma Teresa Cáceres

Official statistics has the requirement to produce and disseminate dis-aggregated information that responds to the management demands ofpublic administration and the society. Cartography is an indispensable

tool to disseminate spatial information. GIS and Statistical Softwareevolution has enabled the approximation of activities of the statisticaland cartographical Public Bodies. In this paper the contribution of theapplication of Operational Research techniques (such as the locationanalysis) to the integration of the Institutes of Statistics and Cartogra-phy of Andalusia is shown.

2 - A parametric analysis on optimal locations for an en-tering firm under delivered price competitionBlas Pelegrin, Statistics and Operations Research, Universityof Murcia, Spain, [email protected], Pascual Fernandez, MaríaD. García

Facility location is an strategic decision for an entering firm whichcompetes on delivered pricing with some pre-existing firms. In thispaper, we analyze the effect on the optimal locations of the enteringfirm produced by variations of the marginal delivered price. For dif-ferent scenarios in the transportation network of the Region of Murcia(Spain), we study how optimal locations can change with deliveredprice, and how the profits obtained by the entering firm and its com-petitors vary with delivered price. Finally, some conclusions are pre-sented.

3 - Extended Hypercube Queueing Models for Stochas-tic Facility Location ProblemsBurak Boyacı, ENAC, EPFL, EPFL ENAC INTER LUTS,GC C2 406 (Bâtiment GC) Station 18, 1015, Lausanne, Vaud,Switzerland, [email protected], Nikolas Geroliminis

In this research, we are modeling the facility location problem for sys-tems with many servers and stochastic demand, such as emergency re-sponse and on-demand transportation systems, by using spatial queues,specifically hypercube queuing models (HQMs). We are proposingtwo extended HQMs and a heuristic method which uses an approachsimilar to dynamic programming. Briefly, the heuristic solves a subproblem for each sub region which is generated by using similaritiesin demand and then combines them accordingly. Preliminary analysisgives promising results for some real life instances.

� TD-19Tuesday, 14:30-16:00RB-2111

HeuristicsStream: Nonlinear ProgrammingInvited sessionChair: Edite M.G.P. Fernandes, Algoritmi R&D Centre, Universityof Minho, School of Engineering, Campus de Gualtar, 4710-057,Braga, Portugal, [email protected]

1 - Using neural networks to map beam angles into in-tensity objective function values in IMRT optimiza-tion problemsJoana Matos Dias, FEUC / INESCC, University of Coimbra,Av. Dias da Silva, 165, 3004-512, Coimbra, Portugal,[email protected], Humberto Rocha

In IMRT (Intensity Modulated Radiation Therapy) optimization prob-lems radiation angles need to be found so that the prescribed radiationdoses are reached. The choice of optimization algorithms is limiteddue to the computational time needed to evaluate a given solution.Neural networks (NN) can be used as a black box mapping from ra-diation angles to intensity problem objective function values. NN canbe used inside several optimization algorithms, providing a fast wayof estimating the objective function values, allowing a near optimalsolution to be found in reasonable computational times.

2 - A study of derivative-free optimization techniquesfor handling the sampling aspect in automated algo-rithm configurationZhi Yuan, IRIDIA, Université libre de Bruxelles, Av. F.Roosevelt 50, CP194/6, 1050, Brussels, Belgium,[email protected], Thomas Stützle

159


Automated algorithm configuration problem (ACP) has become anemerging application of operations research, and has proved its impor-tance. Finding a good setting for numerical parameters of an algorithmcan be regarded as a stochastic continuous optimization problem. Thiswork focuses on the sampling aspect of ACP, and advanced derivative-free optimization techniques such as BOBYQA, CMA-ES, and MADSare applied. We study their behavior on the ACP landscape, and extendthem with specific techniques for handling the simulation stochasticityin the ACP.

3 - Solving Quadratic 0-1 Knapsack Problem Using anArtificial Fish Swarm AlgorithmEdite M.G.P. Fernandes, Algoritmi R&D Centre, Universityof Minho, School of Engineering, Campus de Gualtar,4710-057, Braga, Portugal, [email protected], Md. AbulKalam Azad, Ana Maria A.C. Rocha

The quadratic 0-1 knapsack problem finds a subset of items that yieldsmaximum profit without exceeding the knapsack capacity. Some exactas well as heuristic methods have been proposed for solving this typeof problem. Recently, a population-based artificial fish swarm (AFS)algorithm that simulates the behaviors of fish swarm inside water hasemerged. We propose a binary version of AFS for solving quadratic0-1 knapsack problems. To make infeasible solutions to feasible oneswe use the decoding algorithm. We test the method with a benchmarkset of problems available in literature.

� TD-20Tuesday, 14:30-16:00RB-2113

Decision Theory and Analysis

Stream: Decision Making Modeling and Risk Assess-ment in the Financial SectorInvited sessionChair: Tina Comes, Institute for Industrial Production (IIP),Karlsruhe Institute of Technology (KIT), Hertzstr. 16, 76187,Karlsruhe, Germany, [email protected]

1 - Assessing the Impact of Stock Volatility on the Effi-ciency of Listed Commercial Banks: A ConditionalNonparametric ApproachAnamaria Aldea, Economic Informatics and Cybernetics, TheBucharest Academy of Economic Studies, 010374, Bucharest,Romania, [email protected], Luiza Badin, CarmenLipara

In this paper we employ nonparametric conditional methodology toassess the impact of external factors like stock volatility on the effi-ciency of commercial banks. The approach allows decomposing theimpact of external factors in impact on the attainable set in the input-output space, and/or impact on the distribution of the efficiency scores.We analyze listed commercial banks from FactSet database, as theirreduced activity on financial markets is usually associated with lowerrisk. The results provide a perspective on how listed commercial banksperformed in 2009, after the crisis.

2 - Explaining Results and Uncertainties of a RobustScenario-Based Multi-Criteria Decision Support Sys-tem: Generation of Natural Language ReportsTina Comes, Institute for Industrial Production (IIP),Karlsruhe Institute of Technology (KIT), Hertzstr. 16, 76187,Karlsruhe, Germany, [email protected], Frank Schultmann

In decision support systems, the purpose of generating natural lan-guage reports is explaining the results of the analysis and comple-menting numeric results and visualisations. Natural language reportsadd transparency and clearness to the evaluation. To extend theseapproaches for scenario-based decision support, a framework for thecharacterisation of uncertainties within the scenarios and the identi-fication of the most significant uncertainties that drive the evaluationis presented. This novel approach is used to generate reports, whichdescribe the reliability of scenarios and explain the results and of ascenario-based multi-criteria decision support system.

3 - Implausible alternatives in the elicitation of multi-attribute value functionsRudolf Vetschera, Dept. of Business Administration,University of Vienna, Bruenner Str. 72, A-1210, Vienna,Austria, [email protected], Elisabeth Wolfsteiner,Wolfgang Weitzl

Many elicitation methods for multiattribute value functions requireholistic evaluations of hypothetical alternatives. Generation of such al-ternatives often ignores correlation between attributes present in realalternatives. We conducted an empirical study to test the effect ofimplausible alternatives on response consistency. Using a constraint-based approach, we found that respondents facing implausible alter-natives are more consistent with an additive model, and with statedintra-attribute preferences. All subjects had a low level of consistencywith stated inter-attribute preferences

4 - Modified BIPOLAR Method for Multi-criteria DecisionAnalyses under RiskOlena Sobotka, Operations Research, Wroclaw University ofEconomics, Komandorska str., 118/120, 53-345, Wroclaw,Poland, [email protected], EwaKonarzewska-Gubala

The multi-criteria decision problems with uncertainty outcomes of al-ternatives with known probability distributions are considered. Wepropose the outranking procedure for such alternatives based on themulti-criteria decision method BIPOLAR. The procedure is: 1) to de-fine the multi-criteria preference relations on the set of all alternative’soutcomes by comparing with sets of "good" and "bad" objects definedby decision maker and 2) to outrank the alternatives according to de-fined multi-criteria preference relations using the stochastic dominancerelations.

� TD-21Tuesday, 14:30-16:00RB-2115

Risk Management

Stream: Long Term Financial DecisionsInvited sessionChair: Ulla Hofmann, Institute for Management, UniversityKoblenz, Universitätsstraße 1, 50670, Koblenz, Germany,[email protected]

1 - The general approach to the modeling of the riskmanagement procedures and its specification for in-vestment problemsTatiana Zolotova, Chair f Applied mathematics, StateUniversity of the Ministry of Finance of the RussianFederation, 127006, Moscow, Nastasinsci lane, house 3, case2, Research financial institutions, 127006, Moscow, RussianFederation, [email protected]

The work is devoted to the description of the general model of the riskmanagement, including two submodels: the model of an estimation ofsystem effectiveness and the model of an estimation of the risk of itsfunctioning. The general model can be applied to decision-making inconditions of risk for various complex systems and processes. Prob-lems of management are considered for stochastic processes at thestrategic and share investment, using various estimations of risk. Meth-ods of these problems decision in view of criteria of efficiency and riskare offered.

2 - Growth Optimal Portfolio Insurance — does it payUrsula Walther, FB 1, Berlin School of Economics and Law,Badensche Str. 50-51, 10825, Berlin, Germany,[email protected]

Growth optimal portfolio insurance (GOPI) is a very promising ideabecause it combines the attractive features of growth optimality withcurbing the high risk of such portfolios. Recent research has proposedimplementation rules in discrete time. We explore the empirical poten-tial of GOPI in comparison to classic CPPI and option based portfolio

160


insurance. For this purpose we perform extensive simulations withseveral types of underyings (time-series based and real world). Perfor-mance is evaluated based on the entire return distribution applying alarge set of risk and return figures.

3 - Analysis of Measurement Methods of the UtilityFunction in context of Prospect TheoryUlla Hofmann, Institute for Management, University Koblenz,Universitätsstraße 1, 50670, Koblenz, Germany,[email protected], Thomas BurkhardtIn the literature, we find several experimental methods to measure thevalue function as well as the probability weighting function of prospecttheory. We compare several of those methods with regard to accuracy,practicability, and theoretical foundation. Our aim is to identify accu-rate and practical methods for individual PT-Utility functions, with aparticular focus on the assessment of risk attitudes in investment deci-sions.

4 - Long-term Investment Behaviour Considering Finan-cial Constraints — An Analysis in the Context of In-formation TechnologyStefan Pfosser, FIM Research Center Finance & InformationManagement, University of Augsburg, Universitätsstraße 12,86159, Augsburg, Germany,[email protected], Vasko IsakovicIn line with existing studies, this paper explains the observable invest-ment behaviour of firms in the context of Information Technology (IT)by considering their financial constraints as well as specific aspects ofIT investments. By applying a dynamic optimization model, the scarcefinancial resources are optimally allocated to IT investments using acash reserve, so that their value contribution is maximized. Since weshow that a financially constrained firm can achieve a higher value con-tribution by holding cash, we make an important contribution towardsmaking better long term decisions.

� TD-22Tuesday, 14:30-16:00RB-2117

Experimental Economics and GameTheory

Stream: Experimental Economics and Game TheoryInvited sessionChair: Ulrike Leopold-Wildburger, Statistics and OperationsResearch, Karl-Franzens-University, Universitätsstraße 15/E3,8010, Graz, Austria, [email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - Reputation of arbitrators in bargaining modelsVladimir Mazalov, Karelia Research Center of RussianAcademy of Sciences, Institute of Appied MathematicalResearch,Karelia Research Center, Pushkinskaya st. 11,185910, Petrozavodsk, Karelia, Russian Federation,[email protected], Julia TokarevaWe consider final-offer arbitration procedure in case where there ismore than one arbitrator. Two players, 1 and 2, are interpreted as La-bor and Management and in dispute about an increase in the wage rate.They submit final offers to a Jury with N arbitrators, each of them hasher own assessment presented by some CDF and reputation in Jury.Then we find the assesment distributed with weighted CDF of Juryand select the offer which is closest to this assessment. The Nash equi-librium in this arbitration game is derived. We compare the optimalstrategies with voting procedure solution.

2 - The effects of honesty and social preferences on re-porting behaviorUlrike Leopold-Wildburger, Statistics and OperationsResearch, Karl-Franzens-University, Universitätsstraße 15/E3,8010, Graz, Austria, [email protected], ArletaMietek

Reporting is an important issue in management accounting. In situa-tions characterized by information asymmetry in an organization, e.g.,in the situation of upper and lower management, reporting is necessaryto ensure that the upper management has all the necessary informa-tion. In our experimental study we analyze the reporting behavior indifferent settings. We want to know how an agent’s participation in thereporting process influences his/her willingness to truthful communi-cation of private information, if there are (no) financial incentives tomisrepresentation.

3 - Finding a fair cost allocation mechanism in flexiblehorizontal supply chainsChristine Vanovermeire, University of Antwerp, 2000,Antwerp, Belgium, [email protected],Kenneth SörensenAs companies are faced with more stringent demands of clients andenvironmental laws, new logistics solutions such as horizontal collab-oration among shippers are emerging. The total gain of such initiativeranges from 9%-30%. Moreover, more optimization opportunities ex-ist as companies can be more flexible towards the aspects of delivery(e.g. delivery date) in order to create a better ’fit’ with their partners.However, one of the main implementation issues of such a collabora-tion is dividing those gains. The ability of the Shapley value to giveincentives to be flexible is investigated.

� TD-23Tuesday, 14:30-16:00RB-Delta

Recent Advances of MathematicalProgramming in Supply ChainManagement

Stream: Mathematical ProgrammingInvited sessionChair: Turan Paksoy, Industrial Engineering, Science, SelcukUniversity Industrial Engineering Department, Campus, 42100,Konya, Turkey, [email protected]: Eren Ozceylan, Industrial Engineering, Natural and AppliedSciences, Selcuk University Industrial Engineering Department,Campus, 42031, Konya, Turkey, [email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - An exact algorithm for the green vehicle routingproblemÇagrı Koç, Industrial Engineering, Selcuk University, Turkey,[email protected], Ismail Karaoglan, Sevgi ErdoganThe Green Vehicle Routing Problem (G-VRP) accounts for the addi-tional challenges associated with operating a fleet of Alternative FuelVehicles (AFVs). This study offers a mixed integer programming for-mulation, an effective branch-and-cut algorithm and a simulated an-nealing algorithm for solving the G-VRP. The proposed branch-and-cut algorithm implements several valid inequalities adapted from theliterature for the problem and a heuristic algorithm based on simulatedannealing algorithm to obtain upper bounds.

2 - Artificial Bee Colony Algorithm for Solving Uncapac-itated Facility Location ProblemsMustafa Servet Kıran, Computer Engineering, Science,42075, Konya, Turkey, [email protected], ErenOzceylan, Turan PaksoyArtificial bee colony (ABC) algorithm is first investigated for solvingnumerical optimization problems, inspired waggle dance and foragingbehaviors of honey bee colonies. In this study, an adapted version ofABC is proposed to solve binary optimization problems such as un-capacitated facility location problem (UFLP). The accuracy and per-formance of ABC have been examined on well-known twelve small,medium and large sizes UFLPs and compared with particle swarm op-timization (PSO). The experimental results show that ABC is betterthan PSO in terms of solution quality and robustness.

161


3 - Comparisons of Different Fuzzy Mathematical Pro-gramming Approaches on Supply Chain Network De-sign ProblemNimet Yapici Pehlivan, Statistics, Selcuk University, SelcukUniversity Science Faculty, Campus Konya/TURKEY, 42075,Konya, Turkey, [email protected], Eren Ozceylan, TuranPaksoyIn most realistic supply chain network (SCN) design problems, envi-ronment coefficients, model parameters and objectives are frequentlyimprecise. Fuzzy set theory provides an appropriate framework to de-scribe uncertainty. The aim of this paper is investigating the effects ofdifferent fuzzy mathematical programming approaches on a SCN de-sign problem. A fuzzy mathematical model is developed for a multi-echelon SCN which has uncertainty in capacity, demand and objec-tives. A test problem is presented and comparison of results of differ-ent approaches is made using proposed problem.

4 - Integration of Reverse Supply Chain Optimizationand Disassembly Line BalancingEren Ozceylan, Industrial Engineering, Natural and AppliedSciences, Selcuk University Industrial EngineeringDepartment, Campus, 42031, Konya, Turkey,[email protected], Turan PaksoyWhile reverse supply chain (RSC) optimization includes recycling anddisassembling activities through a supply chain; disassembly line bal-ancing (DLB) problem involves determination a line design in whichused products are disassembled. The aim of this study is to optimizeand design the RSC that minimizes the transportation costs while bal-ancing the disassembly lines in disassembly centers, which minimizesthe total fixed costs of opened stations, simultaneously. A numericalexample and different scenarios are conducted to illustrate the applica-bility of the proposed model.

� TD-24Tuesday, 14:30-16:00CC-A11

Forecasting methods

Stream: ForecastingInvited sessionChair: Aris Syntetos, Centre for Operational Research andApplied Statistics - Salford Business School, University ofSalford, Maxwell Building, The Crescent, M5 4WT, SALFORD,United Kingdom, [email protected]

1 - Boot.EXPOS:Strength and sensibilityClara Cordeiro, CEAUL, University of Algarve, CampusGambelas„ FCT, DM, 8005-139, Faro, Portugal,[email protected], Manuela NevesExponential smoothing methods are the launch platform for modellingand through bootstrap several sample paths for a time series are cre-ated. Since its beginning, Boot.EXPOS algorithm has showed promis-ing results. For dependent data is well-know the bootstrap vocationin obtaining forecast intervals and now we also deviate this attentionfor point forecast proposes. We will illustrate the performance of ourprocedure for some well-know data sets and will emphasize its recentadaptation for forecasting time series with complex seasonal patterns.

2 - Complexity of exponential smoothing models and itsinfluence on forecast quality in retailJohannes Püster, University of Münster - ERCIS,Leonardo-Campus 3, 48149, Münster, NRW, Germany,[email protected], Axel Winkelmann,Justus Holler, Jörg BeckerRetail poses several challenges on sales forecasting. The combinationof broad assortment and high number of stores leads to a high amountof time series, which need to be forecasted on a daily basis at providersof fast moving consumer goods. Appropriate forecasting models needto be of good forecasting quality and of low complexity in order toallow for fast execution. This work presents a structured overview oncomplexity-influencing parameters of exponential smoothing models.Based on a data set of 250 retail time series the influence of modelcomplexity on forecast quality is elaborated.

3 - Estimating autoregressive models in the presenceof missing data and non-normally distributed errortermsKorneel Bernaert, MOSI, VUB, Pleinlaan 2, 1030, Brussels,Belgium, [email protected], An Peeters

Estimation of the parameters of an autoregressive model in the pres-ence of missing data is typically performed using the Kalman filter toobtain maximum likelihood estimators. These rely on the error termbeing normally distributed. This paper constructs estimators using themethod of moments estimators which do not rely on this normality ofthe error term. The performance of the method of moments estimatoris then compared to the maximum likelihood and some often-used adhoc (mean imputation, deletion) estimators in a simulation study for afirst-order autoregressive model.

4 - Neural networks for streamflow series forecasting: Acomparative study between echo state networks andMLP’sHugo Siqueira, Department of Systems Engineering,University of Campinas, College of Eletrical Engeneering,Cidade Universitária Zeferino Vaz, Barão Geraldo District,13083-970, Campinas, São Paulo, Brazil,[email protected], Levy Boccato, Romis Attux,Christiano Lyra

The forecast of seasonal streamflow series is fundamental for the sus-tainable use of water resources and it is a key information for hydro-electric generation planning. Echo state networks, recurrent structurescharacterized by a tradeoff between generation of memory from feed-back and straightforwardness, are a contemporaneous alternative to ad-dress the problem. However, the possible benefits of this approach inthe problem still need proper clarification. This talk gives a contribu-tion by discussing a comparative study of different ESN architecturesand a MLP, in several practical scenarios.

� TD-25Tuesday, 14:30-16:00CC-A23

Pathways toward low carbon transport

Stream: Long Term Planning in Energy, Environmentand ClimateInvited sessionChair: Sandrine Selosse, Centre for Applied Mathematics, MINESParisTech, Rue Claude Daunesse, 06904, Sophia Antipolis,France, [email protected]

1 - GHG emissions cap impact on French biofuel mix us-ing MARKAL/TIMES ModelPaul Hugues, Centre for Applied Mathematics, MINESParisTech, Rue Claude Daunesse, BP 207, 06904, SophiaAntipolis, France, [email protected], EdiAssoumou, Nadia Maïzi

One of the main reasons of biofuels development has been carbon mit-igation. However, greenhouse gas emissions related to biofuel pro-duction are dependent on the whole production pathway as well asland use change. To consider the wide impact of biofuel productionin long-term prospective, a full range of diverse but integrated aspectshave to be considered, such as: biomass supply, transportation, pack-aging, and transformation to biofuel. In this work, we investigate howa CO2 emissions cap or a carbon tax would impact the French pathwayto low emissions, and the biofuel mix.

2 - Impact of a carbon tax on french tranports sector: ATIMES model for mobility.Jean-Michel Cayla, EDF/Mines ParisTech, Paris, France,[email protected]

In order provide insights about realistic pathways to low-carbon trans-ports in 2050, we have developed a TIMES model with a strongly de-tailed module for transports in France. In particular we have developed

162


an original approach that takes households heterogeneity and house-holds investment behavior as well as daily consumption behavior intoaccount. We are then able to better design realistic technology diffu-sion curves. Then we study the impact of a carbon tax in transportssector and its ability to reduce CO2 emissions.

3 - The option value of electrified vehicles from a welfareperspectiveAdrien Vogt-Schilb, CIRED, 75011, Paris, France,[email protected], Manon Solignac, Céline Guivarch

We evaluate how much it is desirable to pay today to have the option ofbuying electrified vehicles (EVs) tomorrow. From exogenous fuel andelectricity prices distributions, we model total costs of ownership andmarket shares of a set of technologies. We compare scenarios whereEVs are available with scenarios without EVs. We quantify how EV’savailability reduces the expected cost of the automotive system and itsexposure to oil prices variability. Adding an economy-wide carbonconstraint on the model, we show that EV’s availability also reducesthe expected carbon price.

4 - Technology and social factors in passenger trans-port: a comparative assessment of different lever-agesFrançois Briens, Centre for Applied Mathematics, MINESParisTech, 06904, Sophia Antipolis, France,[email protected], Edi Assoumou

This prospective research evaluates and compares the sensitivity ofCO2 emissions and energy consumption of the French passenger ve-hicles fleet up to 2035, to different technological and sociological pa-rameters, such as electric and hybrid technology penetration rates, theshares of the different power class in the fleet, the annual averagemileage, and households’ motorisation levels. The results highlightthe importance of sociological factors and advocate for integrated ap-proaches to engage in ambitious pathways toward low energy and car-bon societies.

� TD-26Tuesday, 14:30-16:00CC-A24

OR in Agriculture, Forestry and Fisheries

Stream: OR in Agriculture, Forestry and FisheriesInvited sessionChair: Marcela Gonzalez-Araya, Departamento de Modelación yGestión Industrial, Universidad de Talca, Merced 437, s/n, Curicó,Región del Maule, Chile, [email protected]

1 - Multiple criteria and group decision methods to val-uate ecosystem servicesMarina Segura Maroto, Applied Statistics, OperationsResearch and Quality, Universitat Politècnica de València,Camino de Vera S/n, 46071, Valencia, Valencia, Spain,[email protected], Concepcion Maroto, BaldomeroSegura

The traditional procedure of economic valuation is based on individualutility maximization. Nevertheless, many ecosystem services are con-sidered as public goods, affecting people and groups in society. Mul-tiple Criteria and Group Decision Making have been used mainly inenvironmental decision making, more than to explicit economic valuesfor ecosystem goods and services. This paper presents an exhaustiveup-to-date review with a classification of approaches and applications,pointing out its main strengths and drawbacks to contribute to the val-uation of ecosystem services in the future.

2 - Optimal control models for sustainable fishery man-agementJoao Lauro D. Faco’, Dept. of Computer Science,Universidade Federal do Rio de Janeiro, Av. do PEPE, 1100 /203, 22620-171, Rio de Janeiro, RJ, Brazil, [email protected]

Sustainable management of fisheries is a pressing need for fish increas-ing importance as a high quality protein source. Stocks of the last wildfood are collapsing in many regions of the ocean. Optimal Controltheory is used to build spatial/temporal models for complex marineecosystems problems. Dynamic population interactions for competi-tor/predator/prey multispecies are coupled discrete-time nonlinear dif-ference equations. A hyperbolic smoothing Nonlinear Programmingmethod defines specific clusters; a reduced-gradient method is appliedto search the solution, and to parameters estimation.

3 - A mixed integer linear model for planning harvest ap-plied to apple orchardsMarcela Gonzalez-Araya, Departamento de Modelación yGestión Industrial, Universidad de Talca, Merced 437, s/n,Curicó, Región del Maule, Chile, [email protected],Wladimir Soto-Silva

An optimization model to support planning decisions of harvest in ap-ple orchards is developed. The model seeks to minimize the costs ofhiring and firing work labor, machinery use and loss of fruit qualityby not harvesting in the maturity date, subject to restrictions demandfor packaging plants, processing capacity packing plants, constraintsabout production in each field of the orchard, time window constraintsfor the harvest according to the apple variety planted in each field,among others. This model was applied to three apple orchards ofMaule Region.

4 - Sustainability of agriculture - only a dream consider-ing the world population growth?Joern Hamann, Food quality and safety, University ofVeterinary Medicine, Hannover, FRG, Platanenring 10,24119, Kronshagen, Schleswig-Holstein, Germany,[email protected]

Sustainability of agriculture tries to meet mankind’s needs of thepresent without compromising the future. The agricultural sector - cor-nerstone of food security - has to provide food for almost 35 % morepeople globally by mid-century. Without doubt, this can and will bereached. However, the question is whether it is possible with a sus-tainable agriculture. In any case, a lot of money must be spent onagricultural innovations in order to reach the goal of sufficient food forall especially without any further impairment of biological systems.

� TD-27Tuesday, 14:30-16:00CC-A25

Uncertainties in Decision ProcessesStream: Decision ProcessesInvited sessionChair: Ahti Salo, Systems Analysis Laboratory, Aalto UniversitySchool of Science and Technology, P.O. Box 11100, Otakaari 1 M,00076, Aalto, Finland, [email protected]

1 - Sources and management of uncertainty in invest-ment appraisals influenced by corporate social re-sponsibilityAron Larsson, Computer and System Sciences, StockholmUniversity, 16440, kista, Stockholm, Sweden,[email protected], Sara Vickman, Leif Olsson

In an exploratory study, we assess how managers try to account forcorporate social responsibility in investment appraisals. We focus onhow they manage uncertainty and the source of the uncertainties theyemphasize, and conclude that they are generally exposed to and alsoconsider both aleatory and epistemic uncertainty and need to handleaspects beyond ROI. There is a mix of macro and micro variables toconsider, creating a diversified environment where real options in com-bination with decision analysis modeling could help in structuring theproblem and enlighten different areas and aspects.

163


2 - Bayesian evaluation and selection strategies in port-folio decision analysisEeva Vilkkumaa, Department of mathematics and systemsanalysis, Aalto University, School of science, Systemsanalysis laboratory, Aalto University, P.O.Box 11100, 00076Aalt, Espoo, Finland, [email protected], Juuso Liesiö,Ahti Salo

In project portfolio selection, the projects’ future values are typicallyuncertain. We show how the Bayesian modeling of these uncertain-ties increases the expected value of the selected portfolio, raises theexpected number of optimal projects in the selected portfolio and de-creases post-decision disappointment. We also show that it pays offto re-evaluate only those projects which can be re-evaluated relativelyaccurately and which have particularly uncertain initial value estimatesthat are close to the selection threshold.

3 - Uncertain mine clearance quality and how it affectsthe decision to perform quality control in humanitar-ian deminingTobias Fasth, Department of Computer and Systems Sciences,Stockholm University, Forum 100, SE-164 40, Kista, Sweden,[email protected], Aron Larsson, Love Ekenberg

The IMAS standard for mine actions conforms to the ISO 2859 stan-dard for quality control of cleared land. ISO 2859 is primarily aimedfor use in the sampling of traditional industrial products, and not oncleared land. The sampling plans are dependent on the acceptable qual-ity limit. This quality limit is subject to uncertainty and we specificallyaddress the problem of the deviance in the actual and acceptable qual-ity of cleared land, and how this influence the decision process andthe cost of quality control, leading to the question if whether qualitycontrol should be conducted at all.

4 - Action Research for assessing the impact ofscenario-based multi-criteria interventions on thestrategic decision-making processCamelia Ram, Operational Research Group, London Schoolof Economics, Houghton Street, WC2A 2AE, London,[email protected]

The literature on the combined use of scenarios and multi-criteria de-cision analysis (MCDA) continues to grow. While action research hasbeen advocated as an appropriate method for studying interventionsinvolving MCDA, guidelines on applying the methodology to provideunique insights for future interventions remain limited. This paper ex-plores how action research was used to assess the value of a scenario-based MCDA intervention. The methodology was applied to supportthe evaluation and enhancement of strategic options for a regional cor-poration in a developing country.

� TD-28Tuesday, 14:30-16:00CC-A27

Anticipation and Synchronization

Stream: Models of Embodied CognitionInvited sessionChair: Susie Vrobel, The Institute for Fractal Research,Ernst-Ludwig-Ring 2, 61231, Bad Nauheim, Hessen, Germany,[email protected]

1 - Trust as Embodied AnticipationSusie Vrobel, The Institute for Fractal Research,Ernst-Ludwig-Ring 2, 61231, Bad Nauheim, Hessen,Germany, [email protected]

Trust is defined as a temporal observer-participant extension. This ex-tension is described in terms of succession and simultaneity, i.e. thelength and depth of time: mutually exclusive temporal dimensions asdefined in my Theory of Fractal Time. While succession increasestemporal complexity from an endo-perspective, simultaneity reducesit through embodied anticipation. Successful incorporation of parts ofthe environment leads to the formation of a new systemic whole.

2 - Distortions of spatial phase and temporal-synchronism disorders in amblyopiaUwe Kämpf, Research Unit, Caterna GmbH, Biensdorf 78,D-01825, Liebstadt, Saxonia, Germany,[email protected]

Amblyopia is a functional impairment of visual performance whichis partially due to coherence losses in the cooperation of visualfrequency-filter channels. Chaotic instead of synchronized states in re-verberating feed-forward loops (i.e. a lack of perceptual anticipation)induce disorders of temporal synchronism in the visuo-cortical sys-tem’s neuronal "synfire chains’. The result is a distortion in the visualfilters’ spatial phase. Accordingly, a narrow-band frequency-selectivepattern stimulation is proposed as a supportive treatment of amblyopia.

3 - Effects of accuracy feedback on fractal characteris-tics of time estimationSebastian Wallot, MINDLab, Aarhus University, Jens Chr.Skous Vej 20, 8000, Aarhus C, Denmark,[email protected], Nikita Kuznetsov

The current experiment investigated the effect of visual accuracy feed-back on the variability of time interval estimates. To assess the impactof feedback, the likelihood of receiving visual feedback was manip-ulated. Inter-tap intervals were analyzed with fractal and multifrac-tal methods to quantify complex long-range correlation patterns in thetiming performance. Increased amounts of feedback lead to a decreasein fractal long-range correlations, as well to a decrease in the mag-nitude of local fluctuations in the performance. Also, most data setsshow significant multifractal signatures.

� TD-29Tuesday, 14:30-16:00CC-A29


Stream: DEA and Performance Measurement (con-tributed)Contributed sessionChair: Chris Tofallis, Business School, University ofHertfordshire, College Lane, AL10 9AB, Hatfield, Herts., UnitedKingdom, [email protected]

1 - Constructing rankings in DEA - a goal programmingapproachPaulo Morais, Engenharia Industrial e Gestão, Faculdade deEngenharia da Universidade do Porto; Universidade Lusófonado Porto, Rua Roberto Frias, Rua Augusto Rosa, 24,4200-465, Porto, Portugal,[email protected], Ana Camanho

This paper presents a goal programming model that allows comparingunits assessed with DEA. In classical models, given the fact that eachunit is assessed with a unique set of weights that aims to maximizeits efficiency, the construction of rankings is not possible. Weights re-strictions to establish a common set of weights is widely discussed inliterature. Using the evaluation of quality of life of 206 European citiesas an example, we propose a goal programming model that minimizesthe deviations in relation to a composite indicator score obtained in aprevious stage.

2 - On the Environmental Variables Analysis in TwoStages DEARafael C Leme, Federal University of Itajuba, UNIFEI-IEPG,CP 50, 37500-903, Itajuba, MG, Brazil, [email protected],Anderson Paulo Paiva, Pedro Paulo Balestrassi, PauloEduardo Steele Santos

Benchmarking plays a central role under regulatory scene. DEA isa non-parametric efficiency performance technique usually used forbenchmark analysis. The frontier is defined by the efficient firms thatenvelop the less efficient firms, defining efficiency gap. An issue ariseswhen some environmental variables may influence the results. In this

164


work, we propose an approach based in Design of Experiments andDesirability method to identify the impact of environmental variableson inefficiency gap, and define which most affect inefficiency. Theresults are obtained using real Brazilian data.

3 - Value Efficiency for the FDH modelMerja Halme, Information and Service Economy, AaltoSchool of Economics, Pl 11000, 00076, Aalto, Finland,[email protected], Pekka Korhonen, Juha Eskelinen

An approach to supplement preference information in the FDH modelis introduced. We assume the Decision Maker (DM) evaluates the unitsusing an implicit value function reaching its maximum at his/her mostpreferred unit as in Value Efficiency Analysis (VEA). The value func-tion is assumed to be quasi-concave and we use convex cones to ap-proximate its contours. Pairwise comparisons are used as preferenceinformation. As in VEA the distance of each unit from the value func-tion’s contour passing through the most preferred unit is approximated.We discuss a real case of bank branches.

4 - Economic efficiency of europen air traffic controlsystemsKenneth Button, School of Public Policy, George MasonUniversity, 3351 Fairfax Drive, (MS 3B1), 20124, Arlington,Virginia, United States, [email protected], Rui Neiva

Much analysis has been conducted on the efficiency of airlines, but lesshas been done on air traffic control systems (ANSs). The diversity ofANSs regimes in Europe allows rigorous comparison of 36 Europeansystems from 1998 to 2010. The bootstrapped DEA framework as-sumes a multifaceted output function (safety and delay objectives) withboth physical inputs (traffic volumes, labor, investment, flight hourscontrolled) and policy (ownership and financing systems). Attentionis paid to the relative efficiencies of different types of partnerships andthe implications of financial structures.

� TD-30Tuesday, 14:30-16:00CC-A31

Data confidentiality 2

Stream: Data ConfidentialityInvited sessionChair: José Antonio González Alastrué, Dept. Estadística eInvestigación Operativa, UPC, Edificio C5, Campus Nord, JordiGirona 1-3, 08034, Barcelona, Spain, [email protected]

1 - A Fix and Relax Heuristic for Controlled Tabular Ad-justmentDaniel Baena Mirabete, Statistics and Operations Research,Universitat Politècnica de Catalunya, Spain,[email protected], Jordi Castro

Controlled Tabular Adjustment (CTA) is an emerging protection tech-nique for tabular data. CTA results in a mixed integer linear program-ming problem, which can be challenging even for tables of moderatesize. In this work the Fix and Relax (FR) strategy, based on partitioningthe set of binary variables into clusters to selectively explore a smallerbranch and bound tree, is applied to large CTA instances. We reportcomputational results on a set of real and random CTA instances. FR iscompetitive compared to Cplex in terms of finding a good upper boundin difficult instances.

2 - Potential Breaches of Confidentiality in Statistical Ta-bles containing Magnitude DataMartin Serpell, Department of Computing, University of theWest of England, Frenchay Campus, Coldharbour Lane, BS161QY, Bristol, United Kingdom, [email protected],Jim Smith, Alistair Clark

The purpose of this paper is to highlight potential unintended breachesin confidentiality in published statistical tables that have been pro-tected using cell suppression. In order to protect large tables somecell suppression algorithms employ back-tracking heuristics. Tests

show that although the back-tracking heuristics can handle multi-dimensional non-hierarchical tables they fail to adequately protect hi-erarchical ones. On average having one hierarchical dimension willleave 30% of primary cells unprotected, two will leave 60% unpro-tected and three will leave 90% unprotected.

3 - On data protection for graphs: application to socialnetworksVicenc Torra, IIIA-CSIC, Campus UAB s/n, 08193,Bellaterra, Catalonia, Spain, [email protected], Klara Stokes

Graphs are versatile structures that can be used to represent severaltypes of information. In particular, the structure and the content of so-cial networks can be represented in terms of graphs, where nodes cor-respond to users and edges correspond to the relationships establishedbetween them. Data protection for graphs requires a proper definitionfor disclosure risk and information loss, as well as adequate data pro-tection methods. In this talk we will discuss k-anonymity for graphs,some information loss measures, and the performance of protectionmethods with respect to the measures.

4 - Improving the solution of CTA through valid inequal-itiesJosé Antonio González Alastrué, Dept. Estadística eInvestigación Operativa, UPC, Edificio C5, Campus Nord,Jordi Girona 1-3, 08034, Barcelona, Spain,[email protected], Jordi Castro

Controlled tabular adjustment solves a mixed integer linear problemfor protecting statistical tables against the risk of disclosure, which isa key concern for National Statistical Agencies. In this work we ob-tain some valid inequalities for CTA and show they may be useful insignificantly improving exact (branch and bound) and heuristic (blockcoordinate descent) procedures for CTA. Valid inequalities are derivedfrom a search through the standard constraints of the table, wheneverthey include some sensitive cells.

� TD-31Tuesday, 14:30-16:00CC-A33

Data Mining in Economics and Ecology

Stream: Data Mining, Knowledge Discovery and Artifi-cial IntelligenceContributed sessionChair: Ivan Reyer, Dorodnicyn Computing Centre of RAS, ul.Pererva d. 58, kv. 153, 109451, Moscow, Russian Federation,[email protected]: Renato Bruni, Dip. Ingegneria Informatica, Automatica eGestionale, University of Roma "Sapienza", Via Ariosto 25,00185, Roma, Italy, [email protected]

1 - Predicting the customer churn in Social Networkgames based on multiple criteria decision makingmodelsChiao-Chen Chang, International Business, National DongHwa University, No. 1, Sec. 2, Da Hsueh Rd., Shoufeng,97401, Hualien, Taiwan, [email protected], Yang-ChiehChin

Customer churn has become a critical issue, especially in the competi-tive and mature social network game industry. This study applies roughset theory, a rule-based decision-making technique, to extract rules re-lated to customer churn; then adopts the Decision Making Trial andEvaluation Laboratory (DEMATEL) technique to analyze the relation-ship among characteristics in customer churn; and finally uses the flowgraph, a path-dependent approach, to present the relationships betweenrules and different kinds of churn. Practical and research implicationsare also offered.

165


2 - Modeling partial customer churn: on the value of firstproduct-category purchase sequencesVera Miguéis, DEIG, Faculdade de Engenharia daUniversidade do Porto, Porto, Portugal,[email protected], Dirk Van den Poel, Ana Camanho,João Cunha

Retaining customers has been considered one of the most critical chal-lenges among those included in Customer Relationship Management(CRM). This paper proposes to include in partial churn detection mod-els the succession of first products’ categories purchased as a proxy ofthe state of trust and demand maturity of a customer towards a com-pany in grocery retailing. We use logistic regression as the classifica-tion technique. A real sample of approximately 75,000 new customerstaken from the data warehouse of a European retail company is used totest the proposed models.

3 - Electricity consumption and economic growthcausality in OECD countries: a panel cointegrationapproachKasirga Yildirak, IAM, METU, 06531, Ankara, Turkey,[email protected], Ali Sabri Taylan, CumhurEkinci

This paper examines the long-run, short-run and causal relationship be-tween electricity consumption and economic growth for a set of OECDcountries. We also investigate the presence of any directional causal-ity between real economic activity and electricity consumption. Fortesting the relationship between electricity consumption and economicgrowth, we make use of use panel unit roots, panel cointegration test,panel causality and panel vector autoregressive model.

4 - Predicting the fundamental value of financial assetsby ridge regressionMichael Kamp, IAIS.KD.CAML, Fraunhofer IAIS, SchlossBirlinghoven 1, 53754, Sankt Augustin, NRW, Germany,[email protected], Mario Boley, ThomasGaertner

Assessing the fundamental value of financial assets is crucial for in-vestment as well as for market regulation. This problem is difficult,because present fundamental values depend on disclosed informationand future value developments. Therefore, estimating them is es-sentially a prediction task that can be addressed by machine learn-ing. We show that good predictions can be achieved by performingridge regression based on novel correlation features that capture intra-and inter-dependencies of market segments. Despite their expressivepower, these features can be extracted from public data.

� TD-32Tuesday, 14:30-16:00CC-A34

National Integrated Assessment Models

Stream: Emissions Trading and EnergyInvited sessionChair: Magnus Fröhling, Institute for Industrial Production (IIP),Karlsruhe Institute of Technology (KIT), Hertzstraße 16, D-76187,Karlsruhe, Germany, [email protected]

1 - Environmental assessment of energy scenarios forthe deployment of carbon capture and storage (CCS)on coal fired power plants in the UKKonstantinos Tzanidakis, Department of Power andPropulsion, Cranfield University, Building 183, Room 803,MK430AL, Bedford, United Kingdom,[email protected], Tim Oxley, Helen ApSimon

Looking to 2030 and beyond integrated assessment modelling needs tobe extended to cover new energy technologies. We describe applica-tion of the UKIAM to different scenarios for the introduction of carboncapture & storage (CCS) in the UK up to 2050, and investigate green-house gas emissions and air quality implications for human health and

ecosystems. This involves using a life cycle analysis approach to re-flect the emissions arising from the development and operation of CCSand associated storage systems. The scenarios extend to 2050 and in-clude comparison of both pre- and post-combustion CCS

2 - A simulation-based approach for a national inte-grated assessment model for GermanyPatrick Breun, Institute for Industrial Production, KarlsruheInstitute of Technology (KIT), Hertzstraße 16, D-76187,Karlsruhe, Germany, [email protected], Rebecca Ilsen,Magnus Fröhling, Frank Schultmann

The talk describes the otello model, a national integrated assessmentmodel (IAM) to support clean air policy in Germany. The model aimsat providing a tool to assess instruments for control of SOx, NOx, NH3,and CO2 emissions from industrial production, energy supply, trans-port and residential buildings with respect to sustainability. Differenttypes of policy instruments are evaluated considering criteria from thethree dimensions of sustainability. The simulation-based model, espe-cially the industry sub model, are introduced and examples of scenariocalculations are presented.

3 - Managing without Growth: Italian ScenariosSimone D’Alessandro, Scienze Economiche, University ofPisa, via Ridolfi 10, 56124, Pisa, Italy, [email protected],Giorgio Gallo, Giovanni Bernardo

This paper presents a structural model for the Italian economy by ap-plying system dynamics to scenario analysis. Government affects theperformance of the economy, through the use of instruments such astaxation, labour policies and market regulation. The aim is twofold.First, we investigate the impact of public policies on six main perfor-mance indexes: rate of growth, unemployment, poverty, fiscal position,GHG emission, and inequality. Second, we evaluate whether it is pos-sible to disjoint wellbeing and economic growth.

4 - Computer modeling of urban air quality and relatedhelth effectsPiotr Holnicki, Systems Research Institute of the PolishAcademy of Sciences, Newelska 6, 01-447, Warsaw, Poland,[email protected], Zbigniew Nahorski, Marko Tainio

An important application of air pollution models is air quality manage-ment and decision support. Modelling techniques allow us to assessecological impact of emission sources and benefits of emission abate-ment. Complexity of such forecasting systems involves uncertainty ofthe model outputs. This paper addresses the problem of uncertainty ofemission inventory and the impact on air pollution predictions and ad-verse health effects. The computational experiment was implementedfor Warsaw Metropolitan Area, Poland, with the CALPUFF air disper-sion model used as the forecasting tool.

� TD-33Tuesday, 14:30-16:00CC-A37

OR for Education and SustainableDevelopment

Stream: OR for Development and Developing Coun-triesInvited sessionChair: Alexander Makarenko, Institute for Applied SystemAnalysis, National Technical University of Ukraine "KPI",Prospect Pobedy 37, 03056, Kiev, Ukraine, [email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - Self-organization processes in the European HigherEducation AreaMarek Frankowicz, Faculty of Chemistry, JagiellonianUniversity, ul. Ingardena 3, 30-060, Krakow, Poland,[email protected]

166


Usefulness and applicability of the theory of complexity to character-ize evolution and structural changes in European higher education isdiscussed. Intrinsic dynamics of various processes induced by Bolognareforms is highlighted. Three cases are analyzed in more detail: dy-namics of international mobility flows, the concept of self-extractingquality management systems, and sectoral qualification frameworksviewed as complex adaptive systems. Particular attention is paid topossibilities of applying OR techniques to support decision makingprocesses at the national and European level.

2 - Industrial heritage as an educational polygon for de-velopment strategiesVladimir Hain, Institute of History and Theory ofArchitecture and Monument Restoration, Slovak university oftechnology, Faculty of architecture, Námestie slobody 19, 81245, Bratislava, Slovakia, Slovakia, [email protected],Eva Kralova

Industrial heritage provides one of the most important records on de-velopment of cities and towns last two centuries. The basic purpose ofthe "Educational polygon" is to develop a model and operational toolsto raise the awareness of the importance of valuable buildings as a partof the cultural and technological history to stimulate social, econom-ical and political decision to protect them in a new creative way. Itsprioritize conservation is a necessary aspect of a city development andstudy shows example how to set up an appropriate framework for itsintegration within existing structure.

3 - A hybrid MCDM approach to assess the sustainabil-ity of students’ preferences for university selectionMetin Dagdeviren, Department of Industrial Engineering,Engineering Faculty, Celal Bayar Bulvar, Gazi Unv. MMF,Endustri Muh., 06570, Ankara, Turkey,[email protected], Kabak Mehmet

An integrated approach which employs ANP and PROMETHEE to-gether is proposed. It is concerned with criteria influencing stu-dent choice in Turkey to establish a model for predicting students’preference for universities. The ANP is used to analyze the struc-ture of the problem and to determine weights of the criteria, and thePROMETHEE is used to obtain final ranking, and to make a sensitivityanalysis. The results indicate that three factors, future career prospectsand opportunities, scholarship and university’s social life at the top inthe university selection.

4 - Competitive market vs. central planning economyand long-run growthKenji Miyazaki, Faculty of Economics, Hosei University,4342 Aihara, 194-0298, Machida, Tokyo, Japan,[email protected]

We use a simple model with an endogenous discounting rate and a lin-ear technology to investigate whether a competitive equilibrium has ahigher balanced growth path (BGP) than a social planning solution.The implications are as follows. People with an inclination to comparethemselves with others have an endogenous discounting rate. Suchan inclination affects the economic growth rate in a competitive mar-ket economy. A competitive market economy sometimes achieves ahigher economic growth than a central planning economy, but it some-times fluctuates due to self-fulfilling prophecy.

� TD-34Tuesday, 14:30-16:00CC-A39

OR for Sustainable Living andDevelopment

Stream: Sustainable Living: Cognitive, Social, Eco-nomical, Ecological and World ViewInvited sessionChair: Pedamallu Chandra Sekhar, Department of MedicalOncology, Dana-Farber Cancer Institute, Boston, 02115, Ipswich,MA, United States, [email protected]

Chair: Theodor Stewart, Statistical Sciences, University of CapeTown, Upper Campus, 7701, Rondebosch, South Africa,[email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - Healthy housing environment in small towns inSoutheastern PolandJustyna Kobylarczyk, Cracow University of Technology,Poland, [email protected] threats of contemporary civilization make it more and more diffi-cult to live in the city. Therefore, sustainable development is becominga profound need of modern times. The author’s environmental researchcarried out in some small towns of Southeastern Poland made it pos-sible to enumerate the most important needs of the inhabitants of thisregion. They influence the creation of a healthy housing environmentwhich is the imperative objective of sustainable development. Amongother features, they include safety and privacy as well as the nearnessof nature.

2 - Combined multicriteria and scenario analysis as aidto sustainable developmentTheodor Stewart, Statistical Sciences, University of CapeTown, Upper Campus, 7701, Rondebosch, South Africa,[email protected] of sustainable policies in the developing world context re-quires consideration of many divergent criteria related to balancing so-cial, environmental, equitability and economic issues. Furthermore,this occurs typically in a context of major levels of uncertainty, to alarger extent than in the developed world. We critically examine meansby which multicriteria decision analysis and scenario planning can becombined into an integrated approach to address all issues. Illustra-tions of the use of the integrated approach in the development contextwill be discussed.

3 - The Antecedents and the Benefits of Partnership inPEF (Private Equity Fund) of South Korean Indus-tries: The Sustainability PerspectiveHansuk Lee, Sangmyung Univeristy, Korea, Republic Of,[email protected], Seongtae Hong, In-Young LeeIn recent years, Sustainable Operations Management(SOM) has beenreceiving attention from many researchers. Most of the research onSOM has been limited to conceptual frameworks, case studies andsome empirical papers. Social dimension of sustainability is concernedwith the well being of people and communities. This paper presents astudy aimed at quantifying the antecedents and benefits of partnershipin PEF(Private Equity Fund). With this research we made efforts tostudy the social dimension of sustainability in the corporate social re-sponsibility context.

4 - The effect of organizational communication in in-trapreneurship and business performanceSükran Öktem, Business Administration, Baskent University,Ankara, Turkey, [email protected], Hakan Turgut,Ismail Tokmak, Gözde KubatFor achieving a sustainable competitive advantage in global compe-tition, organizational communication for businesses and intrapreneur-ship has gained importance. To identify the presence of the effect oforganizational communication in both intrapreneurship and businessperformance in Ankara (Turkey), a survey has been carried out in dif-ferent sectors for manufacturing industry companies. Analysis of datafrom the survey with the help of the relations between these variableswill be conducted.

� TD-35Tuesday, 14:30-16:00CC-A41

Robust logistic applications

Stream: OR Applications in IndustryInvited sessionChair: Tobias Winkelkotte, IT, zooplus AG, Sonnenstraße 15,80331, München, Germany, [email protected]

167


1 - Robust Model for Dynamic Multilevel Capacitated Fa-cility Location under uncertaintyVincenzo De Rosa, Chair of Business Administration andLogistics, Universität Erlangen-Nürnberg, Lange Gasse 20,90403, Nürnberg, Germany,[email protected], Marina Gebhard,Jens Wollenweber

A new model formulation for the robust capacitated facility locationproblem for distribution systems under uncertainty is presented. Min-imizing the expectation of the relative regrets across scenarios over amulti-period planning horizon is set as objective. It is achieved bydynamically assigning multi-level production allocations, facility lo-cations and capacity adjustments for uncertain parameter developmentover time. The robustness and effectiveness of the MIP-model is shownby comparison to a deterministic planning approach in an exemplarysetting where the future is assumed certain.

2 - Scheduling Long Distance Transports in a Postal Dis-tribution NetworkTobias Winkelkotte, IT, zooplus AG, Sonnenstraße 15, 80331,München, Germany, [email protected], LiSun

Postal distribution consists of the sorting of big amounts of items andthe transporting over long distances. Transports have to be scheduledsuch that the sortings can be completed in time. The question is, whento send transports from one sorting facility to another, while achievinga certain service level. We present an appropriate mixed-integer linearprogramming model to schedule transports such that a service level isachieved and the costs are minimal. We present a heuristic approachwhich searches a wide range of the solution space by systematicallyvarying the desired service level.

3 - New approach to The Single Track Railway Schedul-ing ProblemMaya Laskova, The Institute of Control Sciences V. A.Trapeznikov Academy of Sciences, Moscow, Russia, RussianFederation, [email protected], AlexanderLazarev, Elena Musatova

We consider the single track railway scheduling problem with two sta-tions. Two subsets of trains are given, where trains from station one goto station two, and trains from station two go to station one. All trainshave the same speed. At most one train from station one or two can beon the track at any time. We consider the single track railway schedul-ing problem with two stations with optimization criteria of minimizingadditional completion time for each trains. We propose a heuristic ap-proach to The Single Track Railway Scheduling Problem.

� TD-36Tuesday, 14:30-16:00CC-A43

QMPD Session2Stream: Quantitative Models for Performance and De-pendabilityInvited sessionChair: Markus Siegle, Computer Science, Universitaet derBundeswehr Muenchen, Universitaet der Bundeswehr Muenchen,85577, Neubiberg, Germany, [email protected]

1 - Queuing model of hop-by-hop overload control in SIPsignaling networkYuliya Gaidamaka, Telecommunication Systems Department,Peoples’ Friendship University of Russia, 115419, Moscow,Russian Federation, [email protected], PavelAbaev, Konstantin Samouylov, Alexandr Pechinkin, SergeyShorgin

Major standards organizations, ITU, ETSI, and 3GPP have all adoptedSIP as a basic signalling protocol for NGN. The current SIP overloadcontrol mechanism is unable to prevent congestion collapse and mayspread the overload condition throughout the network. The purpose

of the paper is to develop a hysteretic model for hop-by-hop overloadcontrol in SIP server networks. A two-class priority queuing systemis considered. The model has a buffer with three thresholds and hys-teretic load control. We obtained the algorithm for calculating the mainperformance measures of the system.

2 - Dependability assessment of large railway systemsGrégory Buchheit, A3SI, Arts et Métiers Paristech, 4, rueAugustin Fresnel, 57078, Metz, France,[email protected], Olaf Malassé, Nicolae Brînzei

The performance evaluation of systems and in particular their avail-ability depends on the dysfunctional behaviors and maintenance poli-cies of its subsystems and components. Our approach is based on au-tomatic generation of large scale Petri Nets, which are colored, timedand hierarchical, from a meta-model leaning on atomic and elementarycomponents of both French and European railway signalling systemswhose behavior respects the specifications that are validated by a for-mal method. Metrics are defined for performance analysis from thedesign, operation and maintenance point of view.

3 - Colored Petri Nets based modeling and simulation ofbandwidth allocationJulija Asmuss, Institute of Telecommunications, RigaTechnical University, Azenes str. 12, Riga, Latvia,[email protected], Gunars Lauks, Viktors Zagorskis

We present the design and simulation of dynamically adaptive band-width allocation protocols in a changing network environment usingColored Petri Nets. The proposed approach as an example of DaVinciarchitecture allows describing how a single physical network can sup-port multiple traffic classes with different performance objectives bymeans of multiple virtual networks constructed over the physical one.We consider the problem of bandwidth allocation as a maximizationproblem for the aggregate performance and analyze simulation resultsfor delay sensitive and throughput sensitive traffics.

4 - SFERA: A Simulation Framework for the Perfor-mance Evaluation of Restart Algorithms in service-oriented systemsAlexandra Danilkina, Institute for Computer Science, FreieUniversität Berlin, Takustr. 9, 14195, Berlin, Germany,[email protected], Philipp Reinecke, KatinkaWolter

Restart is a well-known client-side technique for improving perfor-mance and service availability. We illustrate SFERA framework fea-tures with an evaluation of the optimal restart timeout in a complexSOA system. We simulate different configurations using workflowsand model component response-times by phase-type distributions fit-ted to measurements from a SOA tested. Several service-policies in-cluding ageing servers and infinite servers are considered as well asdifferent arrival processes and restart algorithms. We compare com-pletion times and service availability for different scenarios.

� TD-37Tuesday, 14:30-16:00CC-Act

OR in Health & Life Sciences 4Stream: OR in Health & Life Sciences (contributed)Contributed sessionChair: Fabricio Sperandio, INESC-TEC, Faculdade deEngenharia, Universidade do Porto, Rua Dr. Roberto Frias, s/n,4200-465, Porto, Portugal, [email protected]

1 - Optimal timing of joint replacement using mathemat-ical programming and stochastic programming mod-elsBaruch Keren, Industrial Engineering and ManagementDepartment, SCE - Shamoon College of Engineering,Bialik/Basel Sts., P.O.B. 950, 84100, Beer Sheva, Israel,[email protected], Joseph Pliskin, Yossi Hadad

168


The optimal timing for radical medical procedures as joint replace-ment must be seriously considered. We show that under deterministicassumptions the optimal timing for joint replacement is a solution ofa mathematical programming problem, and under stochastic assump-tions the optimal timing can be formulated as a stochastic program-ming problem. For the case where the patient’s remaining life is nor-mally distributed along with a normally distributed survival of the newjoint, the expected benefit function from surgery is completely solved.This enables practitioners to find the optimal timing.

2 - Probabilistic Risk Assessment methods applicationfor medical device reliabilityRobertas Alzbutas, Lithuanian Energy Institute, KaunasUniversity of Technology, LT-44403, Kaunas, Lithuania,[email protected], Darius Naujokaitis, Vytautas Janilionis,Giedr Alzbutien

In this paper the main techniques and methods of Probabilistic RiskAssessment as well as their application for risk and reliability analysisof medical devices are introduced. The focus is set on the Fault TreeAnalysis (FTA) technique and related quantification methods based onMinimal Cut-Sets (MCS) and Binary Decision Diagram (BDD). Also,an example of probabilistic assessments of simple system failure andcomparison of produced outcomes is presented. The main advantagesand disadvantages of MCS and BDD as well as FTA application forRisk Informed Decision Making are described too.

3 - Integrating probability distribution elicitation withMarkov decision modeling: a case study in the treat-ment of patients with heart failureQi Cao, Department of Epidemiology, University MedicalCentre Groningen, P.O. Box 30001, 9700 RB, Groningen,Netherlands, [email protected], Douwe Postmus, Hans Hillege,Erik Buskens

Early-stage economic evaluations of novel medical equipment arecharacterized by evidence scarcity as data from clinical research is stillmissing. These evaluations must therefore be based on evidence fromless direct sources, such as early bench or animal testing and expertjudgement. In this study, we discuss how the resulting uncertainty inthe model parameters can be incorporated in a probabilistic way byintegrating probability distribution elicitation with Markov decisionmodeling. The method is illustrated in a case study in the treatmentof patients with heart failure.

4 - Robust surgery scheduling via simulation optimiza-tionFabricio Sperandio, INESC-TEC, Faculdade de Engenharia,Universidade do Porto, Rua Dr. Roberto Frias, s/n, 4200-465,Porto, Portugal, [email protected], BernardoAlmada-Lobo, José Borges

Surgery scheduling problems are characterized by strong uncertainty,compromising the results of deterministic optimization methods. Thiswork presents a Simulation Optimization approach to the surgeryscheduling problem at the operational level. The system is composedby a multi-objective optimization search procedure and a stochasticsimulation model, which is used to estimate the performance of alter-native schedules under uncertainty (variable surgery duration, cancel-lations, and emergencies). The approach aims to devise robust sched-ules, better fitting the stochastic nature of the problem.

� TD-38Tuesday, 14:30-16:00HH-Colombus

Multiobjective Optimization for DecisionSupport

Stream: Multiobjective OptimizationInvited sessionChair: Heinz Roland Weistroffer, Information SystemsDepartment, Virginia Commonwealth University, Snead Hall, 301West Main Street, 23284-4000, Richmond, VA, United States,[email protected]

1 - IND-NIMBUS Framework for Interactive Multiobjec-tive OptimizationVesa Ojalehto, Department of Mathematical InformationTechnology, University of Jyväskylä, P.O. Box 35 (Agora),FI-40014 University of Jyväskylä, Finland, Finland,[email protected], Jussi Hakanen, Kaisa Miettinen

We present a framework for interactive multiobjective optimization(IMO) called IND-NIMBUS. IND-NIMBUS is a stand-alone systemfor solving real-world nonlinear multiobjective optimization problemsand contains implementations of several IMO methods. Developmentprinciples of IND-NIMBUS have been interaction with the decisionmaker, computational efficiency, ability to change methods during thesolution process and utilization of the same modules with differentmethods e.g. for visualizing and storing solutions. Applications in-clude radiotherapy treatment planning and chemical engineering.

2 - State of Multiobjective Optimization SoftwareHeinz Roland Weistroffer, Information Systems Department,Virginia Commonwealth University, Snead Hall, 301 WestMain Street, 23284-4000, Richmond, VA, United States,[email protected], Yan Li

We present an overview and a comprehensive survey of current multi-objective optimization (MOO) software. The purpose of this work istwo-fold: For the academic researcher we provide an annotated com-pendium of MOO implementations to show what has been done andwhere there are gaps. For the practitioner interested in solving specificMODM problems, we provide a selection guideline to find the mostsuitable software to meet his or her needs.

3 - On Developing User Interface for Interactive Multiob-jective OptimizationJussi Hakanen, Dept. of Mathematical InformationTechnology, University of Jyvaskyla, P.O. Box 35 (Agora),FI-40014, University of Jyvaskyla, Finland,[email protected], Kaisa Miettinen, Suvi Tarkkanen,Hannakaisa Isomaki

By using interactive multiobjective optimization (IMO), a decisionmaker (DM) progressively provides preference information in orderto find the most satisfactory compromise between the conflicting ob-jectives. In this paper, we consider possibilities the research field ofvisual analytics can provide for supporting the decision making pro-cess and information exchange between IMO methods and a DM. Theimplementation of an IMO method Pareto Navigator is used as an ex-ample to demonstrate concrete challenges of interaction design and wedescribe the incremental development of the user interface.

4 - Laplacian Regular Multiple Criteria Linear Program-ming for Semi-supervised ClassificationZhiquan Qi, Graduate University of Chinese Academy ofSciences, 200062, Beijing, China, [email protected]

In this paper, inspired by the success of RMCLP, we proposed a novelLaplacian RMCLP (called Lap-RMCLP) method for semi-supervisedclassification problem. Unlike other methods based on graph, Lap-RMCLP is a natural out-of-sample extension, which can classify datathat becomes available after the training process, without having to re-train the classifier or resort to various heuristics. Furthermore, by ad-justing the parameter, Lap-RMCLP can convert to RMCLP naturally.All experiments on synthetic and real data sets show that Lap-RMCLPis a competitive method in semi-supervised classification.

� TD-39Tuesday, 14:30-16:00HH-Cousteau

MCDA software for real-world MCDAproblems

Stream: Innovative Software Tools for MCDAInvited sessionChair: Brian Reddy, ScHARR, University of Sheffield, 30 RegentCourt, S1 4DA, Sheffield, United Kingdom,[email protected]

169


1 - a MCDA tool for evaluating the overall comfort on-board the French high speed trains (TGV)Mohamed, Lounes Mammeri, Innovation and reaserch, SNCF,7 square des tilleuls, 92350, le plessis robinson, France,[email protected]

Comfort being one of the main reasons why passengers like travellingby train, it is important for railway operators to be able to estimate theoverall comfort for various types of trains. We describe the first stepsof a study aiming at constructing a tool based on a multiple criteria de-cision aiding approach. We consider the comfort as a judgment basedon a set of complex elements by taking into account the passengers’point of view. We have proposed a hierarchical model of the comfortcomponents and defined the associated scales and some interactionsbetween components.

2 - Eliciting ElectreTri category limits for a group of de-cision makersVincent Mousseau, LGI, Ecole Centrale Paris, Grande voiedes vignes, 92 295, Chatenay Malabry, France,[email protected], Olivier Cailloux, Patrick Meyer

Multicriteria sorting assigns alternatives to predefined ordered cate-gories. We consider the Electre Tri method, which involves 3 types ofparameters: 1) category limits defining the frontiers between consecu-tive categories, 2) weights and majority level specifying which coali-tions form a majority, 3) veto thresholds characterizing discordanceeffects. We propose an elicitation procedure to infer category limitsfrom assignment examples provided by multiple DMs. The procedurecomputes a set of category limits common to all decision makers, withvariable weights and vetoes for each DM.

3 - Investigating mcda approaches for the national insti-tute for health and clinical excellenceBrian Reddy, ScHARR, University of Sheffield, 30 RegentCourt, S1 4DA, Sheffield, United Kingdom,[email protected]

NICE uses evidence-based, health economic approaches to prioritiseinterventions, reduce variation in levels of treatment between regionsand encourage best practice across the NHS. For various reasons itis challenging to quantify, model and describe the multiple effects ofpublic health interventions. As a result there is uncertainty aroundcost effectiveness, complicating decision processes. This presentationwill describe some common problems in this area, explain why MCDAtechniques may help and explain where these approaches are currentlybeing further investigated for future use.

� TD-40Tuesday, 14:30-16:00HH-Livingstone


Stream: Preference LearningInvited sessionChair: Krzysztof Dembczynski, Institute of Computing Science,Poznan University of Technology, Piotrowo 2, 60-965, Poznan,Poland, [email protected]

1 - Improving ranking performance with cost-sensitiveordinal classification via regressionHsuan-Tien Lin, Computer Science and InformationEngineering, National Taiwan University, Taiwan,[email protected], Yu-Xun Ruan, Ming-Feng Tsai

This paper proposes a novel ranking approach, cost-sensitive ordinalclassification via regression (COCR). The approach applies a theoreti-cally sound method for reducing an ordinal classification task to binaryones and solves them with regression. In addition, COCR allows spec-ifying mis-ranking costs to improve the ranking performance. The op-portunity is exploited by deriving a corresponding cost for the popularERR criterion. Evaluations on several large-scale data sets verify thesignificant superiority of the ERR-tuned COCR over commonly usedsimple regression approaches.

2 - A new rule-based method for label rankingMassimo Gurrieri, University of Mons, Rue du Houdain 9,7000, Mons, Belgium, [email protected],Xavier Siebert, Philippe FortempsThis work focuses on an application of preference ranking, where theproblem is to learn a mapping from instances to rankings over a fi-nite set of labels. Our approach is based on a learning reduction tech-nique and provides such a mapping in the form of logical rules: if[antecedent] then [consequent], where [antecedent] contains a set ofconditions (connected by a logical conjunction operator (AND)) while[consequent] consists in a ranking among labels. This approach com-prises five phases: preprocessing, rules generation, post-processing,classification and ranking generation.

3 - Learning preference relations with Kronecker ker-nels: Some theoretical and algorithmic resultsTapio Pahikkala, Turku Centre for Computer Science, 20014,Turku, Finland, [email protected] this talk, we consider a framework for learning various types of pref-erence relations that is based on Kronecker product kernels and theirmodifications. As case studies, we consider tasks of inferring rank-ings of objects and learning to predict nonlinear preferences, as wellas extensions to more complex preference learning problems. Next,we present theorems about the universal approximation properties ofthe considered kernel functions. Finally, we present computationallyefficient learning algorithms for the considered problems and practicalresults on several application domains.

4 - Efficient information retrieval in bioinformatics withconditional ranking algorithmsWillem Waegeman, Department of Applied Mathematics,Biometrics and Process Control, Ghent University, Coupurelinks 653, 9000, Ghent, Belgium,[email protected] domains like multimedia retrieval, bioinformatics, social networkanalysis and game playing, one can find retrieval tasks where the goalconsists of inferring a ranking of objects, conditioned on a target ob-ject query. We recently presented a general machine learning frame-work for learning conditional rankings from various types of relationaland structured data. In this talk we focus on three specific bioinfor-matics applications: retrieval of enzymes w.r.t. functionality, rankingof ligands for computational drug design and modelling interactionsbetween bacterial species.

� TD-41Tuesday, 14:30-16:00HH-Heyerdahl

Dynamical Systems and MathematicalModelling in OR

Stream: Dynamical Systems and Mathematical Mod-elling in ORInvited sessionChair: Selma Belen, Mathematics and Computer, CAG University,Adana-Mersin Karayolu Uzeri, Yenice-Tarsus, 33800, TARSUS /Mersin, Turkey, [email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - Fractional partial differential equations driven byfractional Gaussian noiseMahmoud El-Borai, Mathematics, Faculty of scienceAlexandria University, Dep. of Mathematics Faculty ofscience, Alexandria University Alexandria Egypt, 21511,Alexandria Egypt, Egypt, [email protected], KhairiaEl-Said El-NadiSome fraction parabolic partial differential equations driven by frac-tion Gaussian noise are considered. Initial-value problems for theseequations are studied. Some properties of the solutions are given un-der suitable conditions and with Hurst parameter less than half.

170


2 - A discrete system simulating an e-shop built bottomup from customer behaviorPeter Vojtas, Software Engineering, Charles University,Faculty of Mathematics and Physics, Malostranske nam. 25,118 00, Prague, Czech Republic, [email protected],Ladislav Peska

Our motivation is modeling a web based e-shop from manager’s per-spective. One of problems of such modeling is to find analytical de-scription. We combine database preference querying and our model oflearning user preferences for explicit/implicit users’ behavior. We pro-pose a simulation system with stochastic generated artificial users andmanager’s decision. In our experiments, distribution of user behaviorwas collected in a real travel agency and system response containeditems based partly on advertisement, partly on popularity and partly ontop-k object of each user in person.

3 - Taxonomy of concepts and terms in administrationErwin Reizes, O.R., Fac.Ing./UdelaR,Uruguay, Luis A.deHerrera 1042,ap.1701 (home), ap.1701, 11300, Montevideo,Uruguay, [email protected]

A taxonomy of concepts and terms seems necessary to cope with theBabel of synonyms and homonyms which plagues present System andAdministration Theory and Practice. Therefore, based on these ideas,on OR and on ideas of Linguistic, Logic and Information, a bijectivemapping between concepts and terms is given.

4 - A Hybrid Modelling Approach for Physical NetworkDesign under Demand UncertaintySutthipong Meeyai, Cranfield University, United Kingdom,[email protected]

We consider a modelling approach for physical network design whenit is managed in a decentralised way. An optimization approach gen-erally determines an optimum or near-optimum solution from a ’top-down perspective.’ Instead of designing only from the standard ap-proach, we examine the results from a ’bottom-up perspective’ by anagent-based simulation. With the agent properties we can investigateoperations by a local objective with a set of degree of freedom. We pro-vide a modelling approach for network design with a numerical resultfrom a real-world example.

� TD-42Tuesday, 14:30-16:00BW-Amber

Advances in Stochastic Modeling andSimulationStream: Stochastic Modeling and Simulation in Engi-neering, Management and ScienceInvited sessionChair: Basak Tanyeri, Bilkent University, Turkey,[email protected]

1 - A Comparison of Artificial Neural Network and Multi-nomial Logit Models in Predicting MergersNilgun Fescioglu-Unver, Industrial Engineering, TOBBUniversity of Economics and Technology, Sogutozu Cad.No:43, Sogutozu, 06560, Ankara, Turkey, [email protected],Basak Tanyeri

Predicting who will propose merger bids and who will receive mergerbids is important to measure the price impact of mergers. This studyinvestigates the performance of artificial neural networks and multino-mial logit models in predicting bidder and target candidacy. We usea comprehensive dataset that covers the years 1979 to 2004 and in-cludes all deals with publicly listed bidders and targets. We find thatboth models perform similarly while predicting target and non-mergerfirms. The multinomial logit model performs slightly better in predict-ing bidder firms.

2 - A survey analysis on the investment attitudes of in-dividual investorsBelma Ozturkkal, International Trade and Finance, Kadir HasUniversity, Cibali, Istanbul, Turkey,[email protected]

This study aims to analyze determinants of trading behavior of localinvestors, where a survey of 55 questions on 85 people is used. Thesurvey is composed of four parts: demographic properties, perceivedcharacters, investment preferences, portfolio diversification. The find-ings show that investors are subject to home bias and the surveyedinvestors prefer to invest in local equity market.

3 - Stochastic interventions: Methods to model uncer-taintyShweta Agarwal, Management Science, London School ofEconomics, New Academic Fields, Lincoln’s Inn Fields,WC2A 2AE, London, Great Britain, United Kingdom,[email protected], Gilberto Montibeller

Managers frequently associate controlling risk with managing the un-certainty that generates them. One way to represent this in decisionmodels is by not treating uncertainties as exogenous but by allowingthe probability distributions of state variables to be altered by stochas-tic interventions. We define such an intervention using principles ofcausal reasoning and propose modelling it with generalized condition-ing. The ability to model the effects of probability alteration can en-hance the prescriptive scope of decision analytic tools in risk manage-ment.

4 - Records Properties of Nonstationary Time SeriesAna Elizabeth García Sipols, Estadística e investigaciónoperativa, Universidad Rey Juan Carlos, c/ Tulipan s/n, 28933,Madrid, Spain, [email protected], Clara Simon deBlas, M. Teresa Santos Martin

This paper compares the statistical properties of the records from inde-pendent and identically distributed time series with those of time seriescontaining a single unit root. It is shown that there are important dif-ferences in both the limiting distributions and the convergence rates ofthe associated record counting processes. We prove some propertiesfor the test statistic in the context of the renewal theory and we suggesttwo new candidates to test the hypothesis of random walk with positiveand negative drift.

� TD-43Tuesday, 14:30-16:00BW-Granite

Stochastic and robust optimization

Stream: Stochastic ProgrammingInvited sessionChair: Abdel Lisser, LRI, Universite de Paris Sud, Bat. 650,91405, Orsay, France, [email protected]

1 - Solving SCOPF problems by a new structure-exploiting methodNaiyuan Chiang, School of Mathematics School ofMathematics School of Mathematics School of Mathematics,University of Edinburgh, Room 5620, JCMB, The King’sBuildings, EH9 3JZ, Edinburgh, State, United Kingdom,[email protected]

The aim of this paper is to demonstrate some new approaches to solvethe linearized security constrained optimal power flow (SCOPF) prob-lem by OOPS, which is a modern structure-exploiting primal-dualinterior-point implementation. Firstly, we present a reformulation ofthe SCOPF model, in which most matrices only need to be factorizedonce throughout the IPM iterations. Moreover, we use a precondi-tioned iterative method to solve the corresponding linear system andsuggest several schemes to pick a good and robust preconditioner basedon combining different "active’ contingency scenarios.

171


2 - A Second-Order Cone Programming approach forLinear programs with joint probabilistic constraintsAbdel Lisser, LRI, Universite de Paris Sud, Bat. 650, 91405,Orsay, France, [email protected]

In this this talk, we consider a special case of Linear programmingproblems with joint probabilistic constraints (LPPC). We assume thatthe constraint matrix is random with normally distributed coefficientsand independent rows. To solve this hard problem, we use the piece-wise linear approximation and the piecewise tangent approximation tocome up with upper and lower bounds respectively. For this purpose,we solve two different second-order cone programming (SOCP) prob-lems. Numerical results on randomly generated data are given.

3 - Robust portfolio optimization - impact of copula andestimation choiceJustyna Majewska, Department of Demography and BusinessStatistics, University of Economics in Katowice, 1 Maja 50,40-287, Katowice, Katowice, Poland,[email protected], Grazyna Trzpiot

The paper deals with estimating risk-minimizing portfolios when de-partures from standard assumption of Gaussianity are present. Its aimis to investigate the impact of different copula on optimal portfolios,when different robust alternatives to the classical variance portfoliooptimization problem are used. Specifically, we examine some aspectsof the relationship of dependence structure and robust estimation. Wepresent applications using a data set based on a global portfolio of anemerging market investors located in Central and Eastern Europe.

4 - On reduction of quantile optimization problems withdiscrete distributions to mixed integer programmingproblemsVladimir Norkin, Operations Research, Institute ofCybernetics, Glushkov avenue, 40, 03187, Kiev, Ukraine,[email protected], Andrey Kibzun, Andrey Naumov

In the report we suggest a method of equivalent reduction of a gen-eral quantile optimization problem with discrete distribution of ran-dom data to large scale mixed integer programming problems. Thelatter are supposed to be solved by standard discrete optimizationsoftware. As a particular case a two-stage quantile stochastic op-timization problem is considered. Our main contribution consistsin establishing new conditions of equivalence between two formula-tions. Some generalizations can be found in http://www.optimization-online.org/DB_HTML/2010/05/2619.html

� TD-44Tuesday, 14:30-16:00BW-Marble

Model Generation and Model SelectionStream: Machine Learning and its ApplicationsInvited sessionChair: Michael Khachay, Ural Branch of RAS, Institute ofMathematics and Mechanics, S.Kovalevskoy, 16, 620990,Ekaterinburg, Russian Federation, [email protected]: Vadim Strijov, Russian Academy of Sciences, ComputingCenter, Vavilova 42-268, 119333, Moscow, Russia, RussianFederation, [email protected]

1 - Search for Optimal Composite IndicatorsMiroslav Klucik, Faculty of Economics, VSB-TechnicalUniversity of Ostrava, Havlickovo nabrezi 38, 701 21,Ostrava, Czech Republic, [email protected]

Contemporary approaches to construct composite leading indicatorsusing economic theory and statistical analysis rely mostly on subjec-tive decisions of researchers and analysts. An artificial intelligence tool— genetic programming — can mine all the information available inlarge data sets and help to construct models best fitted to the referencedata. The composition of indicators and the lead of time series are be-ing searched for by symbolic regression. The structure of associatedinformation between reference series and large data sets is shown.

2 - Nonlinear regression model generation and graphtransformationsRoman Sologub, Innovations and High Technology, MoscowInstitute of Physics and Technology, 508, 86, Altufievskoe sh.,Moscow, 127349, Moscow, [email protected]

We investigate a problem of non-linear regression model generation.The practical application is to define regression of strike and maturityof a European option towards its volatility. To derive the volatilitysurface we use symbolic regression. The competitive models are gen-erated as superpositions of expert-given functions. The main problemof the model generation process is to filter the isomorphic models withdifferent structure. The methods of graph transformation are used assolution. The result is a set of models of unique structures. The optimalmodel is selected from this set.

3 - Multiclass classification of cardio-vascular diseasepatients with sample size estimationAnastasia Motrenko, Applied mathematics and management,MIPT, Moscow, Russian Federation,[email protected]

We discuss an algorithm that classifies four groups of patients, dividedby their health condition. Concentrations of proteins in blood cellsare used as features. Our first objective is to select a set of featuresthat will classify the patients making minimum amount of errors. Thisselection is implemented by means of exhaustive search. Two classifi-cation strategies are investigated, "one versus all" and "all versus all".The second objective is the sample size. Amount of data is small, so weevaluate minimum sample size, necessary for statistical significance ofclassification.

4 - Mixture models in the financial time series forecast-ingVadim Strijov, Russian Academy of Sciences, ComputingCenter, Vavilova 42-268, 119333, Moscow, Russia, RussianFederation, [email protected]

To forecast financial time series one needs a set of models of optimalstructure and complexity. The mixture model selection procedures arebased on the coherent Bayesian inference. To estimate the model pa-rameters and covariance matrix, Laplace approximations methods areintroduced. Using the covariance matrix one could split up the data setto form mixture of models and select a model with minimum descrip-tion length.

� TD-45Tuesday, 14:30-16:00BW-Water

Quantitative Approaches in Managerialand Financial Accounting

Stream: Quantitative Approaches in Managerial andFinancial AccountingInvited sessionChair: Matthias Amen, Chair for Quantitative Accounting &Financial Reporting, University of Bielefeld, Universitaetsstrasse25, 33615, Bielefeld, Germany, [email protected]

1 - Optimizing continuous inventoryMatthias Amen, Chair for Quantitative Accounting &Financial Reporting, University of Bielefeld,Universitaetsstrasse 25, 33615, Bielefeld, Germany,[email protected]

Physical inventory exists explicitly or implicitly (IFRS, ISA 501) inall accounting systems. The German Commercial Code (Handelsge-setzbuch, HGB) requires an annual physical inventory for each item ofthe balance sheet. Traditional physical inventory requires a stop of op-erations of the production plant for a certain period around the balancesheet date. To ensure continuous production continuous inventory isalso allowed. We present an optimization approach to determine theschedule of stocktaking in a continuous inventory system during theaccounting period.

172

EURO 25 - Vilnius 2012 TE-01

2 - Meta Managerial Accounting — quantitative ap-proaches for designing managerial accounting sys-temsMarkus Puetz, Chair, Department of Managerial Accountingand Control, WHL Graduate School of Business andEconomics, Hohbergweg 15 - 17, D-77933, Lahr,Baden-Wuerttemberg, Germany, [email protected]

Meta Managerial Accounting (MMA) is used to design effective man-agerial accounting systems for practical applications. In this presenta-tion 3 MMA approaches will be discussed: 1. Analytical and graphtheoretical approach for representation of business and control pro-cesses with open decision networks. 2. First order predicate logicapproach for specification of business and control processes and po-tentials, including handling of verification and validation tasks whichrefer to control methods. 3. Approach for implementation of simula-tion based open decision networks for MMA purposes.

3 - Corporate taxes, capital structure and valuation:Combining Modigliani/ Miller and Miles/ EzzellUlrich Schäfer, Professur für Finanzen und Controlling,Georg-August-Universität Göttingen, Platz der GöttingerSieben 3, 37073, Göttingen, Germany,[email protected], Stefan Dierkes

Valuing a company with Discounted Cash Flow (DCF) approachesrequires assumptions about the financing strategy. In general, purestrategies are considered thus far: Either, according to Modigliani andMiller, debt levels or, following Miles and Ezzell, capital structure tar-gets are pre-determined for all periods. In this paper, it is shown howto value a company for an arbitrary combination of these strategies.The familiar approaches are embedded into a common framework. Forpractical application, a combined financing allows for a more realisticmodeling of a company’s financing strategy.

4 - Electronic transfer of financial data to banks - rea-sons for refusal and possible improvementsKarina Sopp, University of Vienna, Bruenner Straße 72, 1210,Vienna, Austria, [email protected]

Electronic transfer of financial data based on a reporting language likeXML or XBRL gets more and more common in multiple areas. How-ever in the communication between banks and their business customersthe distribution of financial information in alternative formats still re-mains dominant and therefore shows potential for further development.To detect the reasons for the lack of acceptance a survey was carriedout among approximately 2000 Austrian accountants. The results ofthe survey show next to the reasons for the low acceptance the condi-tions which have to be met for a higher acceptance.

Tuesday, 16:30-17:30

� TE-01Tuesday, 16:30-17:30Opera

Celebration of EURO and its PresidentsStream: Keynote, Tutorial and Plenary LecturesPlenary sessionChair: M. Grazia Speranza, President of EURO, EURO Office,ULB CP210/01, 1050, Bruxelles, Belgium,[email protected]: Gerhard Wäscher, President-Elect of EURO, EURO Office,ULB CP210/01, 1050, Bruxelles, Belgium, [email protected]

1 - 40 Years of EURO: History, Applications, Future Po-tentialsHans-Jürgen Zimmermann, Operations Research, AachenInstitute of Technology, Korneliusstr.5, 52076, Aachen,Germany, [email protected] years ago, i.e. in 1972, the world looked quite different from to-day: Logistics were slower and more complicated: The best way totravel from Europe to the USA was by ship, which took between 5and 13 days. Alternatively one could take a "Super Constellation" andgo via Island in 24 hours. There were no electronic booking systems,customs-and passport control at each European Border, but no securitysystems at airports, no terrorism and no hijacking of planes. Commu-nication was as clumsy: No fax, no email, no courier services, a letterbetween Europe and oversees took two to three weeks, each singleletter had to be typed ,and there were neither social nets nor a www.Professionally there existed three large OR-journals (and a number ofsmall national ones), East and West was separated by the Iron Curtain."Large" OR-Societies existed in the USA, Japan and Great Britain andin Europe there were a dozen of smaller national OR Societies. Therewas hardly any communication between these European societies andif one wanted to know what was going on in OR in another Euro-pean country, the best way was, to ask a colleague in the USA. In theUSA Ackoff had stated (in 1979) "American Operations Research isdead though not yet buried". Under these circumstances the presidentsof the European OR societies met at the IFORS conference 1972 inDublin and came to the conclusion, that this situation of OR in Europewas sub-optimal and should be improved. They met again in 1973and 1974 and decided to organize the first European OR conferenceat the end of January 1975 in Brussels. There were 500 participantsand the representatives of 10 European countries. In the framework ofthis conference EURO was founded, 7 European working groups werestarted, the EURO-bulletins was started and it was decided to preparethe publication of a European OR journal (EJOR). In the meantimeEURO has 30 member countries, several professional journals, about30 working groups, several very successful types of events, severalprizes and awards, a very impressive web-page, and has had 25 suc-cessful EURO conferences. Hence: EURO has turned out to be a verysuccessful organization. The success of OR can, however, not only bemeasured by the manifold growth of EURO ! As an applied disciplinethe situation of OR in Europe should also be considered from the pointof view of available OR-tools, from its public visibility, the areas ofapplications, the education in this area, and its relationship to otherdisciplines that have emerged in the meantime and are relevant for OR.In some of these dimensions OR can certainly also be considered to bevery successful. In others there are still or again big challenges thathave to be met, if OR is also to be successful in the future. In this pre-sentation some of these challenges will be considered in more detailand some ways suggested, how they can possibly be met.

2 - Celebration of the EURO PresidentsM. Grazia Speranza, President of EURO, EURO Office, ULBCP210/01, 1050, Bruxelles, Belgium,[email protected], Gerhard WäscherEURO is especially pleased and proud to recognise the 17 formerEURO presidents with a special celebration as part of this plenary ses-sion. To celebrate the 25th EURO conference we look back on thedevelopment of the Association and its continued growth and success.This is a perfect time to reflect on our history and thank former presi-dents for their strong guidance and leadership.

173

WA-01 EURO 25 - Vilnius 2012

Wednesday, 8:30-10:00

� WA-01Wednesday, 8:30-10:00RB-Alfa

Keynote Lecture: Professor Boris Polyak

Stream: Keynote, Tutorial and Plenary LecturesKeynote sessionChair: Leonidas Sakalauskas, Operational Research, Institute ofMathematics & Informatics, Akademijos 4, LT-08663, Vilnius,Lithuania, [email protected]

1 - Robust eigenvector problem and its application toPageRankBoris Polyak, Laboratory of adaptive and robust systems,Institute for Control Science, Profsojuznaya 65, 117997,Moscow, Russian Federation, [email protected]

There are novel approaches to Robust Linear Programming and RobustLeast Squares which convert them into nonsmooth convex optimiza-tion problems. We provide similar technique for robust eigenvectorproblem for stochastic matrices. Such robust formulations are naturalfor numerous ranking problems. The approach is compared with stan-dard ones, such as PageRank. Numerical optimization algorithms fornonsmooth problems arising are addressed.

� WA-02Wednesday, 8:30-10:00RB-Beta

New Models for Project Scheduling

Stream: Project Management and SchedulingInvited sessionChair: Christian Artigues, LAAS, CNRS, 7 avenue du ColonelRoche, 31077, Toulouse Cedex 4, France, [email protected]

1 - An integrated solution procedure for project staffingBroos Maenhout, Business Informatics and OperationsManagement, Ghent University, Tweekerkenstraat 2, 9000,Gent, Belgium, [email protected], MarioVanhoucke

We study a strategic scheduling problem that integrates projectscheduling and personnel staffing in order to determine how to staffand schedule single projects as best as possible. A decision is made onthe best mix of resource types (regular workers, overtime and tempo-ral workers) to carry out a single project. In this perspective, we wantto determine the project schedule and deadline and the usage of thedifferent resources in order to minimize the total cost.

2 - Variable resource consumption and dynamic ac-tivity duration in the resource-constrained projectscheduling problemTorben Schramme, DS&OR Lab Paderborn, University ofPaderborn, Warburger Straße 100, 33098, Paderborn,Germany, [email protected], Leena Suhl, Stefan Bunte

The resource-constrained project scheduling problem (RCPSP) hasbeen discussed in literature for decades, although there was only lit-tle focus on handling dynamic resource consumption for activities. Wewill present an extension of this problem considering dynamic resourceallocation for every execution period of an activity which leads to ac-tivities with a fixed workload but variable duration. We will show somerelated practical problems with such requirements, discuss the difficul-ties of this extension for current solving methods and show some ideashow to deal with them.

3 - Integer programming and constraint propagation forscheduling under energy constraintsChristian Artigues, LAAS, CNRS, 7 avenue du ColonelRoche, 31077, Toulouse Cedex 4, France, [email protected],Pierre Lopez, David RivreauWe consider a scheduling problem where activities are subject to timewindows and "energy" resource constraints. The amount required byeach activity on the energy resource at each time has to be decidedsuch that it lies between a minimum and a maximum once the activityis started and before the activity is finished. The total amount of en-ergy required instantaneously cannot exceed a maximal limit and, foreach activity, the total energy brought must fulfill its demand. To solvethis NP hard problem we propose integer programming and constraintpropagation techniques.

4 - A preprocessing procedure to improve recent ex-act algorithms for the resource-constrained projectscheduling problemAlexander Schnell, Dept of Management, University ofVienna, Austria, [email protected], RichardHartlThis talk deals with exact algorithms for the standard version of theresource-constrained project scheduling problem assuming a singlemode, non-preemption and renewable resources. Thereby, we analyseand enhance Branch and Bound algorithms from the literature com-bining principles from Constraint Programming, Boolean SatisfiabilitySolving and Mixed-Integer Programming. In this context, we proposea preprocessing step to two state-of-the-art algorithms to improve theircomputational performance. The results on instances with 60 (120)jobs show the efficiency of our preprocessing procedure.

� WA-03Wednesday, 8:30-10:00RB-L1


Stream: EURO/ROADEF ChallengeInvited sessionChair: H. Murat Afsar, Industrial Systems, University ofTechnology of Troyes, 12 rue Marie Curie, BP 2060, 10010,Troyes, France, [email protected]: Christian Artigues, LAAS, CNRS, 7 avenue du ColonelRoche, 31077, Toulouse Cedex 4, France, [email protected]: Ender Özcan, Computer Science, University ofNottingham, Jubilee Campus, Wollaton Road, NG8 1BB,Nottingham, United Kingdom, [email protected]: Eric Bourreau, COCONUT, LIRMM, 161 Rue Ada, 34000,Montpellier, France, [email protected]: Emmanuel Guere, Google, 38 Avenue de l’Opera, 75002,Paris, France, [email protected]: Kedad-Sidhoum Safia, LIP6 - UPMC, 4 Place Jussieu,75005, Paris, France, [email protected]: Marc Sevaux, Lab-STICC - CNRS UMR 3192, Universitéde Bretagne Sud - UEB, Centre de Recherche - BP 92116, 2 rue deSaint Maude, 56321, Lorient, France, [email protected]

1 - Constraint-based large neighborhood search for ma-chine reassignmentJochen Speck, Institute of Theoretical Informatics, KarlsruheInstitute of Technology, 76131, Karlsruhe, Germany,[email protected], Felix Brandt, Markus VölkerThe presented approach is basically a large neighbourhood search thatiteratively improves the given solution. In each iteration a subset ofprocesses is released, while all other processes stay on their currentmachine. Released processes are then reassigned to the machines bya constraint program and the first improving solution is selected. Weevaluate different neighbourhood selection strategies, branching tech-niques and optimization to improve performance on big instances. Fur-thermore, we present a network flow formulation that allows to com-pute lower bounds on the necessary costs.

174

EURO 25 - Vilnius 2012 WA-05

2 - Constraint-based large neighborhood search ap-proach based on constraint programming for the ma-chine reassignment problemDeepak Mehta, Cork Constraint Computation Center,University College Cork, Ireland, [email protected], BarryOSullivan, Helmut Simonis

We use a large neighborhood search approach based on constraint pro-gramming (CP) to solve the machine reassignment problem. We re-peatedly select a subset of processes and re-optimize the resulting sub-problem by reassigning machines to these processes using CP. The al-gorithm often finds a solution that is significantly better than the initialone. One advantage of our approach is that we can restore the domainsof the variables efficiently when a subset of processes are removedfrom a subset of machines. Therefore, there is no need to create newvariables or domains for a subproblem.

3 - Combining VNS, Simulated Annealing, and a GreedyHeuristic for the ROADEF/EURO 2012 ChallengeFrederic Roupin, LIPN, CNRS-UMR 7030, Université Paris13, 99 avenue, Jean-Baptiste Clément, 93430, Villetaneuse,France, [email protected], LaurentAlfandari, Franck Butelle, Camille Coti, Lucian Finta, GérardPlateau, Antoine Rozenknop

We describe the resolution method of Senior Team S26 for theROADEF/EURO 2012 Challenge. There are two main algorithms thatare used: Variable Neighborhood Search (VNS) and Simulated An-nealing (SA). Both algorithms explore neighborhoods generated viathree types of moves: Swap, Shift and Ejection Chain. Initial solutionused is either the one provided by Google with the data set, or an alter-native initial solution that we produce with a specific greedy heuristic.

4 - A GRASP approach for the machine reassignmentproblemMichaël Gabay, Laboratoire G-SCOP, 46 avenue Félix Viallet,38031, Grenoble, France,[email protected], Sofia Zaourar

We describe our approach to solve the machine reassignment problem.The main idea is to combine different heuristics to find good qualitysolutions. We apply a GRASP metaheuristic scheme: first, a construc-tive heuristic inspired by vector bin packing problem builds varioussolutions. Then we use local search to improve these solutions. Wetake advantage of problem constraints structure to design moves thatkeep the solution feasible and are very fast to apply.


Combinatorial Optimization: Applications

Stream: Combinatorial OptimizationInvited sessionChair: Stephan Visagie, Department of Logistics, University ofStellenbosch, Posbus 3404, Matieland, 7602, Stellenbosch, SouthAfrica, [email protected]

1 - Selecting subsets of foods to measure nutrient expo-sure: the max_r methodHannelie Nel, Department of Logistics, StellenboschUniversity, Private Bag X1, Matieland, 7602, Stellenbosch,Western Province, South Africa, [email protected], MartinKidd

In human nutrition, the max_r method measures specific nutrient ex-posure. This method calculates the optimal subset of foods that max-imizes the Pearson correlation of a given nutrient intake, for exampleiron, based on the chosen subset of foods with the total nutrient intake(iron, for example) based on all the foods. This paper illustrates the useof metaheuristics to solve this combinatorial optimization problem.

2 - A pallet arrangement algorithm for legal loadsNeil Jacobs, Logistics, Stellenbosch University, 7600,Stellenbosch, Western Cape, South Africa, [email protected]

In road transport it is frequently found that a load of pallets in certainarrangements, contravenes legal axle mass and wheel mass restrictionson a given vehicle. The need thus exists to find legal arrangements.It is possible to determine an optimal position for the centroid of anyload for a given vehicle and let that position be the target point in ar-rangement processes. Arrangement for a pallet load centroid to a targetpoint presents a combinatorial problem, which could be very hard tosolve to optimality. A method which finds a feasible area (as opposedto a target point) for the load centroid is presented. An algorithm ispresented which finds legal arrangements for pallet loads.

3 - An algorithm for the bi-objective integer minimumcost flow problemSalima Nait Belkacem, Economy and Commercialisation desHydrocarbures, University of M’hamed Bougara, 1 avenue del’indépendance, 35000, Boumerdès, Algeria,[email protected]

In this paper, we propose an algorithm that finds more efficient integernon-extreme points in the objective space. This algorithm is based onthe well known method. Our alternative method improved the secondphase of Sedeno-Noda and Gonzalez-Martin’s algorithm published on2001. We show that our method gives more efficient points in the ob-jective space and we give a didactic example.

4 - On the optimization of a unidirectional picking line ina DC.Jason Matthews, Department of Logistics, University ofStellenbosch, South Africa, [email protected], StephanVisagie

A real life order pick operation making use of cyclical picking linesused by Pep stores Ltd, South Africa, is considered. Multiple pickerswalk in a clockwise direction around a conveyor belt past bin locationswhile picking orders. The problem of assigning SKUs to bin locationsand the sequencing of the resulting orders for the pickers is discussed.A strong lower bound on the number of cycles needed for a fixed as-signment of SKUs is established and heuristics for assigning SKUs tobin locations presented.


Topics in combinatorial opimization I

Stream: Discrete Optimization, Geometry & Graphs(contributed)Contributed sessionChair: Eric Bourreau, COCONUT, LIRMM, 161 Rue Ada, 34000,Montpellier, France, [email protected]

1 - A column generation model for Eternity 2Eric Bourreau, COCONUT, LIRMM, 161 Rue Ada, 34000,Montpellier, France, [email protected]

Eternity II is a combinatorial problem: a 2D edge matching puzzle.The goal is to place 256 square tiles on a 16x16 square such that alladjacent tiles have matching colors along their common edge. Despite2 million dollars proposed by Tomy, nobody, between 2007 and 2010was able to find a solution. We will show an innovative way to describethe problem: a polyomino packing problem with additional cardinalityconstraints. We propose a column generation model. We validate it,on smaller instances generated with the same parameter as the original(inaccessible!) instance.

2 - Optimization of a Nonlinear Workload BalancingProblemStefan Emet, Dept of Mathematics and Statistics, Universityof Turku, Vesilinnantie 5, FI-20014, Turku, Finland,[email protected]

175


We consider the optimization of a printed circuit board (PCB) assem-bly unit. The production unit consists of a number of lines of differentmachines in series for inserting electronic components on bare PCBs.The objective is to optimize the usage of the lines by optimally allo-cating the jobs to the machines of each line. The balancing problemis modeled and solved using Mixed Integer Nonlinear Programming(MINLP) techniques. It is shown that the presented model can be con-vexified in an equivalent linear form. A set of test problems is success-fully solved.

3 - Efficient Parallel Algorithm for Solving UNSAT 3-SATand Similar Instances Via Static DecompositionEmir Demirovic, Mathematics, Faculty of Natural Sciences,Bosnia And Herzegovina, [email protected], HarisGavranovic

We rework our SAT solver, D-Sat, to work in a parallel environment.As with D-Sat, the solution space is decomposed into simpler pieces ,i.e., the original problem is decomposed into smaller problems, whichare solved in parallel and combined to obtain the solution of the orig-inal problem. Any existing solver may be used to process the newlydecomposed pieces, as long as it outputs sufficient information aboutthe solution process. Our experimental results show that significantlylower execution times can be achieved for UNSAT 3-Sat and similarinstances.

4 - Clique-Forest partitions of P-4-tidy graphsSulamita Klein, COPPE-PESC, Universidade Federal do Riode Janeiro, Caixa Postal 68511, Cidade Universitária,21941-972, Rio de Janeiro, RJ, Brazil, [email protected],Loana Nogueira, Raquel Bravo, Fabio Protti

A graph G is a P4-tidy graph if, for any induced subgraph H of G iso-morphic to a P-4, there exists at most one vertex outside H formingan induced P4 along with three vertices of H. This work considers theproblem of (C,F)-partitions of P4-tidy graphs, that is, the problem ofdetermining if the set of vertices of a P4-tidy graph can be partitionedinto 2 subsets C and F such that C induces a complete graph and Finduces a forest (acyclic graph). We characterize the P4-tidy-(C,F)graphs by forbidden subgraphs and we present a linear-time algorithmto recognize this class.

� WA-06Wednesday, 8:30-10:00RB-Gamma

Variable Neighborhood Search and hybridmetaheuristicsStream: Variable Neighborhood SearchInvited sessionChair: Dauwe Vercamer, Marketing, Ghent University,Tweekerkenstraat 2, 9000, Gent, Belgium,[email protected]

1 - Variable neighborhood search for Unit commitmentproblemRaca Todosijevic, Mathematical Institute of the SerbianAcademy of Sciences and Arts, 11000, Belgrade, Serbia,[email protected], Igor Crevits, Saïd Hanafi, MarkoMladenovic, Nenad Mladenovic

Unit commitment (UC) problem for thermal units consists of findingoptimal electricity production plan in long time period of T hours. Forthe first time Variable neighborhood search based heuristic is suggestedin this paper as a mean for solving it. In each hour, economic dispatch(ED) problem which is the problem of convex programming is solvedby lambda iteration method. Neighborhood structure Nk is defined byopening or closing k thermal units. Extensive computational resultsare performed on well-known test instances from literature.

2 - A VNS for the Real-Life Asymmetric Large-ScaleMTSP with Stochastic CustomersDauwe Vercamer, Marketing, Ghent University,Tweekerkenstraat 2, 9000, Gent, Belgium,[email protected], Dirk Van den Poel, Birger Raa

When working in B2C environments, distribution logistics often dealwith customers who are not at home. If a customer’s service time issubstantial compared to the travel times, it is clear this impacts thenumber of customers one vehicle can serve in its tour. Taking this intoaccount, we developed in collaboration with a home vending companya VNS meta-heuristic that creates least-cost routes for a fleet of identi-cal trucks. Given the impact of real-world constraints on the solution,we used real travel times. Our method solved instances of up to 30.000customers within reasonable time.

3 - Environmental unit commitment problem with CO2taxesMarko Mladenovic, Numerical Mathematics andOptimization, University of Belgrade Faculty of Mathematics,Serbia, [email protected], Igor Crevits, Saïd Hanafi,Nenad Mladenovic, Raca Todosijevic

The unit commitment problem (UC) consists of optimally schedulingelectrical power plants to produce sufficient power each hour with min-imal fuel expenses. Depending on the fuel used by the plants a certainamount of carbon dioxide (CO2) will be emitted to the atmosphere.Ecological taxes that follow this emission vary from country to coun-try and can be more expensive than the fuel itself. In this paper weoffer a new and more realistic model for solving the UC problem in-cluding CO2 environmental taxes. To solve this new model we usedthe VNS methaheuristic.

4 - A system based on ontologies, agents and meta-heuristics applied to the multimedia service of thebrazilian digital television systemToni Wickert, Informatic, Unisinos - Universidade do Vale doRio dos Sinos, Av. Unisinos - 950, São Leopoldo, RS, Brazil,[email protected], Arthur Gomez

With the advent of the Brazilian Digital Television System the userswill be able to have an interactive channel. Thus, will be possibleto access the multimedia application server, i.e., to send or to receiveemails, to access interactive applications, among others. This paperproposes the development and the implementation of an architecturethat includes a module that suggests the content to the user accord-ing to his profile and another module to optimize the content that willbe transmitted. The implementation was developed using ontologies,software agents and metaheuristics.

� WA-07Wednesday, 8:30-10:00RB-Eta

Logistics and Vehicle Routing

Stream: Vehicle Routing and Logistics OptimizationInvited sessionChair: Juan José Salazar González, Estadística e InvestigaciónOperativa, Universidad de La Laguna (Tenerife), Av. AstrofísicoFrancisco Sánchez, s/n, Facultad de Matemáticas, 38271, LaLaguna, Tenerife, Spain, [email protected]: Hipólito Hernández-Pérez, Estadística e InvestigaciónOperativa, Universidad de La Laguna (Tenerife), C/ AstrofísicoFrancisco Sánchez s/n, 38271, La Laguna, Canarias, Spain,[email protected]

1 - One stage approximation algorithm to solve the2LCVRPJavier Faulin, Department of Statistics and OR, PublicUniversity of Navarre, Los Magnolios Builing. First floor,Campus Arrosadia, 31006, Pamplona, Navarra, Spain,[email protected], Angel, A. Juan, Alba Agustín,Oscar Domínguez

We propose a parameter-less algorithm (MultiRound-2LCVRP meta-heuristic) to solve the two-dimensional loading capacitated vehicleproblem (2LCVRP). From the biased randomization of the Clarke andWright saving matrix, the algorithm merges routes that minimize dis-tance costs and ensure loading feasibility. Our remarkable contribu-tions are the next ones: (1) an optimal solution is reached from each

176


round, (2) the algorithm combine implicitly two completely differentheuristics in only one and (3) for the loading part, any consideredheuristic can be hold.

2 - Large Neighborhood Search for solving the Newspa-per Delivery ProblemBelma Turan, Department of Business Administration,University of Vienna, Chair for Production and OperationsManagement, Brünner Straße 72, 1210, Vienna, Austria,[email protected], Karl Doerner, Richard Hartl,Verena Schmid

We consider a real-world problem of delivering daily and weeklynewspapers. The goal is to find a minimal number of vehicles thatserve the stores at minimum travel costs. Fairness regarding the num-ber of stops per driver is obtained by minimizing the number of stopsper route. The objective is represented as a weighted sum of these threeterms. The problem is formulated as a Multi Depot Multi Period Vehi-cle Routing Problem with Time Windows. A method inspired by ideasof Large Neighborhood Search is developed to solve the problem. Wepresent results on data based on benchmark instances.

3 - A multi-commodity Pickup-and-Delivery ProblemHipólito Hernández-Pérez, Estadística e InvestigaciónOperativa, Universidad de La Laguna (Tenerife), C/Astrofísico Francisco Sánchez s/n, 38271, La Laguna,Canarias, Spain, [email protected], Juan José SalazarGonzález

The "multi-commodity Pickup-and-Delivery Traveling SalesmanProblem" is a generalization of the well-known TSP in which citiescorrespond to customers providing or requiring known amounts of mdifferent products, and the vehicle has a given capacity. It is assumedthat a specified product collected from a customer can be supplied toany other customer that requires a quantity of this product. We intro-duce a Mixed Integer Linear Programming model for the m-PDTSP,discuss a decomposition technique and describe some strategies tosolve the problem. Computational results are presented also.

4 - The Robust Vehicle Routing Problem with StochasticDemandRemy Spliet, Econometric Institute, Erasmus UniversityRotterdam, Burgemeester Oudlaan 50, 3000DR, Rotterdam,Netherlands, [email protected], Adriana F. Gabor, RommertDekker

We consider a vehicle routing problem where a schedule has to be de-signed before demand is known. Next, demand is revealed and we areallowed to adjust the original schedule to ensure capacity constraintsare satisfied. However, it is not allowed to differ too much from theoriginal schedule. Difference is measured by counting arcs that areused in the original schedule, but not in the adjusted schedule. We en-countered this problem at Dutch retail chains. Too many adjustments,distorts operations beyond what managers think is acceptable. Wedevelop a heuristic based on the classical Cluster-First Route-Secondprinciple for the (Capacitated) Vehicle Routing Problem.

� WA-08Wednesday, 8:30-10:00RB-Epsilon

Satisfiability and Hypergraphs

Stream: Boolean and Pseudo-Boolean OptimizationInvited sessionChair: Stefan Porschen, Fachgruppe Mathematik, Fachbereich 4,HTW Berlin, Fachbereich 4, Treskowallee 8, 10318, Berlin,Berlin, Germany, [email protected]

1 - Euler walks in uniform hypergraphsZbigniew Lonc, Faculty of Mathematics and InformationSciences, Warsaw University of Technology, pl. Politechniki1, 00-661, Warsaw, Poland, [email protected], PawelNaroski

We generalize the concept of an Euler walk to uniform hypergraphs.The problem of deciding if a k-uniform hypergraph has an Euler walkis NP-complete, for k larger than 2. We prove that for a broad class ofuniform hypergraphs an obvious necessary condition for existence ofan Euler walk is sufficient as well. We also give a linear time algorithmfor constructing an Euler walk and an Euler tour (if they exist) in thiscase. Motivations of studying Euler walks in hypergraphs come fromthe areas of geographic information systems, computer graphics andinteger programming.

2 - Generalized Complexity of SubsumptionArne Meier, Institut für Theoretische Informatik, Fakultät fürElektrotechnik und Informatik, Universität Hannover,Appelstrasse 4, 30167, Hannover, Germany,[email protected]

Description Logics are a well known concept not only in the area ofdatabases but also in the semantic web in terms of the web ontologylanguage OWL 2. We turn towards the subsumption problem whichhas been referred to as the key inference problem by Nardi and Brach-man. Given two formulae A, B and a terminology T, this problem asksif A implies B in every satisfying model for T. We restrict the allowedquantifiers as well as the Boolean functions and classify the compu-tational complexity of these fragments. The Boolean restrictions aredefined through Post’s lattice using the term of clones.

3 - A satisfiability-based approach for generalized tan-glegrams on level graphsAndreas Wotzlaw, Institut für Informatik, Universität zu Köln,Köln, Germany, [email protected], EwaldSpeckenmeyer, Stefan Porschen

A tanglegram is a pair of trees on the same leaf set with matchingleaves in the two trees joined by an edge. They are used in com-putational biology to compare evolutionary histories of species. Wepresent a propositional logic based formulation of two combinatorialproblems: the planar embedding and the crossing minimization prob-lem. We show that our approach can handle cases with more than two,not necessarily binary, trees defined on arbitrary leaf sets. We concludewith an experimental comparison of our technique and several knownheuristics for solving generalized binary tanglegrams.

4 - Satisfiability thresholds beyond k-XORSATAndreas Goerdt, Fakultät für Informatik, TechnischeUniversität Chemnitz, 09107, Chemnitz, Germany,[email protected], Lutz Falke

We consider random systems of constraint equations of length k whichare interpreted as equations modulo 3: We show for k greater 15 thatthe satisfiability threshold of such systems occurs where the 2-corehas density 1: The density is the number of equations divided by thenumber of variables. We show a similar result for random uniquelyextendible constraints over 4 elements. Our results extend previousresults of Dubois/Mandler for equations mod 2 and k = 3 and Conna-macher/ Molloy for uniquely extendible constraints over a domain of4 elements with k = 3 arguments.

� WA-09Wednesday, 8:30-10:00RB-Zeta

Graphs and networks I

Stream: Graphs and NetworksInvited sessionChair: Nicolas Bourgeois, ESSEC Business School, 95000, Cergy,France, [email protected]

1 - On tightness in the analyzis of Branch and Bound al-gorithms.Nicolas Bourgeois, ESSEC Business School, 95000, Cergy,France, [email protected]

The running time of Branch-and-Bound algorithm is classicallybounded by the size of the recursion tree. Still, this analyzis fails tocapture the fact that instances do not meet worse case at every step.Measure and Conquer was a first answer, showing how this upper

177


bound could be improved for several problems and algorithms. Wedisplay some classes of problems and algorithms where we prove that’Measure and Conquer’ analyzis cannot be tight as well, and purpose anew strategy to decrease the bound again. This includes, for example,the most straightforward algorithms for maximum clique.

2 - Moderatly exponential approximation for FeedbackVertex Set.Emeric Tourniaire, Lamsade, Paris Dauphine University,75000, Paris, France, [email protected]

A feedback vertex set of a graph is a set that hits every cycle. Find-ing a minimal vertex set is one of the original Karp’s problems, it isalso APX-complete. On the other hand, some exponential algorithmsfaster than 2n have recently been issued [Fomin et al., 2009], but theupper bound they found in O(1.74n) remains quite intractable for largeinstance. Following the paradigm defined by [Bourgeois, Escoffier andPaschos, 2010], we design a branching algorithm which can solve thisproblem with (1+epsilon) approximation ratio with smaller exponen-tial time.

3 - Fast algorithms for finding specific subgraphs in Bi-ologyMarc Bailly-Bechet, Biology, LBBE - Université Lyon 1,Lyon, France, [email protected]

Modern biology generates huge volumes of data that can be integratedusing a network framework. One frequent question is to isolate inthese graphs structures of interest. Finding appropriate connected sub-graphs turns out to be a prize-collecting Steiner tree problem, com-putationally intractable. We develop a message-passing, probabilisticand distributed formalism, inspired from statistical physics, and applyour algorithm to various datasets combination in the baker’s yeast. Ouralgorithm can deal with large data sets, run in parallel and be adaptedto other problems in systems biology.

4 - Bipartite finite Toeplitz GraphsSara Nicoloso, IASI-CNR, Viale Manzoni 30, 00185, Roma,Italy, [email protected], Ugo Pietropaoli

Let n, a_1, ..., a_k be distinct positive integers: a finite Toeplitz graphis a graph with n vertices, two of which are connected by an edge iffthe absolute value of the difference of their indices is a_1, a_2, ...,or a_k. We characterize the whole family of bipartite finite Toeplitzgraphs with k=3: the proved result completes the characterization oftheir chromatic number, and is based on a simple characterization ofbipartite finite Toeplitz graphs with k=2. In addition, we characterizesome classes of bipartite Toeplitz graphs with k=4.

� WA-10Wednesday, 8:30-10:00RB-Theta

Non-deterministic models 3Stream: Production Management & Supply ChainManagement (contributed)Contributed sessionChair: Ernest Benedito, Organitzacio d’Empreses, UniversitatPolitècnica de Catalunya, C/ Mare de Deu de Nuria, 41-43, 08017Barcelona, Spain, [email protected]: Frédéric Dugardin, LOSI, University of Technology ofTroyes, 12, rue Marie Curie, 10010, Troyes, France,[email protected]

1 - Performance analysis through stochastic OEE simu-lationWerner Schroeder, Economics and Business Management,Montanuniversität Leoben, Franz-Josef-Strasse 18, 8700,Leoben, Styria, Austria,[email protected], Markus Gram

To reduce hidden losses is an important aim of manufacturing com-panies. The OEE is a well known and widely used operating metricto measure losses. Due to the deterministic nature of the calculation

model, the OEE only is reflecting a static image of the process effi-ciency. Since the influential variables can also be of stochastic natureour approach attempts to simulate these stochastic behaviors. It is ex-amined, in particular, whether it makes more sense reducing the OEEvariability than reducing its average value. The validation of the modelwill be confirmed by industrial application.

2 - A Two Stage Solution Procedure of Stochastic Pro-gramming Problem for Production Planning with Ad-vance Demand InformationNobuyuki Ueno, Dept. of Management Information Systems,Prefectural University of Hiroshima, 1-1-71 Ujina-Higashi,Minami-Ku Hiroshima-City, 734-8558, Hiroshima, Japan,[email protected], Koji Okuhara, Takashi HasuikeUnder demand uncertainty, the production planning problem with inadvance demand information is formulated as a nonlinear stochasticprogramming problem. Feasible direction and cutting plane methodshave been proposed, but they need the gradient of the probability func-tion. We propose a two stage solution procedure. First, an approximatesolution is obtained by solving iteratively a linear programming prob-lem. Then, a near-optimal solution is looked for close to the solution,using problem convexity and meta-heuristics. The proposed procedureobtains near-optimal solutions efficiently.

3 - Stochastic bi-level programming of production plan-ning in a reconfigurable aircraft fuselage assemblyYohanes Kristianto, production, university of vaasa,yliopistonranta 10, 65101, vaasa, vaasa, Finland,[email protected] research is motivated by a real life aircraft manufacturing. Whilethe assembly sequence is not flexible due to the size of the aircraftparts and space constraint of the assembly plant. We solve the prob-lem as a two-stage stochastic program where the production routingand scheduling have to be established in advance by anticipating thefuture variations about manufacturing lead times and the demands. Weobserve that the reformulation of the problem into a shortest path prob-lem reduces the iteration time of the solution by implementing a col-umn generation.

4 - Strategic capacity planning in a single-site produc-tion system considering renewal, maintenanece, in-ventory and cash-flow management under uncer-taintyErnest Benedito, Organitzacio d’Empreses, UniversitatPolitècnica de Catalunya, C/ Mare de Deu de Nuria, 41-43,08017 Barcelona, Spain, [email protected] propose a mixed-integer linear program for strategic capacity plan-ning of a single-site production system, under uncertainty. The modeltakes into account the production planning, the inventory and cash-flow management, and tax payments. A computational experience isdescribed and the results are explained. We propose a robust strategy toreduce the resolution time, by reducing the periods in which a capacitydecision can be taken.

� WA-11Wednesday, 8:30-10:00RB-Iota

Scheduling problems with additionalconstraintsStream: Realistic Production SchedulingInvited sessionChair: Roman Capek, Department of Control Engineering, CzechTechnical University in Prague, Technicka 2, 12135, Prague,Czech Republic, [email protected]

1 - New heuristics for the permutation flowshop problemwith flowtime minimizationRuben Ruiz, Departamento de Estadistica e InvestigaciónOperativa Aplicadas y Calidad, Universidad Politecnica deValencia, Camino de Vera S/N, 46021, Valencia, Spain,[email protected], Quan-Ke Pan

178


In recent years a large number of heuristics have been proposed forthe minimization of the flowtime in the permutation flowshop schedul-ing problem. This paper presents a comprehensive review and com-putational evaluation as well as a statistical assessment of 22 existingheuristics has been carried out. Five new heuristics are presented. Thecomparison results identify the best existing methods and show that thefive newly presented heuristics are competitive or better than the bestperforming ones in the literature for the permutation flowshop problemwith the total completion time criterion.

2 - A soft computing based approach to integrated pro-cess planning and scheduling with setup and ma-chine capacity considerationsFiliz Senyüzlüler, Industrial Engineering, Zirve University,Zirve Üniversitesi Mühendislik Fakültesi, Kızılhisar Kampüsü27260 Gaziantep, 27260, Gaziantep, Turkey,[email protected], Adil Baykasoglu, Turkay Dereli

The benefits gained from integrated process planning and schedulingin manufacturing industry motivated researchers to pay great efforton this subject. The present work investigates a new approach whichmakes use of Baykasoglu’s grammatical representation of generic pro-cess plans and clonal selection algorithm in order to integrate processplanning and scheduling. In this approach Giffler&Thompson Algo-rithm and VIKOR method is employed by considering setup times andmachine capacity constraints. Proposed approach is applied to someliterature problems in order to analyze its performance.

3 - Blocking hybrid flow shop robotic cell schedulingproblem with unrelated parallel machines, machineeligibility constraints and multiple part typesSeyda Topaloglu, Industrial Engineering, Dokuz EylulUniversity, 35160, Izmir, Turkey, [email protected],Atabak Elmi

This paper addresses the robotic scheduling problem in blocking hy-brid flow shop cells with unrelated parallel machines, machine eligibil-ity constraints, multiple part types, and a single robot to convey partsbetween stages. Initially, a mixed- integer programming model is pro-posed to minimize the makespan. Due to the complexity of the model,a simulated annealing based solution approach is developed using anew neighborhood structure based on block properties. The computa-tional results demonstrate that the developed SA algorithm is efficientfor this problem.

4 - Resource constrained project scheduling problemwith alternative process plans and total changeovercost minimizationRoman Capek, Department of Control Engineering, CzechTechnical University in Prague, Technicka 2, 12135, Prague,Czech Republic, [email protected]

We study the RCPSP problem with alternative process plans where thegoal is to minimize the total changeover cost given by all performedchangeovers. A special attention is paid to the model that includesdeadlines and covers a possibility to define more alternative ways howto complete the project. A selection of activities present in the scheduleinfluences both decision (finding a feasible schedule) and optimizationproblem. We propose a heuristic based on local search that is per-formed in a shifting time window. The heuristic is able to handle largescale instances in an efficient manner.

� WA-12Wednesday, 8:30-10:00RB-Omicron

Railway Optimization I

Stream: Optimization in Public TransportInvited sessionChair: Gabrio Curzio Caimi, Netzentwicklung, BLS Netz AG,Genfergasse 11, 3001, Bern, Bern, Switzerland,[email protected]: Marco Laumanns, IBM Research - Zurich, 8803,Rueschlikon, Switzerland, [email protected]

1 - A decomposition approach to real-time trainreschedulingLeonardo Lamorgese, SINTEF, Forskningsveien 1, 0314,OSlo, Norway, [email protected]

Train rescheduling consists in the definition of new schedules and pos-sibly new routes for trains when a deviation from the official timetableoccurs. This must be done quickly and the new solution should min-imize (a measure of) the total deviation. In contrast with the classic"holistic" approach, we show how to decompose the problem into astation problem and a line problem. This decomposition allows the useof different models which can take advantage of the features of specificrailway infrastructure. We solve to optimality a number of real-life in-stances from single-track lines in Italy.

2 - Opportunities and challenges with new railway plan-ning approach in SwedenMalin Forsgren, SICS, Box 1263, 16429, KISTA, Sweden,[email protected], Martin Aronsson, Sara Gestrelius, HansDahlberg

Long lead times in railway planning can give rise to a significant dis-crepancy between the original plan and the traffic eventually operated,resulting in inefficient utilization of capacity. Research shows that therailway sector in Sweden would benefit from a different planning ap-proach in which capacity consuming decisions are pushed forward intime whenever possible. This approach is currently being implementedat Trafikverket, the Swedish Transport Administration. With it followsa number of mathematical opportunities and challenges, some of whichwill be presented in this talk.

3 - A Rapid Branching method for the Vehicle RotationPlanning ProblemMarkus Reuther, Optimization, Zuse-Institut Berlin,Takustrasse 7, 14195, Berlin, Germany, [email protected], RalfBorndörfer, Thomas Schlechte, Steffen Weider

The Vehicle Rotation Planning Problem is to schedule rail vehicles,i.e., the rolling stock, in order to cover the trips of a given timetable bya cost optimal set of vehicle rotations. The Problem integrates severalfacets of railway optimization, i.e., vehicle composition, fleet manage-ment, maintenance constraints, and regularity aspects. We propose anRapid Branching algorithm that we are using to solve very large scaleinstances given by our industrial partner DB Fernverkehr AG, which isthe largest intercity railway operator in Germany.

4 - Railway transportation planning optimizationJean Damay, Innovation & Research, SNCF, 40, avenue desterroirs de France, 75611, Paris, France, [email protected],Francis Sourd

Our contribution consists in the optimization of the passenger railwaytransportation supply in a specific area. Given on this area the pos-sible train routings and the passenger traffic forecasts on a regularday, the model provides to the carrier a transportation plan that re-spects structural hard constraints, minimizes the violation of specificsoft constraints, and optimizes the customer conveyance and the in-herent investment costs. The resolution is based on a decompositionof this large problem and embeds OR technics such as mixed integerprogramming, meta-heuristic, or list algorithms.

� WA-13Wednesday, 8:30-10:00RB-Tau

Uncertainty in Inventory Management

Stream: Inventory ManagementInvited sessionChair: Mario Guajardo, Finance and Management Science, NHHNorwegian School of Economics, NHH, Institutt forForetaksøkonomi, Helleveien 30, 5045, Bergen, Norway,[email protected]

179


1 - Forest biomass planning under uncertaintyMikael Rönnqvist, Département de génie mécanique,Université Laval, G1V 0A6, Québec, Canada,[email protected], Patrik Flisberg, Mikael Frisk

The interest in forest biomass has increased dramatic. Compared to tra-ditional forest logistics, an additional transformation, chipping, mustbe planned. The demand at heating plants is uncertain with seasonalvariations. This makes it difficult to coordinate the best use of re-sources for harvesting, chipping, inventory and transportation. Theinventory management to secure supply is critical. Hence there is aneed for OR based decision support dealing with large scale opera-tions under uncertainty. We report on industrial case studies from theSwedish forest industry.

2 - Safety Stock Placement in Multi-Echelon InventorySystems - A Comparison of the Stochastic-Serviceand Guaranteed-Service ApproachesThomas Wensing, INFORM GmbH, Pascalstr. 23, 52076,Aachen, NRW, Germany, [email protected]

When demand is stochastic, one can think of two basic fulfillment-policies for an inventory system. First, the system may fully capturedemand uncertainty and thus provide a constant (guaranteed) servicetime to the following levels. Second, uncertainty may only partially becaptured, meaning that the system’s supply process will be stochastic,too. We analytically compare these two strategies based on a linearmulti-echelon inventory system and indicate practical implications.

3 - Inventory management and pooling of spare parts inan energy companyMario Guajardo, Finance and Management Science, NHHNorwegian School of Economics, NHH, Institutt forForetaksøkonomi, Helleveien 30, 5045, Bergen, Norway,[email protected], Mikael Rönnqvist

We study the inventory of spare parts problem in an energy company,producer of oil and gas. Spare parts are stored in several plants, whichcontrol their inventory per separate based on a min-max system. Weapply methods to decide control parameters for this system under thefill rate constraint. We distinguish unit-size and lot-size demand cases,based on several distribution models. After the control parameters arefound, we use them to analyze the impact of two main sources of sav-ings: risk pooling among the plants and correcting inventory inaccu-racy coming from duplicate item codes.

� WA-14Wednesday, 8:30-10:00RB-Omega

Workforce scheduling II

Stream: Timetabling and RosteringInvited sessionChair: Lakhdar Djeffal, Universit Hadj-lakhdar, 05000, Batna,Algeria, [email protected]

1 - Solving the Integrated Physician and SurgeryScheduling Problem Under Stochastic DemandChristophe Van Huele, Faculty of Economics and BusinessAdministration, Ghent University, Tweekerkenstraat 2, 9000,Gent, Belgium, [email protected], MarioVanhoucke

We present a stochastic program to solve both the operating theatrescheduling problem as the physician rostering problem. Both problemshave received a huge amount of attention in literature, but the combi-nation remains untackled, largely due to complexity. The stochasticinteger programming formulation is created based on the most impor-tant objectives and restrictions of both problems. We analyze schedulesafter applying variation on the surgeries. Our experiments show someinteresting insights for physician roster schedulers as well as operatingtheatre scheduling managers.

2 - An Optimisation Model for Staff Planning in a HomeCare OrganisationPablo Andrés Maya Duque, Environment, technology andtechnology management, University of Antwerp,Stadscampus, S.B.513, Prinsstraat 13, 2000, Antwerp,Antwerp, Belgium, [email protected], Marco Castro,Kenneth Sörensen, Peter GoosIn this talk, we present the core optimisation component of a decisionsupport system that Landelijke Thuiszorg, a non-profit organisationthat provides home care services for several provinces in Belgium, willimplement in order to assist the regional service planning. The optimi-sation model takes into account assignment, scheduling and routing de-cisions simultaneously, while considering two objectives, namely theservice level and the travelled distance. A solution strategy based ona set partition formulation and a randomised local search is describedand its performance is evaluated.

3 - A Branch and Bound Algorithm for Pharmacy DutyScheduling ProblemOzgur Ozpeynirci, Department of Logistics Management,Izmir University of Economics, Sakarya Cad. No:156,Balcova, 35330, Izmir, Turkey, [email protected] Turkey, the pharmacies provide service during the daytime of week-days. At the nights of weekdays and at the weekends, only on dutypharmacies provide service. The pharmacy duty scheduling (PDS)problem aims to assign duties to the pharmacies in the least costlyway over a planning horizon. The PDS is multi-period facility locationproblem with special side constraints and it is NP-Hard. We proposea branch and bound algorithm and test the performance of the algo-rithm on randomly generated instances. The study is supported by theScientific and Technological Council of Turkey (TUBITAK).

4 - A Mixed Integer Linear Program for Scheduling Prob-lemLakhdar Djeffal, Universit Hadj-lakhdar, 05000, Batna,Algeria, [email protected] work deals with a personal scheduling problem were the staffwishes and degrees of quality regarding their timetable are taken intoaccount. This optimization problem is modelled as a mixed integer lin-ear programming program. A solution algorithm based in branch andprice procedure is then presented. Computational results are presentedshowing the ability of our algorithm to solve realistic instances andcomparison with other solution approaches.

� WA-15Wednesday, 8:30-10:00RB-2101

Vector and Set-Valued Optimization III

Stream: Vector and Set-Valued OptimizationInvited sessionChair: Vicente Novo, Matematica Aplicada, Universidad Nacionalde Educacion a Distancia, Juan del Rosal no. 12, 28040, Madrid,Spain, [email protected]: Elena Molho, Dipartimento di Scienze Economiche eAziendali, Università di Pavia, Via San Felice 5, 27100, Pavia,Italy, [email protected]

1 - Accuracy functions and robustness tolerances invector discrete optimizationYury Nikulin, Department of Mathematics and Statistics,University of Turku, Matematiikan laitos,Turun yliopisto,Vesilinnantie 5, Luonnontieteiden talo, 20014, Turku, Finland,[email protected], Marko M. Mäkelä, Olga KarelkinaA general vector combinatorial optimization problem is consideredwhere individual linear cost functions are subject to small independentperturbations. For the Pareto optimality principle appropriate defini-tions of the worst-case relative regret and robust solution are specified.It is shown that these definitions are closely related to the concept ofaccuracy function. We also present formulae allowing the calculationof robustness tolerances of a single cost vector.

180


2 - On generalized well-posedness for vector optimiza-tionRuben Lopez, Departamento de Matematica y FisicaAplicadas (DMFA), Universidad Catolica de la SantisimaConcepcion, Alonso Ribera 2850, 409-0541, Concepcion,VIII Region, Chile, [email protected]

The aim of this work is to study notions of generalized well-posednessfor vector optimization problems. To do this, we employ a recent no-tion of approximate solution for vector optimization problems coupledwith a notion of variational convergence for vector-valued functions.

3 - A notion of well-posedness in set-valued optimiza-tionElena Molho, Dipartimento di Scienze Economiche eAziendali, Università di Pavia, Via San Felice 5, 27100, Pavia,Italy, [email protected], César Gutiérrez, EnricoMiglierina, Vicente Novo

We study a notion of pointwise well-posedness in the setting of setoptimization. We study the properties of a scalarization of the set opti-mization problem in the sense of Gerstewitz that allows us to character-ize the strict minimizers on a collection of cone-proper sets. We char-acterize the well-posedness of the set optimization problem throughthe well-posedness, in the Tykhonov sense, of an associate family ofscalar optimization problems. Finally, we use a notion of quasicon-vexity for set-valued maps in order to individuate a class of well-posedproblems.

� WA-16Wednesday, 8:30-10:00RB-2103

Conic Optimization: Algorithms andApplications

Stream: Linear and Conic ProgrammingInvited sessionChair: Knut Haase, Institut f. Verkehrswirtschaft, Lehrstuhl BWL,insb. Verkehr, Universität Hamburg, Von-Melle-Park 5, 20146,Hamburg, Germany, [email protected]

1 - Speeding up the spectral bundle method by solv-ing the quadratic semidefinite subproblems with aPSQMR approachChristoph Helmberg, Fakultät für Mathematik, TechnischeUniversität Chemnitz, D-09107, Chemnitz, Germany,[email protected], Kim-Chuan Toh

The spectral bundle method is tuned to solving semidefinite programs(SDP) with large semidefinite matrix variables having constant orbounded trace. For efficient bundle sizes, solving a quadratic semidef-inite subproblem by interior point methods formed the bottleneck sofar. We report on our experience with a preconditioned symmetricquasi minimal residual (PSQMR) approach for computing the New-ton step in this interior point method like in the package QSDP. On ourtest instances this results in significant savings. However, results arenot yet satisfactory for cutting plane approaches.

2 - New results in copositive optimizationMirjam Duer, Mathematics, University of Trier, 54286, Trier,Germany, [email protected]

Copositive programming (i.e., minimization of a linear function overthe cone of copositive matrices) is a highly useful tool to model non-convex quadratic problems, possibly even including binary variables.In this talk, we discuss some recent developments, such as copositivitytests, cutting plane algorithms and the relationship to Parrilo’s approx-imation hierarchy.

3 - A Semidefinite Optimization Approach to Multi-RowFacility LayoutPhilipp Hungerländer, University of Klagenfurt, Austria,[email protected], Miguel Anjos

The multi-row facility layout problem is concerned with placing de-partments along one or several rows so as to optimize objectives suchas material handling and space usage.Significant progress has beenmade in recent years on solving single-row problems to global op-timality using semidefinite optimization. In our talk we present theextension of the semidefinite programming approach to multi-row lay-out. Our computational results show that the proposed semidefiniteoptimization approach provides high-quality global bounds in reason-able time for double-row instances with up to 16 departments.

4 - A Branch-and-Price Approach for Sales Force De-ploymentKnut Haase, Institut f. Verkehrswirtschaft, Lehrstuhl BWL,insb. Verkehr, Universität Hamburg, Von-Melle-Park 5,20146, Hamburg, Germany,[email protected], Sven MüllerThe objective of sales force deployment is to maximize the total profit.In literature valid upper bounds are not provided. We propose a semi-infinite binary model. The LP-relaxation is solved by column genera-tion. For the optimal objective function value of the LP-relaxation anupper bound is provided. Branch-and-Price is used to obtain a tightgap for the objective function value of the optimal integer solution.We propose contiguity constraints ensuring contiguous territories. Thelargest instance comprises 50 potential locations and 550 sales cover-age units: 1273 sec., gap < 0.01.

� WA-17Wednesday, 8:30-10:00RB-2105

Global Optimization 2

Stream: Global OptimizationInvited sessionChair: Herman Mawengkang, Mathematics, The University ofSumatera Utara, FMIPA USU, KAMPUS USU, 20155, Medan,Indonesia, [email protected]: Dmitry Krushinsky, Department of Operations, Universityof Groningen, Nettelbosje 2, 9747 AE, Groningen, Netherlands,[email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - Modeling Super-flexibility Sustainable DistributionCentre of a Supply ChainRonsen Purba, Mathematics Dept., STMIK MikroskilMedan/Graduate School of Mathematics, University ofSumatera Utara, Jalan Thamrin No. 140 Medan NorthSumatera, Medan, Indonesia, [email protected] supply chain is a network that performs the procurement of rawmaterial, the transportation of raw material to intermediate and endproducts, and the distribution of end products to retailers or customers.Super-flexibility is defined as the capacity to transform by adaptingto new realities. So far, superflexibility has not been included in themodel of sustainable suply chain. In this paper we address a new opti-mization model to decide distribution centres based on super-flexibilityand environmental consideration.

2 - Fish Processed Production Planning Under Uncer-tainty Considering QualityTutiarny Naibaho, Mathematics, Quality University, FMIPAUSU, 20155, Medan, North Sumatera Province, Indonesia,[email protected] goal in production planning is to meet customer demand. Themajor decisions are production and inventory levels for each productand the number of workforce in each planning period. This paper con-siders the management of small traditional business at North Sumat-era Province which performs processing fish into several local seafoodproducts. The inherent uncertainty of quality to be met, leads the pro-duction planning problem to a chance constraint programming model.We use scenario generation based approach for solving the model.

181


3 - On multi-objective black box optimization of expen-sive objectivesVytautas Jancauskas, Vilnius University, LT-01513, Vilnius,Lithuania, [email protected], Antanas Zilinskas,Panos PardalosThe problems are supposed non-convex, gradients not available; pos-sible number of computations of objectives is relatively small. There-fore metrics used to evaluate algorithms should be defined correspond-ingly. The selection of test functions representing real world expensivemulti-objective problems is discussed. Prospective of generalizationof the algorithms, supposed to the single-objective global optimizationof expensive problems, to the multi-objective optimization is analyzedincluding Lipshitz approach, algorithms based on statistical models,radial basis functions based algorithms.

4 - Optimality and multiobjectiveness of cell formationin group technologyDmitry Krushinsky, Department of Operations, University ofGroningen, Nettelbosje 2, 9747 AE, Groningen, Netherlands,[email protected], Boris Goldengorin, Jannes SlompIn this talk we present an exact model for cell formation based on theminimum multicut problem. We show that minimisation of intercellmovement may not be appropriate as a sole goal. Several alternativeobjectives are proposed together with the ways of inserting them intothe considered model. An industrial example is used for illustrativepurposes.

� WA-20Wednesday, 8:30-10:00RB-2113

Portfolio Decision Analysis

Stream: Decision Making Modeling and Risk Assess-ment in the Financial SectorInvited sessionChair: Marius Radulescu, Institute of Mathematical Statistics andApplied Mathematics, Casa Academiei Române, Calea 13Septembrie nr.13, Bucharest 5, RO-76100, 050711, Bucharest,Romania, [email protected]

1 - Absolutely optimal portfoliosGheorghita Zbaganu, Faculty of Mathematics and Informatics,University of Bucharest, Academiei 14„ Bucharest, Romania,[email protected], Marius RadulescuConsider n risky assets. A portfolio is a n-dimensional vector whosecomponents represent the proportions of the sum owned by the investorthat are invested in assets. A portfolio x is called absolutely optimal ifit realizes the maximum of the expected utility of its return for everyconcave utility function. The aim of the paper is to give characteri-zations of those financial markets for which a given portfolio is abso-lutely optimal. Here by a financial market we understand a randomvector whose components are the random variables associated with therate of return of assets.

2 - Structured Portfolio Management under AmbiguityJean-luc Prigent, ThEMA, University of Cergy-Pontoise, 33,Bd du Port, 95011, CERGY-PONTOISE, France,[email protected], Hachmi Ben AmeurIn this paper, using results about martingale theory and convex analy-sis, we determine the general optimal portfolio payoff within ambigu-ity. We extend results of Leland (1980), Pflug and Wozabal (2007)and generalize the robust optimization of Ben-Tal and Nemirovski(1998,1999) by introducing the ambiguity index based on relative en-tropy. Our findings have important applications in financial engineer-ing for the optimal design of financial structured portfolios.

3 - Determination of the Optimal Weights in a CurrencyPortfolio with Sharpe Ratio Maximizing ApproachCelal Barkan Güran, Management Engineering, IstanbulTechnical University, Ulus Sok. 11/8 Sezer Apt, Suadiye,34740, Istanbul, Turkey, [email protected], OktayTas

In this paper we design a portfolio whose components are all currenciesthat are officially traded in Turkey. Based on the last five years dailydata, the currencies lying outside the efficient frontier is excluded fromthe analysis and then Sharpe Ratio function, whose independent vari-ables are the weights of the corresponding currencies, is constituted.Finally, the optimal weights maximizing this Sharpe Ratio function isfound by using numerical optimization methods. This study can beconsidered as one of the leading portfolio optimization works aboutthe currencies traded in Turkey.

4 - Applications of portfolio theory to production plan-ning for fish farmsMarius Radulescu, Mathematical Statistics, Institute ofMathematical Statistics and Applied Mathematics, CasaAcademiei, Calea 13 Septembrie, nr. 13, 050711, Bucharest,Romania, [email protected], Constanta ZoieRadulescu, Sorin Radulescu

The paper presents several single period portfolio selection models forproduction planning in fish farms. The aim of the models is to findoptimal fish production plans. A fish production plan is a couple (x,y)formed by a binary matrix (denoted by x) that describes the alloca-tion of fish species to pools and a vector y that describes the moneyallocations to pools. The objective functions and the constrains of themodels are the financial risk and the expected return of the fish pro-duction plans. In order to solve the models heuristic algorithms andcomputer simulation is used.

� WA-21Wednesday, 8:30-10:00RB-2115

Financial Service Management

Stream: Long Term Financial DecisionsInvited sessionChair: Heinz Eckart Klingelhöfer, Managerial Accounting andFinance, Tshwane University of Technology, Building 12, Room117, Private Bag X680, 0001, Pretoria, Gauteng, South Africa,[email protected]

1 - Long-term Simulation of Investment StrategiesGeraldine Tchegho, Faculty 7: Business Studies &Economics; Department of Finance, University of Bremen,Hochschulring 4, 28359, Bremen, Germany,[email protected], Thorsten Poddig

Pension funds and life insurance companies face the problem of long-term return guarantees, either as a promised return rate or any kindof capital guarantee. Investors need to know how the return-risk-profiles behave in the long run. We investigate vector autoregressiveand Markov switching models as well as a simple random walk modelas benchmark to simulate asset returns using US macro economic datafrom 1871-2011. We use a rolling estimation window to assess perfor-mance and show that the different asset return simulation models leadto considerably varying results.

2 - Investments into ServicesHeinz Eckart Klingelhöfer, Managerial Accounting andFinance, Tshwane University of Technology, Building 12,Room 117, Private Bag X680, 0001, Pretoria, Gauteng, SouthAfrica, [email protected]

The paper investigates the characteristics of services and develops atwo-step approach for their valuation. The first step delivers the in-dividually optimal service programme. The second step allows forvaluating investments into new services and for identification of thedeterminants of the investments price ceiling. The may be interpretedas (corrected) net present values of the payments and the interdepen-dencies arising from changes in the optimal programs. We use sensi-tivity analysis and example to support these findings and to get moreinformation on the effects of these determinants.

182


3 - When to cut a treeFritz Helmedag, Economics, Technical University ofChemnitz, Chair VWL II, Thueringer Weg 7, 09107,Chemnitz, Germany, [email protected]

For about two centuries, the optimal rotation period in forestry hasbeen debated intensively. According to the meanwhile prevalent opin-ion, the so-called Faustmann condition solves the problem because itmaximizes the present value of woodland. The result, however, con-tradicts the ’principle of maximum yield’, i.e. it is productively ineffi-cient. The article develops Faustmann’s approach further and providesan objective function suited to reconcile available cutting rules.

4 - When to cut a tree given sustainability constraintsThomas Burkhardt, Campus Koblenz, IfM, UniversitaetKoblenz-Landau, Universitätsstr. 1, 56070, Koblenz,Germany, [email protected]

In a remarkable paper Helmedag analyzes the problem when to cut atree and derives an optimality condition which is independent of theinterest rate, as opposed to the well known Faustmann approach. Wediscuss this approach and claim that a proper inclusion of sustainabilityconstraints resolves the contradiction.

� WA-23Wednesday, 8:30-10:00RB-Delta

Optimization in Machine Learning

Stream: Convex OptimizationInvited sessionChair: Ronny Luss, Sierra Project Team, INRIA, France,[email protected]: Vadim Strijov, Russian Academy of Sciences, ComputingCenter, Vavilova 42-268, 119333, Moscow, Russia, RussianFederation, [email protected]

1 - Application of cutting plane methods in machinelearningVojtech Franc, Department of Cybernetics, Czech TechnicalUniversity in Prague, FEE, Technicka 2, 16627, Prague,Czech Republic, [email protected]

Cutting plane methods are optimization techniques that incrementallyconstruct an approximation of a feasible set or an objective function bylinear inequalities called cutting planes. Numerous variants of this ba-sic idea are among standard tools used in convex nonsmooth optimiza-tion. Recently, cutting plane methods have seen growing interest in thefield of machine learning. We describe basic theory behind these meth-ods and show several of their applications to solving machine learningproblems. We concentrate on learning structured output classifiers andtheir applications in computer vision.

2 - Sparse Rank-One Matrix Approximations: ConvexRelaxations, Direct Approaches, and Applications toText DataRonny Luss, Sierra Project Team, INRIA, France,[email protected], Marc Teboulle

The sparsity constrained rank-one matrix approximation problem is adifficult mathematical optimization problem arising in a wide array ofuseful applications. We survey a variety of approaches including con-vex relaxations and direct approaches to the original nonconvex for-mulation. Convex relaxations are solved by applying fast first-ordermethods, while the direct approach builds on the conditional gradientmethod. Numerical experiments and applications with text data will begiven.

3 - Optimizing support vector regression parameters byusing global search algorithmAlexey Polovinkin, Computational Mathematics andCybernetics, Nizhny Novgorod State University, NizhnyNovgorod, Russian Federation,[email protected], Konstantin Barkalov, NikolaiZolotykh, Iosif Meyerov, Sergey Sidorov

The performance of Support Vector Regression (SVR) significantly de-pends on insensitive zone thickness, a penalty factor and kernel func-tion parameters. The most popular algorithms for finding optimal pa-rameters of SVR use optimization of cross-validation error functionthat is multiextremal in general case. We use global search algo-rithm (based on information-statistical approach for global optimiza-tion) to solve multiextremal optimization problem. The algorithm re-duces multidimensional problem to equivalent one-dimensional prob-lem by applying Peano-type space-filling curves.

� WA-24Wednesday, 8:30-10:00CC-A11

Data Mining and Decision Making

Stream: Data Mining and Decision MakingInvited sessionChair: Benjamin Gotthardt, Department of Mathematics andScientific Computing, Uni Graz, 8010, Graz, Styria, Austria,[email protected]

1 - Development of Traffic Accidents Prediction Modelwith Neural NetworksMuhammed Yasin Çodur, Engineering/civil, AtaturkUniversity, Ataturk University Engineering Faculty CivilEngineering department no :224, 25240, Erzurum, Turkey,[email protected], Ahmet Tortum

Road traffic accidents continue to be a major problem in Turkey. Itis important to clarify the relationship between traffic accidents andvarious influencing factors in order to reduce the number of traffic ac-cidents. The data was collected on the Turkey/Erzurum Highways. Inorder to minimize the uncertainty of the data, neural network theorywere applied. The neural network theory can provide fair learning per-formance by modeling the human neural system mathematically. Inconclusion, this study focused on the practicability of the neural net-work theory for traffic safety analysis.

2 - Multi product Newsvendor-problem - solution withNCP and a non-trivial effectBenjamin Gotthardt, Department of Mathematics andScientific Computing, Uni Graz, 8010, Graz, Styria, Austria,[email protected], Marc Reimann

This talk investigates a multi product Newsvendor-problem with twolimited capacity sources. Under uncertain demand the solution con-tains the production strategies (speculative, reactive, dual or no pro-duction) and the produced quantities for each product. Until now onlymodels governed by one limited capacity are known in literature. In thesolution proces a non-trivial effect was found: With decreasing levelof one capacity (other one kept constant), a back-and-forth switch be-tween production strategies occurs. Due to the special structure of theNewsvendor an NCP is used to solve it.

3 - A cash flow and profitability monitoring system forretailers in the greek pharmaceutical industryGeorge Marinakos, Engineering Sciences, University ofPatras, 26500, Patras, Greece, [email protected],Sophia Daskalaki, Theodor Drinias, Kostas Tsekouras

We present a cash flow and profitability monitoring solution designedfor Greek pharmacies. It relates accounting entries with credit timeand so reflects the times of cash flows for a business unit. The equa-tions take into account attributes that affect liquidity. Monthly inflowsand outflows are represented by equations that eventually build the liq-uidity curve and the cash flow balance overtime. The model has anambitious and useful purpose, to inform and consult the owners of thebusiness units and the other members of the pharmaceutical chain andthus reduce financial risk for the chain.

183


4 - Preference Modelling and Market Price Forecastingwith Causal-Retro-Causal Neural NetworksHans Georg Zimmermann, Corporate Technology CT T,Siemens AG, Otto-Hahn-Ring 6, 81730, München, Germany,[email protected], Ralph Grothmann,Christoph Tietz

Forecasting of market prices is a basis of rational decision making.Recurrent neural networks offer a framework for modelling temporaldevelopments. Causality explains the present state of a system by fea-tures, which are prior to the current state. Looking for alternatives, weremember that markets are human made dynamical systems. In mi-croeconomics we describe human behaviour with utility functions. Ifwe know the human reward function, we could describe the behaviourof the market participants and thus the market by retro causal equa-tions. This approach improves the forecasting accuracy.


Future role of nuclearStream: Long Term Planning in Energy, Environmentand ClimateInvited sessionChair: Nicklas Forsell, CMA, CMA, France,[email protected]

1 - Future challenges for the French power generationparadigmEdi Assoumou, Centre de Mathematiques Appliquees, MinesParisTech, Sophia Antipolis, France,[email protected], Vincent Mazauric, NadiaMaïzi

In the late 80’s France shifted from fossil fuel consumption to electric-ity due to its nuclear power plants electricity producing capacity. Wepropose to examine the consequences of this shift in terms of long termenergy policy in France. We apply an updated TIMES-FR model thatoptimizes energy systems with explicit descriptions of the technologiesused. Through scenarios that reflect different nuclear policies, we il-lustrate the impact of future energy choices on the evolution of Frenchelectricity generating system for the 2050 horizon and assess the ques-tion of reliability of the system

2 - Looking Japan energy future after the 11th March2011’s earthquakeAdrien Wacziarg, EPI, EDF, 75, Paris, France,[email protected], Nadia Maïzi, PrabodhPourouchottamin

The scope of the study is to investigate the consequences of the GreatEast Japan Earthquake on the energy future of Japan and elsewhere,while conducting prospective studies. One important focus will beJapan’s possible long term evolution, beyond the current post-crisisclimate. This will lead to a better understanding of how other coun-tries, most notably western will change and the parameters that comesinto play.

3 - Critical Analysis of "Energy 2050" report: Anoverview of the French electricity mixRenaud Dudouit, MINES ParisTech, 06904, Sophia AntipolisCedex, France, [email protected]

In wake of the Fukushima accident and national targets to reduce thegreenhouse gases emissions, Questions are being raced concerninghow the French electricity mix will evolve. "Energy 2050’ report triesto answer these questions by studying 4 scenarios of nuclear produc-tion: extension of the operation of today’s nuclear power plants, accel-eration of 3rd generation (European Pressurized Reactor) deployment,reduction of nuclear power capacity, and exit of the nuclear power.This study will analyze these different scenarios and evaluate their im-pact on the future French electricity mix.

4 - An evaluation of the European CCS potentialOlivia Ricci, Orleans Economic Laboratory, University ofOrleans, 45067, Orleans, France, [email protected],Sandrine Selosse

Carbon negative biofuels could be obtained if a carbon capture and ge-ological storage (CCS) chain is applied to their production. This studyevaluates the potential of biofuels-CCS in the transport sector in orderto limit temperature increase as far below 2C as possible. This analysisis conducted with the multiregional TIAM-FR optimization model.


OR and Environmental Management

Stream: OR and Environmental ManagementInvited sessionChair: Ralf Gössinger, Business Administration, Production andLogistics, University of Dortmund, Otto-Hahn-Str. 6, 44227,Dortmund, Germany, [email protected]: Michael Kaluzny, Business Administration, Production andLogistics, University of Dortmund, Germany,[email protected]

1 - A Modified Cross Entropy Method for the Optimiza-tion of an Environmentally Sustainable Supply ChainAli Eshragh Jahromi, School of Mathematical Sciences, TheUniversity of Adelaide, North Terrace Campus, 5005,Adelaide, South Australia, Australia,[email protected], Behnam Fahimnia

We develop an optimization model that minimizes the overall supplychain costs while explicitly incorporating multiple environmental per-formance measures including carbon emission, energy consumptionand waste generation. A modified Cross-Entropy solution approach isdesigned to determine the optimal production and distribution strate-gies for the proposed environmentally sustainable supply chain. Lastly,numerical results from the model implementation in a real world casestudy are analyzed and interesting practical and managerial implica-tions are discussed.

2 - Resilience of the smart meter-enabled electricitysupply chainsBehzad Samii, Operations and Technology ManagementCenter, Vlerick Research, Reep 1, 9000, Gent, Belgium,[email protected], Hakan Ümit

Controlling demand is one of the most effective —yet hard to reach-ways to achieve resilience in any supply chain. In electricity supplychains, fortunately, the real-time consumption data provided by thesmart meters can be instrumental in designing mechanisms to flattendemand (i.e. resilience by variability reduction) during the peak hours.In this study, we use actual consumption data (recorded by the smartmeters in two Belgian tows during the summer of 2011) to proposeand compare subscribers’ segmentation mechanisms based on the 15-minute and 24-hour meter reading frequencies.

3 - Value co-creation system for B2B service: A casestudy on after-sales service of electric power indus-tryNobuhiko Nishimura, Faculty of Economics, NagasakiUniversity, 4-2-1, Katafuchi, 850-8506, Nagasaki, Japan,[email protected]

The service proposal process in B2B sector still has been carried outmainly by experienced service engineers. The objective of this re-search is to establish efficient value co-creation process of the B2Bservice for the after-sales service of electric power plants. A numberof service proposal documents were disassembled into elemental ser-vice components using the text mining. Each service proposal wascharacterized as the integration of each elemental service component.Then the causal relation of the criteria to the business circumstanceswas modeled with Bayesian network approach.

184


4 - Energy price models: Regime switching with cluster-ing techniquesBaptiste Salasc, R&D, Air Liquide, Centre de rechercheClaude & Delorme, 1, chemin de la porte des Loges B.P 126,78354, Jouy-en-Josas, France,[email protected], Steven Gabriel, Yohan Shim

Energy prices present different commonly known smooth features, likemean reversion, troubled by highly volatile periods in the occurrenceof unexpected jumps, spikes or structural breaks. The purpose of thiswork is to investigate the possibility to apply clustering methods usedin data mining (k-mean, k-median) to the detection and reproduction ofone of the specific features of energy prices: the regime switching. Theapplication of clustering techniques to energy price regime switchingmodels improved significantly the robustness of the underlying meanreversion model parameters estimation


Decision Support Systems

Stream: Decision Support SystemsInvited sessionChair: Pascale Zaraté, Institut de Recherche en Informatique deToulouse, Toulouse University, 118 route de NarBonne, 31062,Toulouse, France, [email protected]

1 - Decisions, processes, and decision processes forenabling risk sharing and performance creation atthe buyer-supplier interfaceKristian Rotaru, Accounting and Finance, Monash University,Australia, 900 Dandenong Rd, 3145, Caulfield East, Victoria,Australia, [email protected], CarlaWilkin, Leonid Churilov

We investigate how, through alignment of risk-performance objectivetrade-offs and underlying business processes, risk sharing and perfor-mance creation can be enabled at the buyer-supplier interface in col-laborative supply chain processes. This is achieved by extending theRisk-Aware Value-Focused Process Engineering modeling methodol-ogy to formally represent value creation mechanisms that are drivenby both performance improvement and risk minimization objectives ofbuyers and suppliers and align these mechanisms with process compo-nents at the buyer-supplier interface.

2 - Modelling to generate alternatives using biologically-inspired algorithmsJulian Scott Yeomans, OMIS, Schulich School of Business,York University, 4700 Keele Street, SSB S338, M3J 1P3,Toronto, Ontario, Canada, [email protected],Raha Imanirad

In solving real-world programming applications, it is preferable to gen-erate numerous alternatives providing disparate perspectives. Theyshould possess near-optimal measures to all known modelled objec-tive(s), but be different in their system structures. Biologically-inspiredmodelling-to-generate-alternative algorithms are particularly efficientat creating multiple solution alternatives that both satisfy required sys-tem performance criteria and yet remain maximally different in theirdecision spaces. Efficacy of some bio-inspired methods is demon-strated on some applications.

3 - The Role of product factors, sellers’ factors and atti-tude toward risk on e-store purchasingArik Sadeh, Management of Technology, Holon Institute ofTechnology, 52 Golomb Street, P.O. Box 305, 58102, Holon,Israel, [email protected]

The willingness to purchase products in e-stores is examined. Threeaspects of purchasing are explored: (1) the role of product factors, (2)sellers’ factors and (3) the attitude of buyers toward risk. The productfactors are quality and price of a given product while accurate prod-uct description, on-line seller accountability, and privacy and security

are sellers’ factors. Portfolios of these five factors are examined withrespect to motivation to purchase in e-stores using an experimental de-sign scheme. The attitude of buyers toward risk is examined usingZukerman’s sensation questionnaire.

4 - A multicriteria decision scheme for water pipe re-placement prioritizationYoussef Tlili, GESTE, ENGEES, 1 Rue Quai Koch, 67000,Strasbourg, France, [email protected], Amir Nafi

The goal of this study is to identify a reliable replacement policy forpipes of a water distribution system. The paper deals with two topics:1) Asset management of water networks which is based on the priori-tization of pipes that require an urgent intervention for the rehabilita-tion or replacement. 2) Decision-making concerning the classificationof pipe candidates for renewal through the comparison between twokinds of aggregation methods: a) sophisticated methods or models, b)simple understandable schemes.


OR Military Applications

Stream: OR in Military, Safety and Security Applica-tionsInvited sessionChair: Ana Isabel Barros, Military Operations, TNO, POBox96864, 2509 JG, The Hague, Netherlands, [email protected]: Herman Monsuur, Faculty of Military Sciences,Netherlands Defence Academy, P.O. Box 10.000, 1780 CA, DenHelder, Netherlands, [email protected]

1 - Taking uncertainty into account in Unmanned AerialVehicle tour planningLanah Evers, Defence, Security and Safety, TNO, P.O. Box96864, 2509 JG, The Hague, Netherlands,[email protected], Kristiaan Glorie, Suzanne van der Ster,Ana Isabel Barros, Herman Monsuur

Unmanned Aerial Vehicles (UAVs) are used to gather information forboth civilian and military purposes by capturing imagery of specificlocations in the area of interest. We model this problem as the orien-teering problem. To cope with environmental dynamics, we introducethe Two Stage Orienteering Problem (TSOP), which accounts for theeffect of fuel uncertainty on the tour information gain. Using stochas-tic programming we formulate and solve the TSOP and show that theresulting tours have a higher expected tour information gain than toursresulting from a robust optimization approach.

2 - Multi objective decision analysis for fightersquadrons flight scheduling problemMehmet Durkan, Turkish Staff College, Turkey,[email protected]

Flight scheduling in fighter squadrons is a hard and complicated prob-lem which comes with a dynamic environment and multiple decisionmakers. Using pilots as machines and missions to be flown as jobs, itcan be solved like an assignment problem. In this research, the ValueFocused Thinking method is applied to build a decision analysis modelto help decision makers in fighter squadrons evaluate the mission-pilotmatches. The model is used not only for evaluating matches but alsofor ordering assignments to see priorities and proved to be helpful andaccelerated the assignment matching process.

3 - Determining maintenance manpower requirementsfor aircraft unitsNicole van Elst, TNO - Defense, Security and Safety, POBox96864, 2509 JG, The Hague, Netherlands,[email protected], Wouter Noordkamp

When introducing a new weapon system the required maintenancemanpower levels is an important issue. This paper describes how theDutch manpower that is required to guarantee future F-35 operations

185


has been calculated using simulation. Due to the nature of defense op-erations it is nearly impossible to compensate for shortages. Thus Mea-sures of Effectiveness (MOEs) to determine whether the manpowerlevels are adequate may not be based on averages. We describe whichMOEs are used and how the required number of replications has beendetermined, since confidence intervals could not be used.

4 - Lifetime management of defence fleet capabilityPatrick Tobin, Arts and Sciences, Australian CatholicUniversity, St Patricks Campus, Victoria Pde Fitzroy, 3065,Melbourne, Victoria, Australia, [email protected]

Efficient management of defense fleet capability and effective mainte-nance and replacement of assets over a variety of time horizons is animportant operational issue affecting various types of platform. Manycommon features exist in studying all of these assets (platforms) in-cluding efficient maintenance on a scheduled basis, dealing effectivelywith unscheduled maintenance and addressing long term sustainabilityof capability which meet budget targets and use constraints. We lookat various approaches to the problem. Difficulties arise in each type ofapproach which tackles this aspect.


Data Mining

Stream: Optimization and Data MiningInvited sessionChair: Erik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39, 85577,Neubiberg, Germany, [email protected]

1 - Exact and heuristic algorithms based on SupportVector Machine for Feature Selection with applica-tion to Financial ProblemsRenato De Leone, School of Science and Technologies,Università di Camerino, via Madonna delle Carceri 9, 62032,Camerino, MC, Italy, [email protected], Sonia DeCosmis

SVMs are tool belonging to the class of supervised learning methodsthat can solve Regression and Feature Selection Problems. Robustnesscan be incorporated to address the issue of uncertainties and boundedvariability in the data.In this talk we will present new heuristic algo-rithms for Robust Support Vector Regression and Feature Selection.Wewill show that, the use of Simulated Annealing allows to efficientlyselect a subset of features,moreover, the use of robustness slightly de-creases the quality of the solution in the training set while improvingthe quality for the test set.

2 - Arrhythmia Classification via Mathematical Program-mingEmre Çimen, Industrial Engineering, Anadolu Universty,Anadolu Üniversitesi Iki Eylül Kampüsü EndüstriMühendisligi no:106, 26000, Eskisehir, Turkey,[email protected], Gurkan Ozturk

Heart diseases are killing millions every year. A lot of people losetheir lives due to failure of diagnose heart disease earlier. Therefore,early detection of arrhythmia which is significant portion of heart dis-eases can be defined as a classification problem. The aim of this studyis to develop a arrhythmia classifier based on mathematical program-ming which consists of k-means clustering algorithm and polyhedralconic functions. Proposed classifier is tested on Mit — Bih and UCIArrhythmia databases.

3 - Proposal for generation of the three-way PerceptualMap using non-metric Multidimensional Scaling withclustersMoacyr Machado Cardoso Junior, Production Department,Aeronautics Institute of Technology - ITA, R LicinioRodrigues Alves 114, Chácara Jafet, 08730661, Mogi dasCruzes, São Paulo, Brazil, [email protected], Rodrigo Scarpel

The proposal for improvement of the Three-Way perceptual map usingMDS involves the separation of the judges in homogeneous clusters,so that more than one perceptual map of the group of judges will begenerated. The validation of clusters was performed with an Ex-anteindicator of quality. The Final configurations obtained for each clus-ter were tested for effective separation of centroids within objects. Theseparation ratio generated an Ex-post indicator of quality of the separa-tion of clusters. The results showed that the use of two or three clusterswere optimal in most cases.


Scheduling Applications

Stream: Emerging Aspects of Production Planning inContinuous Process Industries: Theory, Optimization,and PracticeInvited sessionChair: Krystsina Bakhrankova, Applied economics, SINTEF -Technology and society, Box 4760 Sluppen, S. P. Andersens veg 5,7465, Trondheim, Norway, [email protected]

1 - Single machine scheduling with due dates and per-ishable raw materialsJean-Charles Billaut, University of Tours, 37200, Tours,France, [email protected], Federico DellaCroce, Patrick Esquirol, Jean-François Tournamille

We consider a single-machine scheduling problem. Each job is char-acterized by duration, a due date and a quantity, corresponding to araw material. All jobs require the same product, which is character-ized by a volume per vial and life duration after opening. We searchfor a schedule, minimizing the quantity of lost product and satisfyinga given value of the maximum lateness. This NP-hard problem comesfrom a chemotherapy production, where perishable active agents arerequired for the realization of medical drugs. Efficient solution meth-ods based on integer linear programming are proposed.

2 - Optimum cost blending application in flour millsMehmet Akif Sahman, Selcuk University, Turkey,[email protected], Abdullah Oktay Dundar, AdemAlpaslan Altun

The quality and trade-off optimization of the product produced at flourmills is a significant problem. In this study, the most appropriate costblending decision in producing flour suitable to the quality sought bycustomers from among various varieties and qualities of wheat wasrealized by using linear programming (the Big M method). The appli-cation was coded from among the object-oriented programming lan-guages, Visual C#. Based on the results obtained, it was observed thatthe linear program produced optimal value solutions that were rapidand decisive.

3 - Production Scheduling in Batch Process IndustriesUsing Timed Automata ModelsSubanatarajan Subbiah, Dept. of Biochemical and chemicalengineering, Technische Universität Dortmund, Emil-Figgestrasse 70, 44227, Dortmund, NRW, Germany,[email protected], ChristianSchoppmeyer, Sebastian Engell

The approach to model complex scheduling problems in the processindustries by timed automata (TA) and to solve them using reachabil-ity analysis is presented. The machines and the jobs are modeled assets of TA in a modular fashion and are composed to form a globalautomaton (GA) - directed graph. The GA has an initial node whereno task has been started and at least one target node where all tasksto produce the products have been finished. Each path from the initialnode to a target represents a schedule and a cost optimal reachabilityanalysis is performed to compute the optimal schedule.

186




Stream: DEA and Performance Measurement (con-tributed)Contributed sessionChair: Cecilio Mar-molinero, Kent Business School, University ofKent, KBS, University of Kent, CT2 7PE, Canterbury, Kent,[email protected]

1 - On the properties of a linear transformation of vari-ables in Data Envelopment AnalysisAbolfazl Keshvari, Aalto University School of Economics,Finland, [email protected], Pekka KorhonenWe consider the problem of making a linear transformation of vari-ables in DEA. When a linear transformation is nonsingular, no prob-lems arise provided the efficient frontier is re-defined. Instead, prob-lems ensue, if a linear transformation is singular. This is the case, whenwe reduce the number of variables. Using the geometrical structure ofDEA, we show how different sets of variables can change the domi-nance cones and affect the efficient frontier. We will consider variousdimension reduction methods applied in DEA and point out that a userhas to be very careful in choosing the method.

2 - Classifying Inputs and Outputs Based on TOPSISMethod Using Modified DEA ModelSahand Daneshvar, Industrial Engineering, EasternMediterranean University, North Cyprus (via Mersin 10Turkey), 5555, Gazimagusa, Famagusta, Turkey,[email protected], Gokhan IzbirakIn conventional data envelopment analysis, it is assumed that the inputversus output status of any particular performance measure is known.In some situations, finding the status of some variables from the inputor output point of view is very difficult; these variables are treated asboth inputs and outputs and are called flexible measures. In this paperthe CCR Model, modified by facet analysis, and TOPSIS Method areused for classifying inputs and outputs based on Cook and Zhu (2007),for specifying the input versus output status of flexible measures.

3 - Weights in the multi-activity DEA modelCecilio Mar-molinero, Kent Business School, University ofKent, KBS, University of Kent, CT2 7PE, Canterbury, Kent,[email protected], Fabiola Portillo, Diego PriorThe standard DEA model assumes that the units to be assessed per-form a single activity by using inputs to generate outputs. The multi-activity DEA model allocates efficiency scores to each one of the ac-tivities that are jointly performed, but there is a cost: the formulationbecomes non-linear. We discuss the difference between the standardand linearized model within a micro-economic theoretical framework,and use Karush-Kuhn-Tucker theory to demonstrate that the linearizedand the standard model only produce the same results if the units beingassessed are 100% efficient at all activities.


Sustainable Development in CivilEngineering and Multi-attribute DecisionMaking

Stream: OR for Sustainable DevelopmentInvited sessionChair: Tatjana Vilutiene, Department of Construction Technologyand Management, Vilnius Gediminas Technical University,Sauletekio ave. 11, LT01001, Vilnius, Lithuania,[email protected]

1 - Logistics significance of wood product manufactur-ing on competitiveness based managementKristine Fedotova, Civil construction and real estateeconomics, Riga Technical University, Meza 1/7-212,LV-1048, Riga, Latvia, [email protected], InetaGeipele, Sanda GeipeleTopicality is associated with the important contribution of wood prod-ucts manufacturing in Latvian economy. Important role in develop-ment of wood products manufacturing has improvement of wood prod-ucts manufacturing that includes acceptance of strategic managementdecisions and solutions for wood resource flow optimization through-out wood products added value chain, as well as an interference withrelated industries. Solutions of logistics issues play leading role in in-creasing manufacturing management by providing options of productmanufacturing cost optimization.

2 - Structural Reliability Analysis and Decision SupportApplying Probabilistic MethodsRobertas Alzbutas, Lithuanian Energy Institute, KaunasUniversity of Technology, LT-44403, Kaunas, Lithuania,[email protected], Gintautas DundulisA probability-based modelling approach was developed to analyze re-liability of critical structures and to support decisions for managing ofuncertainty. If parameters of structural model are uncertain, then thisrandomness is taken into account. As example the parameters impor-tant for structural integrity are analysed. The finite element method isused for the deterministic analysis of strength and then applying proba-bilistic methods the parameters under uncertainty were taken into con-sideration and probabilities of structural failure were estimated.

3 - Model of structure solution selection for the sustain-able building’s designKlaus Holschemacher, Faculty of Civil Engineering andArchitecture, HTWK Leipzig, Karl-Liebknecht-Strasse 132,04277, Leipzig, Germany,[email protected], Ernestas Gaudutis,Jolanta TamosaitieneThe aim of sustainable development is to minimize the ecological dam-age resulting from infrastructure creation, operation and maintenance,reduce the ecological footprint of human activities. Many conceptsof sustainable design aren’t properly estimated in high rise buildingsdesign process. The objective of this research is to demonstrate howsimulation can be used to reflect fuzzy inputs, which allows more com-plete interpretation of model results. Authors discuss on advantagesand disadvantages of presented sustainable high rise building structuralsystem selection model.

4 - Selection of the rational modernization measures:Case of art school modernization in Birstonas cityTatjana Vilutiene, Department of Construction Technologyand Management, Vilnius Gediminas Technical University,Sauletekio ave. 11, LT01001, Vilnius, Lithuania,[email protected], Ceslovas IgnataviciusSignificant reduction of energy costs, improvement of building indoorcomfort conditions, reduction of pollution of the environment, im-provement of aesthetic appearance of the building and extending thelifetime of the building are important issues of modernization. Thisarticle presents the analysis of energy saving measures for art schoolin Birstonas, Lithuania. Objective of the study is to select appropriateenergy saving measures by comparison of packages of energy savingmeasures.


Solution concepts for TU-games

Stream: Cooperative Game TheoryInvited sessionChair: Ilya Katsev, St. Petersburg Institute for Economics andMathematics, Russian Academy of Sciences, Tchaikovsky st. 1,191187 St. Petersburg, Russia, 195067, Saint-Petersburg, RussianFederation, [email protected]

187


1 - The [0,1]-nucleolus in 3-person cooperative TU-gamesNadezhda Smirnova, International Banking Institute, RussianFederation, [email protected], SvetlanaTarashnina

We investigate a new solution concept of a cooperative TU-game calledthe [0,1]-nucleolus in a class of 3-person TU-games. We construct ana-lytical formulas for calculation of the solution. Early it has been provedthat the [0,1]-nucleolus contains the prenucleolus, the SM-nucleolusfor arbitrary 3-person TU-games and the modiclus for balanced games.So, the obtained formulas are hold for them as well.

2 - The lexicographic prekernelElena Yanovskaya, Game Theory and Decision Making,St.Petersburg Institute for Economics and Mathematics, RAS,Tchaikovsky st. 1, 191187, Saint-Petersburg, RussianFederation, [email protected]

The lexicographic prekernel of a cooperative game with transferableutilities lexicographically minimizes the maximal surpluses of oneplayer over another. It is a non-empty efficient solution contained inboth the least core and the prekernel, and it may not contain the prenu-cleolus (Yarom, 1981). In the pesentation a combinatorial characteri-zation of vectors belonging to the lexicographic prekernel is given. Thecharacterization is a weak analog of the known Kohlberg’s (Kohlberg1971) characterization of the prenucleolus with the help of balancedcollections of coalitions.

3 - The SD-prenucleolus for TU gamesIlya Katsev, St. Petersburg Institute for Economics andMathematics, Russian Academy of Sciences, Tchaikovsky st.1, 191187 St. Petersburg, Russia, 195067, Saint-Petersburg,Russian Federation, [email protected]

We introduce and characterize a new solution concept for TU games.The new solution is called SD-prenucleolus and is a lexicographicvalue although is not a weighted prenucleolus. The SD-prenucleolussatisfi?es several desirable properties and is the only known solu-tion that satis?fies core stability, strong aggregate monotonicity andnull player out property in the class of balanced games. The SD-prenucleolus is the only known solution that satisfi?es core stability,continuity and is monotonic in the class of veto balanced games

4 - Cooperative games on accessible union stable sys-temsRene van den Brink, Econometrics, VU UniversityAmsterdam, De Boelelaan 1105, 1081 HV, Amsterdam,Netherlands, [email protected], Encarnación Algaba, ChrisDietz

We consider restricted cooperation in cooperative games where the setof feasible coalitions (i) contains the emptyset, (ii) is union stable (i.e.the union of every two nondisjoint feasible coalitions is feasible), and(iii) satisfies accessibility (i.e. every feasible coalition has at least oneplayer such that without this player the coalition is feasible). This gen-eralizes several known set systems such as the sets of connected coali-tions in an undirected graph and antimatroids. We provide results onthese structures, their dual structures, their supports and an associatedrestricted game.


Sports and OR

Stream: OR in SportsInvited sessionChair: Luciano Mercadante, School of aAplied Science,University of Campinas, Pedro Zaccaria St, 1300, Jd Santa Luzia,Limeira, SP, Brazil, 13484350, Limeira, São Paulo, Brazil,[email protected]

1 - A proposal for redesign of the FedEx Cup playoff se-ries on the PGA TOURChris Potts, School of Mathematics, University ofSouthampton, Highfiled, SO17 1BJ, Southampton,Hampshire, United Kingdom, [email protected],Nicholas Hall

We propose a new design for The TOUR Championship, the final eventof the FedEx Cup playoff series on the PGA TOUR, whereby the tour-nament is a strongly seeded match play event. We validate our designby estimating the probability that a player with a particular rank en-tering The TOUR Championship finishes in various positions in theFedEx Cup standings. A match play format is chosen because it iscloser to the spirit of a playoff event than stroke play. We discuss howour proposal achieves the various objectives outlined by PGA TOURCommissioner Tim Finchem about a possible redesign.

2 - The pooling of sports teams in parallel competitionsDennis Van den Broeck, MOSI, VUB, P.P. Rubensstraat 9,1880, Kapelle-op-den-Bos, Belgium, [email protected]

Although the organization of sports competitions is widely discussedin literature, less attention was devoted to the phenomenon of parallelcompetitions at some fixed hierarchical level. This is often the case incountries covering huge geographical areas or in the organization ofregional sports. The term pooling is used for the process of groupingteams together in separate (parallel) competitions. In this contributionmodels based on mathematical optimization are given to achieve fairresults. Applications in Belgian football are discussed and recommen-dations are given.

3 - Forecasting in-play match outcome in One-day inter-national CricketMuhammad Asif, Center of Operational Research andApplied Statistics, University of Salford, 86 Ayres Road, M167GP, Manchester, United Kingdom,[email protected], Ian McHale

We present a model to forecast in-play match outcome in One-day in-ternational cricket. We use the modified Duckworth/Lewis method toestimate the remaining wickets resources at each given stage of the in-nings. These resources can further be used as an important covariatefor logistic model.

4 - Patient transportation and mountain running: look-ing for optimal solutionsMartin Bracke, Department of Mathematics, University ofKaiserslautern, Gottlieb-Daimler-Str. 48, 67663,Kaiserslautern, Rheinland-Pfalz, Germany,[email protected]

Mathematical Modelling with students at high school and undergradu-ate/graduate level has long been our focus. This paper introduces tworeal-world problems, namely the planning of efficient patient trans-portation in large hospitals and the design of optimal strategies inmountain running. We are going to outline some important findingsfrom different student groups. Moreover, we shortly discuss similari-ties and differences of the various approaches as well as common chal-lenges of the modelling teams.


Societal Complexity, Healthcare andSustainable Development

Stream: Methodology of Societal ComplexityInvited sessionChair: Dorien DeTombe, Methodology of Societal Complexity,Chair Euro Working Group, P.O.Box 3286, 1001 AB, Amsterdam,Netherlands, [email protected]: Eizo Kinoshita, Urban Science Department, MeijoUniversity, 4-3-3 Nijigaoka, 509-0261, Kani, Gifu, Japan,[email protected]

188


1 - How to handle societal complexityDorien DeTombe, Methodology of Societal Complexity, ChairEuro Working Group, P.O.Box 3286, 1001 AB, Amsterdam,Netherlands, [email protected]

In the world of today there are many complex societal problems suchas climate change and credit crisis, traffic and pollution. Policy makershandle these problems globally or locally depending on the problem.When policy makers are not trained during their academic educationthey should ask for scientific support on handling complex societalproblems. A scientific methodology for handling complex societalproblems is developed in the field of Methodology of Societal Com-plexity. Applying this methodology leads to a more stable and sustain-able changes of complex societal problems.

2 - A framework for asking questions about the state ofthe art: the Methodology of Societal ComplexityStephen Taylor, Champlain Regional College, Retired, 5320Avenue MacDonald, Apt 207, H3X 2W2, Cote Saint-Luc,Quebec, Canada, [email protected]

Predicting the future of humankind is difficult and complex. The im-mediate future continues the past, but, while many scenarios providepossible futures, the dominant theme is change. Mediating change willbe the predominant occupation of future societal leaders, and manychanges are rooted in complex societal problems. This session pro-vides a framework for discussion of the current state and future ofthe study of complex societal problems, focusing attention on aspectsof the COMPRAM Methodology. The participants’ contributions willguide leaders of the field find pathways to its future.

3 - Globalization or isolation?-Ricardo’s ModelEizo Kinoshita, Urban Science Department, Meijo University,4-3-3 Nijigaoka, 509-0261, Kani, Gifu, Japan,[email protected]

Through the attempt to clarify justifiable economic circumstanceswhere a policy of globalization and/or a policy of national isolationis proven valid, the author revealed in this thesis that there are twosuch economic phases: An economy where the theory of comparativeadvantage, proposed by David Ricardo, is applicable and an economywhere the theory is not applicable. The author applied his original ap-proach to prove the validity of Ricardo’s comparative advantage theory,and found that the theory can be justified only when a macro economyis in the primal problem phase.

4 - Critical systems practice in generating creativeknowledgeSlavica P. Petrovic, Faculty of Economics, University ofKragujevac, D. Pucara 3, 34 000, Kragujevac, Serbia,[email protected]

Critical Systems Practice (CSP), as a metamethodology, is based oncritical systems thinking. CSP consists of four phases: creativity (toidentify key dimensions of the problem situation), choice (to selecta systems methodology/ies, methods for tackling the problem situa-tion), implementation (to make and implement proposals for change)and reflection (to assess the intervention and learn about the problemsituation and the methodology/ies, methods). Through paradigm diver-sity protection, CSP seeks to provide creative knowledge for managingcomplex organizational and societal problems.

� WA-37Wednesday, 8:30-10:00CC-Act

OR in Health & Life Sciences 5Stream: OR in Health & Life Sciences (contributed)Contributed sessionChair: Ana Viana, INESC TEC/ISEP, Campus da FEUP, Rua Dr.Roberto Frias, 4220-465, Porto, Portugal, [email protected]

1 - Operations Research for pediatric care systems: anopportunity for the personal medical advisor in an-tibiotic treatment

Jelena Hadzi-Puric, Department for Computer Science,Faculty of Mathematics, Studentski trg 16, 11000, Beograd,Serbia, Serbia, [email protected], Jeca Grmusa

Pediatric drug dosing should be based on knowledge of the physiolog-ical characteristics of drugs and pharmacokinetic parameters obtainedduring clinical trials. Evidence of safety and efficiency cannot be ex-trapolated from adults to children (especially for antibiotics therapy).The objectives of this paper are to review the quantitative features ofchosen antibiotics, to introduce appropriate mathematical model andsolver. We have also designed an extensible Web application that reg-ularly updates the database with dosing changes, adverse reactions ofdrugs and biochemical parameters.

2 - An accurate model of Critical Care Unit throughqueueing theoryIzabela Komenda, Mathematics, Cardiff University,Senghennydd Road, CF24 4AG, Cardiff, United Kingdom,[email protected], Jeff Griffiths, Vincent Knight

The random behaviour of arrivals and lengths of stay within a CriticalCare Unit (CCU) make for a complex system. As such dealing withthe growing constraints on the healthcare system at the acute level isdifficult. In this study queueing theory is used to develop a new math-ematical model of patients’ flow. Predictions from the model are com-pared to the observed performance of the Units in the real world, andthe sensitivity of the model to changes in Unit size is explored.

3 - Providing healthcare for elderly in smart homesRadu Prodan, Automation, Technical University Cluj-Napoca,Gheorghe Baritiu Street, no.26-28, room 350, 400027,Cluj-Napoca, Cluj, Romania, [email protected], IoanNascu

The main objective of this paper is that of proposing a smart environ-ment which monitors the elderly people continuously in their daily liv-ing at home and in case of happening of an accident or a major healthproblem sends an emergency call for help. The paper concentrates onimplementation details and practical considerations of merging the var-ious technologies into a functional system. (Acknowledgment: TUCN,paper supported by contract no. POSDRU/107/1.5/S/78534, projectco-funded by the European Social Fund through the Sectorial Opera-tional Program Human Resources 2007-2013)

4 - A compact formulation for the Kidney ExchangeProblemXenia Klimentova, INESC TEC, 4000, Porto, Portugal,[email protected], Miguel Constantino, AnaViana, Joao Pedro Pedroso, Abdur Rais, Filipe Alvelos

In recent years several countries set programs that allow exchange ofkidneys between two or more incompatible patient-donor pairs. Theseprograms are commonly referred to as Kidney Exchange Programs(KEPs) and can be modeled as integer programs (IPs). Previouslyproposed IP formulations are exponential either in the number of con-straints or of variables. In this talk, a new compact IP formulation forthe KEP will be presented. The new description and computationalresults will be compared to other known formulations.

� WA-38Wednesday, 8:30-10:00HH-Colombus

Multiobjective Optimization in LocationProblemsStream: Multiobjective OptimizationInvited sessionChair: José Rui Figueira, Instituto Superior Tecnico, TechnicalUniversity of Lisbon, Av. Cavaco Silva, Tagus Park, 2780 - 990Porto Salvo, 2780 - 990, Lisbon, Portugal, [email protected]: Christiane Tammer, Mathematics and Computer Science,Martin-Luther-University Halle-Wittenberg, Theodor-Lieser-Str.5, D-06120, Halle, Germany,[email protected]

189


1 - Robust multicriteria optimization problemsElisabeth Köbis, Faculty of Natural Sciences II,Martin-Luther-Universität Halle-Wittenberg, Institute ofMathematics, 06099, Halle (Saale), Saxony-Anhalt, Germany,[email protected]

In this talk, we propose a class of robust multicriteria optimizationproblems. We show that well-known robust optimization problemscorrespond to certain scalarizations of robust multicriteria optimiza-tion problems by proving that solutions of the corresponding scalarproblem are weakly and, under uniqueness conditions, strictly Paretooptimal for the robust multicriteria problem. Additionaly, we showthat we get well-known concepts of robustness by using the Tammer-Weidner-functional. Furthermore, we propose applications for robustmulticriteria optimization problems in location theory.

2 - Location Problems with Future FacilitiesAndrea Wagner, Mathematics, Martin-Luther-UniversityHalle, Theodor-Lieser-Straße 5, 06120, Halle/Saale, Germany,[email protected], Kathrin Klamroth,Christiane Tammer

The talk deals with median location problems where the goal is to es-tablish one new facility with respect to a set of existing facilities aswell as a set of future facilities. Future facilities are equivalent to theexisting ones concerning their usage but they do not exist yet. In caseof polyhedral distances the problem will be linearized. For the moregeneral case of an arbitrary norm, duality statements will be given us-ing a generalized Lagrange function. Further it will be shown that thedual of the linearized problem and the Lagrange dual problem coincidein case of polyhedral norms.

3 - An algorithm for solving multicriteria location-costoptimization problemsShaghaf Alzorba, Mathematics, Martin-Luther-University,Richard-Paulick-Str.13, 1210, 06124, Halle, Sachsen-Anhalt,Germany, [email protected]

We investigate a multicriteria location - cost optimization problem,where distances as well as cost functions are to be minimized. First, us-ing a decomposition method we solve the vector-valued location part.A convex partition algorithm for the solution set of the location prob-lem is derived in order to minimize the linear (cost vector-valued) partover this partition. Finally, we present the implementation of the algo-rithm in Matlab.

4 - Robustness in Multiobjective OptimizationJonas Ide, Fakultät für Mathematik, Georg-August-UniversitätGöttingen, Germany, [email protected], AnitaSchöbel, Matthias Ehrgott

Robust Optimization considers problems with uncertain objectivefunction or feasible set. Due to various applications, many differentinterpretations of robustness exist. These various concepts so far onlyapply to single objective functions. Hence, the question arises howto extend these robustness concepts to multiobjective optimization. Inthis talk we present some robustness concepts for multiobjective prob-lems. We will concentrate on strictly robust efficiency, discuss howto compute these robust solutions and show how this concept can beapplied in portfolio optimization.

� WA-39Wednesday, 8:30-10:00HH-Cousteau

Dynamic Programming 1

Stream: Dynamic ProgrammingInvited sessionChair: Lidija Zadnik Stirn, Biotechnical Faculty, University ofLjubljana, Vecna pot 83, 1000, Ljubljana, Slovenia,[email protected]

1 - Decision support model based on group AHP anddynamic programming for optimal regulation of pro-tected areas

Lidija Zadnik Stirn, Biotechnical Faculty, University ofLjubljana, Vecna pot 83, 1000, Ljubljana, Slovenia,[email protected]

Management of protected areas which calls for an equilibrium of eco-nomic, ecological, technological and social objectives is based onlong-term, sustainable and multiple use (agriculture, forestry, water,CO2, etc.) principles. Several stakeholders with conflicting interestsare included through participatory planning of development possibili-ties. The model established on new developed group AHP methods anddynamic programming will be used to solve the treated managementproblem. The acceptable consistency of the aggregated group intervaljudgments is ascertained on individual crisp judgments.

2 - A dynamic programming approach to analyze the de-velopment of non conventional oil supply under un-certainty: application to the Canadian oil sandsFrederic Lantz, IFP-School, 228, avenue Napoleon Bonaparte,92852, Rueil-Malmaison, France, [email protected]

The aim of this paper is to model the development of the non con-ventional Canadian crude oil under uncertainty and increasing envi-ronmental cost. First, a LP model has been set-up in a deterministicapproach. Second, a dynamic programming model has been devel-oped with several crude oil prices and CO2 taxes level.The simulationfor 2030 with the LP model point out that the Canadian tar sands couldrepresent a significant crude oil supply. However, the results of thedynamic programming model point out the negative impact of uncer-tainties on the development of the resource.


This paper proposes a parallel procedure for dynamic multi-objectiveTSP For each of the objectives, a multi-start local search algorithmruns on several processors for the dynamic TSP. A solution attractordrives the local search trajectories to converge into a small region in thesolution space. This procedure identify the solutions from the solutionattractor. This procedure not only generates a set of Pareto-optimal so-lutions, but also provides the structural information about each of thesolutions, which can aid a decision-maker in choosing the best com-promise solution in real-time.

4 - Bi-criteria optimization problem of binary objectsflow servicing by stationary service processor withstorage containerAnastasia Kuimova, Department of Computer Sciences,Control Systems and Telecommunications, Volga StateAcademy of Water Transport, 5a Nesterova St., 603950,Nizhni Novgorod, Russian Federation,[email protected]

The mathematical model of binary objects flow servicing in one servic-ing processor system with storage container is considered. The prob-lem of synthesizing effective service policies subject to two indepen-dent criteria is formulated. This problem is proved to be NP-hard. Itis solved based on Pareto optimization concept. Total set of Pareto-optimal service policies is synthesized using the bi-criteria dynamicprogramming based algorithm. The model gives a description of masswater-transport control processes such as diesel fuel deliveries to theNorthern Territories of the Yamal Peninsula.

� WA-40Wednesday, 8:30-10:00HH-Livingstone


Stream: Preference LearningInvited sessionChair: Willem Waegeman, Department of Applied Mathematics,Biometrics and Process Control, Ghent University, Coupure links653, 9000, Ghent, Belgium, [email protected]

190


1 - Preference elicitation for interactive learning of Opti-mization Modulo Theory problemsAndrea Passerini, disi, University of Trento, via sommarive 5,38100, Povo di Trento, Italy, [email protected], PaoloCampigotto, Roberto BattitiWe consider unknown decision maker preferences as weighted com-binations of predicates in a certain theory of interest. We developedan iterative approach alternating a search and a learning stage: (i) thecurrent approximation of the preference function is optimized (Opti-mization Modulo Theory), providing a set of candidate solutions; (ii)the preference function is refined according to the DM feedback onthese candidates. The learning stage relies on a sparse ranking algo-rithm aimed at automatically selecting the relevant predicates and theircombination, among a catalogue of candidate ones.

2 - On the characterization of a ranking procedure basedon a natural monotonicity constraintMichael Rademaker, Departement of MathematicalModelling, Statistics and Bioinformatics, Ghent University,Coupure links 653, 9000, Gent, Belgium,[email protected], Bernard De BaetsIn previous work, we formulated a new ranking procedure in the tra-ditional context where each voter has expressed a linear order relationor ranking over the candidates. The final ranking of the candidates istaken to be the one which best adheres to a natural monotonicity con-straint. For a ranking a>b>c, monotonicity implies that the strengthwith which a>c is supported should not be less than the strength withwhich either one of a>b or b>c is supported. Though some propertiesof the ranking procedure have been determined, a formal characteriza-tion is still ongoing.

3 - Preference learning for automated reasoningEvgeni Tsivtsivadze, Radboud University, Heyendaalseweg135, 6525 AJ, Nijmegen, Netherlands, [email protected],Daniel Kuehlwein, Tom de Ruijter, Twan van Laarhoven,Josef Urban, Tom HeskesLearning reasoning techniques from previous knowledge is a largelyunderdeveloped area of automated reasoning. As large bodies of for-mal mathematical libraries are becoming available, the state-of-the-artpreference learning and ranking methods can help us to address a cor-nerstone task for automated theorem provers, namely premise selec-tion and ranking. In this talk I will briefly describe our recent resultson development and application of the preference learning and rankingalgorithms to the automated reasoning domain.

4 - Learning from preferences in the context of Case-Based Reasoning (CBR)Patrice Schlegel, Mathematics and Computer Science,Philipps-Universität Marburg, Hans-Meerwein-Straße 6,35032, Marburg, Hesse, Germany,[email protected] research has shown the potential of exploiting knowledge con-tained in preference data. Our goal is to apply this concept in the fieldof Case-Based Reasoning, where it complements the underlying rea-soning paradigm of solving problems by considering the stored solu-tions of similar problems. We aim at liberating the method from therigidity and limitations of the problem-solution tuple by replacing thesolution with preferences over a space of candidate solutions. Thereby,it becomes more flexible while accommodating for the uncertain andapproximate nature of case-based problem solving.

� WA-41Wednesday, 8:30-10:00HH-Heyerdahl

Young scientists doing research in OR

Stream: Young People for System Theory, Optimizationand EducationInvited sessionChair: Alexis Pasichny, Students’ Science Association, NationalTechnical University of Ukraine, 37, Peremohy av., 03056, Kiev,Ukraine, [email protected]

1 - Heteroscedastic processes in finance modellingBohdan Pukalskyi, Students Science Association, NationalTechnical University of Ukraine, Kovalskyi provulok 5,03057, Kyiv, Ukraine, [email protected] exists a big amount of different models to explain financial pro-cesses. Heteroscedastic modeling has important place among them.This paper analyses different heteroscedastic models in application tomodern stock price forecasting to find the most adequate models forbetter understanding of the market situation.

2 - Comparative analysis of methods for prediction of fi-nance and economic processes.Yaroslav Shevchenko, Institute of Applied System Analysis,National Technical University of Ukraine, Kovalsky lane 5,03085, Kyiv, Ukraine, [email protected] prediction of nonlinear nonstationary processes a number of difficul-ties appears in selecting mathematical models relying on the quality ofpredictions.This paper makes a comparative analysis of several math-ematical models for price formation on stock exchange, so we’ll findthe most appropriate model from proposed according to the quality ofprediction.

3 - Mathematical methods for regional sustainable de-velopment modeling based on SD measurement met-ricsAlexej Orlov, Students Science Association, Institute forapplied system analysis of the National Technical Universityof Ukraine, Peremohy ave. 37, 03056, Kyiv, Ukraine,[email protected] of a system of factors (indices and indicators, based on expertand statistical data) for measuring the sustainable development allowsturning the sustainability study to the practical tool which is used instrategic planning and decision-making. Sustainable development ona regional level is one of urgent issues in governmental policies in de-veloped countries. There are different metrics used for sustainable re-gional development estimation and modeling in the focus of this talk;some cases of their application are considered and analyzed as well.

� WA-42Wednesday, 8:30-10:00BW-Amber

Robust Optimization and RandomizedMethodsStream: Stochastic Modeling and Simulation in Engi-neering, Management and ScienceInvited sessionChair: Erik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39, 85577,Neubiberg, Germany, [email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]: Ayse Özmen, Scientific Computing, Institute of AppliedMathematics, Middle East Technical University, KazimkarabekirMah. 2048. Sok No:5/5, Etimesgut, 06790, Ankara, Turkey,[email protected]

1 - A robust optimization approach to production plan-ning under non-compliance risksBan Kawas, IBM Research – Zurich, 8803, Rueschlikon,Zurich, Switzerland, [email protected], MarcoLaumanns, Eleni PratsiniIn various industries, production companies face compliance risks asgovernmental authorities enforce good manufacturing practices for theprotection of consumers and the environment. As failing inspectionscan be costly, companies should measure and manage their exposureto non-compliance risks. We study an adversarial decision model thataddresses these risks and aims to find a strategy reflecting a suitablerisk-return trade-off. Within the model, data uncertainty is addressedusing robust optimization techniques, and a compact formulation as arobust MIP is derived.

191


2 - Fast sample average approximation for minimizingConditional-Value-at-RiskEduardo Moreno, Faculty of Engineering and Sciences,Universidad Adolfo Ibañez, Avda. Diagonal Las Torres 2640Of:532C, Peñalolen, Santiago, Chile,[email protected], Daniel Espinoza

Sample Average Approximation approach allows to solve stochasticoptimization problems, like minimizing CVaR, using linear program-ming. However, the number of samples required to get good approx-imated solutions tends to be very large. In this work we propose anautomatic aggregation scheme to exactly solve linear programs withCVaR objective function using sample average approximations with avery large number of scenarios. Our results shows that this aggrega-tion scheme is in average between 3-100 times faster than the standardLP-equivalent formulation.

3 - Stochastic optimization of nonlinear problems withconstraints and differentiable functions: The waitand see methodKherchi Hanya, ENSSEA, 16000, Alger, Algeria,[email protected]

It’s a matter to do a stochastic optimization of nonlinear problems withconstraints and differentiable functions by using a passive method,namely, the method of wait and see based on the work established instochastic linear programming. This technique consists in studying thebehavior of the optimum as the random effect of parameters of the ob-jective function. We will try to establish the existence of a generalprobabilistic law managing the behavior of the optimum, and then wecan have the necessary information such as the mean and variance.

4 - Randomized Clustering: Weakening Necessary Con-ditions for the Confidence Interval of the True Num-ber of ClustersMikhail Morozkov, Sankt-Petersburg State University,198504, St. Petersburg, Russian Federation,[email protected], Oleg Granichin, Zeev (Vladimir)Volkovich

Finding the true number of clusters in a data set has proved to be oneof the most challenging problems in cluster analysis. In this paper wedescribe approach where the true number of groups is picked as theslope discontinuity of the underlying index function. Randomized al-gorithm is proposed to find confidence interval for such point. We pro-vide relaxed conditions for necessary conditions to find correspondinginterval with a priori fixed level of confidence. The ideas of scenarioapproach are used to significantly reduce computational intensity. Sev-eral simulation examples are given.

� WA-43Wednesday, 8:30-10:00BW-Granite

Stochastic programming in industry I

Stream: Stochastic ProgrammingInvited sessionChair: Laureano Fernando Escudero, Dept. de Estadística eInvestigación Operativa, Universidad Rey Juan Carlos, c/Tulipan,S/n, 28933, Mostoles (Madreid), Spain,[email protected]

1 - Risk averse measures in Stochastic mixed 0-1 Opti-mizationLarraitz Aranburu, Economía Aplicada III, UPV/EHU, SanBartolomé 28, 1.A, 48300, Gerniika-Lumo, Bizkaia, Spain,[email protected], Laureano Fernando Escudero,María Araceli Garín, Gloria Perez, Gloria Perez

We present several risk averse measures for stochastic mixed 0-1 op-timization as alternatives to the objective function expected value op-timization. Most of the proposed strategies are extensions of the twostage case studied in the literature to the multistage environment. Ad-ditionally, two innovative multistage strategies are proposed: a mix-ture of CVaR and first-order SDC strategies and a mixture of CVaR

and second-order SDC strategies. As a pilot case the optimization ofthe immunization strategy in fixed-income security portfolios is con-sidered under some sources of uncertainty.

2 - Scenario cluster partitioning in the Lagrangian basedproceduresAitziber Unzueta, Applied Economy III, UPV/EHU,Lehendakari Aguirre 83, 48015, Bilbao, Spain,[email protected], Laureano Fernando Escudero, MaríaAraceli Garín, Gloria Perez

A scenario Cluster based Lagrangian Decomposition scheme for ob-taining strong lower bounds to the optimal solution of two-stagestochastic mixed 0-1 problems is introduced. The aim consists of re-solving the Lagrangian dual problem via solving scenario cluster sub-models once the nonanticipativity constraints have been dualized. Wereport some computational experience to compare the performance ofseveral Lagrange multiplier updating schemes for different number ofscenario clusters. The results show that the cluster partitioning outper-forms the traditional scenario partitioning.

3 - Parallel computing via break stage scenario cluster-ing for multistage stochastic programmingGloria Perez, Applied Mathematics and Statistics andOperational Research, Universidad del País Vasco, Scienceand Technology Faculty, Barrio Sarriena s/n, 48940, Leioa,Spain, Spain, [email protected], María Merino, LaureanoFernando Escudero, María Araceli Garín, Unai Aldasoro

Break Stage Scenario Clustering is introduced for decomposing mixedinteger optimization problems, where nonanticipativity constraints arepartially relaxed. We analyze the effect of using parallel programmingin several strategies. Main results are based on our BFC-MS procedure.We have implemented a C++ code that uses the CPLEX optimizerwithin COIN-OR for solving the independent MIP submodels withMPI. The experience shows the speedup obtained in the computationalcluster ARINA provided by the SGI/IZO-SGIker at the UPV/EHU.

4 - Vehicle Routing with Soft Time Windows andStochastic Travel Times: A Column Generation andBranch-and-Price Solution ApproachMichel Gendreau, MAGI and CIRRELT, ÉcolePolytechnique, C.P. 6079, succ. Centre-ville, H3C 3A7,Montreal, Quebec, Canada, [email protected],Duygu Tas, Nico Dellaert, Tom Van Woensel, Ton de Kok

We study a vehicle routing problem with stochastic travel times. Foreach customer, a soft time window allows early and late servicing.The objective is to minimize the sum of transportation costs (total dis-tance traveled, number of vehicles used and total expected overtime ofdrivers) and service costs (penalties for time-window violations). Weapply a column generation procedure. The master problem is a clas-sical set partitioning problem. The pricing subproblem corresponds toan elementary shortest path problem with resource constraints. Integersolutions are obtained by branch-and-price.

� WA-44Wednesday, 8:30-10:00BW-Marble

Data Analysis and Its Applications

Stream: Machine Learning and its ApplicationsInvited sessionChair: Michael Khachay, Ural Branch of RAS, Institute ofMathematics and Mechanics, S.Kovalevskoy, 16, 620990,Ekaterinburg, Russian Federation, [email protected]

1 - Forecasting algorithm for short time series with pre-liminary classificationIrina Yatskiv, Computer Science, Transport andTelecommunication institute, Lomonosova 1, LV-1019, Riga,Latvia, [email protected]

192


Algorithm for short time series forecasting on the basis of ARIMAmodel and classification are offered. A set of time series with lengthsthat is enough for forecast are grouped in homogeneous classes. The"time series-centroid" (TSC) for each class is determined and the esti-mation of parameters of ARIMA model has been performed for it. Theshort time series refer to one of the classes using the Mahalonobis dis-tance. Then, a forecast on the basis a model for the TSC of determinedclass is built and then corrected by application the distance betweenTSC and short time series.

2 - A robust algorithm for sequential ANOVA with incom-plete dataCarmen Anido, Economic Analysis: Quantitatve Economy,Autonoma University of Madrid, Faculty of Economics,Cantoblanco., 28049, Madrid, Spain, [email protected],Teofilo Valdes

We present an algorithm for ANOVA analysis of data under the follow-ing robust conditions: (a) the data is not received in batch but sequen-tially, (b) the data may be imprecise, (c) the error distributions mayvary within the class of the strongly unimodal distributions and (d) thevariance of the errors is unknown. The algorithm updates the analysisas soon as a new data is received. This updating is simple, needs littlecomputational requirements, yields the estimates of the effect param-eters and ends up estimating the asymptotic covariance matrix of theparameter estimates.

3 - A clustering method based on Independent Compo-nent AnalysisTakashi Onoda, System Engineering Lab., CRIEPI, 2-11-1,Iwado Kita, Komae-shi, 201-8511, Tokyo, Japan,[email protected]

The clustering result in k-means clustering method depends heavily onthe chosen initial clustering centers. We propose a clustering methodbased on the independent component analysis. The proposed methoddoes not need initial clustering centers. We evaluate the performanceof our proposed method and compare it with some clustering methodsby using artificial datasets and benchmark datasets. The experimentsshow that the normalized mutual information of our proposed methodis better than the normalized mutual information of k-means clusteringmethod, KKZ method and k-means++ clustering method.

4 - Dissimilarity Based on Cluster Discernibility on At-tribute SubsetsYoshifumi Kusunoki, Osaka University, Japan,[email protected], Tetsuzo Tanino

Clustering is to classify given objects into clusters using their at-tributes. In this research, we focus on clustering for objects with nom-inal attributes. For such objects, logical expressions such as "attributea1 equals to v" are suitable to describe features of clusters. However,clustering methods using dissimilarity for object pairs do not necessar-ily output clusters having simple and compact logical expressions. So,we propose dissimilarity using cluster discernibility on attribute sub-sets. We apply the proposed dissimilarity to clustering, and examinethem by numerical experiments.

� WA-45Wednesday, 8:30-10:00BW-Water

Non-Airline Revenue ManagementApplications

Stream: Advanced Inventory Control and PricingStrategiesInvited sessionChair: Jochen Gönsch, Department of Analytics & Optimization,University of Augsburg, Universitätsstraße 16, D-86159,Augsburg, Germany, [email protected]

1 - Network capacity control and road-based freighttransportationJörn Schönberger, University of Bremen, 28359, Bremen,Germany, [email protected]

Network capacity control (NCC) tools identify the most profitable pro-posals where the profitability is determined by incorporating a forecastabout the expected remaining demand. Inaccurate demand forecastscompromise the transfer of NCC tools from the airline industry into theroad haulage sector. However, road haulage product portfolios consistto a large part of flexible products with two or more execution modes.We investigate if the large number of flexible products contributes tothe compensation of inaccurate demand forecasts and enable the re-useof NCC tools in road haulage.

2 - Operational capacity management in the car rentalindustryClaudius Steinhardt, Department of Analytics &Optimization, University of Augsburg, Universitätsstraße 16,86159, Augsburg, Germany,[email protected], Jochen Gönsch

Car rental companies are often able to spontaneously adjust their fleetsize within a certain range in order to match market needs. We presentan approach that integrates these fleet decisions with revenue manage-ment’s capacity control. To decide on the fleet size, we propose anLP-formulation that is used in a hierarchical planning context witha rolling horizon, allowing for the combination with arbitrary ap-proaches of standard capacity control. To reveal the potential of theapproach, we conduct extensive simulation studies based on real-worlddata from a European car rental company.

3 - Dynamic pricing and efficient management of instal-lation and maintenance resourcesRupal Rana, Business School, Loughborough University,Epinal Way, Loughborough, United Kingdom,[email protected]

We consider a problem faced by many companies that use a fixed num-ber of employees to provide two services an installation service and amaintenance service. We propose to join resources of employees anduse dynamic prices for installations in accordance to the demand formaintenance, and we explore, e.g., how changes in various problemparameters affect the optimal policy structure. We investigate the eco-nomic benefits of a joint strategy and find that dynamic pricing andcapacity flexibility works best.

4 - An EMSR approach for revenue management withplanned upgradesJochen Gönsch, Department of Analytics & Optimization,University of Augsburg, Universitätsstraße 16, D-86159,Augsburg, Germany, [email protected],Sebastian Koch, Claudius Steinhardt

We consider the capacity control problem with integrated upgrading.The optimal dynamic program for this problem is hard to solve evenin the single-leg case, as multiple hierarchical resource types must beconsidered simultaneously. Therefore, we propose a new heuristic ap-proach that generalizes the well-known EMSR-a procedure. A simula-tion study compares the performance to other capacity control methodsincorporating upgrades. The results show that our approach signifi-cantly outperforms existing methods including the successive planningapproaches widely used in commercial RM systems.

193

WB-01 EURO 25 - Vilnius 2012

Wednesday, 10:30-12h00

� WB-01Wednesday, 10:30-12h00RB-Alfa

Keynote Lecture: Professor MatteoFischettiStream: Keynote, Tutorial and Plenary LecturesKeynote sessionChair: Elena Fernandez, Universitat Politecnica de CAtalunya,Barcelona, Spain, [email protected]

1 - On the role of randomness in exact tree search meth-odsMatteo Fischetti, DEI, University of Padua, via Gradenigo6/a, 35100, Padova, Italy, Italy, [email protected], MicheleMonaci

High-sensitivity to initial conditions is generally viewed as a drawbackof tree search methods, as it leads to an erratic behavior to be mitigatedsomehow. In this talk we address the opposite viewpoint. Our workinghypothesis is that erraticism is just a consequence of the exponentialnature of tree search, that acts as a chaotic amplifier, so it is largelyunavoidable. We discuss possible ways to actually take advantage ofrandomness, and present computational experiments on a testbed ofdifficult MILP instances.

� WB-02Wednesday, 10:30-12h00RB-Beta

Preemption, Lower Bounds andRobustness in Project Scheduling

Stream: Project Management and SchedulingInvited sessionChair: Christoph Schwindt, Institute of Management andEconomics, Clausthal University of Technology, Julius-Albert-Str.2, 38678, Clausthal-Zellerfeld, Germany,[email protected]

1 - Incorporating costs in the RCPSP with pre-emptionFrancisco Ballestin, Department of Statistics and OR, PublicUniversity of Navarra, Campus Arrosadia, S/N, 31006,Pamplona, Spain, [email protected]

The classical models of project scheduling problems suppose that ac-tivities are non-preemptable. However, some project environments al-low this assumption to be relaxed. Almost all published papers pre-suppose that activities can be restarted at no additional cost, but thisis not true in some practical situations. We define new models thatincorporate costs in the RCPSP with preemption. The cost might beof different types; this paper focuses in what we think are the mostplausible and interesting ones. We offer algorithms capable of offeringquality solutions for each of the proposed models.

2 - Lower and upper bounds for preemptive projectscheduling with generalized precedence relation-shipsChristoph Schwindt, Institute of Management and Economics,Clausthal University of Technology, Julius-Albert-Str. 2,38678, Clausthal-Zellerfeld, Germany,[email protected], Tobias Haselmann

We study a project scheduling problem where the activities can be in-terrupted at any point in time and generalized precedence relationshipsbetween the activities have to be taken into account. After providing aformal problem statement, we introduce a novel MILP formulation andan optimization model combining a large-scale linear program with a

single machine scheduling problem. Moreover, we report on the resultsof an experimental performance analysis comparing upper bounds ob-tained with the MILP model and lower bounds that arise from solvingthe linear program by column generation.

3 - Lower bounds for a Fixed Job Scheduling Problemwith an equity objective functionDamien Prot, Automatique et Productique, Ecole des Minesde Nantes, 4, rue Alfred Kastler, 44307, Nantes Cedex 3,France, [email protected], Tanguy Lapegue,Odile Bellenguez-Morineau

We are interested in a problem that consists in assigning a set of tasksto a set of workers with specific skills. Given a targeted working loadfor each worker, we want to maximize the fairness between workers;more precisely, our objective is to minimize the difference between thehighest and the lowest worker’s gap value, which is defined as the dif-ference between the targeted working load and the real working load.We present specific lower bounds for this problem and compare themwith the one obtained by solving an integer linear program with anindustrial software.

4 - Increasing schedule robustness by task groupingMichel Wilson, Software Technology, Delft University ofTechnology, Mekelweg 4, 2628 CD, Delft, Netherlands,[email protected], Cees Witteveen

An important problem in scheduling is ensuring robustness of solutionsin case of execution delays. We propose a new method, task grouping,and apply it to resource-constrained project scheduling. Using thismethod tasks that must be executed sequentially can be grouped, buttheir definitive order is determined at execution time such that delayscan sometimes be mitigated. As a consequence, our method gener-ates a set of execution options for a schedule. Using the well-knownPSPLIB instances, we show that our method can reduce the impact ofdelays on the execution of the schedule significantly.

� WB-03Wednesday, 10:30-12h00RB-L1


Stream: EURO/ROADEF ChallengeInvited sessionChair: Christian Artigues, LAAS, CNRS, 7 avenue du ColonelRoche, 31077, Toulouse Cedex 4, France, [email protected]: Ender Özcan, Computer Science, University ofNottingham, Jubilee Campus, Wollaton Road, NG8 1BB,Nottingham, United Kingdom, [email protected]: Emmanuel Guere, Google, 38 Avenue de l’Opera, 75002,Paris, France, [email protected]: Kedad-Sidhoum Safia, LIP6 - UPMC, 4 Place Jussieu,75005, Paris, France, [email protected]: H. Murat Afsar, Industrial Systems, University ofTechnology of Troyes, 12 rue Marie Curie, BP 2060, 10010,Troyes, France, [email protected]: Eric Bourreau, COCONUT, LIRMM, 161 Rue Ada, 34000,Montpellier, France, [email protected]: Marc Sevaux, Lab-STICC - CNRS UMR 3192, Universitéde Bretagne Sud - UEB, Centre de Recherche - BP 92116, 2 rue deSaint Maude, 56321, Lorient, France, [email protected]

1 - Google/Roadef Challenge: a 100-line LocalSolvermodel model qualifies for the final roundRomain Megel, e-lab, Bouygues SA, 75008, Paris, France,[email protected], Bertrand Estellon, Thierry Benoist,Julien Darlay, Frédéric Gardi, Karim Nouioua

The machine reassignment problem proposed by Google is a verylarge-scale optimization problem. We solved it with LocalSolver,the first math programming software combining the simplicity of useof a model-and-run solver and the power of local-search techniquesfor combinatorial optimization (www.localsolver.com). In practice it

194

EURO 25 - Vilnius 2012 WB-05

means that we merely translated the problem definition into a straight-forward 0-1 model and then launched LocalSolver on this model.Ranked 25th among 82 teams from 33 countries, LocalSolver is thesole math programming solver qualified for the final round.

2 - Adaptive Local Search for Google Machine Reassign-ment problemEmir Demirovic, Mathematics, Faculty of Natural Sciences,Bosnia And Herzegovina, [email protected],Mirsad Buljubasic, Haris Gavranovic

We present a hybrid method to efficiently solve Google Machine Re-assignment problem. We study, implement, combine and empiricallyexamine different local search neighborhoods to solve the set of avail-able instances. We present results obtained with the solver respectingthe given computational time of 5 minutes. The solutions obtained withlonger runs and/or with relaxed constraints will also be reported. Wealso develop a decomposition method for the given problem improv-ing overall numerical efficiency of the method. Some of the obtainedresults are proven to be optimal or near optimal.

3 - A Hybrid Large Neighborhood and Local Search forthe Machine Reassignment ProblemThibaut Vidal, CIRRELT - Université de Montréal &ICD-LOSI - Université de Technologie de Troyes, 12, rueMarie Curie, BP 2060, cedex, 10010, TROYES, France,[email protected], Hugues Dubedout, Renaud Masson,Julien Michallet, Puca Penna, Vinicius Petrucci, AnandSubramanian

A hybrid metaheuristic combining Local Search (LS), Large Neigh-borhood Search (LNS) and integer programming is proposed for themachine reassignment problem. The LS performs relocations and ex-changes of processes with similar resource consumptions and depen-dency requirements among machines in the cluster. The LNS itera-tively and adaptively selects a subset of decision variables whose op-timized values are given by a compact integer programming formu-lation. The method produces competitive results on the first set ofinstances provided by the ROADEF 2012 challenge organizers.


New Ideas in Integer Programming

Stream: Combinatorial OptimizationInvited sessionChair: Francois Soumis, GERAD, 3000 Cote Ste-Catherine, H3T2A7, Montreal, Québec, Canada, [email protected]

1 - A fast and accurate algorithm for stochastic integerprogramming, applied to stochastic shift schedulingRémi Pacqueau, GERAD, 3000 chemin de la Cote SainteCatherine, H3T 2A7, Montreal, Quebec, Canada,[email protected], Francois Soumis, LeHoang

Stochastic programming can yield significant savings over determin-istic approaches. However, stochastic approaches always lead to verylarge problems, since recourse must be computed for every scenario.There is no fast and exact method for solving such problems. In thisarticle, the L-shaped method is improved in two ways: a Benders cutsdynamic management algorithm for the master problem and a mul-tithreaded implementation to solve the subproblems. Those two im-provements yield a heuristic able to solve a 10 million variables IPproblem in less than 5 minutes, with a very good accuracy, and enablethe resolution of larger instances.

2 - Correction of an infeasible mixed integer linear pro-gramming problem towards feasibilityDmitry Borodin, Optimization, OM partners, Belgium,[email protected], Viktor Gorelik

This research provides a practical method for solving infeasible mixed-integer programming models by correcting their constraints. The ini-tial infeasible MILP is decomposed into two subproblems finally yield-ing to the correction problem which is deduced to an LP in case ofusing the uniform correction criterion. The solution of the correctedMILP shows how the constraints (e.g. resources) of the initial infeasi-ble model can be changed in a minimal way to make the model feasiblethat can mean a technological update for the modeled production pro-cess (e.g. energy consumption reduction).

3 - Integral simplex using decompositionFrancois Soumis, GERAD, 3000 Cote Ste-Catherine, H3T2A7, Montreal, Québec, Canada, [email protected],Issmail Elhallaoui, Zaghrouti AbdelouahabSince the early ’70s, several authors proposed, without much success,adaptations of the simplex algorithm to reach an optimal integer solu-tion to the set-partitioning problem, with a sequence of basic integersolutions. We present an algorithm that iteratively finds improving in-teger solutions up to optimality by solving small subproblems. Theo-retical background and experimental results on problems up to 1 600constraints and 500 000 variables are presented. Larger problems aresolved in time between 10 and 20 minutes.

4 - A hybrid solution to configurate H.264/AVC videoCODEC for the Brazilian Digital TVIris Linck, University of the Sinos Valley, Av. Unisinos, 950,93022000, Sao Leopoldo, Rio Grande do Sul, Brazil,[email protected], Arthur GomezThis article presents a study focused on finding good solutions to im-prove the H.264/AVC CODEC configuration for encoding video tothe Brazilian Digital Television System (DBTS). A hybrid algorithmcalled Simulator of Metaheuristics applied to a CODEC (SMC) wasdeveloped and it utilizes the concepts of Tabu Search and Genetic Al-gorithm. SMC will search for the best CODEC configuration by usingsix parameters: bitrate, framerate, quantization parameters of the B, Pand I slices; and the number of B slice in a Group Of Pictures. Thegoal is to get a compressed video with a good picture quality.


Topics in combinatorial opimization II

Stream: Discrete Optimization, Geometry & Graphs(contributed)Contributed sessionChair: Robert Manger, Department of Mathematics, University ofZagreb, Bijenicka cesta 30, 10000, Zagreb, Croatia,[email protected]

1 - Fixing rules, Reduction and Optimal resolution forthe Knapsack Problem with SetupsAbdelkader Sbihi, Axe Logistique-Terre-Mer-Risque, Ecolede Management de Normandie, Le Havre, France, 30, rue deRichelieu, 76087, Le Havre Cedex, France,[email protected] knapsack problem with setups (KPS) is a non standard hard vari-ant of the knapsack problem with setups costs and constraints and withmany real-life applications. We present a two-phase optimal approachfor KPS: (i) the first one is a reduction based phase obtained by fix-ing the variables to their optimal values in the solution via upper/lowerbounding rules and (ii) in the second phase, we apply an exact methodfor the reduced problem composed by the remaining free variables. Weconsidered both B&B and DP and we give a comparison study of thetwo variants of the approach.

2 - Adding an Edge between Two Levels of a CompleteK-ary Linking Pin Structure Minimizing Total Dis-tanceKiyoshi Sawada, Department of Policy Studies, University ofMarketing and Distribution Sciences, 3-1,Gakuen-nishi-machi, Nishi-ku, 651-2188, Kobe, Japan,[email protected]

195


This study proposes a model of adding relation to an organizationstructure which is a complete K-ary linking pin structure where everypair of siblings in a complete K-ary tree of height H is adjacent suchthat the communication of information in the organization becomes themost efficient. When an edge between a node with a depth M and itsdescendant with a depth N is added, an optimal pair of depth (M,N)*is obtained by minimizing the total distance which is the sum of thenumber of edges in shortest paths between every pair of all nodes inthe complete K-ary linking pin structure.

3 - Minimizing the costs of evacuation paths by decom-posing network flowsJan Peter Ohst, Mathematics, TU Kaiserslautern, Germany,[email protected], Stefan Ruzika

Many models and algorithms for large scale routing and flow optimiza-tion are known. The solution of such problems is naturally a flow on anetwork. However to implement this flow a decomposition into pathsis required. While it is straight forward to find an arbitrary decompo-sition the choice of the decomposition can have a tremendous effecton the properties of the single paths. The problem of finding optimaldecompositions is NP-hard but has not been investigated in detail. Aninteger program for such problems as well as an approximation algo-rithm on series-parallel graphs are presented.

4 - Implementing the work function algorithm by net-work flows and flow cost reductionRobert Manger, Department of Mathematics, University ofZagreb, Bijenicka cesta 30, 10000, Zagreb, Croatia,[email protected], Tomislav Rudec

This paper deals with the work function algorithm (WFA) for solv-ing the on-line k-server problem. The WFA is usually implementedby network flow techniques including the flow augmentation method.In the paper we develop a new implementation of the WFA, which isagain based on network flows, but uses simpler networks and the costreduction method. We also present experiments showing that our im-plementation is considerably faster than the conventional one.

� WB-06Wednesday, 10:30-12h00RB-Gamma

Industrial Organization

Stream: Dynamical Systems and Game TheoryInvited sessionChair: Luis Ferreira, Matemática, Escola Superior de EstudosIndustriais e de Gestão, IPP, ESEIG, Rua D Sancho I, número 981,4480-876, Vila do Conde, Porto, Portugal,[email protected]

1 - Flexibility in a Stackelberg leadershipLuis Ferreira, Matemática, Escola Superior de EstudosIndustriais e de Gestão, IPP, ESEIG, Rua D Sancho I, número981, 4480-876, Vila do Conde, Porto, Portugal,[email protected]

We study the effects of product differentiation in a Stackelberg modelwith demand uncertainty, only for the first mover. We do an ex-anteand ex-post analysis, in terms of product differentiation and of the de-mand uncertainty, of the profits of the leader and of the follower firms.We show that even with small uncertainty about the demand, the fol-lower firm can achieve greater profits than the leader, if their productsare sufficiently differentiated.

2 - Parallel vs. Sequential Interaction Protocols for theMultilateral Negotiations in Distributed, Multi-agentEnvironmentPiotr Palka, The Faculty of Electronics and InformationTechnology, Warsaw University of Technology, ul.Nowowiejska 15/19, 00-665, Warsaw, mazowieckie, Poland,[email protected]

In e-commerce complex markets there are numerous limitations andother ties between market participants and objects of trade. Frequentlyoccurring constraints arising from a structure of the problem, as wellas preferences of participants, cause inability to reach a satisfactoryoutcome using bilateral negotiations. The solution is either the tradeon centralized auctions or the multilaterally negotiated agreement. Inthe paper we consider two interaction protocols for the multilateral ne-gotiation. One is based on the sequential message passing, the otherassumes parallel communication.

3 - Symmetric rendezvous search on a hexagonJohn Howard, Operational Research, London School ofEconomics, Houghton Street„ London, WC2A 2AE, UnitedKingdom, [email protected]

In symmetric rendezvous search on a hexagon two players are initiallyplaced at two different vertices two edges apart. Each period they mustmove to one of the adjacent vertices. They try to meet in the leastpossible expected time using the same (mixed) strategy.

This problem is unsolved, so we modify it by giving the players extrainformation. We are then able to solve the modified problem, therebygiving a lower bound to the original problem. The solution requiresan interesting analysis of the behaviour of two intertwined discrete dy-namical systems.

4 - Lorenz and lexicographic maximal allocations forbankruptcy problemsJavier Arin, Economic Theory, University of the BasqueCountry, L. Agirre 83, 48015, Bilbao, Spain,[email protected]

A bankruptcy problem consists of a set of claimants who must dividebetween them an infinitely divisible good, the endowment, that is notsufficient to satisfy their claims in full. We introduce egalitarian crite-ria to solve bankruptcy problems. We focus on two egalitarian criteria:the Lorenz and lexicographic criteria. Talmud Rule is one of the mostimportant sharing rules. We prove that this rule is also a lexicographicmaximizer. This analysis allows a new sharing rule: Lexmax Rule. Wealso characterize the sets of Lorenz maximal vectors of awards/losses.

� WB-08Wednesday, 10:30-12h00RB-Epsilon

Boolean Optimization in Graph Theory

Stream: Boolean and Pseudo-Boolean OptimizationInvited sessionChair: Martin Milanic, UP IAM and UP FAMNIT, University ofPrimorska, Muzejski trg 2, 6000, Koper, Slovenia,[email protected]

1 - Forbidden induced subgraph characterizations ofgraph classesGuillermo Durán, University of Buenos Aires, Argentina,[email protected]

In this talk we present several results about partial characterizations ofdifferent graph classes by forbidden induced subgraphs. First, we ana-lyze three classes related to perfect graphs: clique-perfect, coordinatedand balanced graphs. Furthermore, we study two important classes ofintersection graphs: circular-arc and circle graphs. We will show themain results on these topics obtained for our group in the last 10 years.Joint works with F. Bonomo, M. Chudnovsky, L. Grippo, M. Groshaus,M.Lin, M. Safe, F. Soulignac, G. Sueiro, J. Szwarcfiter and A. Wagler.

2 - A generalization of the P4-free graph sandwich prob-lem to multigraphsEndre Boros, RUTCOR, Rutgers University, 08854,Piscataway, New Jersey, United States,[email protected], Vladimir Gurvich

An n-multigraph is an n-edge colored complete graph. D is a 3 edgecolored triangle, and P is a 2 colored complete graph on 4 vertices, suchthat each color defines a P4. In this paper we show a polynomial timealgorithm to decide if a given n-multigraph contains a P- and D-free

196


subgraph. This result generalizes the P4-free graph sandwich problemintroduced by Golumbic, Kaplan, and Shamir in 1995. As motivationand application, we consider n-person positional game forms. It isknown that such game forms are in a one-to-one correspondence withP- and D-free n-graphs.

3 - The Price of Connectivity for Vertex CoverOliver Schaudt, Institut für Informatik, AFS, Universität zuKöln, Weyertal 80, 50931, Cologne, Germany,[email protected]

We study a graph parameter called the Price of Connectivity for Ver-tex Cover (PoC). It was recently introduced by Cardinal et al. and isdefined as the ratio of the minimum size of a connected vertex coverand the minimum size of a vertex cover. We prove structural resultsfor the PoC. These results concern PoC-critical, PoC-perfect and PoC-near-perfect graphs.

4 - Equistable graphs: conjectures, results, and con-nections with Boolean functionsMartin Milanic, UP IAM and UP FAMNIT, University ofPrimorska, Muzejski trg 2, 6000, Koper, Slovenia,[email protected]

Equistable graphs form a graph class generalizing the well known classof threshold graphs, and are closely related to general partition graphsand triangle graphs. No combinatorial characterization of equistablegraphs is known, and the complexity status of recognizing equistablegraphs is open. In this talk, we will present an overview of resultsand conjectures on equistable graphs and related classes. Wheneverapplicable, connections to corresponding notions from the theory ofBoolean functions will be discussed. Joint works with V. Levit, S.Miklavic, J. Orlin, G. Rudolf and D. Tankus.

� WB-09Wednesday, 10:30-12h00RB-Zeta

Optimization Problems on Graphs

Stream: Graphs and NetworksInvited sessionChair: Mirjana Cangalovic, Laboratory for Operational Research,Faculty of Organizational Sciences, University of Belgrade, JoveIlica 154, Belgrade, Serbia, Serbia, [email protected]

1 - An algorithm for finding the most probable cut setMirko Vujosevic, Faculty of Organizational Sciences,University of Belgrade, ul. Jove Ilica 154, 11000, Belgrade,Serbia, [email protected], Dragana Makajic-Nikolic,Nebojsa Nikolic

A new approach for finding the most probable cut set is proposed. Thestructure of the system is firstly modeled as a fault tree with randomprimary events as leafs, intermediate events as logic functions of pri-mary events, and the top event in the root. The top event is also a logicfunction of the primary events. A cut set is a set of primary eventswhich cause the occurrence of the top event. Given probabilities ofthe primary events, the problem is to find the most probable cut set.An equivalent reverse Petri net is made and its properties are used fordeveloping a new algorithm.

2 - Strong metric dimension of generalized Petersengraphs GP(n,1)Mirjana Cangalovic, Laboratory for Operational Research,Faculty of Organizational Sciences, University of Belgrade,Jove Ilica 154, Belgrade, Serbia, Serbia, [email protected],Jozef Kratica, Vera Kovacevic Vujcic

Generalized Petersen graphs GP(n,k) were introduced by Coxeter in1950. Recently, the metric dimension and the minimal doubly re-solving set of GP(n,k) have been investigated. In this paper we con-sider a related problem of determining the strong metric dimension forGP(n,1) and prove that it is equal to n.

3 - A Set Covering Approach to Solving HeterogeneousVehicle Routing ProblemMilan Stanojevic, Faculty of Organizational Sciences,University of Belgrade, Jove Ilica 154, 11000, Belgrade,Serbia, [email protected], Bogdana Stanojevic

We consider a vehicle routing problem with limited heterogeneous fleetof vehicles having various capacities, fixed costs and variable costs.Our approach provides approximate solution. It consists of two phases.In first phase we generate set of "good" routes for all types of vehiclesfrom the fleet. For this step we use randomized enhanced savings al-gorithm. In second phase we solve to optimality set covering problemusing routes from the first phase. Varing number of generated routeswe can trade between solving time and solution quality. Computationresults are presented.

4 - On the dominating tree problemAndré Rossi, Lab-STICC - UMR 3192, Université deBretagne-Sud, Centre de Recherche, BP 92116, 56321,Lorient, France, [email protected]

The problem under consideration is called the dominating tree prob-lem. Let G be an edge weighted, simple, connected and undirectedgraph. We want to build a tree of minimum cost on G such that eachvertex is either covered by the tree, or is adjacent to a vertex that iscovered by the tree. The cost of every edge is assumed to be strictlypositive. We provide a simple proof of NP-hardness, and a series ofcases where the problem can be solved in polynomial time. In addi-tion, we present a cutting plane approach to this problem.

� WB-10Wednesday, 10:30-12h00RB-Theta

Supply Chain Management 1

Stream: Production Management & Supply ChainManagement (contributed)Contributed sessionChair: Rifat Gürcan Özdemir, Industrial Engineering Department,Istanbul Kültür University, Atakoy Campus, Atakoy-Bakirkoy,34156, Istanbul, Turkey, [email protected]

1 - Managing and Measuring Business PerformanceOtilija Sedlak, Quantitative Methods in Economics, Faculty ofEconomics Subotica, Ivana Sarica 14., Segedinski put 9-11,24000, Subotica, Vojvodina, Serbia, [email protected],Zoran Ciric

Project performance has become a significant issue, especially as poorperformance in high profile information systems projects. The busi-ness management critical success factors: cost, quality, time, customerexpectations. In this paper will be evaluated an integrative frameworkof performance for managing and measuring to bring together numer-ous and evolving strands. We believe it is important to investigatewhether innovation theories can be generalized and empirical findingsare applicable in different economic contexts.

2 - Optimization in supply chain decision makingMarija Cileg, Quantitative Methods in Economy, Faculty ofEconomics, Segedinski put 9-11, 24000, Subotica, Serbia,[email protected], Tibor Kis

Performance improvement of companies participating in supply chainsis necessary to achieve harmonization, co-operation, coordination andefficiency. Partial optimization can be implemented in every stage ofthe chain, while general optimization deals with the whole of a givensupply chain. In this paper one possible approach to optimization prob-lems in a specific case is presented with creation of economic param-eters necessary for realization of partial tasks. A modification of thegoal programming model is presented as well.

197


3 - A Trade Credit Model for Supply Chain CoordinationAbhishek Chakraborty, Operations Management, IndianInstitute of Management Calcutta, Diamond Harbour Road,Joka, 700104, Kolkata, West Bengal, India,[email protected], Ashis Chatterjee

This paper considers a single vendor and single buyer supply chainmodel where the vendor allows a trade credit to the buyer to synchro-nize his production cycle with the ordering cycle of the buyer. Thevendor is being benefitted from the reduction in his set up costs whilethe buyer is being compensated for the extra inventory that he is car-rying through the trade credit offered by the vendor. Trade credit isdeveloped as a mechanism for achieving supply chain coordination inthe weak form.

4 - An Analysis of Supply Chain Related Graduate Pro-grams in EuropeY. Ilker Topcu, Industrial Engineering, Istanbul TechnicalUniversity, Istanbul Teknik Universitesi, Isletme Fakultesi,Macka, 34357, Istanbul, Turkey, [email protected], EmelAktas, Sezi Cevik Onar, Des Doran

There is a continuous interest in analysing and designing SCM edu-cation. Parallel to increasing impact of globalisation, more and morecompanies operate in larger supply chains and there is an increasingneed for skilled employees trained in SCM. Motivated by lack of stud-ies on graduate level supply chain education, this research aims to de-termine the current status of supply chain related programs offered atgraduate level in Europe. Our findings suggest that programs are clus-tered according to the weight of various knowledge and skill areas suchas SCM, managerial or quantitative skills.

� WB-11Wednesday, 10:30-12h00RB-Iota

Other realistic scheduling problems

Stream: Realistic Production SchedulingInvited sessionChair: Alper Gucumengil, Foreign Relations Department, TurkishStatistical Institute, Yucetepe mah., Necatibey cad. No:114Cankaya, Ankara, Turkey, [email protected]

1 - Due-date assignment problems with common flow-allowanceGur Mosheiov, Business School, The Hebrew University ofJerusalem, Mount Scopus, Jerusalem, 91905, Jerusalem,Israel, [email protected], Baruch Mor

In due-date assignment problems with a common flow-allowance, thedue-date of a given job is defined as the sum of its processing time anda job-independent constant. We study first the minsum version of theproblem. In this case, the scheduler has the option to perform a main-tenance activity which is rate modifying, i.e., improves the processingtimes of the following jobs. We consider a number of versions of thissetting. We also study the minmax version of the problem, with ex-tensions to position-dependent processing times, and to the (NP-hard)setting of parallel machines.

2 - Approximate algorithms for one-machine schedulingwith interfering jobsPaz Perez Gonzalez, Industrial Management, University ofSeville, Camino de los descubrimientos s/n, 41092, Sevilla,Spain, [email protected], Jose M. Framinan

In this paper, we analyze the problem of scheduling two sets of jobson one machine with the objectives of total completion times for bothsets. This problem is known to be NP-hard considering different ap-proaches (Pareto, epsilon-constrained and linear convex), and severalalgorithms have been proposed in the literature. We analyze the ap-proaches, compare the corresponding solution procedures, and presentnew heuristics for the problem.

3 - Flowshop scheduling with flexible operations:Throughput optimizationHakan Gultekin, Industrial Engineering, TOBB-University ofEconomics and Technology, Endustri Muhendisligi Bolumu,TOBB Ekonomi ve Teknoloji Universitesi, Sogutozu Cad.No:43, 06560, Ankara, Turkey, [email protected] classical flowshop scheduling, processing times of the parts are as-sumed to be fixed, predefined parameters. This assumption oversim-plifies the problem for the systems consisting of flexible machines ca-pable of performing different operations. In such systems, some ofthe operations that a part requires can be processed by alternative ma-chines. In order to maximize the throughput rate the assignment ofthese operations to the machines must be determined. In this study wedevelop solution procedures for problems arising in 2 and 3-machineflowshops producing identical or multiple parts.

4 - Project Management Techniques: The Case of Agri-cultural Statistical ProjectAlper Gucumengil, Foreign Relations Department, TurkishStatistical Institute, Yucetepe mah., Necatibey cad. No:114Cankaya, Ankara, Turkey, [email protected],Hüseyin TatlidilA general framework on Project Management(PM) approach is out-lined and PM techniques,namely Gantt, CPM and PERT are scruti-nized in this paper.The immediate focus of study,however,is on theassessment of an agricultural statistical project which is one of theprojects carried out as part of preparatory work in the adaptation pro-cess of Turkey to European Statistical System with reference to PMtechniques.By way of assessment of completed project,the aim is toimprove current PM approach for international statistical projects andto offer suggestions in adopting PM techniques in future projects.

� WB-12Wednesday, 10:30-12h00RB-Omicron

Railway Optimization II

Stream: Optimization in Public TransportInvited sessionChair: David Canca, School of Engineers, University of Seville.,Av. de los Descubrimientos s/n, Isla de la Cartuja, 41092, Seville,Spain, [email protected]

1 - The management of rail system breakdowns for re-ducing travel demand impactsLuca D’Acierno, Dipartimento di Ingegneria dei Trasporti,Università degli Studi di Napoli "Federico II", Via Claudio,21, 80125, Napoli, Italy, [email protected], Mariano Gallo,Bruno Montella, Antonio PlacidoIn urban context, the improvement of rail systems represents a usefultool to promote the use of public transport and reduce environmen-tal impacts of private car use. However, in the event of breakdowns,since faulty trains cannot usually be overtaken and their removal couldpose extreme difficulties, re-establishing the regular service could in-volve inconveniently long travel times. Therefore, in this framework,we analyse effects of management strategies for different levels of de-graded services in order to define the best solution to adopt to minimisetravel demand discomfort.

2 - New approaches for Wagon Routing in RailroadFreight TrafficRobert Voll, Institute of Transport Logistics, TU DortmundUniversity, Leonhard-Euler-Strasse 2, 44227, Dortmund,Germany, [email protected], Uwe ClausenWe consider a consolidation problem from railroad freight traffic. Theproblem is modeled as a fixed charge multicommodity flow problem.We present two new solution approaches based on a column genera-tion formulation. The first one is a Branch-and-Price-algorithm. Thesecond method is a heuristic algorithm. We present computational re-sults for a couple of test instances we constructed in cooperation withour industrial partners. Results from our computations are comparedto solutions provided by CPLEX.

198


3 - Analysis of the variability of travel conditions andflows along a transit lineVincent Benezech, Ecole des Ponts ParisTech, ENPC -LVMT, 6-8 avenue Blaise Pascal, 77455, Marne-la-Vallée,France, [email protected], Fabien Leurent

One major issue in transit planning pertains to service reliability. Thepaper’s objective is to provide a stochastic model of traffic variabilityand passenger exposure along a transit route. Headways are treatedas random variables. Their distribution is estimated using data fromthe Transilien operator. Analytical formulas for the mean and varianceof passenger loads on any section of the line, as well as average pas-senger exposure to crowding in the case of log-normal headways arederived. In the general case, a zip-unzip algorithm computing gen-eralised travel times is described, for inclusion in a complete trafficassignment model.

4 - A methodology to analyze quality of railway timeta-blesDavid Canca, School of Engineers, University of Seville., Av.de los Descubrimientos s/n, Isla de la Cartuja, 41092, Seville,Spain, [email protected], Eva Barrena, Encarnación Algaba,Alejandro Zarzo

Railway scheduling and timetabling are two of the classical stages inthe hierarchical railway planning process and they perhaps representthe step with major influence on user’s quality of service perception.This paper considers the problem of timetable determination consid-ering a dynamic behavior of passenger demand considering user andservice provider points of view. Aspects like the influence of train ca-pacity are discussed. To illustrate the methodology an application tothe C5 line of Madrid rapid transit system is presented.

� WB-13Wednesday, 10:30-12h00RB-Tau

Transportation and Logistics

Stream: OR Applications in the Automotive IndustryInvited sessionChair: Achim Koberstein, Business Administration,Goethe-University of Frankfurt, Grueneburgplatz 1, 60323,Frankfurt am Main, Germany, [email protected]: Martin Grunewald, Institute of Automotive Managementand Industrial Production, Technische Universität Braunschweig,Katharinenstraße 3, 38106, Braunschweig, Germany,[email protected]

1 - Choosing charge carriers to guarantee the efficientmaterial supply of production areasDominik Berbig, Institut für Fördertechnik undLogistiksysteme, KIT, Gotthard-Franz-Straße 8, 76131,Karlsruhe, Germany, [email protected], Kai Furmans, MichaelaKöker, Maximilian Altefrohne

To guarantee efficient material supply, it is crucial to choose adequatecharge carriers for every part number: The larger the carrier, the largeris the required space, the number of contained pieces and, as a re-sult, the fixed capital costs. The smaller it is, the higher the necessarytransportation frequency. This trade-off shows that choosing a chargecarrier is subject to many parameters. Currently, this choice is oftenbased on the size only. This means wrong choices and wasted capital.Thus, we present an approach to solve this problem, a new model andpossible benefits.

2 - The impact of decision making strategies in differentnetwork types on stock fluctuationMeisam Nasrollahi, Industrial Engineering, University ofTehran, 32th - Masjed Jame St., Fasa - Fars - Iran,7461833749, Fasa, Fars, Iran, Islamic Republic Of,[email protected], Jafar Razmi, Reza Ghodsi

The knowledge about interdependency between decision makingstrategies and the stock fluctuation in a supply chain is important toidentify those parts of a supply network which has the highest po-tential for improving total uncertainty. Two scenarios are suggested:each supply partner designs its own material handling system and net-work configuration organizes an integrated material handling policy forall supply partners. A multi-objective model is developed to comparethese scenarios and identify under which situations lasting uncertaintyreduction can be achieved.

3 - An integrated inventory-transportation system withperiodic pick-ups and leveled replenishmentMartin Grunewald, Institute of Automotive Management andIndustrial Production, Technische Universität Braunschweig,Katharinenstraße 3, 38106, Braunschweig, Germany,[email protected], Thomas Volling, Thomas Spengler

In this paper we develop a combined inventory-transportation system.The principle idea is to integrate a simple replenishment policy witha routing component to derive operationally consistent standard routesas a basis for milk-run design. Therefore we combine stochastic ve-hicle routing with a replenishment policy which makes use of safetystock to level the variability propagated into transportation operations.To evaluate the approach, we compare its performance with stochasticvehicle routing as well as sequential vehicle routing and safety stockplanning.

� WB-14Wednesday, 10:30-12h00RB-Omega

Timetabling and Transport

Stream: Timetabling and RosteringInvited sessionChair: Hagai Ilani, Industrial Eng. and management, ShamoonCollege of Engineering, 84 Jabotinsky st., 77245, Ashdod, Israel,Israel, [email protected]

1 - Algorithm for Solving Two-Stations Railway Schedul-ing ProblemElena Musatova, Institute of Control Sciences V. A.Trapeznikov Academy of Sciences, Moscow, RussianFederation, [email protected], Alexander Lazarev

There are two railway stations and a set of freight cars which arecharacterized by departures and destinations, weight and significance.Each car arrives to the departure in some release time and should bedelivered to the destination before a due date. There are some restric-tions on number of cars in a train. It is necessary to form trains andschedules that minimize a cost function. A polynomial algorithm isproposed for solving the described problem.

2 - An Integrated Optimization Model for Timetablingand Machinist Assignment in Light Rail Transit Sys-temsSelmin DanisOncul, Management Engineering, IstanbulTechnical University, ITU Isletme Fakultesı, Macka-Istanbul,34367, Istanbul, Turkey, [email protected]

This study proposed a novel model to solve the timetabling prob-lem combining with resource allocation of a light rail transit system.Firstly, a detailed literature review is carried out, then extensive inter-views have been realized with authorized experts. Mathematical mod-els are built and solved by mixed-integer programming using CPLEXalgorithm. Validation is made by comparing model results with real-life application. Sensitivity analysis is also performed by changingparameter values such as train capacity and also weights of objectivefunction. Finally all the results are analyzed.

3 - Real-time Control Procedures for TransitAlessandro Sales, Civil engineering, Università degli StudiRoma Tre, via vito volterra, 62, 00146, Roma, Italy,[email protected], Stefano Carrese

199


In every high frequency transport system, the problem of regularityis critical. Randomness can increase passenger’s travel time. We fo-cus the attention on the real-time deadheading problem, because dead-heading is the only way that allows to re-establish the correct headwaywithout increase of the lap time. When a vehicle is deadheaded, it runsempty from a terminal skipping a number of stations, typically in orderto reduce expected large headways at later stations. The objective is todetermine the optimal number of stations to skip in order to minimizepassenger’s total travel time.

4 - A General Two-directional Two-campus TransportProblemHagai Ilani, Industrial Eng. and management, ShamoonCollege of Engineering, 84 Jabotinsky st., 77245, Ashdod,Israel, Israel, [email protected], Elad Shufan, Tal Grinshpoun

In the present work we extend a reduction of a simple case of a trans-portation problem between two destinations to the shortest path prob-lem. Our reduction is valid for a general two directional case wherea set of n requests for a ride between two destinations has to be ex-ecuted by a fleet of K vehicles of given capacities, in order to mini-mize the total waiting time of all the passengers. Our approach canbe implemented to the case where passengers have different levels ofimportance.

� WB-15Wednesday, 10:30-12h00RB-2101

Theory and algorithms of bilevelprogramming I

Stream: Variational Inequalities and Bi-Level ProblemsInvited sessionChair: Stephan Dempe, Mathematics and Computer Sciences,Technische Universitaet Freiberg, 09596, Freiberg, Germany,[email protected]

1 - Fuzzy bilevel optimization problemAlina Ruziyeva, Fakultät für Mathematik und Informatik, TUBergakademie Freiberg, Akademiestr. 6, 09596, Freiberg,Germany, [email protected], Stephan Dempe

The optimal solution of a fuzzy optimization problem is described asthe set of Pareto optimal solutions of a corresponding multiobjectiveoptimization problem. The preferable fuzzy optimal solution is sup-posed to have a maximal membership function value, i.e. it has thehighest potential being realized by the follower. In the talk fuzzybilevel optimization problems are considered in which the preferablefuzzy optimal solution of the fuzzy lower level problem is used. Aimof the talk is to describe a solution algorithm for this problem using thestability region of this solution.

2 - Necessary optimality conditions in pessimisticbilevel programmingAlain B. Zemkoho, Mathematics and Computer Sciences,Technical University Bergakademie Freiberg, Akademiestr. 6,09599, Freiberg, Saxonia, Germany,[email protected]

We consider the "pessimistic version" of bilevel programming pro-grams. Employing advanced tools of variational analysis and gener-alized differentiation, we provide a general frameworks ensuring theLipschitz continuity of the value functions. Several types of lowersubdifferential necessary optimality conditions are derived. We alsoderive upper subdifferential necessary optimality conditions of a newtype. Certain links are established between the obtained necessary op-timality conditions for the pessimistic and optimistic versions in bilevelprogramming.

3 - Vanishing Stress Constraints in Topology Optimiza-tion of Mechanical StructuresWolfgang Achtziger, Department of Mathematics, Universityof Erlangen-Nuremberg, Chair of Applied Mathematics 2,Cauerstrasse 11, 91058, Erlangen, Germany,[email protected]

We consider the optimization of topology of mechanical structures in adiscretized finite element setting. These problems express shape opti-mization through a problem of optimal material distribution on a givenelement grid. Local stress constraints are essential for a realistic mod-eling. The resulting problem is a large-scaled mathematical programwith vanishing constraints (MPVC). In the recent past several numeri-cal recipes have been proposed for the treatment of MPVCs. The talkpresents some of these approaches and reports on their success in thenumerical treatment of topology problems.

4 - Branch and bound method for the competitive facilitylocation problemAndrey Melnikov, Sobolev Institute of Mathematics, RussianFederation, [email protected], Vladimir Beresnev

We deal with the competitive facility location problem, where twoplayers, the leader and the follower, sequentially open their facilitiesin finite discrete space. Every consumer chooses one opened facilityaccording to his own preferences. The problem consists in openingthe leader’s facilities so as to obtain the maximum profit, taking intoaccount the follower’s reaction. The problem is formulated in termsof the bilevel integer programming. An exact procedure based on theB&B method has been developed. Computational results for the ran-domly generated instances are discussed.

� WB-16Wednesday, 10:30-12h00RB-2103

Linear and Quadratic Programming

Stream: Linear and Conic ProgrammingInvited sessionChair: David Bartl, Department of Mathematics, University ofOstrava, 30.dubna 22, 701 03, Ostrava, Czech Republic,[email protected]

1 - On the complexity of steepest descent algorithms forminimizing quadratic functionsClovis Gonzaga, Dept. of Mathematics, Federal Univ. ofSanta Catarina, Cx. postal 5210, Florianópolis, SC, Brazil,Florianópolis, SC, Brazil, [email protected]

In this talk we examine patterns in the distribution of step lengths insteepest descent algorithms for minimizing a convex quadratic func-tion. We show how a large number of short steps are needed, and howthese relate to the much smaller number of large steps. We note that theorder in which the step lengths is used is irrelevant, and show a worstcase example with a small number of variables. We also conceive abrute force algorithm which is in a certain way optimal.

2 - Convex quadratic programming applied to the stabil-ity number of a graphMaria F Pacheco, Instituto Politécnico de Bragança - ESTiG,Quinta de Santa Apolónia, Gab. 112, 5301-857, Bragança,Portugal, [email protected], Domingos Cardoso, Carlos J. Luz

We deal with graphs whose stability number can be determined by aconvex quadratic program and describe algorithmic techniques for thedetermination of maximum stable sets in such graphs (except thereis an induced subgraph with least adjancency eigenvalue and optimalvalue of the convex quadratic program not changing if the neighbor-hood of any vertex is deleted). Such a graph is called adverse. As-suming that every adverse graph has convex-QP stability number, analgorithm for the recognition of graphs with convex-QP stability num-ber is introduced and applied to determine maximum matchings.

3 - A discrete version of Farkas’ Lemma and Duality The-orem for homogeneous linear programmingDavid Bartl, Department of Mathematics, University ofOstrava, 30.dubna 22, 701 03, Ostrava, Czech Republic,[email protected]

We report a discrete version of Farkas’ Lemma, which is formulated inthe setting of a module over a linearly ordered commutative ring (e.g.

200


the ring of the integer numbers). Inspired by the trick of the homog-enization, used to prove Gale’s Theorem of the alternative, which wealso show, we introduce the concept of the primal and dual problem ofhomogeneous linear programming. We present the optimality condi-tion for the primal and dual problem, respectively. In conclusion, weshow that the Strong Duality Theorem holds for the problems.

4 - Optimization for switching control systemShahlar Meherrem, Mathematics, Yasar University, Izmir,Turkey, Selcuk Yasar Camp, agacli yol , No.35-37, Bornova,Izmir, Turkey, [email protected], Refet Polat

A class of nonsmooth optimal switching control problems is consid-ered (nonsmoothness posed on the minimizing functional). The costfunctional is to be minimized. Necessary optimality conditions are de-rived by normal cone and Dubovitskii-Milyutin theory: a) treat localminima via the empty intersection of certain sets in the primal space;b) approximate the above sets by convex cones with no intersections;c) arrive at dual necessary optimality conditions in the form of an ab-stract Euler equation. The main result is a new optimality condition forthe nonsmooth switching control system.

� WB-17Wednesday, 10:30-12h00RB-2105

DC programming and DCA 1

Stream: Nonconvex Programming: Local and GlobalApproachesInvited sessionChair: Tao Pham Dinh, INSA Rouen, 76131, Rouen, France,[email protected]

1 - A Continuous Optimization Approach for the GeneralArt Gallery ProblemMahdi Moeini, Institute of Operating Systems and ComputerNetworks, Technical University of Braunschweig, IBR,Algorithms Group„ Muhlenpfordtstr. 23, 38106,Braunschweig, Germany, [email protected], AlexanderKroeller, Christiane Schmidt

Suppose that an art gallery is in the shape of a polygon. The classicalArt Gallery Problem (AGP) seeks for determining the minimum num-ber of guards that are sufficient to oversee the entire art gallery. It hasbeen proven that this problem is NP-hard even for very special cases.We present a continuous optimization approach based on DC (Differ-ence of Convex functions) programming techniques to solve the AGPproblem. The numerical experiments have been carried out on somebenchmark data sets and the results will be presented.

2 - A DC Programming technique for sparse regulariza-tion in least-squares and logistic regressionsMamadou Thiao, LITA, LITA- Université de Lorraine, 57000Ile du saulcy, Metz, France, [email protected],Hoai An Le Thi, Tao Pham Dinh

This talk addresses a DC programming technique for solving ell_2 -ell_0 simultaneous regularization problem in least-squares and logis-tic regressions. We first consider the original ell_2 - ell_0 -regularizedproblems and we reformulate them as DC programs with the help ofnew exact penalty techniques. The resulting DC programs possess verynice DC structures and are quite adapted to convex DC relaxation tech-niques. Experiments in the context of feature selection are performedand the results on real-world data sets illustrate the potential and showthe usefulness of this method.

3 - A DC programming approach for clustering massivedata setsHoai An Le Thi, Computer Science, University of Lorraine,Ile du Saulcy„ 57 045, Metz, France,[email protected], Ta Minh Thuy, LydiaBoudjeloud-Assala

In this paper, we present a robust nonconvex optimization method forclustering massive data using data streams. Our approach is based onDC (Difference of Convex functions) programming and DCA (DC Al-gorithms) that have been successfully applied in various fields of ap-plied sciences. The method consists of two phases. Phase 1: the datais divided into subsets on which an efficient DCA for clustering is in-vestigated. Phase 2: another DCA for weighted clustering on the set ofcenters obtained by phase 1 is studied. The numerical results on realdata sets show the efficiency of our method.

4 - Solving Generalized Orienting Problem using DCAAnh Son Ta, LMI, INSA de Rouen, LMI, INSA de Rouen,+33, Rouen, France, [email protected], Hoai An Le Thi,Tao Pham Dinh, Djamel Khadraoui

Orienteering problem is well-known as a NP-hard problem in trans-portation with many applications. This problem aims to find a pathbetween a given set of control points, where the source and destina-tion points are specified with respect to maximize the total score ofcollected points and satisfy the distance constraint. In this paper, wefirst analyze the structure of a generalized orienting problem and a newsolution method, based on DC programming and DCA, is introduced.Preliminary numerical experiments are reported to show the efficiencyof the proposed algorithm.

� WB-18Wednesday, 10:30-12h00RB-2107

Discrete LocationStream: Network and Discrete LocationInvited sessionChair: Antonio Manuel Rodriguez-Chia, Estadistica e IO,Universidad de Cádiz, Facultad de Ciencias, Pol. Rio San Pedro,11510, Puerto Real (Cadiz), Cadiz, Spain,[email protected]

1 - The p-center problem with capacities and failureaheadAlfredo Marín, Departamento de Estadística e InvestigaciónOperativa, University of Murcia, Facultad de Matemáticas,Campus de Espinardo, 30100, Murcia, Spain, [email protected],Inmaculada Espejo, Antonio Manuel Rodriguez-Chia

This work deals with the p-center problem, where the aim is to mini-mize the maximum distance between any user and his reference cen-ter, considering that in case of failure of any center, the capacities ofthe remaining centers must suffice to satisfy the demand of all users,taking into account that the users will have as a reference the closestsurviving center. The problem is of interest when locating emergencycenters and, at the same time, taking precautions against emergencieswhich can cause failure of the center it self. We consider differentformulations for the problem

2 - Assigning channels in cellular networks by neuralnetworksEnrique Dominguez, Dept. of Computer Science,E.T.S.I.Informatica - University of Malaga, Campus Teatinoss/n, 29071, Malaga, Spain, [email protected], Jose Muñoz

Wireless communication constitutes one of the fastest growing indus-try segments in recent years, encompassing a number of applicationdomains such as the cellular networks. Efficient allocation of chan-nels for wireless communication in different network scenarios hasbecome an extremely important topic of recent research. The mainchallenge lies in the fact that the channel allocation problem (CAP) isNP-complete. In many practical situations, sometimes we may need tobe satisfied with a near-optimal solution because of a maximum allow-able time limit is imposed for allocation of channels.

3 - A p-median model with distance selectionStefano Benati, Department of Sociology, University ofTrento, Via Verdi 26, 38100, Trento, Italy,[email protected], Sergio García Quiles

201


We are introducing an extension of the p-median model in which theunits distance function, defined in an m-dimensional space, is thesum of elementary distances calculated on spaces of dimension 1 (thismodel originate from cluster analysis). We let the researcher selectonly q coordinates out of the m dimensional axes, so we are allow-ing different and possible distance definitions between the units. Wepropose many ILP formulations to the model and compare their prop-erties.

4 - Single Allocation Capacitated Ordered Median HubLocation ModelsAntonio Manuel Rodriguez-Chia, Estadistica e IO,Universidad de Cádiz, Facultad de Ciencias, Pol. Rio SanPedro, 11510, Puerto Real (Cadiz), Cadiz, Spain,[email protected], Justo Puerto, Ana BelRamos-Gallego

In this paper we consider new formulations, based on the ordered me-dian objective function, for capacitated hub location problems withnew distribution patterns induced by the different user’s roles withinthe supply chain network. This approach introduces some penalty fac-tors associated with the position of an allocation cost with respect to thesorted sequence of these costs. First we present formulations for thisproblem, and then develop stronger formulations by exploiting proper-ties of the model. The performance of all these formulations is com-pared by means of a computational analysis.

� WB-19Wednesday, 10:30-12h00RB-2111

Predictive Modelling in Finance andInsuranceStream: Simulation Methods in FinanceInvited sessionChair: Aysegul Iscanoglu Cekic, Statistics, Selcuk University,KONYA, Turkey, [email protected]

1 - Simulation Study for Modified Maximum LikelihoodEstimations in Poisson RegressionMuslu Kazım Körez, Statistic Department, Faculty ofScience, Selcuk University, 42225, Konya, Turkey,[email protected], Evrim Oral, Yunus Akdogan, CoskunKus

For most of the generalized linear models (GLM), the maximum likeli-hood (ML) equations involve nonlinear functions of the parameters andthey are intractable. Solving these equations by iterations can be prob-lematic for reasons of convergence to wrong values.In this study,Wecompare the ML and MML estimators with respect to their biases andmean square errors. Moreover, we provide the coverage probabili-ties of the confidence intervals (based on ML and MML estimators)of Poisson regression parameters by using a Monte Carlo simulationstudy. We also investigate the revised MML estimators proposed byOral (2005) with respect to their initials values. We show that ordi-nary least squares estimators yield approximations as good as thoseproposed by Oral (2005).

2 - An Application of Wavelet Regression on Risk Anal-ysisAsır GenÇ, Statistic, Selcuk University, Konya, Turkey,[email protected], Tarik Yilmaz

In this study we make a risk analysis for a software company whichplans to develop and launch a new product. In the analysis we firstlymodel the financial statements of the company. The model includesboth stochastic and deterministic variables. Therefore, we make a sim-ulation study and we provide 10-year cash flow tables for the company.Then a risk analysis procedure is implemented. Moreover, we apply awavelet regression smoothing procedure to the obtained values and wefind new risk values. Finally, these new risk values are compared withNet Present Value and a decision plan is given.

3 - Predictive Power of Generalized Additive Models inInsuranceAysegul Iscanoglu Cekic, Statistics, Selcuk University,KONYA, Turkey, [email protected]

In insurance an event or a claim is usually related with various num-ber of risk factors. The classification techniques are used to modelthe impact of such factors on occurance of such events or claims. Thegeneralized additive model is one of these techniques. In this study, weimplement the generalized additive models with Conic quadratic pro-gramming approach (GAM-CQP) to the insurance framework. We alsocompare the predictive performance of GAM-CQP with the classicalgeneralized additive models and some other techniques, e.g. general-ized linear models and semi-parametric regression.

4 - An Application of Wavelet Regression in ExchangeRate EstimationTarik Yilmaz, Statistics, Science Institute, Selcuk UniversitesiAlaaddin Keykubat Kampusu, Selcuklu, 42020, KONYA,Turkey, [email protected], Asır GenÇ

In this study, our main aim is to forecast exchange rate in an effec-tive way. For this purpose we propose a new technique which is basedon wavelet regression. In order to evaluate the effectiveness of thetechnique, the EURO(EUR) - U.S. Dollar (USD) parity data is used.Effectiveness of the new process is also shown by backtesting.

� WB-20Wednesday, 10:30-12h00RB-2113

Financial Optimization 1

Stream: Financial OptimizationInvited sessionChair: J. E. Beasley, Mathematical Sciences, Brunel University &JB Consultants, Department of Mathematical Sciences, KingstonLane, UB8 3PH, Uxbridge, Middlesex, United Kingdom,[email protected]

1 - A Risk-Return Approach to Enhanced IndexationFrancesco Cesarone, Economics, University of Rome 3,Faculty of Economics, Via Silvio D’Amico 77, 00145, Rome,Italy, [email protected], Renato Bruni, Andrea Scozzari,Fabio Tardella

Enhanced Indexation is the problem of selecting a portfolio that gener-ates excess return with respect to a benchmark index. We propose a lin-ear bi-objective optimization model for enhanced indexation that max-imizes average excess return and minimizes underperformance withrespect to the index over a given observation period. The efficientfrontier for this problem can be easily computed with standard LPsolvers. We describe interesting theoretical features of the model andwe present some experimental results for well-known financial datasets showing promising out-of-sample performance.

2 - Portfolio selection model based on technical, funda-mental and market value analysisTea Poklepovic, Faculty of Economics Split, Domovinskograta 54, 21000, Split, Croatia, [email protected], BrankaMarasovic, Zdravka Aljinovic

The aim of this paper is to develop decision-making model for select-ing optimal portfolio in stock exchange based on technical and fun-damental analysis. In the first step of the model the set of stocks areselected based on technical analysis. In the second step optimal port-folio will be formed from selected stocks using multi criteria method.The method is based on PROMETHEE II approach and different ac-counting criteria and those based on market values are used. The se-lected model has been applied in Zagreb Stock Exchange (ZSE) as areal case.

202


3 - Computing the Nondominated Surface in Tri-criterionPortfolio SelectionRalph E. Steuer, Terry College of Business, University ofGeorgia, Department of Banking and Finance, Brooks Hall,30602-6253, Athens, GA, United States, [email protected]

Motivated by the desire to extend Markowitz portfolio selection to anadditional linear criterion (dividends, etc.), we demonstrate an exactmethod for computing the nondominated set of a tri-criterion programthat is all linear except that one of its objectives is to minimize a con-vex quadratic function. With the nondominated set of the resulting pro-gram being a surface composed of curved platelets, a multi-parametricalgorithm is devised for computing the platelets so that they can begraphed precisely. In this way, graphs can display all tri-criterion non-dominated solutions at one time.

4 - Portfolio rebalancing with an investment horizon andtransaction costsJ. E. Beasley, Mathematical Sciences, Brunel University & JBConsultants, Department of Mathematical Sciences, KingstonLane, UB8 3PH, Uxbridge, Middlesex, United Kingdom,[email protected]

We consider the problem of rebalancing an existing financial portfo-lio, where transaction costs (fixed and/or variable) have to be paid ifwe change the amount held of any asset. We indicate the importanceof the investment horizon when rebalancing such a portfolio and il-lustrate the nature of the efficient frontier that results when we havetransaction costs. We model the problem as a mixed-integer quadraticprogram with an explicit constraint on the amount that can be paid intransaction cost. Results are presented for the solution of test problemsinvolving up to 1317 assets.

� WB-21Wednesday, 10:30-12h00RB-2115

Data Analysis in the Financial Sector

Stream: Numerical Methods in FinanceInvited sessionChair: Patricia Herranz, Economy, Quantitative Methods andEconomic History, Universidad Pablo de Olavide, Ctra. Utrera km1, 41013, Seville, Spain, [email protected]

1 - Sampling Frequency and Stochastic Diffusion Pro-cesses of Term Structure of Interest Rates and Ex-change RatesA. Can Inci, College of Business - Finance, Bryant University,1150 Douglas Pike, 02917, Smithfield, RI, United States,[email protected]

Recent studies conjecture that higher frequency data provide a betterseparation of continuous and jump components of a dynamic process.Stochastic processes such as jumps may not be fully utilized sincethey may be smoothed out with low frequency monthly or quarterlydata. This study examines a sophisticated multi-country multi-statenonlinear stochastic model along with affine versions in the context ofsampling frequency. Empirical performances of the models are docu-mented and compared across models and across sampling frequencies.

2 - Seasonal allocation decision based on a multi-objective optimizationFellipe Santos, UFMG, 30580530, Belo Horizonte, Brazil,[email protected], Adriano Lisboa, DouglasVieira, Rodney Saldanha, Marcus Lobato

This paper discusses and formulates a simplified model of the Brazil-ian energy market , from where a bi-objective optimization problemfor revenue and risk is derived in order to define optimal portfolios.The resulting problem is a sub-differentiable analytical problem withguarantees to be solved to optimality by cutting-plane methods. Somecase studies are presented in order to test the formulation using theellipsoidal algorithm.

3 - Information acquisition of new technology for main-tenance and replacement investment decisionsThomas Yeung, Department of Industrial Engineering &Automatic Control, Ecole des Mines de Nantes / IRCCyN, 4,rue Alfred Kastler B.P. 20722, La Chantrerie, 44307, Nantes,France, [email protected], Khanh Nguyen, BrunoCastanierAn important mission of managers is to determine the maintenance andreplacement investment plan for equipment under technological evolu-tion. The arrival time and profitability of new technology are uncertain;however information on this process may be acquired. We formulatea partially observable Markov decisions process to decide whether toobtain additional information as well as the action for the asset (wait,maintain, replace) based on its state and the information available onnew technology. We provide structural properties of the optimal policyas well as numerical examples.

4 - One approach to reduce some aspects of costs as-sociated with long term carePatricia Herranz, Economy, Quantitative Methods andEconomic History, Universidad Pablo de Olavide, Ctra. Utrerakm 1, 41013, Seville, Spain, [email protected], Ana M. MartinCaraballo, M. Manuela Segovia-Gonzalez, Guerrero FlorNew technological and scientific advances have allowed that in somecountries increases life expectancy but, unfortunately longevity causesthat many people have disabilities that limit or even make it impossibleto do activities of daily living. In this paper we analyze the relationshipbetween the different diseases and different degrees of disabilities usedunder Spanish law. In this way the government could implement pre-vention policies that reduce the costs associated with long term careand improve the quality of life of disabled people. We are using thenew survey data EDAD 2008.

� WB-22Wednesday, 10:30-12h00RB-2117

Fuzzy Optimization

Stream: Fuzzy Optimization - Systems, Networks andApplicationsInvited sessionChair: Erik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39, 85577,Neubiberg, Germany, [email protected]: Yoshiki Uemura,655-1,Yuzaki,Kawanishichou,Shikigun,636-0202,Nara,Japan,636-0202, Nara, [email protected]: Kateryna Pereverza, Students Science Association,National Technical University of Ukraine, Kyiv, Scherbakova str.,42/44, app. 54, Kyiv, Ukraine, [email protected]

1 - Involving fuzzy orders for multi-objective linear pro-grammingOlga Grigorenko, Mathematics, University of Latvia, ZelluStreet 8, LV-1002, Riga, Latvia, [email protected] our work we propose to involve fuzzy order relations for solv-ing MOLP problems. We use fuzzy orders to describe the ob-jective functions. Further the global fuzzy order relation is con-structed by aggregating the individual fuzzy order relations. Thus theglobal fuzzy relation contains the information about all objective func-tions. In our approach we can naturally use compensatory aggrega-tion functions and even more we can use weights to show the prefer-ence of objective functions. This work is supported by ESF project2009/0223/1DP/1.1.1.2.0/09/APIA/VIAA/008.

2 - Forecasting Enrollments Based on Particle SwarmOptimization and Neural Network Based Fuzzy TimeSeriesOzer Ozdemir, Statistics, Anadolu University, AnadoluUniversity Faculty of Science, Department of Statistics,Eskisehir, Turkey, [email protected], MemmedagaMemmedli

203


Many forecasting studies used enrollments of the University of Al-abama to show comparison results easily by applying the concept offuzzy time series. So, in this paper, a forecasting study based on fuzzytime series and particle swarm optimization is presented to find theproper model for this data set. Also, a new approach for neural networkbased fuzzy time series is taken into account for forecasting because ofneural networks’ capabilities in handling nonlinear relationships. Allresults are compared with other studies proposed in the literature toshow the performances of all models.

3 - Selection of appropriate smoothing parameter fordifferent sample size in nonparametric fuzzy localpolynomial modelsMunevvere Yildiz, Statistics, Anadolu University, Turkey,[email protected], Memmedaga Memmedli

In this paper, we considered the relationship between the smoothingparameter value and sample size as a simulation study in nonpara-metric fuzzy local polynomial regression. For this aim, we developedfuzzy version of generalized cross-validation criteria (GCV) for se-lecting smoothing parameter in nonparametric fuzzy local polynomialmodels. The appropriate smoothing parameters are selected by GCVcriteria for different sample size and then performances of the mod-els are compared using these appropriate smoothing parameters withsample sizes.

4 - Dynamic Clustering to Improve Marketing Decisionsin Retail StoresRichard Weber, Department of Industrial Engineering,University of Chile, Republica 701, 2777, Santiago, Chile,[email protected], Georg Peters

Clustering is a common technique for customer segmentation. In real-life situations, however, customer behavior changes over time and sodo customer segments. We present a methodology for dynamic roughclustering and apply it to synthetic as well as to purchase data froma retail store. Comparing clustering techniques reveals the particularadvantages rough clustering offers. The results obtained from analyz-ing purchase data incrementally provide insides into changing buyingbehavior which permits to anticipate Marketing decisions, such as e.g.promotions or demand-specific advertisement.

� WB-23Wednesday, 10:30-12h00RB-Delta

Recent Advances on Proximal Methods IStream: Convex OptimizationInvited sessionChair: Hector Ramirez, Mathematical Engineering Department,Universidad de Chile, Avda. Blanco Encalada 2120, Piso 5,Santiago Centro, Santiago, RM, Chile, [email protected]

1 - Dual convergence for penalty algorithms in convexprogrammingMiguel Carrasco, Universidad de los Andes, Santiago,7620001, Chile, [email protected], Felipe Alvarez, ThierryChampion

We consider an implicit iterative method to solve a convex program-ming problem, which combines inexact variants of the proximal pointalgorithm with parametric penalty functions. For this primal itera-tive method is possible to associate a dual approximate multiplier se-quence. Under some conditions on the parameters, we show the con-vergence of the approximate multiplier sequence towards a particularoptimal solution of the dual problem

2 - Squared Hessian Riemannian metrics on convex setsand gradient-like optimization methodsCristopher Hermosilla, Equipe COMMANDS, ÉcolePolytechnique, INRIA-Saclay and ENSTA ParisTech, France,[email protected], Felipe Alvarez

We investigate a class of variable metric techniques to solve a con-vex constrained optimization problem. We investigate squared Hes-sian metrics induced Legendre type functions and we characterize thegeodesic curves of this kind of manifold. We propose a family ofgradient-like algorithms for constrained optimization based on linearsearch along the geodesics. We are particularly interested in findingconditions ensuring value convergence. We present two alternativecharacterization for this algorithm and we introduce a sort of dualityscheme based on Legendre change of coordinates.

3 - Convergence of the coupling of an inexact general-ized proximal algorithm with barrier methodsJulio López, Matemática, Universidad Técnica Federico SantaMaría, Av. Vicuña Mackenna 3939, San Joaquín, Santiago,Metropolitana, Chile, [email protected]

Here, we introduce a barrier-type penalty for the nonlinear inequal-ity constraints, then endowing the linear equality and conic constraintswith the Riemannian structure induced by the Hessian of an essen-tially smooth convex function, and finally considering the flow gen-erated by the Riemannian penalty gradient vector. We investigate thewell-posedness of the resulting ODE and we prove that the value of theobjective function converges to the optimal value. This convergence isextended to the sequences generated by a coupling of a generalizedproximal method with parametric barrier schemes.

4 - Solving Network Flow Problems with General Non-Separable Convex Costs using a Two-phase GradientProjection AlgorithmKwong Meng Teo, Industrial & Systems Engineering,National University of Singapore, Singapore,[email protected], Trung Hieu Tran

We propose an algorithm to solve network flow problems with gener-alized non-separable convex costs — problems often encountered inpractical applications such as multi-commodity flows, traffic assign-ment and telecommunications problems. The algorithm can be im-plemented using general solvers such as CPLEX. Finally, we evaluateand compare the performance of the algorithm with other common ap-proaches, using several industrial size problems.

� WB-24Wednesday, 10:30-12h00CC-A11

Data Mining in Early Warning Systems

Stream: Data Mining in Early Warning SystemsInvited sessionChair: Antonio Rodrigues, CIO-FCUL, University of Lisbon,Edificio C6, Campo Grande, 1749-016, Lisboa, Portugal,[email protected]

1 - Detection of seasonal changes in climate data bycomparative analysis; A case study for TurkeyTülay Akal, Statistics, Middle East Technical University,Turkey, [email protected], Fidan Fahmi, Elcin KartalKoc, Vilda Purutcuoglu Gazi, Cem Iyigun, CeylanYozgatligil, Inci Batmaz

In this study, we consider to investigate plausible climate changes inTurkey by clustering techniques implemented in data mining field. Forthis purpose we choose a homogenous set of stations from the Cen-tral Anatolia region. We apply bi-plot, SAMMON mapping, fuzzy,k-means, and hierarchical clustering methods for benchmark meteoro-logical variables monthly recorded in Turkey. We evaluate the resultsin two periods: i) from 1950 to 1980 and from 1980 to 2010 separately,ii) from 1950 to 2010 together. We also use paired statistical tests todetect seasonal changes in these periods.

2 - Developing precipitation models for continental cen-tral Anatolia, TurkeyFulya Aykan, statistics, METU, atapark mah. 1335 sok.,23/16 atapark /keçiören, ankara, Turkey,[email protected], Elcin Kartal Koc, Ceylan Yozgatligil,Cem Iyigun, Vilda Purutcuoglu Gazi, Inci Batmaz

204


Precipitation models help for deciding future scenarios of early warn-ing systems for droughts. In this study, we develop precipitation mod-els of continental central Anatolia, where drought has been a recurrentphenomenon for the last several decades. In the study, the monthly pre-cipitation is modeled by using Time Series Regression and Multivari-ate Adaptive Regression Splines methods using 13 regressor variablessuch as monthly mean pressure, monthly mean humidity, monthlymean vapor pressure and some temperature indices, measured for 30years period (1980-2010).

3 - New statistical methods for climate change investi-gationsGregory Gurevich, Industrial Engineering and Management,SCE - Shamoon College of Engineering, Balik Bazel, 84100,Beer Sheva, Israel, [email protected], Yossi Hadad

The main goal of this study is to present and apply recently devel-oped nonparametric change point detection and estimation techniquesfor confirming patterns of regional changes in temperature. Utilizingthese methods, monthly temperatures (mean, mean maximum, meanminimum) and diurnal temperature ranges in different regions of Israelhave been investigated over a period of 37-years, (1967-2003). MonteCarlo experiments were carried out to obtain the p-values of the pro-posed tests. The results were supported also by traditional statisticalmethods.

4 - Surveillance and security risk minimization in portsAntonio Rodrigues, CIO-FCUL, University of Lisbon,Edificio C6, Campo Grande, 1749-016, Lisboa, Portugal,[email protected]

The protection of physical infrastructures from terrorist/criminal at-tacks or from natural/technological disasters heavily relies on earlywarning detection capabilities from sensor networks. For ports andharbours, surveillance optimization is an especially complex problem,and different types of sensors should be used complementarily. Wepropose a methodology to estimate risk maps, both ex-ante and ex-postthe inclusion of a possible surveillance solution. Furthermore, we dis-cuss how the cost vs. risk optimization of solutions can be supportedby the estimation of response surfaces.


Smart cities and Smart grids

Stream: Long Term Planning in Energy, Environmentand ClimateInvited sessionChair: Edi Assoumou, Centre de Mathematiques Appliquees,Mines ParisTech, Sophia Antipolis, France,[email protected]

1 - Demand response in smart grid systemsAbhilasha Aswal, Information Technology, InternationalInstitute of Information Technology Bangalore, India,[email protected], Sunil Kumar Vuppala, G. N.Srinivasa Prasanna

The general optimization problem in a smart grid system with de-mand/price uncertainty is a min-max problem. In the presence of de-mand response (DR) mechanisms, we show that if all the appliances onthe grid are controllable, then the optimizations are polynomial timelinear minimizations. We discuss algorithms to find the optimal so-lution in both the general case and DR case. We also discuss algo-rithms for cases where only a subset of appliances can be controlled -the number of uncertain variables is reduced but not eliminated com-pletely. Our results are based on both analysis and simulation.

2 - Smart grids and prospective modeling for the elec-tricity sectorStephanie Bouckaert, Center for applied Mathematics, MinesParisTech, 06560, Sophia Antipolis, France,[email protected], Edi Assoumou,Nadia Maïzi

Our daily lives are facilitated by an increasing number of electric ap-pliances. However using electricity as two implications: a real timebalance between total supply and demand, and a high environmentalimpact. In the transition toward sustainable energy systems smart gridshave the potential to transform balancing with responsive demands andto enable higher shares of renewables. Including these operational ad-vantages in long term model is challenging. This communication de-scribes a methodology to quantify the long term impact of smart gridon future prospects for the electricity sector.

3 - Towards Interactive Urban Planning with LocalSearch TechniquesBruno Belin, LINA, France, [email protected],Marc Christie

The task of urban planning consists in the spatial organization of thevarious blocks, referred to as urban forms, that constitute a city (resi-dential, commercial, industrial, ...). The spatial organization is guidedby a collection of rules related to environmental, social, transportation,and energy questions. We propose a tool assist designers in this task bymodelling the problem as a local search optimisation task.To achievelarge-scale interactive editing, we introduce a dynamic list of banishedcandidates and demonstrate its effectiveness to reduce the explorationof neighborhoods.

4 - Sustainable development and territory planning: aprospective approach to rethink citiesSteve Lechevalier, Center for applied Mathematics, MinesParisTech, 1 rue claude Daunesse, 06560, Sophia Antipolis,France, [email protected], SandrineSelosse

The "serious game" ECOTYPE aims to help cities to communicateabout issues of land use in a fun and interactive form. This is to designand implement an interactive device involving prospective and repre-sentations of local development plan in an immersive space where thepublic is invited to play the future of its territory. In a geodesic room,the public will make choices about land use thanks to an interface.Then, consequences will be represented in the present and the futurethanks to lights and sound.


Multi-Criteria Decision Making andApplications 1

Stream: Multi-Criteria Decision Making and Environ-mental ManagementInvited sessionChair: Marta Castilho Gomes, CESUR, Instituto SuperiorTécnico, Av. Rovisco Pais, 1049-001, Lisboa, Portugal,[email protected]

1 - On the generation of positional voting systems withvariable scoresBonifacio Llamazares, Economía Aplicada, Universidad deValladolid, Avda. Valle de Esgueva 6, 47011, Valladolid,Spain, [email protected]

Scoring rules are a well-known class of positional voting systemswhere fixed scores are assigned to the different ranks. However, themain drawbacks of this decision rules is the choice of the scoring vec-tor, since the winners may change according to the scores used. Forthis reason, several methods have been suggested so that each can-didate may be evaluated with the most favorable scoring vector forhim/her. In this work we propose new positional voting systems thatallow to use different scoring vector for each candidate and avoid someshortcomings of the methods proposed in the literature.

205


2 - Rank-scaled Integral Indicators of Ecological ImpactMikhail Kuznetsov, Moscow Institute of Physics andTechnology, Russian Federation,[email protected] compare objects or alternative decisions one must evaluate a qual-ity of each object. An integral indicator is a real-valued scalar which iscorresponded to the object. We can construct an integral indicator hav-ing a set of objects and features and a quality criterion. We use the setof features and expert-given indicators in the rank scales. That is, everyobject is ranked by the certain feature. We consider every rank-scaledfeature as a polyhedral cone. To find final ranking we use Minkowskisum of this cones. This methodology is used for the Croatian ThermalPower Plants evaluation.

3 - Fast order recommendation separately for each ex-plicit customer of large product systemPeter Vojtas, Software Engineering, Charles University,Faculty of Mathematics and Physics, Malostranske nam. 25,118 00, Prague, Czech Republic, [email protected],Alan EckhardtOur motivation is a personalized recommendation for a web based e-shop (no aggregated advices like average user, most popular ...). Ourinterest is in support of big number of users (each considered individ-ually) for big number of alternatives (usually simple completely de-scribed products - not easy in web scaled applications). Our input isexplicit user rating of a sample of products. We propose a data miningmethod, which based on sample rating, predicts rating (order) of allproducts. We calculate user’s personal objectives and utility and out-put top-k recommendations. We evaluate our experiments wrt. speed(usually < 1 sec.) and order violations (e.g. Kendal tau rank).

4 - A bi-objective linear programming model for haz-ardous medical waste managementMarta Castilho Gomes, CESUR, Instituto Superior Técnico,Av. Rovisco Pais, 1049-001, Lisboa, Portugal,[email protected], Joaquim R. DuqueInterest in medical waste management has raised in recent years acrossEurope. In this work a bi-objective mixed-integer linear programmingmodel for hazardous medical waste (HMW) management is presented.It accounts for construction and operating costs of facilities (eitherfor waste decontamination or incineration), transportation costs andthe environmental assessment of solutions using the Eco-indicator 99methodology. The problem of expanding the HMW incineration ca-pacity in Portugal is addressed and real data used for an economic-environmental trade-off analysis (with a Pareto curve).


Web-based Spatial-temporal CollaborativeDecision Support Systems

Stream: Decision Support SystemsInvited sessionChair: Fatima Dargam, SimTech Simulation Technology, RiesStrasse 120, 8010, Graz, Austria, [email protected]

1 - Scalarising methods in DSS WeboptimVassil Guliashki, Optimization and Decision Making, Inst. ofInformation and Communication Technologies - BAS, Acad.G. Bonchev str., block 2, 1113, Sofia, Bulgaria,[email protected], Leoneed Kirilov, Krassimira Genova,Boris StaykovWe present a Decision Support System for solving Multiple - ob-jective Optimization Problems. Its basic characteristics are: user-independent, multisolver-admissibility, method-independent, hetero-geneity, and web-accessibility. Core system module is an original gen-eralized method including most MOP methods — reference point ap-proach, reference direction approach, interval approach and etc. Thechoice of suitable scalarising function is made automatically depend-ing on the Decision Maker’s dialog. The system provides an API in-terface for external use by third party developers.

2 - Hope: a web-based SDSS for urban managementJoao Coutinho-Rodrigues, Dep. Civil Engineering, Universityof Coimbra, Faculty of Sciences & Tech. andINESC-Coimbra, Polo II, 3030-290, Coimbra, Portugal,[email protected], Lino Tralhão, Eduardo ManuelNatividade-Jesus

Given the urban population growth and the increase of complexity ofurban systems, engineers and decision makers have to face new dailychallenges in urban management. OR techniques combined with mod-ern information technologies may provide new avenues to tackle suchchallenges. An ongoing development of a GoogleMaps-based pro-totype - supporting user graphical edition and report through a webbrowser in a PC (or in a smartphone generating in situ georeferencedreports with alphanumeric and image contents) - embedding a multi-criteria models base running in an algorithm server, will be presented.

3 - Criteria weighting considering the interaction with in-stitution’s goalsSheu-Hua Chen, Distribution Management Dept., NationalChin-Yi University of Technology, 35, Lane, Section 1,Chung-Shan Road,Taipin, 411, Taichung, Taiwan,[email protected], Hong Tau Lee

Conventionally, to engage new teacher the institution’s goals have to beconsidered first, and then the criteria and their relative weights are de-termined. In this research, we claim the determined weights for thesecriteria may not appropriately reflect the preset priority of the goals.This research proposes a faculty selection mechanism that was devel-oped based on the analytic network process to handle such an interde-pendent situation. The nature of the proposed mechanism gives deci-sion makers the chance to recheck for possible bias in criteria weight-ing.


OR Military and Security Applications

Stream: OR in Military, Safety and Security Applica-tionsInvited sessionChair: Ana Isabel Barros, Military Operations, TNO, POBox96864, 2509 JG, The Hague, Netherlands, [email protected]

1 - Interdiction in queueing networksTom van der Mijden, Twente University, Netherlands,[email protected]

In the literature on network interdiction focus is on time independentmodels, although one can imagine situations in which a successful in-terdiction can only take place in certain time windows. In our researchwe will consider time dependent network interdiction by consideringan interdiction game played on a network of queues. An operatorroutes customers through the network towards a destination, while aninterdictor attempts to intercept these customers during their servicetimes at the queues. We analyze optimal strategies for both the opera-tor and the interdictor.

2 - A Smuggling Game with the secrecy of smuggler’sinformationRyusuke Hohzaki, Department of Computer Science, NationalDefense Academy, 1-10-20 Hashirimizu, 239-8686,Yokosuka, Kanagawa, Japan, [email protected]

This report deals with a two-person zero-sum multi-stage smugglinggame. Customs and a smuggler are allowed to take an action of patroland smuggling, respectively, within the limited number of chances.Customs obtains reward by the capture of the smuggler and the smug-gler gets reward by the success of smuggling. Almost all past re-searches modeled their games by complete information game. In thisreport, we deal with a game with asymmetric information, where onlyCustoms does not know past behavior of his opponent. We evaluatethe value of information by deriving Bayesian equilibrium.

206


3 - Modelling covert social networksAna Isabel Barros, Military Operations, TNO, POBox 96864,2509 JG, The Hague, Netherlands, [email protected], Bobvan der Vecht

Social Network Analysis has shown its potentials and shortcomings inthe analysis of covert networks in a military intelligence and criminalinvestigation setting. In practice however, deciding whether a relationbetween two individuals in a covert network is present or not given theavailable information is not an easy task. Often partial and/or unre-liable and/or contradicting information is available. In this paper wewill introduce a structured method to merge information taking intoaccount the reliability of the available sources used and context knowl-edge available.


Support Vector Machines. New results

Stream: Optimization and Data MiningInvited sessionChair: Theodore Trafalis, Industrial Engineering, University ofOklahoma, 202 West Boyd, Rm 124, 73019, Norman, OK, UnitedStates, [email protected]

1 - Feature Ranking for Support Vector Machines withSecond-Order Cone ProgrammingSebastian Maldonado, School of Engineering and AppliedSciences, Universidad de los Andes, Av. San Carlos deApoquindo 2200, Las Condes, Santiago, Chile,[email protected]

This work addresses the issue of feature ranking for linear and kernel-based SVMs considering Second Order Cone Programming formula-tions. These formulations provide a robust and efficient framework forclassification, while an adequate feature selection process avoid errorsin estimation of mean and covariances. Our approach is based on a se-quential backward elimination, proposing different contribution mea-sures to determine the feature relevance. Experimental results on reallife microarray data sets demonstrate the effectiveness and efficiencyof our approach.

2 - A collaborative recommendation using edgeweighted SimRankAyaka Inoue, Graduate School of Science and Technology,Sophia University, Kioi-cho 7-1, Chiyoda-ku, 1028554,Tokyo, Tokyo, Japan, [email protected], YuichiroMiyamoto

We deal with a recommendation method based on the collaborative fil-tering. Our recommendation approach is an elementary one based onvoting by similar customers. We adopt BipartiteSimRank as a measureof similarities between customers (items). Our main contribution is anextension of SimRank by introducing edge weights for practical use inelectric commerce. We also report a computational result using a real-life data that includes a purchase history, personal information and siteaccesses of each customer.

3 - An Adaptive Weighted Kernel Technique for OnlineTraining with Imbalanced DataTheodore Trafalis, Industrial Engineering, University ofOklahoma, 202 West Boyd, Rm 124, 73019, Norman, OK,United States, [email protected], Nicolas Couellan

Kernel methods are widely useful for data classification problems. Inthe case of imbalanced datasets kernel techniques perform poorly inpredicting the minority class for imbalanced data. Imbalanced datasetsexist in several application areas such as medical diagnosis problems.We develop an incremental weighted scheme of classifying imbalanceddata online and provide computational results for several imbalanceddata sets. Our results show that the proposed incremental scheme out-performs a batch version of weighted SVM both in terms of minorityclass prediction accuracy and training speed.



Stream: DEA and Performance Measurement (con-tributed)Contributed sessionChair: Juha Eskelinen, Aalto School of Economics, 00076, Aalto,Finland, [email protected]

1 - Scale efficiency analysis of Brazilian logistic serviceproviders industryCarlos Ernani Fries, Department of Production and SystemsEngineering, Federal University of Santa Catarina, CampusTrindade - CTC/EPS, C.P. 476, 88010-970, Florianópolis,Santa Catarina, Brazil, [email protected], Mônica M. M.Luna, Antonio G.N. NovaesThe Brazilian market for logistics services has experienced since the90s a considerable increase in competition. To attain greater efficiencyLSPs have adopted new strategies such as specialization of offered ser-vices or spreading out activities through firm growth. It is questioned ifthe limits of growth of LSP-scale enterprises have already been reachedor if there is room for further expansion. DEA evaluation of scale ef-ficiency of LSP in Brazil based on secondary data from 2004-2011contributes to the definition of most promising scale for companies op-erating in this market.

2 - Estimating the longevity of breeding individuals inbeef cattle herdsJuliana Cobre, Instituto de Ciências Matemáticas e deComputação, Universidade de São Paulo, AvenidaTrabalhador são-carlense, 400, 13566-590, São Carlos, SãoPaulo, Brazil, [email protected], Erlandson Saraiva,Sabrina CaetanoBeef cattle breeding is particularly important in the Brazilian meatmarket. The longevity maximization of breeding individuals gener-ates higher revenues, reducing involuntary culling and also allowinggenetic grow. The goal of this study is to model variables risk, calvinginterval and age of cow at last calving, and to obtain the permanencerate of cows in the herd (long duration). We propose a joint frailtyparametric model. This modelling considers calving dates, presence ofcensored data and covariates. Computational results on a Nelore cattledata set validate this study.

3 - Intertemporal Sales Efficiency Analysis of a BankBranch Network: A Stochastic Nonparametric Ap-proachJuha Eskelinen, Aalto School of Economics, 00076, Aalto,Finland, [email protected], Timo KuosmanenWe propose a new approach to assess the development of sales ef-ficiency over time. It uses a stochastic nonparametric approach,StoNED. This allows us to eliminate from our efficiency estimates theimpacts of both stochastic noise and changes in demand and exoge-nous conditions. The project is conducted in close collaboration with aretail bank. The monthly efficiencies of the branches of the case bankwere estimated over several years. Our approach is applicable to theperformance management of financial institutions and other networkbased sales organizations.

4 - An automatic democratic approach to weight settingapplied to the new human development indexChris Tofallis, Business School, University of Hertfordshire,College Lane, AL10 9AB, Hatfield, Herts., United Kingdom,[email protected] selection for a multi-dimensional index is a thorny problem.This is often bypassed by the ’agnostic’ option of equal weights, asin the Human Development Index — a ranking of nations by the UNDevelopment Programme. Whatever weights are chosen, some nationswill feel disadvantaged. We propose a 2-step approach to avoid arbi-trary weights: (1) Find the most advantageous set of weights for eachnation in turn - we use multiplicative DEA as the new HDI is based ona weighted product. (2) Regress the associated optimal scores on thecomponents to find a single weight set.

207




Stream: DEA and Performance Measurement (con-tributed)Contributed sessionChair: Sanjeet Singh, Operations Management, Indian Institute ofManagement Calcutta, DH Road, Joka, 700104, Kolkata, WestBengal, India, [email protected]

1 - Measuring Team PerformancesKei Ogiwara, Operating Research, 84-4 Aza EbinokuchiTsuchiya, 015-0055, Yurihonjo City, Japan,[email protected], Tsutomu Mishina, Alejandra GomezPadilla

Teams are organized at any given time to pursue high levels of outputin common business settings. Forming an effective team depends onwho is involved in it and how the results can be evaluated as the fi-nal team outputs. Each employed member brings the necessary skillsto the team to accomplish individual targets. Generally teams havemultiple targets, and the team success rests on how to deploy suitablecombinations of the members’ skills; these issues are treated as teamfunctions. Managing a team is another factor to success. Consideringjointly team functions and management is required.

2 - Group cohesion and organizational performanceCristina-Petronela Durneac, Sociology, S.N.S.P.A., Romania,[email protected]

This paper aims to make a contribution to understanding the level ofcohesion and performance in economic and financial organizations inRomania. The analysis was performed in 12 economic and financial or-ganizations from Bucharest. Based on quantitative data obtained froman organizational survey, the general goal of the research is to make acontribution to the in-depth exploration, better understanding and ex-planation of the existence and the importance of group cohesion onorganizational performance.

3 - DEAHP Approach for Manpower Performance Evalu-ationSanjeet Singh, Operations Management, Indian Institute ofManagement Calcutta, DH Road, Joka, 700104, Kolkata,West Bengal, India, [email protected], Remica Aggarwal

The manpower in a company or organization constitutes an importantand essential asset to an organization. In this paper we have integratedDEA to generate local weights of alternatives from pairwise compari-son of judgment matrix used in the Analytic Hierarchy Process (AHP)for a three attribute system for measuring performance of manpowerat different levels of managerial hierarchy. Multiple expert judgmentshave been considered for weight determination of the attributes. DEA-AHP (DEAHP) has been proposed in this paper as an alternative to thetraditional methods of weight derivation in AHP.


OR in Sustainable Urban Development

Stream: OR for Sustainable DevelopmentInvited sessionChair: Vida Maliene, School of the Built Environment, LiverpoolJohn Moores University, Byrom Street, L3 3AF, Liverpool, UnitedKingdom, [email protected]

1 - Sustainable and hhealthy housing: exploring stake-holders’ viewsAgne Prochorskaite, School of the Built Environment,Liverpool John Moores University, Liverpool John MooresUniversity, Byrom Street, L3 3AF, Liverpool, UnitedKingdom, [email protected], Vida MalieneSustainable housing policy and initiatives need greater alignment withhealth and well being agenda. However, implementation of this is ahighly complex problem as multiple objectives and different stake-holders’ views need to be considered. The paper presents a sustainableand healthy housing criteria system based on a comprehensive reviewof existing codes, standards and literature on housing and health. Theaim of this is to allow evaluation and ranking of dwellings according todifferent stakeholder priorities of sustainability, health and well-beinghousing features.

2 - Development of a multicriteria model for comprehen-sive assessment of residential unitsJana Selih, Department of Civil and Geodetic Engineering,University of Ljubljana, Jamova 2, 1000, Ljubljana, Slovenia,[email protected] change of the buyers’ position in the current economic recessionconceived a need for a tool used to assess various features of the resi-dential unit, and make the decision regarding the purchase on a ratio-nal basis. The proposed model is based on five main criteria: location,architectural design, technical quality, living comfort and energy ef-ficiency. The sub-criteria enable thorough but still robust descriptionof the property of the unit. The applicability of the proposed modelis exhibited by the parametric analysis, where different units withinSlovenian territory are assessed.

3 - Sustainable development problems of the apartmentpolicy in LatviaSanda Geipele, Riga Technical University, Mezha Street 1/7,212 room, LV-1048, Riga, Latvia, [email protected],Ineta Geipele, Iveta StamureThe aim of the research is to state the problems of the Social economicenvironment within the apartment policy in Latvia. Results discov-ered would be the base to elaborate suggestions for optimization ofdwelling technologies. The mathematic, monographic and statisticalforecast methods were applied. The general evaluation of economicaland ecological situation of two housing funds and technologies of thedwelling houses elaborated with defining principles of rent price form-ing justifies chances of the sustainable development of social economicenvironment in Latvia.

4 - Model for municipal real estate strategic manage-mentEgl Klumbyt, Department of Civil Engineering Technologies,Kaunas University of Technology, Krasevskio 30-25,LT-50280, Kaunas, LITHUANIA, Lithuania,[email protected], Rasa ApanavicieneMunicipalities are usually the largest real estate owners and managers.Only about 30% of major European cities are able to provide amountand value of their real estate portfolios. The Lithuanian Free MarketInstitute has introduced its first Index of Municipalities in Lithuaniaand states, that none of the Lithuanian municipality has been devel-oped its real estate management strategy. The article presents the newmodel helping to manage municipal real estate management effectivelyby taking into account the priorities of strategic economic and socialdevelopment tendencies of the region.


Games and decision making

Stream: Cooperative Game TheoryInvited sessionChair: Sirma Zeynep Alparslan Gok, Mathematics, Faculty of Artsand Sciences, Suleyman Demirel University, Faculty of Arts andSciences, Suleyman Demirel University, Department ofMathematics, 32260, Isparta, Turkey,[email protected]

208


1 - Coalition formation in a bargaining game with a fixedpayments contract and delegationHaruo Imai, KIER, Kyoto University, Yoshida HOnmachi,Sakyo, 606-8501, Kyoto, Japan, [email protected]

Coalition formation for bargaining is examined when coalitional con-tracts are limited to delegation to a member in return of fixed pay-ments (adopted in the uniqueness literature). Specifically, SSPE ofthree stage games (coalition formation, coalitional bargaining, and bar-gaining game) are investigated. Such equilibria exist and nontrivialcoalition structures emerge. We show that for order-independent equi-libria, the softest player tends to be favored, but such equilibria maynot exist, by means of a CRRA example. Iterating above stages, alsowe show that a nested coalition structure emerges.

2 - The impact of issue salience and distance in proba-bilistic spatial voting indexesTom Blockmans, MOSI, Vrije Universiteit Brussel, Pleinlaan2, 1050, Brussel, Belgium, [email protected], FrankPlastria, Stefano Benati, Guiseppe Vittucci Marzetti

We present a generalization of a probabilistic spatial voting index. Theimpact of the distance function on the results of the model is exam-ined, comparing between (squared) Euclidean and Manhattan distance.A distinction is made between the results that are weighted and un-weighted by issue salience, representing the parties’ policy priorities.Instead of using the bliss points of the parties as a proxy for the coali-tion, we use an alternative method for estimating the coalition consen-sus. Empirical applications illustrate the influence these three factorshave on the voting indexes.

3 - Blocks of flats renovation gameAgne Reklaite, Vilnius University, Lithuania,[email protected]

In this paper a renovation of blocks of flats is studied. A game is con-structed, which allows government to organise full-scale renovation,effectively use the resources of residents and have relatively low gov-ernment expenses on renovation. In the presented game the decisionsof individual residents lead to an efficient equilibrium. Vilnius’ centralheating data is analysed and optimal strategies are built for residentsof several types of apartment blocks. The results indicate that resi-dents’ decisions highly depend on the expected heating price but maybe influenced by other factors.

4 - On the interval Shapley value: two new characteriza-tionsSirma Zeynep Alparslan Gok, Mathematics, Faculty of Artsand Sciences, Suleyman Demirel University, Faculty of Artsand Sciences, Suleyman Demirel University, Department ofMathematics, 32260, Isparta, Turkey,[email protected]

The Shapley value, one of the most common solution concepts in Oper-ations Research applications of cooperative game theory, was definedand axiomatically characterized in different game-theoretic models. Inthis study we focus on the Shapley value for cooperative games wherethe set of players is finite and the coalition values are compact inter-vals of real numbers. We give two new characterizations of the intervalShapley value with the aid of the properties of efficiency, symmetryand strong monotonicity and by using interval dividends.


Optimization Modeling I

Stream: Software for OR/MSInvited sessionChair: Bjarni Kristjansson, Maximal Software, Ltd., BoundaryHouse, Boston Road, W7 2QE, London, United Kingdom,[email protected]

1 - Solving Constraint Programming problems withAIMMSGuido Diepen, AIMMS, Schipholweg 1, 2034 LS, Haarlem,Netherlands, [email protected]

Besides the support for the traditional Mathematical Programming like(Non-)Linear Programs and Mixed Integer Programs, AIMMS nowalso supports Constraint Programming (CP). Typically CP performsvery well on highly combinatorial problems, which can be difficult tomodel/solve with mathematical programming. AIMMS supports bothcommon global constraints from CP (e.g., count/alldifferent), as wellas specific scheduling constraints, which are useful when modelingscheduling problems in AIMMS.

2 - LocalSolver: black-box local search for combinato-rial optimizationJulien Darlay, e-lab, Bouygues SA, 32 avenue Hoche, 75008,Paris, France, [email protected], Thierry Benoist,Bertrand Estellon, Frédéric Gardi, Romain Megel, KarimNouioua

We present LocalSolver 2.0, a black-box solver for combinatorial op-timization based on local-search techniques. It can handle very largenonlinear problems with millions of 0-1 variables on a standard com-puter. LocalSolver offers simple APIs as well as an efficient modelinglanguage for fast prototyping. It has been tested on classical bench-marks and succeeded the first phase of Google ROADEF/EURO chal-lenge. Moreover LocalSolver is used in several industrial applications:maintenance planning, energy optimization, TV media planning.

3 - Deploying MPL Optimization Models on Servers andMobile PlatformsBjarni Kristjansson, Maximal Software, Ltd., BoundaryHouse, Boston Road, W7 2QE, London, United Kingdom,[email protected], Sandip Pindoria

The IT industry is currently undergoing a major shift, away from tra-ditional standalone applications, to new platforms such as tablet com-puters and mobile phones. In this presentation, we will demonstratingfor the first time, a new server-based version of MPL OptiMax, thatmakes writing mobile applications relatively quick and easy process.We will demonstrate how to integrate optimization models seamlesslywith online data, and then deploying them on a server for servicingboth web and mobile clients, using standard programming languages,such as CSharp, VB, C/C++ or Python.

� WB-37Wednesday, 10:30-12h00CC-Act

OR in Health & Life Sciences 6Stream: OR in Health & Life Sciences (contributed)Contributed sessionChair: Vassilis Kostoglou, Department of Informatics, AlexanderTEI of Thessaloniki, P.O. Box 141, 57400, Thessaloniki, Greece,[email protected]

1 - Improve OR utilization and reduce number of re-quired bedsTheresia Van Essen, University of Twente, Netherlands,[email protected], Johann Hurink, Erwin Hans

Due to increasing health care costs, hospitals are forced to reduce thenumber of beds on the nursing wards while still increasing their pro-duction. In this presentation, we solve this problem by creating an op-erating room (OR) schedule which maximizes the OR utilization andminimizes the number of required beds. The solution approach con-sists of two steps: (1) generate surgery blocks that maximize OR uti-lization and satisfy several resource constraints and (2) assign blocksto ORs and days such that the number of required beds is minimized.The solution approach is tested on real-life data.

209


2 - Optimal experimental design for a pure birth processwith incomplete informationAli Eshragh Jahromi, School of Mathematical Sciences, TheUniversity of Adelaide, North Terrace Campus, 5005,Adelaide, South Australia, Australia,[email protected]

Our goal is to estimate the rate of growth of a population governed bya stochastic model. We may choose n time points at which to count thenumber of individuals present, but due to detection difficulties, or con-straints on resources, we are able only to observe each individual withfixed probability p. We discuss the optimal times at which to make ourn observations in order to maximize Fisher Information. For computa-tional and analytical reasons which will be discussed, we specificallyfocus on the cases n=1 and n=2, presenting both theoretical and nu-merical findings.

3 - Implementation of an OR model for the comparisonof higher technological education specialtiesVassilis Kostoglou, Department of Informatics, AlexanderTEI of Thessaloniki, P.O. Box 141, 57400, Thessaloniki,Greece, [email protected], Michael Vassilakopoulos,Christos Koilias

This work focuses on the comparison of the specialties provided bythe higher technological education regarding graduates’ employmentstatus and vocational prospects. A survey OR model is introducedconsisting of original data collection from 5,183 Greek graduates cor-responding to 45 specialties through a structure questionnaire and tele-phonic interviews. Bivariate, multivariate, and cluster analysis iden-tified the statistically significant differences, leading to findings andproviding guidelines for the selection of a field of studies that leads toa more promising professional career.

� WB-38Wednesday, 10:30-12h00HH-Colombus

Multiobjective Optimization andTransportation

Stream: Multiobjective OptimizationInvited sessionChair: Anatoly Levchenkov, Riga Technical University, 1658,Riga, Latvia, [email protected]: Mikhail Gorobetz, Institute of Industrial Electronics andElectrical Engineering, Riga Technical University, 1 Kalku Street,Riga, Latvia, [email protected]

1 - Multi-Objective Road Pricing: A Game Theoretic andMulti-Level Optimization ApproachAnthony Ohazulike, Applied Mathematics, University ofTwente, Drienerlolaan 5, 7522 NB, Enschede, Netherlands,[email protected], Georg Still, Walter Kern, Eric vanBerkum

Using a game theoretical approach, we develop a pricing schemethat internalizes multiple traffic externalities. Further, we extend thesingle authority road pricing scheme to a scheme with multiple ac-tors/stakeholders or regions. Road users’ interests are represented inthe upper and the same level as the decision makers, thus, makingthem active players in the toll setting game. Having shown that pureNash equilibrium (NE) toll may not exist among the stakeholders (withlikely opposing objectives), we design a mechanism that induces NEwhich coincides with system optimum.

2 - Multi-objective optimization with an immune algo-rithm for a railway safety control systemAndrew Mor-Yaroslavtsev, Riga Technical University,Kronvalda bulv.1, LV-1010, Riga, Latvia, [email protected],Anatoly Levchenkov

This paper describes an immune clonal selection algorithm used to-gether with a negative selection algorithm for use in an intelligentrailway electric vehicle safety system. Embedded devices collect data

about the rolling stock location and status, communicate it to the serverand other devices, and use it to avoid dangerous situations. The authorsexamine the means of communication between the embedded devicesand their effectiveness. The authors review data analysis methods usedto detect, predict and control undesirable rolling stock travel condi-tions.

3 - A Multi-Objective Minimum Cost Flow Problem to De-sign Safe Walking-Routes for School ChildrenKen-ichi Tanaka, The University of Electro-Communications,1-5-1, Chofugaoka, Chofu, Tokyo, Japan, 182-8585, Tokyo,Japan, [email protected], Ryuhei Miyashiro, YuichiroMiyamoto

For school children to walk safely to and from school, it is importantto walk in groups, not alone. However, walking together may forcechildren to walk longer distances than those of their shortest paths. Amulti-objective minimum cost flow problem is presented in which boththe total walking distance and the total distance walked alone for chil-dren are minimized. Pareto optimal solutions show that the total dis-tance walked alone can be greatly reduced by making children slightlydeviate from their shortest paths. Heuristically obtained solutions fora real road network are also analyzed.

4 - Use of Adaptive Control Systems in Multi-CriteriaTasks in Electric Transport ControlAndrejs Potapovs, Faculty of Power and ElectricalEngineering, Riga Technical university, Riga Kalku 1,LV1658, Riga, Riga, Latvia, [email protected]

Nowadays, there are very topical issues, which are connected with thesecurity of rail transport routing and it is caused by the fact that trainintensity and the speeds, which raise the requirements for an obser-vation of a train movement timetable and for the safety level of alltrain movement, are constantly increasing. The following assumptionis proposed: a programmable logic controller (PLC) with a controlprogram on the basis of the adaptive control algorithms, smooth andprecise braking of a train in multi-criteria motion conditions in a sin-gle train anti-collision system.

� WB-39Wednesday, 10:30-12h00HH-Cousteau

Dynamic Programming 2

Stream: Dynamic ProgrammingInvited sessionChair: Martina Hesse, Chair of Production and Lgistics,Georg-August-Universität Göttingen, Platz der Göttinger Sieben 3,37073, Göttingen, Germany,[email protected]

1 - A Dynamic Classification and Prediction Model forRoad Departure Warning SystemsAndre Possani Espinosa, Department of Digital Systems,Instituto Tecnológico Autónomo de México, Rio Hondo No.1,01080, Mexico City, D.F., Mexico, [email protected],Marta Cabo Nodar, Edgar Possani

One of the main tasks identified in the SAFESPOT IP Project, co-funded by the European Union, was the development of a Road De-parture Prevention Application. This application relies on the identi-fication and prediction of vehicle trajectories in order to provide in-advance warnings to drivers and therefore prevent deadly road acci-dents. Continuing with this work, we will present an application of adynamic adaptive model evaluated with real and simulated data. Re-sults obtained are compared with previously implemented techniques.

2 - Some applications of Optimal Control in SustainableFishing in the Baltic SeaDmitriy Stukalin, Universität Greifswald, GreifswaldUniversity, Rathenau-Str., 47, 17489, Greifswald, Germany,[email protected]

210


Problems related to the implementation of optimal control theory forthe three-dimensional dynamic model cod-herring-sprat (the fish pop-ulations management problem) are presented. They belong to a classof models called Lotka-Volterra models influenced by several natu-ral processes and human activities. Interacting species, which inhabitin a common habitat with limited resources, will be considered. Theproblem of optimal harvest policy is solved for the controls of variousclasses and their behaviour is analyzed.

3 - Supporting Tailor-made Performance ManagementSystems by System Dynamics — A Multiple CriteriaBased Software EvaluationWolfgang Ossadnik, Department Business Management /Managerial Accounting, University of Osnabrueck,Rolandstrasse 8, 49069, Osnabrueck, Germany,[email protected], Ralf Kaspar

A fundamental task of Management Accounting is the development,implementation and continuous improvement of Performance Manage-ment Systems (PMS). These can be visualized by strategy maps whichlink key performance indicators according to relevant cause-and-effectrelationships. Based on the dynamic character of PMS the use of Sys-tem Dynamics (SD) is necessary to take into account temporarily de-laying and retroactive influences. In our study, we show the process ofdynamic modeling in performance management as well as the resultsof a multi-criteria evaluation of SD software products.

4 - Modeling carbon storage in cascading usage sys-tems of wood and derived timber productsMartina Hesse, Chair of Production and Lgistics,Georg-August-Universität Göttingen, Platz der GöttingerSieben 3, 37073, Göttingen, Germany,[email protected], Jutta Geldermann

The carbon storage period of forests can be extended by a material useof wood and derived timber products: the storage effect of forests iscontinued in wooden products. A serial connection of multiple mate-rial applications including re- and upcycling loops with a final ener-getic use describes the idea of cascading biomass resources. Relevantparameters and data sources for modeling carbon flows and sinks arepresented and the adequacy of multi-agent and dynamic system mod-eling approaches is discussed.

� WB-40Wednesday, 10:30-12h00HH-Livingstone


Stream: Preference LearningInvited sessionChair: Willem Waegeman, Department of Applied Mathematics,Biometrics and Process Control, Ghent University, Coupure links653, 9000, Ghent, Belgium, [email protected]

1 - Preference-based reinforcement learningWeiwei Cheng, Mathematics and Computer Science,University of Marburg, Germany, [email protected]

We discuss the integration of two sub-fields of machine learning,namely preference learning and reinforcement learning (RL). Whileconventional RL methods are restricted to deal with numerical rewards,in many applications the feedback signals are qualitative but not quan-titative. Therefore, building on novel methods for preference learning,we equip the RL agent with qualitative policy models. We build onan existing method for approximate policy iteration based on roll-outsand make use of a specific type of preference learning methods, thelabel ranking method.

2 - Preference-based policy learningRiad Akrour, INRIA Saclay Île-de-France, 91405, OrsayCedex, France, [email protected], Marc Schoenauer,Michèle Sebag

The expert-robot interaction is framed as a Learning-to-rank problem:the robot demonstrates a policy, that is ranked by the expert w.r.t. toprevious policies. Active Ranking and Reinforcement Learning arecombined: the robot learns an estimate of the expert’s preferences, anddemonstrates the most informative policy w.r.t. this estimate. The dif-ficulty comes from the fact that the preference estimate is learned fromthe behavioral space, while the active learning criteria are defined onthe parametric space. The resulting approach is discussed w.r.t. InverseRL on standard benchmarks.

3 - A simple algorithm for multi-label rankingKrzysztof Dembczynski, Institute of Computing Science,Poznan University of Technology, Piotrowo 2, 60-965,Poznan, Poland, [email protected], WojciechKotlowski, Eyke Hüllermeier

We discuss the problem of rank loss minimization in the setting ofmulti-label classification. In this setting, the rank loss counts the num-ber of label pairs, in which a relevant label does not precede an ir-relevant label in the predicted ranking. We show that a simple al-gorithm that trains a binary classifier for each label independently iscompetitive to the more complex state-of-the-art approaches that relyon minimization of pairwise convex surrogates of the rank loss. Dur-ing the talk, we will give theoretical arguments that justify this simpleapproach.

� WB-41Wednesday, 10:30-12h00HH-Heyerdahl

System Dynamics: demographics andmethodology

Stream: System Dynamics Modeling and SimulationInvited sessionChair: David Wheat, Geography, University of Bergen, NedreFjellsmug 6, 5018, Bergen, Norway, [email protected]

1 - A Simulation Model of Demography & Economy inLithuaniaDavid Wheat, Geography, University of Bergen, NedreFjellsmug 6, 5018, Bergen, Norway, [email protected]

Lithuania’s shrinking labor force is a much-discussed result of emigra-tion and more deaths than births. Yet, the long-run macroeconomicscenario is uncertain. If demographic dynamics prove problematic foreconomic growth, the reinforcing feedback effects will further discour-age family formation and encourage emigration—a vicious circle. Willcounteracting feedback effects be strong enough to avoid economicstagnation? This paper examines this question with a system dynamicsmodel that integrates cross-disciplinary theories with data from Lithua-nia’s economy and demography.

2 - A Simulation-based Analysis of Trends in Immigra-tion and CrimeRichard McDowall, University of Bergen, VilhelmBjerknesvei 57, 5081, Bergen, Norge, Norway,[email protected]

The connection between immigration and crime is a contentious is-sue in debates over immigration policy in European countries, in partbecause there is little consensus over the extent of immigrant criminal-ity. This paper contributes an endogenous explanation of crime trendsin Oslo, using a dynamic simulation model adapted from a cohort-survival model of immigrant populations and criminal behavior. Theultimate contribution of this study is the provision of a tool to aid crim-inal justice capacity planning and immigration policy design.

3 - Diagramming a Feedback Model of EmigrationRasa Cincyte, University of Bergen, Norway,[email protected], David Wheat

Research on the emigration issue reveals the absence of a unified the-ory to guide data collection and analysis, as well as policy design. Re-cent contributions narrow this literature gap by identifying emigrationdrivers that cross disciplinary boundaries and form feedback loops.

211


This paper extends the endogenous perspective with causal loop di-agramming—a system dynamics tool. Another contribution of thispaper is a conceptual framework to facilitate developing a simulationmodel for empirically testing a unified emigration theory and policyoptions in the Lithuanian context.

4 - Modelling the Mechanisms of Behaviour Change atthe Population level in a Social Marketing ContextBrian Dangerfield, Centre for Operations Management,Management Science & Statistics, University of Salford,Salford Business School, Maxwell Building, The Crescent,M5 4WT, Manchester, United Kingdom,[email protected], Norhaslinda Zainal Abidin

We need to better understand the drivers and mechanisms ofpopulation-level human behaviour change when the objective is toachieve a societal benefit. A review is made of the recent contributionswhich system dynamics has offered for the design, comparison andevaluation of societal policies in a social marketing context. Specifi-cally, in the case of childhood obesity, a description is given of how asystem dynamics model might be used to aid the design of policies topromote beneficial changes in eating habits.

� WB-42Wednesday, 10:30-12h00BW-Amber

Energy Systems Planning

Stream: Stochastic Modeling and Simulation in Engi-neering, Management and ScienceInvited sessionChair: Erik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39, 85577,Neubiberg, Germany, [email protected]

1 - Planning Critical Resources for Offshore Oil FieldDevelopment under UncertaintyLeonidas Sakalauskas, Operational Research, Institute ofMathematics & Informatics, Akademijos 4, LT-08663,Vilnius, Lithuania, [email protected], Leonardo Moraes, LuizCarlos Sousa, Roger Rocha, Adilson Elias Xavier

This work considers the problem of planning critical resources for off-shore oil field development for a horizon of one year. Field develop-ment activities comprise a complex mix of resources that are intercon-nected in a way that their management is a challenging task due to theuncertainties of the activities durations. In an oil field developmentone of the key elements are the rigs, however they can only operateeffectively if they are supplied by a certain amount of resources on theright time. We model this problem using stochastic programming andcompare with the deterministic model.

2 - A decision rule approach to medium-term hy-dropower scheduling under uncertaintyPaula Rocha, Department of Computing, Imperial CollegeLondon, SE7 2AZ, London, United Kingdom,[email protected], WolframWiesemann, Daniel Kuhn

We present a multistage stochastic optimisation model for the medium-term scheduling of a cascaded hydropower system. Electricity spotprices change on a much shorter time scale than the hydrological dy-namics of the reservoirs in the cascade. We exploit this property to re-duce computational complexity: we partition the planning horizon intohydrological macroperiods, and we account for intra-stage price vari-ability by using price duration curves. Moreover, we restrict the spaceof recourse decisions to those affine in the observable data, therebyobtaining a tractable approximate problem.

3 - Refinery Operations Planning under UncertaintyLeonardo Nascimento, Federal University Rio de Janeiro,Brazil, [email protected], Helder Venceslau, AdilsonElias Xavier, Virgilio Jose Martins Ferreira Filho, LeonidasSakalauskas, Roger Rocha

Oil refining is a series of processes that aim to separate the crude oilinto pre-standardized fractions. The way these processes can be com-bined result in a variety of schemes where each one can be used asa production plan. This work presents a methodology, based uponstochastic programming (SP) that support the decision makers in themid-term operations planning of an oil refinery. Results generated byrunning a multi-period two-stage SP model are used to measure the im-pact on the economic efficiency when not considering the randomnessof the demand and the receipt of crude oil.

� WB-43Wednesday, 10:30-12h00BW-Granite

Extended Two-Stage Linear StochasticPrograms

Stream: Stochastic ProgrammingInvited sessionChair: Rüdiger Schultz, Mathematics, University ofDuisburg-Essen, Lotharstr. 65, D-47048, Duisburg, Germany,[email protected]

1 - Nonlinear Two-Stage Stochastic ProgramsRüdiger Schultz, Mathematics, University of Duisburg-Essen,Lotharstr. 65, D-47048, Duisburg, Germany,[email protected], Tobias Wollenberg

The paradigms of two-stage stochastic linear programming readily ex-tend to two-stage decision making under nonanticipativity in general.Our talk picks up this abstract observation and elaborates nonlineartwo-stage models of that kind. We report on applications of nonlineartwo-stage stochastic programing in areas as different as supply net-works for grid-based scarce commodities and shape optimization withlinearized elasticity.

2 - Some Remarks on Linear Stochastic Bilevel Pro-gramsCharlotte Henkel, Mathematics, University ofDuisburg-Essen, Forsthausweg 2, 47057, Duisburg, Germany,[email protected]

Compared to linear stochastic two-stage programs, linear stochas-tic bilevel problems (LSBP) exhibit a strongly increased complexety.Starting from a deterministic linear bilevel problem, we derive struc-tural properties for LSBPs using state-of-the-art parametric optimiza-tion techniques. As an outcome, we obtain rather weak analytical re-sults. This significantly effects risk measures and solution algorithmsfor this kind of problem. We emphazise our results by instructive ex-amples.

3 - Decomposition Methods for Stochastic Programswith Dominance Constraints Induced by Linear Re-courseDimitri Drapkin, Mathematics, University of Duisburg-Essen,Forsthausweg 2, D-47057, Duisburg, Germany,[email protected]

We develop linear programming equivalents for two-stage stochasticmodels with linear recourse and dominance constraints of first and sec-ond order. In the favourable case, where only continuous variables arepresent in the second stage, cutting-plane decomposition algorithmsare proposed and discussed along with the computational results.

4 - A stochastic programming approach to optimizationof information retrievalVitaliy Vitsentiy, International Information, Ternopil NationalEconomic University, Ternopil, Ukraine,[email protected]

The research problem of optimization of the expected relevance of re-trieved documents in search sessions with feedback is considered. Wepropose to use Multistage Stochastic Programming for this. To conveythe endogenous uncertainty with gradual resolution, non-anticipativityconstraints in a "tight" linear relaxed form were developed using hy-brid optimization methods. In experiments with simulated database of

212


documents, the proposed approach showed significantly better retrievaleffectiveness comparatively with traditional information retrieval andtwo-stage stochastic programming.

� WB-44Wednesday, 10:30-12h00BW-Marble

Collective Learning Procedures

Stream: Machine Learning and its ApplicationsInvited sessionChair: Michael Khachay, Ural Branch of RAS, Institute ofMathematics and Mechanics, S.Kovalevskoy, 16, 620990,Ekaterinburg, Russian Federation, [email protected]: Vadim Strijov, Russian Academy of Sciences, ComputingCenter, Vavilova 42-268, 119333, Moscow, Russia, RussianFederation, [email protected]

1 - Geometrical properties of connected search spacesfor binary classification problemOleksandr Frei, MIPT, Norway, [email protected] overfitting of machine learning algorithms is often caused by ex-cessive complexity of the search space. In this work we study geo-metrical properties of connected search spaces in the special case ofbinary classification problem. By introducing the technique of randomassignment of labels to data points, we show a couple of new proper-ties of splitting and connectivity profiles of an arbitrary search space.These properties will be demonstrated for the set of linear classifiers inthree-dimensional space and for the set of conjunctive rules.

2 - Approximation for Minimum Committee ProblemKonstantin Kobylkin, Optimization, IMM UB RAS, SophyaKovalevskaya st., 16, 620990, Ekaterinburg, Sverdlovskaya,Russian Federation, [email protected] consider a problem giving an approximation for NP-hard mini-mum committee problem for two sets on the plane, the special case oflearning problem: given sets A and B, a sequence of linear functionsis called a committee if for every point a of A (b of B) there exists asubsequence of functions each of which is positive (negative) over a(over b) with length exceeding a half of the sequence length; the prob-lem is to find the smallest length committee for A and B. Experimentsshow the approximation problem to be reducible to the same one forsets having less cardinality.

3 - Combinatorial theory for ensemble of algorithmsAndrey Ivakhnenko, Moscow Institute of Physics andTechnology, Mocsow, Russian Federation, [email protected] combinatorial theory (CT) of machine learning was recently devel-oped by K. Vorontsov. CT provides sharper estimations of generaliza-tion ability (GA) comparing to statistical theory of machine learning.In particularly, it establishes some new overfitting bounds for logicalconjunction rules. Those bounds were used to improve learning meth-ods for logical classification algorithms on real-world data sets. In thiswork we extend this approach to the ensembles of algorithms. We useCT to find few simple algorithms with high GA and then combine themto ensemble by weighted voting.

4 - Computational complexity and approximability ofseveral geometric covering problems.Michael Khachay, Ural Branch of RAS, Institute ofMathematics and Mechanics, S.Kovalevskoy, 16, 620990,Ekaterinburg, Russian Federation, [email protected],Maria PoberiyThe known Point Covering problem of finding the minimum cardinal-ity cover of a given finite subset of the plane by straight lines (Min-2PC) is NP-hard in the strong sense. We extend this result to the k-dimensional case (for all fixed k>2). By polynomial reduction of the(k-1)PC to the kPC problem we show that the latter is NP-completein the strong sense. This reduction can be transformed into an L-reduction of the Min-(k-1)PC to the Min-kPC problem. Therefore,the Min-kPC problem is Max-SNP-hard for all k>2 and consequentlyhave no PTAS, unless P=NP.

� WB-45Wednesday, 10:30-12h00BW-Water

Customer-Oriented Revenue Management

Stream: Advanced Inventory Control and PricingStrategiesInvited sessionChair: Catherine Cleophas, Wirtschaftswissenschaften, FreieUniversität Berlin, Garystr. 21, 14195, Berlin, Germany,[email protected]

1 - Selling to Heterogeneous Customers of UncertainValue considering both Opportunity Costs and Mis-classification in a Dynamic Revenue ManagementApproachMichael Mohaupt, Dresden University of Technology,Germany, [email protected], AndreasHilbertOpportunity costs are very common practice in revenue management,but loose significance as reference value if the true worthiness of abooking request cannot be observed at the time of decision for ac-ceptance or denial. In customer-oriented revenue management thisproblem may arise if customers differ in their future buying behav-ior (frequency, product choice) but cannot be classified into the ap-propriate customer segment unambiguously. By considering expected(mis)classification we extend a bid-price control and show advance-ments both in classification accuracy and revenues via simulation.

2 - Deterministic Aggregation for Discrete Choice Mod-elsJonas Rauch, JRG MathOpt, University of Heidelberg,Interdisciplinary Center for Scientific Computing, ImNeuenheimer Feld 368, 69120, Heidelberg, Germany,[email protected], Sebastian SagerWhile widely used in simulation, complex discrete choice models arepresently not as common in optimization, mainly due a lack of effi-cient, accurate aggregation methods. Most current approaches involveMonte Carlo simulation, which scales poorly with desired accuracy.We propose the use state-of-the-art numerical methods to computechoice probabilities deterministically for models with a linear utilityfunction. Moreover, we present a way to efficiently compute deriva-tives of choice probabilities w.r.t. various parameters, making themethod especially suitable for optimization purposes.

3 - Customer-Oriented Extensions of Revenue Manage-ment Models and the Value of IntuitionCatherine Cleophas, Wirtschaftswissenschaften, FreieUniversität Berlin, Garystr. 21, 14195, Berlin, Germany,[email protected] spread of a customer-oriented view of revenue management moti-vates the extension of traditional demand models. In current practice,extensions of the model — when not solved analytically — are leftto the intuition of analysts. This paper categorizes customer-orientedmodel extensions and their requirements and suggests the system-atic use of simulations to evaluate the cost and benefit of extendingRM models. Finally, it proposes introducing intelligent agents to therevenue management process to complement traditional analytic ap-proaches.

4 - Filtering techniques to improve forecasting fromsmall numbers in revenue managementPhilipp Bartke, Wirtschaftsinformatik, Freie UniversitätBerlin, 14195, Berlin, Germany,[email protected] RM methods incorporate sophisticated customer choice mod-els that call for the estimation of a variety of parameters. Estimatingthese parameters based on often sparse historical data becomes a keychallenge previously been referred to as the problem of small numbers.This paper proposes the application of filtering techniques well studiedin signal processing and control theory but rarely mentioned in RM lit-erature so far. In an agent-based simulation system, the performance ofthe proposed method in life-like scenarios is evaluated and comparedto more traditional methods.

213

WC-01 EURO 25 - Vilnius 2012


� WC-01Wednesday, 12:30-14:00RB-Alfa

Keynote Lecture: Professor JonathanCaulkinsStream: Keynote, Tutorial and Plenary LecturesKeynote sessionChair: Richard Hartl, Dept of Management, University of Vienna,Bruenner Str. 72, A-1210, Vienna, Austria,[email protected]

1 - Providing a Scientific Basis for Managing IllegalDrugs & MarketsJonathan Caulkins, H. John Heinz III School of Public Policy& Management, Carnegie Mellon University, 5032 ForbesAvenue, 15289, Pittsburgh, United States,[email protected] happens when formal training in operational analysis meets aclassic "wicked problem" with multiple conflicting objectives, com-peting agency priorities, and abundant data that nevertheless do notanswer the questions that really matter? This talk gives one answer,drawing on 20+ years experience working on a particularly fascinatingproblem — policy regarding control of illegal markets, notably thosefor illegal drugs.The tools of operations research, industrial organizations, and eco-nomics can be harnessed to provide an empirical, scientific basis fordrug policy making. The models discussed in this talk focus on druginitiation (product diffusion), price responsiveness (elasticity of de-mand), and operation of the illegal supply chain, both during normaltimes and when the distribution network is disrupted. Correspondingdata are drawn from epidemiological studies, forensic laboratory anal-ysis, undercover buys, and extensive interviews with incarcerated drugsmugglers and dealers in Australia, France, the UK, and the US. Thesemodels provide the foundation for estimating the cost-effectivenessof different strategies for controlling drug use and associated socialharms.However, I will not tell an entirely happy tale of Operations Research(OR) as panacea, limited only by the outside world’s ignorance of itspower. Rather, I will tell a story of a glass that is half full and yet alsohalf empty.For example, these issues are truly "OR problems" in the sense that ORideas and insights underpin key dynamics and in the sense that knowl-edge of OR empowers better understanding and solutions. However,they are also problems of and for economics, sociology, political sci-ence, demography, psychology, medicine, and other disciplines. ORcan and should partner with other disciplines and perspectives whengrappling with fundamental strategy choices related to drug policy.Likewise, the adversarial nature of the policy making process chal-lenges an optimization-based way of looking at the world.The objective of this talk is to reflect on OR’s success and limitationsin this and some allied domains to draw meta lessons for how expertscan effectively use mathematical models and operational analysis toproduce insights that inform decision makers in comparably messy do-mains.

� WC-02Wednesday, 12:30-14:00RB-Beta

Multi-Project Scheduling and StochasticProject Scheduling

Stream: Project Management and SchedulingInvited sessionChair: Rainer Kolisch, TUM School of Management, TechnischeUniversitaet Muenchen, Arcisstr. 21, 80333, Muenchen, Germany,[email protected]

1 - A modified branch and cut approach for ResourcePortfolio Problem under Relaxed Resource Dedica-tion policyUmut Besikci, Industrial Engineering, Bogazici University,Bogazici Universitesi Guney Kampus Muhendislik Fakultesi,Bebek, 34342, Istanbul, Turkey, [email protected],Ümit Bilge, Gündüz Ulusoy

Multi-project scheduling and management problem constitutes an im-portant part of the project scheduling literature. In this study we dealwith Resource Portfolio Problem (RPP) under Relaxed Resource Dedi-cation (RRD) policy. The Branch and Cut (B&C) procedure of CPLEXis modified with different feasible solution heuristics, branching strate-gies and valid inequalities. The proposed modifications utilize the in-sights gained from the previous studies we have carried out for RPP.The experimental results show that the proposed modifications havesignificant effect on solution quality.

2 - Dynamic stochastic resource-constrained multi-project schedulingRainer Kolisch, TUM School of Management, TechnischeUniversitaet Muenchen, Arcisstr. 21, 80333, Muenchen,Germany, [email protected], Thomas Fliedner,Philipp Melchiors, Walter Gutjahr

We consider the problem where project arrive stochastically over time.Projects have a determistic network but stochastic activity durations.The objective is to minimize the average weighted makespan of theprojects. We present solutions procedures which have been proposedfor this problem and compare their performance in an experiemtalstudy.

3 - Assorted sensitivity results in stochastic project net-worksAnand Paul, University of Florida, United States,[email protected]

We present some results pertaining to sensitivity issues in stochasticproject networks. First, we discuss how the impact of activity corre-lation on mean project time depends on network topology. Second,we discuss a metric that quantifies the impact of activity variance onmean project completion time; prior analytical results in the literatureare qualitative rather than quantitative. Finally, we discuss conditionsunder which an increase in mean activity duration results in an increasein mean project duration, keeping activity variances fixed.

� WC-03Wednesday, 12:30-14:00RB-L1


Stream: EURO/ROADEF ChallengeInvited sessionChair: Christian Artigues, LAAS, CNRS, 7 avenue du ColonelRoche, 31077, Toulouse Cedex 4, France, [email protected]: Ender Özcan, Computer Science, University ofNottingham, Jubilee Campus, Wollaton Road, NG8 1BB,Nottingham, United Kingdom, [email protected]: Emmanuel Guere, Google, 38 Avenue de l’Opera, 75002,Paris, France, [email protected]: Eric Bourreau, COCONUT, LIRMM, 161 Rue Ada, 34000,Montpellier, France, [email protected]: Kedad-Sidhoum Safia, LIP6 - UPMC, 4 Place Jussieu,75005, Paris, France, [email protected]: H. Murat Afsar, Industrial Systems, University ofTechnology of Troyes, 12 rue Marie Curie, BP 2060, 10010,Troyes, France, [email protected]: Marc Sevaux, Lab-STICC - CNRS UMR 3192, Universitéde Bretagne Sud - UEB, Centre de Recherche - BP 92116, 2 rue deSaint Maude, 56321, Lorient, France, [email protected]

214

EURO 25 - Vilnius 2012 WC-05

1 - Offline, enhanced, adaptive large neighborhoodsearch for the machine reassignment problemAlberto Delgado, IT University of Copenhagen, RuedLanggaards Vej 7, 2300, Copenhagen, Denmark, [email protected],Yuri Malitsky, Dario Pacino, Kevin Tierney

We present an approach to solve the machine reassignment problemthat combines complete and incomplete approaches into an offline, en-hanced, adaptive large neighborhood framework. Random selection,services, and locations are the operators to create neighborhoods, whilea constraint programming and MIP model, and a Tabu Search algo-rithm are used to re-optimize them. The selection of the neighborhoodoperator is made based on online information, while the approach tore-optimized them is selected based on online information and offlinelearning.

2 - A hybridization of Simulated Annealing and samplelocal search for solving the machine reassignmentproblemMarcus Ritt, Departamento de Informática Teórica,Universidade Federal do Rio Grande do Sul, Brazil,[email protected], Gabriel Portal, Leonardo deMiranda Borba, Alexander J. Benavides, Luciana S. Buriol

We propose two heuristics based on local search with two neighbour-hoods: the first obtains a neighbour by moving a process to anothermachine, and the second by swapping two processes on different ma-chines. The first heuristic is a sample local search. It randomly choosesone of the best neighbours in the first neighbourhood, and uses the sec-ond to escape local minima. The second heuristic is a Simulated An-nealing and uses only the first neighbourhood. The hybrid approachruns Simulated Annealing with two sets of parameters in parallel, andthe sample local search in the remaining time.

3 - A co-evolutionary hyper-heuristic for ROADEF/EUROChallenge 2012 Machine Reassignment problemWojciech Jaskowski, Institute of Computing Science, PoznanUniversity of Technology, ul. Piotrowo 2, 60-985, Poznan,Poland, [email protected], Piotr Gawron, MarcinSzubert, Bartosz Wieloch

We propose a co-evolutionary hyper-heuristic for solving combina-torial optimization problems. Our algorithm learns how to solve aproblem by selecting and generating perturbation operators using thenature-inspired concept of arm races between potential solutions andtests. It co-evolves a population of operators (solutions) with a pop-ulation of states in the problem search space (tests). We evaluate theperformance of the proposed method on the ROADEF/EURO Chal-lenge 2012: Machine Reassignment problem using various operators:local-search improvements, dedicated heuristics and MIP solver.

4 - A hybrid algorithm for machine reassignmentArnaud Liefooghe, LIFL - CNRS - INRIA, Université Lille 1,Lille, France, [email protected], François Legillon,François Clautiaux, El-ghazali Talbi

We present an algorithm for the machine reassignment problem of theROADEF/EURO 2012 challenge. Our approach is divided into twophases, applied in a sequential way. First, an iterated local search al-gorithm is performed. It is based on a best-improvement hill-climber,whose neighborhood consists of all the moves that reassign a givenprocess to another machine. Second, we aim at re-balancing the over-load cost over machines. We use an integer programming solver toimprove the incumbent solution locally by fixing some processes ontheir current machine, and solving the initial model.


Combinatorial Optimization: Applications

Stream: Combinatorial OptimizationInvited sessionChair: Jan van Vuuren, Department of Logistics, University ofStellenbosch, Private Bag X1, Matieland, 7602, Stellenbosch,Western Cape, South Africa, [email protected]

1 - Minimizing setup time in colour printing schedulesAlewyn Burger, Department of Logistics, StellenboschUniversity, Private Bag X1, Matieland, 7600, Stellenbosch,Western Cape, South Africa, [email protected], NeilJacobs, Jan van Vuuren, Stephan VisagieIn this talk a scheduling problem is considered which arises in thefood industry where packaging requires overlay printing with multi-ple colours. Printing machines used for this purpose typically accom-modate a small number of colours simultaneously, called a pallet. Tominimise operation time it is desirable to schedule print jobs so thatthe number of pallet changes, as well as the number of cartridges thathave to be washed between pallet changes is minimized. We solvethe problem exactly for small problem instances and heuristically forlarger instances.

2 - The cost of link failure in network protectionAnton de Villiers, Logistics, Stellenbosch University, 53Crown Street Vierlanden, Durbanville, 7550, Cape town,Western Province, South Africa, [email protected],Alewyn Burger, Jan van VuurenIf the vertices of a network G denote facilities to be protected, and thelinks between facilities denote routes along which protecting guardscan move, a secure deployment is an assignment of guards, at mostone per facility, to the facilities, with the property that, for every fa-cility v, there is either a guard at v or a guard at a facility adjacentto v who, after moving to v, leaves the resulting deployment secureagain. In this talk, bounds are established on the smallest number ofguards required for a secure deployment on G after randomly removinga specified number of links from G.

3 - Variable selection in NIR spectroscopy data using ahybrid PSO algorithmMartin Kidd, Dipartimento di Elettronica, Informatica eSistemistica (DEIS), University of Bologna, Bologna, Italy,[email protected] chemometrics, near-infrared (NIR) spectroscopy is used to obtainspectral data on samples for the purpose of predicting various proper-ties of these samples. NIR data are typically used in conjunction withmultivariate calibration methods in order to model these properties ofgiven samples, and a common obstacle is the presence of noise in thedata. The use of heuristic methods to perform variable selection inNIR data has proved to be useful in the past, and in this talk a two-phase particle swarm optimisation algorithm is presented as a possiblesolution approach to this problem.

4 - An optimisation-based sugarcane harvest schedul-ing decision support system for commercial growersin South AfricaJonas Stray, Textilhögskolan, Hogskolan i Boras,Bryggaregatan 17, SE-501 90, Boras, Vastra Gotaland,Sweden, [email protected], Jan van VuurenCommercial sugarcane growers in South Africa revise their seasonalharvesting schedules continuously since factors such as environmen-tal events, sucrose content, growth rate and cost of harvesting changeover time. Our decision support system (DSS) incorporates regres-sion analysis of sugar content and growth, costing and risk models forthe events. An asymmetric travelling salesman problem with time-dependent costs model sequences the fields by maximising profit. TheDSS is analysed through a case study, and results indicate that the DSSmay be useful in large-scale operations.


Topics in integer and mixed integerprogramming I

Stream: Discrete Optimization, Geometry & Graphs(contributed)Contributed sessionChair: Frank Fischer, Mathematics, Chemnitz University ofTechnology, Reichenhainer Str. 39, 09107, Chemnitz, Germany,[email protected]

215


1 - Investigation of properties and solving of combina-torial optimization problems with linear constraints.Igor Grebennik, Systems engineering, Kharkiv NationalUniversity of Radio Electronics, flat 130, Geroev Truda str,45, 61129, Kharkiv, Ukraine, [email protected],Oleksii Baranov

An optimization problem on a combinatorial set such as permutations,combinations as well as a special combinatorial set with linear con-straints is considered. As a result of inclusion mapping of the com-binatorial set into Euclidean space and a convex extension of objec-tive function to convex set an equivalent optimization problem is con-structed. Estimates of minima for the convex functions on the combi-natorial sets under linear constraints are proposed. The estimates arebased on linear combinatorial problems with constraints and used inthe branch-and-bound algorithm. Examples are given.

2 - On Mixed Integer Random Convex ProgramsGiuseppe Calafiore, DAUIN, Politecnico di Torino, CorsoDuca degli Abruzzi, 24, 10141, Torino, Italy,[email protected], Daniel Lyons, Lorenzo Fagiano

We consider mixed-integer (MI) optimization problems subject to Nrandomly drawn convex constraints. We provide explicit bounds on thetails of the probability that the optimal solution found under the N con-straints will become infeasible for the next random constraint. First,we prove that the number of support constraints (constraints whose re-moval strictly improves the optimal objective) in general convex MIproblems is bounded by a number depending only on the dimensionof the decision vector. Then, we show that the tails of the violationprobability are bounded by a binomial distribution.

3 - The SCIP Optimization Suite - It’s all in the bag!Gerald Gamrath, Zuse-Institute Berlin, Takustr. 7, 14195,Berlin, Germany, [email protected]

We present the SCIP Optimization Suite, a tool for modeling andsolving optimization problems. It consists of the modeling languageZIMPL, the LP solver SoPlex, and the constraint integer programmingframework SCIP. Besides being one of the fastest MIP solvers avail-able in source code, SCIP is able to solve a much wider range of opti-mization problems including non-convex MINLP and provides supportfor branch-and-price. We report on current developments and new fea-tures in the upcoming release 3.0, including enhanced MINLP support,exact integer programming, and parallelization.

4 - Dynamic Graph Generation in Lagrangian RelaxationFrank Fischer, Mathematics, Chemnitz University ofTechnology, Reichenhainer Str. 39, 09107, Chemnitz,Germany, [email protected],Christoph Helmberg

Models based on time expanded networks, which are used in manyapplications, grow very quickly when the number of time steps in-creases which makes it impossible to solve the models with standardmethods. We propose a dynamic graph generation technique that canbe applied in Lagrangian relaxation or column generation approaches.Instead of working on the full model, special cost structures allow tokeep only small sub-networks in memory but without loosing any in-formation. Numerical tests on a large scale train timetabling problemmodel demonstrate the effectiveness of dynamic graph generation.

� WC-06Wednesday, 12:30-14:00RB-Gamma

Mathematical EconomicsStream: Dynamical Systems and Game TheoryInvited sessionChair: Luis Ferreira, Matemática, Escola Superior de EstudosIndustriais e de Gestão, IPP, ESEIG, Rua D Sancho I, número 981,4480-876, Vila do Conde, Porto, Portugal,[email protected]

1 - On the convergence to Walrasian prices in randommatching Edgeworthian economiesLuis Ferreira, Matemática, Escola Superior de EstudosIndustriais e de Gestão, IPP, ESEIG, Rua D Sancho I, número981, 4480-876, Vila do Conde, Porto, Portugal,[email protected]

We show that for a specific class of random matching Edgeworthianeconomies, the expectation of the limiting equilibrium price coincideswith the equilibrium price of the related Walrasian economies. Thisresult extends to the study of economies in the presence of uncertaintywithin the multi-period Arrow-Debreu model, allowing to understandthe dynamics of how beliefs survive and propagate through the market.

2 - A cooperative version of one group pursuit gameYaroslavna Pankratova, Mathematics of economic research„International Banking Institute, Nevsky pr. 60,Saint-Petersburg, Russian Federation, [email protected],Denis Kuzyutin

We consider a nonzero-sum group pursuit game with one evader andm pursuers. At each time moment the players choose their directionsof motion. Each player knows the moment t and his own and all otherplayers’ positions. Moreover, the pursuer knows the directions cho-sen by the evaders at the moment t. We propose a rule to associate acorresponding cooperative game to a considered nonzero-sum pursuitgame. We prove that in the constructed cooperative game the core isnot empty and time-consistent. Finally, we give a number of examplesthat illustrate the main results.

3 - On the Experimental Investigation of the Stock-Exchange Game ModelIgor Katin, Systems Analysis Department, Vilnius UniversityInstitute of Mathematics and Informatics, Akademijos str. 4,LT-08663, Vilnius, Lithuania, [email protected], Joana Katina,Jonas Mockus

A simple SEGM was introduced by J. Mockus in 2003 to simulate thebehavior of several stockholders using fixed buying-selling margins atfixed bank yield in the context of Nash equilibrium. In this paper, anextended USEGM is investigated. USEGM includes the transactioncosts to reflect the reality better. To represent users that prefer lin-ear utility functions, USEGM adds the AR-ABS(p) model, minimizingthe absolute values, to the traditional AR(p) model. The objective ofUSEGM is not forecasting, but simulation of financial time series thatare affected by predictions of the participants.

� WC-07Wednesday, 12:30-14:00RB-Eta

Rich vehicle routing problems

Stream: New Trends in Vehicle RoutingInvited sessionChair: Jorge E. Mendoza, Institute of Applied Mathematics,Université Catholique de l’Ouest, 3 Place André Leroy, 49008,Angers, France, [email protected]: Victor Pillac, Automatique - Productique, Ecole des Minesde Nantes, 4 Rue Alfred Kastler, La Chantrerie, 44300, Nantes,France, [email protected]

1 - A Unified Hybrid Genetic Search for Vehicle RoutingProblemsThibaut Vidal, CIRRELT - Université de Montréal &ICD-LOSI - Université de Technologie de Troyes, 12, rueMarie Curie, BP 2060, cedex, 10010, TROYES, France,[email protected], Teodor Gabriel Crainic, MichelGendreau, Christian Prins

A unified hybrid genetic search is proposed to address multi-attributevehicle routing problems. This meta-heuristic relies on problem-specific route evaluation and assignment operators, along with genericimplementations of Split procedures, local-search improvement meth-ods, and genetic components. Furthermore, a bi-criteria evaluation of

216


individuals, driven by both contribution to population diversity and so-lution quality, enables to enhance exploration capabilities and reducethe risks of premature convergence. High-quality results are reportedfor a wide range of vehicle routing variants.

2 - Constraint-based Construction Heuristics for RichVehicle Routing ProblemsFelix Brandt, Logistics Systems Engineering, FZIForschungszentrum Informatik, 76131, Karlsruhe, Germany,[email protected], Anne Meyer

Constraint programming (CP) is successfully applied to rich vehiclerouting problems (RVRP). However, during the construction phase CPis used as a mere satisfiability checker, if at all, not using its power forfinding an actual solution. In this work we want to show how CP canbe applied during a construction phase. Therefore, we develop branch-ing techniques analogous to established construction heuristics fromliterature. Furthermore, we present our first results and compare theperformance of different approaches.

3 - Generalized disaggregation algorithm for the vehi-cle routing problem with time windows and multipleroutesRita Macedo, LAMIH, Université de Valenciennes, Portugal,[email protected], Saïd Hanafi, FrançoisClautiaux, Cláudio Alves, J. M. Valério de Carvalho

We address the VRP with multiple routes and time windows. For thisvariant of the VRP, there is a time interval within which every cus-tomer must be visited, and every vehicle is allowed to perform morethan one route within the same planning period. We propose a generaldisaggregation algorithm that improves the exact approach describedin (Macedo et al., 2011), and describe a novel rounding rule and a newnode disaggregation scheme based on different discretization units. Inaddition, we propose a new integer model for the problem, which canalso assess if a solution is feasible or not.

4 - A Mixed Integer Linear Programming Model For Solv-ing Large-Scale Integrated Location-Routing Prob-lems For Urban Logistics Applications at Groupe LaPosteMatthias Winkenbach, Kuehne Foundation Endowed Chair inLogistics Management, WHU - Otto Beisheim School ofManagement, Burgplatz 2, 56179, Vallendar, Germany,[email protected], Paul Kleindorfer, BernardLemarié, Christèle Levêque, Alain Roset, Stefan Spinler

We present a large-scale static, deterministic mixed inter linear pro-gramming model solving an integrated location-routing problem in thecontext of urban logistics services (ULS). We aim at determining anoptimal infrastructure and fleet design for an urban consolidation andtransportation network. The model is meant to quantitatively evaluatepotential strategic moves of postal operators into the domain of ULS.Based on operating data from French La Poste, we analyse the keydrivers and sensitivities of optimal infrastructure and fleet design tosupport the development of profitable ULS.

� WC-08Wednesday, 12:30-14:00RB-Epsilon

Boolean Functions and Their Applications

Stream: Boolean and Pseudo-Boolean OptimizationInvited sessionChair: Endre Boros, RUTCOR, Rutgers University, 08854,Piscataway, New Jersey, United States,[email protected]

1 - On the Polyhedral Set of Bipartite TSPBela Vizvari, Industrial Engineering, Eastern MediterraneanUniversity, Gazimagusa, Mersin 10, Turkey,[email protected]

The Bipartite TSP appears in many industrial applications, e.g., to op-timize the route of an assembly robot. In many cases the numericalsolution of TSP is based on the facets of its polyhedral set. The mostimportant constraints are the degree inequalities and the subtour elim-ination constraints. The simplest facet-defining inequalities after theseones are the comb inequalities in the general case. However in the bi-partite case the comb inequalities are not facet-defining, as it will beproven in this presentation. The case of clique-tree inequalities is alsodiscussed.

2 - Some Results on Threshold Separability of BooleanFunctionsGiovanni Felici, Istituto di Analisi dei Sistemi ed Informatica,Consiglio Nazionale delle Ricerche, Viale Manzoni 30,00185, Roma, Italy, [email protected], Endre Boros

We consider the conditions for the existence of a Linear ThresholdFunction that separates two sets of Boolean vectors obtained by dis-cretization of real valued data. Problems of this type are found in Su-pervised Learning and Data Mining. We state a combinatorial neces-sary and sufficient condition for the existence of such function whenpoints belong to the plane, and show how, when formulated as a lin-ear program, this condition is equivalent to the Farkas’ Lemma. Suchequivalency provides insights for our problem in larger dimension thatcan be used in discretization algorithms.

3 - Hardness Results for Approximate Pure Horn CNFFormulae MinimizationEndre Boros, RUTCOR, Rutgers University, 08854,Piscataway, New Jersey, United States,[email protected], Aritanan Gruber

We show that for a pure Horn Boolean function on n variables, unlessP = NP, it is not possible to approximate in polynomial time (in n) theminimum numbers of clauses and literals to within factors of 2(log(1-o(1))n) even when the inputs are restricted to 3-CNFs with O(n(1+e))clauses, for some small e > 0. Furthermore, we show that unless theETH is false, it is not possible to obtain constant factor approximationsfor these problems even in sub-exponential time (in n).

� WC-09Wednesday, 12:30-14:00RB-Zeta


Stream: Maritime TransportationInvited sessionChair: Magnus Stålhane, Industrial Economics and TechnologyManagement, NTNU, Alfred Getz veg 3, 7491, Trondheim,Norway, [email protected]

1 - On routing and risk minimization in offshore heli-copter transportationIrina Gribkovskaia, Faculty of Economics, Informatics andSocial Sciences, Molde University College - SpecializedUniversity in Logistics, Postbox 2110, 6402, Molde, Norway,[email protected], Oyvind Halskau, Fubin Qian

Helicopters are often used for transporting crews to and from offshoreinstallations. Helicopter transportation is regarded as one of the mainrisks in the offshore industry and the oil companies want to minimizethis risk. We show how the expected number of fatalities can be re-duced by applying different routing strategies. We present trade-offsbetween the number of helicopters used and the expected risk usingreal data from the Norwegian offshore industry.

2 - A combined pickup and delivery and inventory rout-ing problemMagnus Stålhane, Industrial Economics and TechnologyManagement, NTNU, Alfred Getz veg 3, 7491, Trondheim,Norway, [email protected], Henrik Andersson, MarielleChristiansen

217


We present a combined pickup and delivery and inventory routingproblem from the tramp shipping industry. The fleet is heterogeneouswith each ship having a different load capacity, speed and cost struc-ture. The inventories are paired, and there are additional optional car-goes available. The objective is to maximize the revenue of transport-ing spot cargoes, while keeping the inventory levels within their limits.Two path flow formulations are presented, and the problem is solvedusing a branch-and-price method. Computational results will be pre-sented.

3 - A base integer programming model and benchmarksuite for liner shipping network designFernando Alvarez, Goteborggata 9, 0566, Oslo, Norway,[email protected], Berit Dangaard Brouer, Christian EdingerMunk Plum, David Pisinger, Mikkel M. Sigurd

The potential for making cost effective and energy efficient liner ship-ping networks using OR is huge and neglected. We believe accessto domain knowledge and data is an entry barrier to approach the linershipping network design problem. The purpose of the benchmark suiteproject is to provide access to the liner shipping domain for indepen-dent OR researchers. We present the domain of liner shipping in re-lation to mathematical programming. An IP model and a suite of in-stances based on real life data from the largest liner shipping company,Maersk Line, and several other stakeholders is presented. Finally, com-putational results yielding the first solutions to the benchmark suite arepresented.

� WC-10Wednesday, 12:30-14:00RB-Theta

Humanitarian Logistics

Stream: Supply Chain ManagementInvited sessionChair: Maria Besiou, Kuehne Logistics University, Brooktorkai20, 20457, Hamburg, Germany, [email protected]

1 - The environmental impact of humanitarian opera-tionsJurgita Balaisyte, INSEAD, Boulevard de Constance, 77305,Fontainebleau, none, France, [email protected],Ruth Carrasco-Gallego, Luk Van Wassenhove

The negative environmental impacts of relief operations arise at thedifferent stages of relief supply chain: from procurement, stock posi-tioning, transportation to waste management. It is important to regardenvironmental issues in humanitarian supply chain decisions. As aresult a growing number of humanitarian agencies are attempting tointegrate environmental considerations in their activities. We map thevalue chain and identify its economic, environmental and social issuesand look for ways to integrate environmental aspects in to humanitar-ian supply chains.

2 - The Effects of Unsolicited Donations in Disaster Re-lief OperationsNha-Nghi Huynh, Kuehne Logistics University, KuehneLogistics University, Brooktorkai 20, 20457, Hamburg,Hamburg, Germany, [email protected], SandraTranschel, Maria Besiou, Luk Van Wassenhove

Why do beneficiaries not get aid items they require or why do interna-tional humanitarian organizations (IHO) deliver these not fast enough?A reason is the chaotic setting in which IHO operate. But what af-fects high-priority (HP) deliveries more are material convergences atdisaster entry points caused by unsolicited donations. They consumeexcessive resources and create serious bottlenecks. We present a queu-ing model that illustrates the effects of unsolicited donations on theHaiti relief operation and shows how systematic coordination couldsignificantly increase the throughput of HP goods.

3 - Funding strategies in humanitarian operations for re-lief and development

Alfonso Pedraza-Martinez, Operations and DecisionTechnologies, Indiana University, 1309 10th Street, KelleySchool of Business, 47405, Bloomington, INDIANA, UnitedStates, [email protected], Arian Aflaki

We model funding strategies for relief and development programs ina multi-donor market with donation uncertainty. Donors decide howmuch to donate to each program; a humanitarian organization decidesbetween 4 different funding strategies: 1) total earmarking; 2) partialearmarking; 3) conditional earmarking; 4) no earmarking. By com-paring these four strategies we offer insights to humanitarian organiza-tions on how to improve their levels of funding while keeping opera-tional efficiency high.

4 - Decentralisation and Earmarked Funding in Humani-tarian Logistics for Relief and DevelopmentMaria Besiou, Kuehne Logistics University, Brooktorkai 20,20457, Hamburg, Germany, [email protected],Alfonso Pedraza Martinez, Luk Van Wassenhove

We study the International Humanitarian Organizations dual missionof relief and development in decentralized, stochastic operations withdifferent levels of earmarked funding, using system dynamics method-ology. Focusing on equity and efficiency we find that a system withlocal procurement and short lead time may take longer to respond thana system with global procurement and high lead time. Our resultshave important implications for humanitarian fleet management prac-tice. Specifically, what organizational structure to adopt in differentorganizational environments.

� WC-11Wednesday, 12:30-14:00RB-Iota

Realistic parallel machines problems

Stream: Realistic Production SchedulingInvited sessionChair: Olga Chub, Faculty of Applied Mathematics andManagement, Kharkiv national university of radioelectroniks,Lenina str., 18, 43, 61166, Kharkiv, Ukraine,[email protected]

1 - Minimising tardiness in parallel machine schedulingwith additional mold type restrictionsAmelia White, Engineering Science, The University ofAuckland, 70 Symonds Street, Auckland, Auckland, NewZealand, [email protected], Andrea Raith

We study a parallel machine scheduling problem with sequence-dependent setup times. The jobs to be produced have due dates andrequire a specific mold type to be installed on a machine for produc-tion. The setup time for these molds is significant and there is onlya small number of each type of mold available. We present a newtime-indexed IP formulation to minimise overall job tardiness, withconstraints to model both the setup times of the molds on the machinesand the restriction on the number of machines which can produce jobsof the same type at the same time.

2 - MIP models and algorithms for earliness/tardinessscheduling problems on parallel machinesRosiane de Freitas Rodrigues, Institute of Computing, UFAM/ UFRJ, Brazil, [email protected], Rainer Xavier deAmorim, Bruno Cardoso Dias

Scheduling problems with earliness and tardiness penalties are pre-sented, each one or both together, involving parallel machines, and in-dependent jobs with distinct processing times and priorities, all of themNP-hard problems. Classical MIP models with improvements, and ex-act and approximate approaches have been developed with tests on -based on - OR-Library instances. Significant results were obtained,reaching optimal solutions when known, and having competitive per-formance compared to the works available in the literature.

218


3 - Multiobjective parallel machine scheduling usingmemetic algorithmsFelipe Baesler, Industrial Engineering, Universidad delDesarrollo, Av Sanhueza 1750, Concepcion, Chile,[email protected], Oscar Cornejo

This work presents a multiobjective optimization algorithm for the par-allel machine scheduling problem based on the memetic algorithmsparadigm (MA). This algorithm combines genetic evolution with lo-cal search in the same way as traditional MA, but with the use ofindependent populations for each objective, as well as a mechanismto find compromised solutions. The algorithm was tested and com-pared against four multiobjective techniques available in the literature,MOGA, SPEA, NSGA II and MOGLS. Two objectives were consid-ered for this problem, Cmax minimization and total tardiness mini-mization.

4 - Mathematical model and method for optimal plan-ning several resourcesOlga Chub, Faculty of Applied Mathematics andManagement, Kharkiv national university of radioelectroniks,Lenina str., 18, 43, 61166, Kharkiv, Ukraine,[email protected]

The resource planning multicriteria model provided that the demandfor enterprise services is a deterministic function has been considered.Enterprise functioning process is defined as a continuous sequenceof projects. Some projects can (and should) be run simultaneously.This fact causes of distribution of several restricted resources (time,finances, machines, human resources etc.) as between the works ofa single project, and between the works of various projects. Solutionmethod based on the interpretation of work as a geometric object anduse the optimal geometric design theory toolkit.

� WC-12Wednesday, 12:30-14:00RB-Omicron

Solution Methods at Seaport ContainerTerminalsStream: Container Terminal OperationsInvited sessionChair: Erhan Kozan, Mathematical Sciences, QueenslandUniversity of Technology, GPO Box 2434, 4001, Brisbane,Queensland, Australia, [email protected]

1 - An Integrated Approach to Optimising Container Pro-cesses at Multimodal Seaport TerminalsErhan Kozan, Mathematical Sciences, Queensland Universityof Technology, GPO Box 2434, 4001, Brisbane, Queensland,Australia, [email protected], Brad Casey

A Multimodal Seaport Container Terminal (MSCT) consists of sub-systems that require optimisation of the operations within them, aswell as synchronisation of the movements of machines and contain-ers between them. We present an integrated approach. Mathematicalmodels will be presented for the Storage Area Subsystem (SAS) andIntra-terminal Transportation Subsystem (ITTS). Initial solutions areimproved through an innovative hyper-heuristic algorithm that inte-grates the SAS and ITTS solutions together and optimises them viameta-heuristic techniques.

2 - Yard crane scheduling by using a genetic algorithmat seaport container terminalsMehmet Ulas KoyuncuoGlu, Industrial Engineering,Pamukkale University, Pamukkale University InformationCentre, 20070, Denizli, Turkey, [email protected], OsmanKulak, Olcay Polat, Mustafa Egemen Taner

Effective use of container terminal equipments is the most importantfactors.Scheduling the yard cranes has an important role in improvingthe efficiency of terminals.In this paper, a novel approach is presentedabout free/restrictive deployment rules and also customer priority anddeployment rules are examined in combination.For the purpose of min-imizing total completion time of the operations, a mixed-integer linear

mathematical model is proposed to solve the problems of assigningone or more yard cranes and of routing by using a genetic algorithmand developing heuristics based approach.

3 - Empty container management and the linkage be-tween global and local allocations in export orientedcountries: case of ThailandMontathip Chanpum, Institute of Transport and LogisticsStudies, The University of Sydney, Room 201-5 C37,Newtown Campus, 2006, Sydney, NSW, Australia,[email protected]

The availability of empty containers is a challenge for shipping linesin deficit areas of export oriented regions. This paper considers emptycontainer management at a global scale and its linkage to empty al-location at a local scale in deficit areas by taking Thailand as a casestudy. We conducted in-depth interviews with scholars and key rel-evant persons from private and public sectors. A framework linkingempty container allocations between the global and local scales is pro-posed, which will help to facilitate container movement and improvethe efficiency of the global supply chain.

4 - Modeling and Optimal Management of Equipments ofthe BMT Containers Terminal (Bejaia’s Harbor)Fazia Aoudia-rahmoune, Operational Research, LaboratoryLAMOS University of Bejaia, Laboratory LAMOS Universityof Bejaia, Traga Ouzemmour, 06000, Bejaia, Algeria,[email protected], Sofiane Soualah, Aichouche Oubraham

The performance of a containers terminal is measured by the stopovertime and the transit cost. Indeed, in a world where,in addition to ri-valry,the least movement is charged and where the time takes often un-expected dimensions,reducing the time and cost of treatment becomesthe main objective at the expense of large investment in equipment andmanagement techniques. In this work, we were interested in the un-loading process at BMT, and we determined the optimal combinationsof equipment to be used. we develop a non-Markovian queues modeland solve the problem by simulation.

� WC-13Wednesday, 12:30-14:00RB-Tau

Emerging Issues in Order Fulfillment

Stream: OR Applications in the Automotive IndustryInvited sessionChair: Bart MacCarthy, Operations Management Division,Nottingham University Business School, NG8 1BB, Nottingham,United Kingdom, [email protected]: Philip Brabazon, Portsmouth Business School, RichmondBuilding, Portland Street, PO1 3DE, Portsmouth, UnitedKingdom, [email protected]

1 - Auction-based coordination of retailers and manu-facturer in the sales of configurable productsAndreas Matzke, Institute of Automotive Management andIndustrial Production, Technische Universität Braunschweig,Katharinenstr. 3, 38106, Braunschweig, Germany,[email protected], Thomas Volling, Thomas Spengler

We propose a mechanism to improve channel coordination in the salesof configurable products, assuming given capacities for product op-tions. The approach combines two selling mechanisms. At first, thecommon list price mechanism between a manufacturer and multipleretailers is adopted for initial configuration of products. Second, ex-cess capacities are sold as product upgrades by an auction mechanism.A numerical study evaluates the benefits of the approach under variousconditions. As compared to static list prices, the approach allows forhigher revenues in the supply chain.

219


2 - A review of works on the planning of capacities andorders in build-to-order automobile productionThomas Volling, Institute of Automotive Management andIndustrial Production, Technische Universität Braunschweig,Katharienenstr. 3, 38106, Braunschweig, Germany,[email protected], Martin Grunewald, Andreas Matzke,Thomas SpenglerWe provide an overview about operations research (OR) models andapplications in build-to-order automobile production. We focus on twoimportant planning objects which have not been considered in priorreviews: the planning of capacities and orders. To bridge the gap be-tween conceptual works on the one hand and quantitative contributionson the other we provide a framework for the structuring of planningtasks. 45 works have been identified in the literature and are classi-fied according to the proposed framework. Open issues that should beaddressed by OR are discussed based on that.

3 - Optimal design of open pipeline order fulfilment pro-cesses in the volume automotive sectorPhilip Brabazon, Portsmouth Business School, RichmondBuilding, Portland Street, PO1 3DE, Portsmouth, UnitedKingdom, [email protected], Bart MacCarthyOrder fulfillment processes used by volume automotive producers arecharacterized by high variety and large scale. Simulation is used tostudy the effect of 11 principal control parameters on performancemetrics in different customer markets. The experimental study usesan NOLH design and a CHAID factor analysis. The parameter set-tings minimizing the cost of fulfillment in a market do not necessarilyminimize the customer-centric metrics of lead time or the degree ofcompromise experienced by customers. The results of the study showthe complexity in this type of order fulfillment process.

� WC-14Wednesday, 12:30-14:00RB-Omega

Timetabling and People

Stream: Timetabling and RosteringInvited sessionChair: Sara McComb, Purdue University, United States,[email protected]

1 - Deterministic Optimization of Examination Timeta-blesSiti Khatijah Nor Abdul Rahim, School of Computer Science,University of Nottingham, Jalan Broga, 43500, Semenyih,Selangor, Malaysia, [email protected], AndrzejBargiela, Rong QuWe introduce a new optimization method for the examinations schedul-ing problem by performing permutations of slots and reassignmentsof exams upon the feasible schedules. The permutations minimizethe number of students taking exams in adjacent slots and guaranteea significant reduction of the cost which is further reduced by reas-signing exams to other slots. Our method delivers competitive resultscompared to other constructive methods in the literature and maintainsthe same deterministic optimization pattern on all benchmark datasetsproving the generality and consistency of the method.

2 - The Classroom Allocation Problem in UniversityTimetablingAntony Phillips, Engineering Science, University ofAuckland, 70 Symonds St, 1010, Auckland, New Zealand,[email protected], Hamish Waterer, MatthiasEhrgott, David RyanUniversity timetabling is a large resource allocation problem, in whichboth times and rooms are determined for each class meeting. Due tothe difficulty of the problem, it is often solved in two stages: timetablegeneration followed by room allocation. This paper looks at ex-act optimisation-based methods for allocating rooms to a timetablein which times are preassigned. Computational results on full sizedatasets are presented, and the trade-off between different approachesis discussed.

3 - Decision Support System for Modification of Timeta-bles After Completion of Registration Process Involv-ing Backlog SubjectsSanjeev Kumar Guleria, Office of Dean, Academic Affairs,Thapar University, P.B. No. 32, 147004, Patiala, Punjab,India, [email protected], A.k. Lal, Daljit Singh Bawa

The classes of backlog subjects are registered in the free slots of thenormal subjects. A free slot in a section of normal subjects is occupiedin some other section for studying a backlog subject. The methodologyinvolves the identification of all students who will get affected by thischange followed by the identification of clashes this change will maketo any of the affected students. The proposed change is made only ifno student suffers due to the proposed change.

4 - Examining Team Communication Using Question-naires, Simulation, and OptimizationSara McComb, Purdue University, United States,[email protected], Ralitza Vozdolska, Deanna Kennedy

Communication is essential for coordinating team activities and exe-cuting a team’s project. The relationship between communication andsuccessful performance, however, is complex. We examine this rela-tionship, as well as the impact of information exchange, team size, in-terdependence among teammates, and media choice on it. The resultsfrom questionnaire data, simulation, and optimization suggest a curvi-linear relationship dependent upon information exchange efficiency,communication requirements as a function of team composition, andwhether exchanges are synchronous or asynchronous.

� WC-15Wednesday, 12:30-14:00RB-2101

Discrete bilevel optimization andVariational inequalities

Stream: Variational Inequalities and Bi-Level ProblemsInvited sessionChair: Diana Fanghaenel, Fachbereich Elektrotechnik/Informatik,Universität Kassel, Wilhelmshöher Allee 73, 34121, Kassel,Germany, [email protected]

1 - Global Optimization of Mixed-Integer Bilevel Pro-gramming ProblemZhaohui Xu, Fakultät für Mathematik, Technische UniversitätChemnitz, Reichenhainer Str. 41/718 D-09107, Chemnitz,Sachsen, Germany, [email protected]

This paper is concerned with mixed-integer nonlinear bilevel program-ming problem, which has a nonlinear objective function and one pa-rameter on the right-hand sides of the constraints in the lower levelproblem. We propose an algorithm via an approximation of the op-timal value function of the lower level problem to solve the bilevelprogramming problem globally.

2 - The Branch-and-Sandwich Algorithm for Mixed-Integer Nonlinear Bilevel Programming ProblemsPolyxeni-Margarita Kleniati, Chemical Engineering, ImperialCollege London, South Kensington Campus, SW7 2AZ,London, United Kingdom,[email protected], Claire Adjiman

We extend our newly introduced algorithm for general bilevel pro-gramming problems, named Branch and Sandwich (B&S), to the dis-crete case. Auxiliary bounding problems are constructed in order tobound the inner optimal value function and provide constant boundcuts for the outer bounding problems. The originally proposed KKT-based relaxations for the inner-upper and outer-lower bounding prob-lems are applicable with respect to the continuous inner variables andfor fixed integer inner variables. A hierarchical branching scheme withadditional lists of nodes, as developed in B&S, is investigated.

220


3 - A bilevel problem with supermodular objective func-tion in the lower levelDiana Fanghaenel, Fachbereich Elektrotechnik/Informatik,Universität Kassel, Wilhelmshöher Allee 73, 34121, Kassel,Germany, [email protected]

We investigate a bilevel programming problem with continuous vari-ables in the upper level and a parametric supermodular maximizationproblem in the lower level. We analyze the structure of this bilevelproblem to transform it into an equivalent one-level combinatorialproblem. Furthermore, we discuss the properties of the equivalentproblem.

4 - Existence results for quasivariational inequalitiesDidier Aussel, Lab.PROMES UPR 8521, University ofPerpignan, Technosud, Rambla de la thermodynamique,66100, Perpignan, France, [email protected]

A quasi-variational inequality is a variational inequality in which theconstraint set depends on the current point. Based on fixed pointand convergence techniques we prove various existence results un-der weak assumptions on the set-valued operator defining the quasi-variational inequality, namely quasimonotonicity and lower or uppersign-continuity. Applications to quasi-optimization and traffic networkproblem are also considered.

� WC-16Wednesday, 12:30-14:00RB-2103

Linear programming: Algorithms andApplications

Stream: Linear and Conic ProgrammingInvited sessionChair: S.M.Atiqur Rahman Chowdhury, Department ofMathematics, University of Dhaka, 1000, Dhaka, Bangladesh,[email protected]

1 - The primal-dual column generation methodPablo Gonzalez-Brevis, School of Mathematics, University ofEdinburgh, United Kingdom,[email protected], Jacek Gondzio, PedroMunari

In this talk we will introduce a column generation method which usesa primal-dual interior point method to obtain well-centred non-optimalsolutions of the restricted master problems. We will show that themethod converges to an optimal solution of the master problem eventhough non-optimal solutions are used in the course of the procedure.Computational experiments will be presented using linear-relaxed re-formulations of three classical integer programming problems. Theresults indicate that the method reduces the number of iterations andrunning times when compared with standard approaches.

2 - Compact Formulations of the Steiner TravelingSalesman ProblemSaeideh D. Nasiri, Mathematics & Statistics and OperationalResearch, STOR-i DTC, Lancaster University, B44, Fyldecollege, Lancaster University, LA1 4YF, Lancaster,Lancashire, United Kingdom,[email protected], Adam Letchford, DirkOliver Theis

The Steiner Traveling Salesman Problem (STSP) is a variant of theTraveling Salesman Problem (TSP) that is particularly suitable whendealing with sparse networks, such as road networks. The standard in-teger programming formulation of the STSP has an exponential num-ber of constraints, just like the standard formulation of the TSP. Onthe other hand, there exist several known compact formulations of theTSP, i.e., formulations with a polynomial number of both variables andconstraints. In this study, we show that some of these compact formu-lations can be adapted to the STSP.

3 - Computational experiments with a cutting plane al-gorithm for a time-indexed formulationLotte Berghman, FEB, KU Leuven, Naamsestraat 69, 3000,Leuven, Belgium, [email protected], FritsSpieksmaWe consider an unrelated parallel machine scheduling problem wherethe processing cost of each job is an arbitrary function of its startingtime. It is shown that valid inequalities known from the literature canbe applied to our time-indexed formulation. A new class of valid in-equalities that contains variables related to multiple machines is pre-sented. All these inequalities are incorporated in a cutting plane al-gorithm. Experiments show how the different classes of inequalitiescontribute to closing the gap between the linear programming relax-ation and the integer optimum.

4 - A Computer Technique for Solving Linear Program-ming Problems with Bounded VariablesS.M.Atiqur Rahman Chowdhury, Department ofMathematics, University of Dhaka, 1000, Dhaka, Bangladesh,[email protected] Programming problems (LPPs) with upper bounded variablescan be solved using the Bounded Simplex method (BSM), without theexplicit consideration of the upper bounded constraints. The upperbounded constraints are considered implicitly in this method whichreduced the size of the basis matrix significantly. In this paper, I devel-oped MATHEMATICA codes for solving such problems. A completealgorithm of the program with the help of a numerical example hasbeen provided. Finally a comparison with the built-in code has beenmade for showing the efficiency of the developed code.

� WC-17Wednesday, 12:30-14:00RB-2105

Nonconvex Programming

Stream: Nonconvex Programming: Local and GlobalApproachesInvited sessionChair: Hoai An Le Thi, Computer Science, University of Lorraine,Ile du Saulcy„ 57 045, Metz, France,[email protected]

1 - Solving the Multidimentional Assignment Problemvia the Cross-Entropy methodManh Nguyen Duc, LMI, INSA de Rouen, Avenue del’Université, 76801, Saint-Étienne-du-Rouvray Cedex, Rouen,France, [email protected], Hoai An Le Thi,Tao Pham DinhThe multidimensional assignment problem (MAP) is a higher dimen-sional version of the linear assignment problem, where we find tuplesof elements from given sets, such that the total cost of the tuples isminimal. The MAP has many recognized applications such as dataassociation, target tracking and resource planning. While the linear as-signment problem is solvable in polynomial time, the MAP is NP-hard.In this work, we propose a new approach based on the Cross-Entropy(CE) method for solving the MAP. The efficiency of our method willbe shown by testing for the large-scale problems.

2 - On the Permuted Perceptron Problem - A New formu-lation and approachHoai Minh Le, Université Paul Verlaine Metz, LITA, UFRMIM, Ile du Saulcy, 57000, Metz, France,[email protected], Hoai An Le Thi, Tiru ArthanariThe permuted perceptron problem (PPP) has received renewed interestdue to its application in Cryptanalysis. The problem can be cast intothe DC (Difference of Convex functions) programming framework andsolved by DC Algorithm as shown in: ’Solving the Perceptron Problemby deterministic optimization approach based on DC programming andDCA’, Le Thi et al., Proc. of the IEEE Conference INDIN 2009. Inthis work we give a new formulation to PPP by using a parametricfow problem. An algorithm to solve the parametric fow problem and aprobabilistic search of the parameter space are developed.

221


3 - Rare event simulations applied to hierarchical sensorplanning with partially known subprocess modelsFrédéric Dambreville, DGA, 16bis Av. Prieur de la Cote d’Or,94110, Arcueil, France, [email protected]

Information collection in the context of Intelligence process usuallytakes the form of a hierarchical, typically bi-level, planning problem.We are especially interested in the planning at the coordination level:a constraint oriented multisensor planning associated with a variety ofmono-sensor control optimizations. In this contribution, we will modelour sensor models an approximation noise. Real sensor models willprovide indications for successive corrections of the planning process.This paper studies strategies for optimizing these plan corrections.

4 - optimisation in water management with passive aer-ation of tanksNataliya Metla, AMS (MODSI), CRP Henri Tudor, 29 av.John F. Kennedy, L-1855, Luxembourg-Kirchberg,Luxembourg, [email protected]

A multi-objective formulation is proposed for optimal design of a smallwater distribution system with a special issue on aeration of reservoirslinked to water quality. Model Predictive Control approach is used tocreate a sequence of corresponding optimization problems over finitetime horizont. A specific feature belongs to the aeration componentwhich is non-convex. A practical approach to cope with this problemwill be discussed. Finally, based on realistic consumption scenariosnumerical results for control strategy will be shown.

� WC-18Wednesday, 12:30-14:00RB-2107

Location Routing

Stream: Network and Discrete LocationInvited sessionChair: Maria Albareda Sambola, Statistics and OperationsResearch, Technical University of Catalonia, C. Colom, 11, 08222,Terrassa, Spain, [email protected]

1 - Adaptive large neighborhood search for the two-echelon capacitated location routing problemVera Hemmelmayr, Vienna University of Economics andBusiness (WU), Nordbergstrasse 15, 1090, Vienna, Austria,[email protected], Claudio Contardo, TeodorGabriel Crainic

We present an adaptive large neighborhood search heuristic (ALNS)for the two-echelon location routing problem (2E-CLRP). The 2E-CLRP is an important problem arising in freight distribution and ad-dresses simultaneous location-routing within two-tiered systems. Thenumber and location of capacitated facilities has to be decided on bothlevels while also routing a fleet of vehicles on each level. Computa-tional results conducted on several sets of instances are presented. Wecompare the performance of the ALNS to several other methods pro-posed in the literature.

2 - A Particle Swarm Optimization Approach for theLocation-Routing Problem with Simultaneous Pickupand DeliveryIsmail Karaoglan, Department of Industrial Engineering,Selcuk University, 42075, Konya, Turkey,[email protected], Fulya Altiparmak

This paper considers a Location-Routing Problem with SimultaneousPickup and Delivery which is defined as finding locations of the depotsand designing vehicle routes in such a way that pickup and deliverydemands of each customer must be performed with same vehicle andthe overall cost is minimized. We propose a hybrid heuristic approachbased on particle swarm optimization, simulated annealing and integerprogramming formulation to solve the problem. Computational resultsindicate that the proposed approach is able to find optimal/good qualitysolutions in a reasonable computation time.

3 - Sparsing arrival times in discrete locationMaria Barbati, Business and Management EngineeringDepartment, University of naples Federico II, P.le Tecchio 80,80078, Naples, Italy, Italy, [email protected], GiuseppeBruno, Alfredo MarínWe consider a discrete two-stage location problem where the mini-mum difference between two consecutive time arrivals of customers toa depot has to be balanced. Customers have to be allocated before toa facility and after to the depot. In order to avoid the congestion ofthe depot the flow of customers will be the most balanced in the time.We propose different Integer Programming formulations and we iden-tify different valid inequalities. Computational experiments have beenperformed and the results have been discussed.

4 - The reliable p-median problem with at-facility serviceMaria Albareda Sambola, Statistics and Operations Research,Technical University of Catalonia, C. Colom, 11, 08222,Terrassa, Spain, [email protected], Yolanda Hinojosa,Justo PuertoThis work deals with a facility location model where service is pro-vided at the facility sites. It is assumed that failures on facilities canoccur and, as a consequence, customers may need to follow a paththrough the opened facilities until finding one that is available to pro-vide them service. The goal of the problem is to locate p facilities sothat the expected total travel cost of the customers is minimized. Twoalternative mathematical programming formulations for this problemare presented and an approximation based on network flows is ana-lyzed.

� WC-19Wednesday, 12:30-14:00RB-2111

Simulation in EstimationStream: Simulation Methods in FinanceInvited sessionChair: Ahmet Çalık, Statistics, Selcuk University, SelçukUniversity Science Faculty, Campus Konya, 42075, Konya,Turkey, [email protected]: Aysegul Iscanoglu Cekic, Statistics, Selcuk University,KONYA, Turkey, [email protected]

1 - Estimation of Stress-Strength Reliability for Expo-nential Geometric DistributionAydın Karakoca, Department of Statistics, Selcuk University,Faculty of Arts and Sciences, Konya, Turkey,[email protected], Ismail Kınacı, Adem YılmazIn this study we are interested in estimation of the stress-strength re-liability R=P(X>Y), X and Y are independent exponential-geometric(EG) random variables. The confidence intervals of R based on maxi-mum likelihood and bootstrap methods are derived. An extensive com-puter simulation is used for comparing the performances of these con-fidence intervals.

2 - Simulation Study for the Comparison of MaximumLikelihood and Modified Maximum Likelihood Esti-mators in Logistic RegressionYunus Akdogan, Statistics Department, Science Faculty,Selcuk University, Selcuk University, 42225, Konya, Turkey,[email protected], Evrim Oral, Muslu Kazım Körez,Coskun KusThe maximum likelihood (ML) equations are generally intractable andtherefore the ML estimators are elusive. To rectify this situation, Tikudeveloped the method of modified maximum likelihood (MML) esti-mation in 1967. The method is now well established and gives es-timators which are simple, unique and highly efficient. In fact, theyare asymptotically fully efficient. Tiku and Vaughan (1997) used themethod to extend the techniques of binary regression to nonlogisticdensity functions. In this study, we consider the ML and MML estima-tors in logistic regression. We compare the ML and MML estimatorswith respect to their mean square errors via a Monte Carlo simulationstudy. We also investigate the coverage probabilities of confidenceintervals of logistic regression parameters based on ML and MML es-timators via simulations. We investigate the revised MML estimators

222


proposed by Tiku and Vaughan (1997) with respect to their initialsvalues as well. We show that using least squares estimators as initialvalues provides approximations as good as those proposed by Tiku andVaughan (1997). REFERENCES

Tiku, M. L. and Vaughan, D. C., 1997, Logistic and nonlogistic densityfunctions in Binary Regression with nonstochastic covariates, Biomet-rical Journal, 39, 883-898.

3 - Estimation of stress-strength parameter for expo-nential distribution based on masked dataDemet Sezer, Statistics, Selcuk University, Selcuk UniversityFaculty of Science, Department of Statistics, Konya, Turkey,[email protected], Ismail Kınacı

In this study, we estimate the stres-strength parameter(R) for serial sys-tems with two components based on masked data. In particular, wecompute the maximum likelihood and Bayes estimates of R. The con-sidered systems consist of two independent components having non-identical complementary exponential lifetime distributions. Also inthe numerical simulation study a comparison between Bayes and max-imum likelihood estimates is introduced.

4 - Fuzzy AHP/DEA Approach For Relative Efficiency OfPrivate Universities In TurkeyAhmet Çalık, Statistics, Selcuk University, Selçuk UniversityScience Faculty, Campus Konya, 42075, Konya, Turkey,[email protected], Nimet Yapici Pehlivan, AhmetPekgör

In the study we apply integrated multi-criteria decision making(MCDM) approach, including the fuzzy analytic hierarchy process(FAHP) and data envelopment analysis (DEA) model. FAHP method isfuzzy extension of AHP applied to solve decision making problems infuzzy environment. DEA method is a MCDM tool which solve prob-lems that don’t require prior information for ranking the decision mak-ing units. In application part, integrated FAHP/DEA is applied to pri-vate universities in Turkey. At first, for determining weights of criteria,FAHP is used and then DEA is applied to rank universities.

� WC-20Wednesday, 12:30-14:00RB-2113


Stream: Financial OptimizationInvited sessionChair: Mariya Nikitina, International Banking Institute, RussianFederation, [email protected]

1 - Investment timing with fixed and proportional costsof external financingMichi Nishihara, Graduate School of Economics, OsakaUniversity, 1-7, Machikaneyama, Toyonaka, 560-0043,Osaka, Japan, [email protected], Takashi Shibata

We develop a dynamic model in which a firm exercises an option toexpand production with cash balance and costly external funds. Whilerelated papers explain their results only by numerical examples, weanalytically prove the following results. In the presence of only a pro-portional cost of external financing, the firm with more cash balanceinvests earlier; however, the presence of both proportional and fixedcosts leads to a non-monotonic relation between the investment timeand cash balance. Our results can potentially account for a variety ofempirical results.

2 - On optimal strategies for long-term businessEfim Bronshtein, Numerical Mathematics and Cybernetics,Ufa State Aviation Technical University, K.Marx str., 12,USATU, 450000, Ufa, Russian Federation,[email protected]

The problem of optimal allocation of resources, that the company re-ceived at time n = 1,2,... on the consumption and the production isconsidered. The company’s activities we describe as the production

functions for all time moments. The production functions are mono-tonically increasing, strictly concave, bounded and differentiable. Weuse the consumption in the discounted form as an indicator of the ef-fectiveness. The solution is given for stable and depressive companies.

3 - Longevity Risk Management and Financial UneasyPremonitionTadashi Uratani, Industrial and System Engineerig, HoseiUniversity, Kajinocho 3-7-2, Koganei, 184-8584, Tokyo,Japan, [email protected]

We consider longevity risk in aspect of national budget and social secu-rity expenditure. Longevity is breaking national budget system throughpension and health-care deficits in Japan. The national debt is accumu-lated in commercial banks as a major asset that becomes their crisis insome conditions. We will examine the theoretical possibility of ap-plication of longevity bond and derivatives to the risk management ofpension budgetary system.

4 - On Minimizing the Ruin Probability by InvestmentMariya Nikitina, International Banking Institute, RussianFederation, [email protected]

A generalization of collective risk model is considered. We proposea model in view of the possibility of investing part of the surplus in arisky asset. The assumption allows minimizing the probability of ruinof an insurance company by investment. This generalization of classi-cal model was computer simulated. The exact solutions and asymptoticapproximations to the probability of ruin were obtained.

� WC-21Wednesday, 12:30-14:00RB-2115

Ratings of Banks and Companies

Stream: Numerical Methods in FinanceInvited sessionChair: Wissem Ennouri, Industrial Management, High Institute ofIndustrial Management of Sfax-Tunisia, 74 Mecca Street, 3001,Sfax, Sfax, Tunisia, [email protected]

1 - Cluster Analysis of Banks in Croatia Regarding theirCharacteristicsAnte Rozga, Faculty of Economics, Quantitative methods,University of Split, Matice Hrvatske 31, 21000, Split, Croatia,[email protected], Ksenija Dumicic, Blanka Škrabic Peric

Croatia privatized her banks to increase their financial performance.As a result 95% of the banking sector is owned by foreigners. Clusteranalysis used eight indicators. The analysis is done for 2003 and 2008.There were 18 most important banks used. For 2003 two clusters wereobtained, almost completely divided into domestic and foreign ownedbanks but one. First cluster had worse financial indicators suggestingsuccessful privatization. Results for 2008 show one change in clus-ter membership, but the difference in their performance has narrowedbetween two clusters, which is good result.

2 - Effectiveness of Credit RatingsIvan Lichner, Economic modeling and analysis, Institute forEconomic Research SAS, Sancova 56, 81105, Bratislava,Slovakia, [email protected]

Credit rating agencies (CRAs) play an important role in the global fi-nancial markets because many market participants use their rating as-sessments. As CRAs have a history of failing to foresee the real fu-ture development this paper aims to find answer whether or not shouldfinancial markets trust CRAs. To find answer to this question the rat-ings of sovereign debt of selected EU countries will be compared withValue-At-Risk measure based on fundamental economic factors. Theanalysis focuses on the ability of ratings to capture economic perfor-mance of assessed country.

223


3 - Small and Medium-Sized Businesses through theCredit Crunch: Evidence from the UKMeng Ma, Business School, The University of Edinburgh, 29Buccleuch Place, EH8 9JS, Edinburgh, United Kingdom,[email protected], Galina Andreeva, Jake Ansell

Access to credit is a vital condition for SMEs survival and devel-opment, especially for young start-up enterprises. The credit crisishas put the problem of SME financing to the fore, with governmentsputting considerable pressure on banks to lend to SMEs. This pre-sentation looks at the credit risk modelling of the UK SMEs via lo-gistic regression applied to a large proprietary dataset, and exploresthe behaviour of small businesses during the credit crisis and beyond(2007-2010) across industrial sectors, geographical regions, financialsituation and for start-ups versus mature businesses.

4 - Pricing Default and Financial Distress Risks in For-eign Currency Corporate Loans in TurkeyAycan Yilmaz, Institute of Applied Mathematics/ FinancialMathematics, Middle East Technical University, Istanbul,Turkey, [email protected], ISil Erol

Latest trends show that non-financial corporations are heavily financedby foreign exchange loans (fx-loans). The paper contributes to the lit-erature on contingent claims pricing by developing a structural modelto value ‘the exchange rate risk’ and ‘the interest rate risk’ of fx-loansat the corporate level under volatile economic conditions in Turkey.The paper uses explicit finite difference method and evaluates both fi-nancial distress and default risks. The results suggest that currencydepreciation and decline in interest rates result in financial distress andtrigger defaults.

� WC-22Wednesday, 12:30-14:00RB-2117

Fuzzy Scheduling and Fuzzy QualityManagement

Stream: Fuzzy Optimization - Systems, Networks andApplicationsInvited sessionChair: Erik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39, 85577,Neubiberg, Germany, [email protected]: Alexej Orlov, Students Science Association, Institute forapplied system analysis of the National Technical University ofUkraine, Peremohy ave. 37, 03056, Kyiv, Ukraine,[email protected]

1 - A comparison study of Wilcoxon and fuzzy WilcoxontestsGultekin Atalik, Statistics, Anadolu University, AnadoluUniversity Faculty of Science Department of Statistics,Eskisehir, Turkey, [email protected], SevilSenturk

Parametric tests depend on the population distributions to be normaland independently. Nonparametric tests do not require assumptionswhich parametric tests require. Parametric and nonparametric testsalso handle crisp values in statistics. In the real world, data may notbe measured so precisely. In these situations, fuzzy sets’ approachesare applied the data sets. In this study, Wilcoxon and Fuzzy Wilcoxontests are compared with each other and all results are indicated clearly.

2 - A design for sampling plans by attributes under fuzzyenvironmentEbru Turanoglu, Industrial Engineering, Selcuk University,Department of Industrial Engineering Selcuk University,Campus, Selcuklu, Konya, 42031, Turkey, 42031, Konya,Turkey, [email protected], Cengiz Kahraman

In the design of accurate acceptance sampling plan, it is usually neededto know the actual quality level entailed by the customer. Particularlyin production environments, it may not be easy to define acceptancesampling parameters including fraction of nonconforming, acceptancenumber, or sample size as crisp values. The purpose of this paper isto develop a new design procedure for single and double samplingplans by attributes based on the fuzzy set theory by formulating thisproblem as a multi-objective mathematical programming consideringproducer’s and consumer’s risks.

3 - Task Flow Control in the GRID-SystemVlad Kucher, Educational-scientific complex - Institute forapplied system analysis, National Technical University of theUkraine (KPI), 04111, Kiew, Ukraine,[email protected]

The model of the planning process for the task performance in theGRID-system with the multilevel hierarchical structure is introduced.The fuzzy-set approach and the application of the principle of squarestowage into the band of a given width, as well as the choice of an ag-gregate system as a mathematical model of the planning process, haveallowed the development of an algorithm, which minimizes the time oftask performance as well as balances workload of resource suppliers.

4 - The analysis of influence of a choice of a class ofmembership functions on quality of work of recogni-tion systemNigar Aliyeva, Institute of Information Technologies,National Academy of Sciences of Azerbaijan, Baku,Azerbaijan, [email protected], Elviz Ismayilov

Was analyzed effectiveness of the hand-printed recognition system us-ing neuro-fuzzy sets with membership functions from different classes.For pattern features was constructed smooth membership functions,and also known triangular, trapezoidal, Gaussian and bell membershipfunctions. Analyzed results of numerical experiments of recognitionsystem with membership functions from different classes. Revealedcomparative advantage of smooth membership functions offered byauthors.

� WC-23Wednesday, 12:30-14:00RB-Delta

Recent Advances on Proximal Methods IIStream: Convex OptimizationInvited sessionChair: Hector Ramirez, Mathematical Engineering Department,Universidad de Chile, Avda. Blanco Encalada 2120, Piso 5,Santiago Centro, Santiago, RM, Chile, [email protected]

1 - Lagrangian-penalization algorithm for constrainedoptimizationJuan Peypouquet, Mathematics, Universidad Tecnica FedericoSanta Maria, Av Espana 1680, 2340000, Valparaiso,Valparaiso, Chile, [email protected]

We present a proximal algorithm that provides solutions to constrainedoptimization problems - constrained variational inequalities - whereobservations of the state variable are required to belong to some closedconvex set in a Hilbert space. The algorithm combines an exact penal-ization procedure involving a bounded sequence of parameters, withthe predictor-corrector multiplier method. As applications, we de-scribe how the algorithm can be used to find sparse solutions of linearinequality systems and solve partial differential equations by domaindecomposition.

2 - Interior Proximal Algorithm with Variable Metric forSecond-Order Cone Programming: Applications toStructural OptimizationHector Ramirez, Mathematical Engineering Department,Universidad de Chile, Avda. Blanco Encalada 2120, Piso 5,Santiago Centro, Santiago, RM, Chile,[email protected], Felipe Alvarez, Julio López

224


We propose an inexact interior proximal type algorithm for solvingconvex second-order cone programs. The proposed algorithm uses avariable metric and an appropriate choice of a regularization param-eter. This choice ensures the well-definedness of the proximal algo-rithm and forces the iterates to belong to the interior of the feasible set.Thus, under suitable assumptions, it is proven that each limit point ofthe sequence generated by the algorithm solves the problem. Finally,computational results applied to structural optimization are presented.

3 - A Projected Subgradient -type method for Equilib-rium ProblemsSusana Scheimberg, COPPE/ Engenharia de Sistemas eComputação-Instituto de Matemática,COPPE/PESC-IM,Universidade Federal do Rio de Janeiro, Caixa Postal 68511„Bloco H/319, 21941-972, Rio de Janeiro, RJ, Brazil,[email protected], Paulo Sergio Santos

We consider an explicit algorithm for solving Equilibrium Problemsin a finite dimensional space. We combine reflections onto suitablehyperplanes to achive feasibility with a projected subgradient-type it-eration. The algorithm has a low computational cost per iteration andsome numerical results are reported.

4 - Proximal Alternating Direction Methods CombiningQuadratic Regularization with Divergence KernelsNatalia Ruiz, Mathematical Engineering, Universidad deChile, Blanco Encalada 2120, fifth floor, santiago, RegiónMetropolitana, Chile, [email protected], Felipe Alvarez

We will introduce a new alternating minimization algorithm for solvinga convex program with separable structure involving linear couplingand non negativity constraints. Our algorithm consists of two prox-imal minimization steps applied to an augmented Lagrangian, whichare performed alternatively, using quadratic regularization for the vari-able defined in the whole euclidean space and divergence type for nonnegativity one; then a proximal maximization step is performed on thedual variable. The global and full convergence of primal-dual variablesto optimal solutions are established.

� WC-25Wednesday, 12:30-14:00CC-A23

OR and Environmental Management -Decision ProcessesStream: OR and Environmental ManagementInvited sessionChair: Susanne Wiedenmann, Chair of Production and Logistics,University of Goettingen, Platz der Goettinger Sieben 3, 37073,Goettingen, Germany,[email protected]

1 - Supply Planning under Uncertainty for the IndustrialUse of Renewable ResourcesSusanne Wiedenmann, Chair of Production and Logistics,University of Goettingen, Platz der Goettinger Sieben 3,37073, Goettingen, Germany,[email protected], JuttaGeldermann

We investigate the supply planning of a processor of agricultural goodswhich are required as raw materials in industrial production. The useof linseed and linseed oil is one example that promises to be relevant inthe future. Problems arise from uncertain quality, quantity and price.A two stage stochastic program is used as a decision support and itsresults are evaluated. We use data from linseed and linseed oil for val-idation.

2 - Hydro-turbines Optimization: Especification and Op-erationJéssica Pillon Torralba Fernandes, Energy Department,UNICAMP, Rua Roxo Moreira 1406 apto 19, 13083592,Campinas, São Paulo, Brazil, [email protected]

This paper aims to present a methodology focused at the optimiza-tion of the Brazilian hydroelectric power plants potential, based onthe specification of hydroelectric turbines to power plants in construc-tion. This methodology choose and compare the efficiency curves ofKaplan and Propeller turbines, showing that, with an appropriate op-eration it is possible to obtain technical and economic improvementsto the plants. The mathematical model is a non-linear mixed integerand aims to maximize the plant generation. The resolution is done byGenetic Algorithms.

3 - Integrated Risk Management Applications in Elec-tricity RetailingFehmi Tanrisever, Industrial Engineering, TU/e, Netherlands,[email protected], Geert Jongen

This paper examines the risk management practices of an electricityretailer in the Dutch electricity market. We derive the optimal futureshedging policy for the retailer in the presence of demand and pricerisk using base- and peak-load futures contracts. We also develop anARIMA-GARCH model for the electricity spot prices and test the ef-fectives of our hedging policies. We show that, on average, by hedgingwith base-load contracts the firm may reduce the variance of its cashflows by 91%. Including peak-load contracts into the hedging portfolioincreases the efficiency of hedging to 95%.


Criteria Weighting in Decision Support

Stream: Multi-Criteria Decision Making and Environ-mental ManagementInvited sessionChair: Alexander Bolshoy, Evolutionary Biology, University ofHaifa, Mt. Carmel, 39105, Haifa, Israel,[email protected]: Mikhail Kuznetsov, Moscow Institute of Physics andTechnology, Russian Federation, [email protected]

1 - A comparison of weight distributions in multi-attribute value theoryJyri Mustajoki, Department of Automation Science andEngineering, Tampere University of Technology, P.O.Box692, 33101, Tampere, Finland, [email protected]

In multi-attribute value theory, various approaches to model uncertain-ties with distributions on importance weights have been presented inthe literature. We carry out a simulation study in which we analyze theeffects of three different weight distributions in terms of the marginalweight and weight ratio distributions they produce. According to ourresults, there are clear differences in the spread of weights the differentdistributions generate. We also discuss on what grounds and on whattypes of applications the use of each distribution can be justified.

2 - "Hit and Run" for the efficient generation of weightsGert van Valkenhoef, Faculty of Economics and Business,Universty of Groningen, PO Box 800, 9700 AV, Groningen,Netherlands, [email protected], Tommi Tervonen,Nalan Bastürk, Douwe Postmus

Multi-criteria decision analysis is often based on weighted aggregationof per-criterion scores, where weights represent preferences. To enablespecifying preferences as linear constrains on the weights, simulation-based methods sample from a uniform distribution on the feasibleweight space. Efficient samplers are known for special cases, but nomethod can efficiently sample weights with general linear constraints.We show how the "Hit and Run" sampler can be used to perform thistype of sampling in fast polynomial time. We also discuss the problemof diagnosing convergence of such samplers.

225


3 - Non-profit project portfolio evaluation and selection:A multicriteria approachNassim Dehouche, LAMSADE- Université Paris-Dauphine,Place du Maréchal de Lattre de Tassigny, 75775, Paris, Ile deFrance, France, [email protected], DanielVanderpooten

We study settings in which a review committee is to select a portfo-lio of the best projects among a set of proposals. We propose a newtwo-stage multiple criteria parameterizable rule-based sorting method-ology. First, an evaluation process is carried out, in which the focusis on an accurate capture of the knowledge created during the workof the committee. Then, a selection process takes place, in which thepreferences of the decision maker regarding the portfolio are modeledin a linear program, the resolution of which specifies values for theparameters of the rule-base.

4 - Maximization of a Sum of Kendall’s Tau Rank-Correlation CoefficientsAlexander Bolshoy, Evolutionary Biology, University ofHaifa, Mt. Carmel, 39105, Haifa, Israel,[email protected]

Given data on a set of objects, while every object is defined by a set ofcharacters, the object ranking problem is to rate each object accordingto its characters. The input data is a matrix R[n, m] with an element(object, character). We study rating problem in context of molecularevolution. The objective is to find an optimal scale which most ac-curately recovers the across-object characters. As solving of this opti-mization problem is NP-hard, we propose an original heuristic method,a special simulation annealing scheme.


Knowledge Management & DecisionSystems

Stream: Decision Support SystemsInvited sessionChair: Shaofeng Liu, School of Management, PlymouthUniversity, PL4 8AA, Plymouth, Devon, United Kingdom,[email protected]

1 - Extending the analysis of the EURO Working Groupon DSS Research Collaboration Network (EWG-DSSCollab-Net V.2)Pascale Zaraté, Institut de Recherche en Informatique deToulouse, Toulouse University, 118 route de NarBonne,31062, Toulouse, France, [email protected], Fatima Dargam,David Dadenne, Isabelle Linden, Shaofeng Liu, Rita Ribeiro,Wei Sun

EWG-DSS-Collab-Net explores the group’s research collaborationsince 1989. It considers a hybrid methodology for data collec-tion, manual and automatic, incorporating web mining of electronicdatabases. It embeds an enhanced model of publication structure, ex-tending the network statistics to collaboration trend analysis, with co-authorships and co-citations to further illustrate the dynamics of publi-cations over time. These features reveal the growth of the community,find researchers for reviewing papers and projects, and encourage col-laboration not only among existing but also new members.

2 - Applying ontology in the analysis of a DSS ResearchCollaboration NetworkFatima Dargam, SimTech Simulation Technology, RiesStrasse 120, 8010, Graz, Austria,[email protected], Antonio C. S. Branco,Alexandre Rademaker, Renato Rocha Souza

This work specifies ontology for the EWG-DSS-Collab-Net network.This network was designed to show the research collaboration dynam-ics in the EWG-DSS. It considers a publication relationship structured

model with authors-publications and multiple keywords-topics. Thenetwork analysis shows co-authorships and co-citations overtime. Inorder to refine the identification of publications that are closely relatedto a given topic, an ontology model is specified. This way, a com-mon vocabulary of classifications of the publications main areas canbe defined and matched with the key-words.

3 - Ranking the seawater desalination methods by usingmulti criteria decision making approachesAmin Esmaeeli, DEpartment of Civil Engineering, Tabrizuniversity, 51579-13573, Tabriz, Iran, Islamic Republic Of,[email protected], Hooman Baghban Asgharinejad, Mahdi Zarghami

In this paper we compared common desalination alternatives such asReverse Osmosis (RO), Mechanical Vapor Compression (MVC), MultiStage Flash (MSF), Multi Effect Distillation-Thermal Vapor Compres-sion (MED-TVC) by using the multi criteria decision making methodsbased SAW and TOPSIS. The assessment criteria are energy require-ment, water production cost, environmental impact, and maintenanceand water quality. According to the outcomes of research for the ca-pacity of 1200 m3/day, RO method is the best decision by the SAWmethod and MVC becomes the best alternative by the TOPSIS method.


OR Military Applications II

Stream: OR in Military, Safety and Security Applica-tionsInvited sessionChair: Ana Isabel Barros, Military Operations, TNO, POBox96864, 2509 JG, The Hague, Netherlands, [email protected]

1 - Convoying against piracyRien van de Ven, Netherlands Defence Academy, Postbus10000, 1780 CA, Den Helder, Netherlands,[email protected], Jack Vermeulen

The last years have seen an increase in the incidence of piracy near So-malia. The Somali East Coast and the Gulf of Aden are high-risk areas.The Gulf of Aden is an important shipping lane due to its location. Thefact that it is a relatively narrow area means that traffic flows in onlytwo directions. So convoying merchant vessels is one of the solutionsto the problem of piracy. This paper presents a mathematical modelhow to optimize the number of merchant vessels that can be protectedby a warship. Through simulation effectiveness of such a system canbe quantified.

2 - Counter improvised explosive devices gameToru Komiya, Dept. of Computer Science, Japan NationalDefense Academy, Hashirimizu 1-10-20, 2398686, Yokosuka,kanagawa, Japan, [email protected], Akinori Yasue

In Iraq or Afghanistan peace keeping operation (PKO), many soldiersare victimized by Improvised Explosive Devices (IED). We have mod-eled the counter IED operation. A patrol car drives along the roadwhere IED may be planted. Terrorist can explode hidden IED accord-ing to his ability. He can also decide number, planting place and powerof the bomb in advance. On the other hand, the PKO vehicle passesbeside the IED without knowing it, but he can decide physical and op-erational parameters of the vehicle beforehand. By using the gamingcontext, we decide the optimal strategies on both sides.

3 - A NATO Guide for Judgement-based OperationalAnalysisIngrid Bastings, Military Operation Research, TNO Defence,Safety and Security, PO Box 96864, 2509JG, The Hague,Netherlands, [email protected], Diederik J.D.Wijnmalen

The SAS Panel of NATO/RTO tasked a group to write a Code of BestPractice for Judgement-based OA in Defence Decision Making. Itsaim is not only to create an understanding of what judgement-based

226


OA is, how it can address messy defence problems and how they canbe coped with, but also how to conduct a judgement-based OA studyin order to maximise the validity, credibility and acceptance of sucha study and its outcomes. We will clarify some of the analyst- andclient-oriented highlights of the Guide.

4 - Topology of Military C2 Systems and its relation tocyber warfareHerman Monsuur, Faculty of Military Sciences, NetherlandsDefence Academy, P.O. Box 10.000, 1780 CA, Den Helder,Netherlands, [email protected]

We relate aggregate qualities found in the military C2 literature to theaxiomatic approach in network theory. We address multi-layered net-works and metrics for network control and for the feedback of opera-tional links, as well as the merging of network metrics and C2 socio-technical covariates. We also apply our findings to cyber warfare.


Optimization and Data Mining (I)

Stream: Optimization and Data MiningInvited sessionChair: Concha Bielza, Inteligencia Artificial, UniversidadPolitécnica de Madrid, Campus de Montegancedo, 28660, Madrid,Spain, [email protected]: Rubén Armañanzas, Artificial Intelligence Department,Technical University of Madrid, Facultad de Informática, Campusde Montegancedo, 28660, Madrid, Spain, [email protected]

1 - Mathematical Models of Supervised Learning andApplication to Biomedical ProblemsMario Guarracino, Istituto di Calcolo e Reti ad AltePrestazioni, 80131 , Napoli, Italy,[email protected], Roberta De Asmundis

Supervised learning models are applicable in many fields of science.Among them, there are Support Vector Machines (SVM), which ex-hibit accurate solutions and low training time. SVM are based on thestatistical learning theory and provide the solution by minimizing aquadratic type cost function. SVM, in conjunction with the use ofkernel methods, provide non-linear classification models, namely sep-arations that cannot be expressed using inequalities on linear combina-tions of parameters. We describe here some applications to biomedicalproblems, with the support of numerical experiments.

2 - Incorporating Partially Labeled Information into Ker-nel Spectral ClusteringCarlos Alzate, IBM Research, Ireland,[email protected], Johan Suykens

Kernel Spectral Clustering (KSC) corresponds to a primal-dual for-mulation to clustering allowing out-of-sample extensions and modelselection in a learning setting. The primal is defined in terms of high-dimensional feature maps and the dual is an eigenvalue problem of aweighted kernel matrix. We extend the core KSC model by addingextra terms in the objective function related to the labeled informationtogether with regularization constants. The resulting dual is no longeran eigenvalue problem but a linear system. Simulation results show theapplicability of the proposed method.

3 - Restating clinical impression of severity index forParkinson’s disease using just non-motor criteriaRubén Armañanzas, Artificial Intelligence Department,Technical University of Madrid, Facultad de Informática,Campus de Montegancedo, 28660, Madrid, Spain,[email protected], Pablo Martinez-Martin, ConchaBielza, Pedro Larrañaga

Clinical impression of severity index (CISI) for Parkinson’s disease(PD) is an index to assess PD severity based on four clinical domains.Since this is a continuous value, a categorization policy is proposed

by solving an optimization problem. Using this encoding, differentcomparisons between CISI and other non-motor indexes or items areaddressed using wrapper item subset selection and estimation of distri-bution algorithms. Results show how some of the non-motor items arevery relevant, achieving good classification performances when usedto predict the CISI severity index.



Stream: DEA and Performance Measurement (con-tributed)Contributed sessionChair: Ahti Salo, Systems Analysis Laboratory, Aalto UniversitySchool of Science and Technology, P.O. Box 11100, Otakaari 1 M,00076, Aalto, Finland, [email protected]

1 - Measuring the Operation Efficiency for Bus Transit inTaiwan with Undesirable OutputsChao-Chung Kang, Department of Business Administrationand Graduate Institute of Management, ProvidenceUniversity, 200, Chung-Chi, Rd., 433, Shalu, Taichung,Taiwan, [email protected]

This paper studies the operating efficiency for bus transit with undesir-able outputs by slack-based data envelopment analysis (SBM-DEA) inTaiwan. Data of Taipei over 2007-2010 is drawn for the study. Vehicle-kilometer and operation revenue are taken as desirable outputs; vehicleemission as undesirable output; labor, vehicle, and fuel are the inputs.We compare the differences in efficiency evaluation produced by DEAand SBM-DEA. The effects in relative efficiency of desirable and un-desirable output can suggest practical strategies for improving the op-erating efficiency of bus transit.

2 - Factors defining Effectiveness of automatic vehiclelocation systemsDaniil Opolchenov, Department of Electronics, Transport andTelecommunication Institute, Lomonosova street 1, LV-1019,Riga, Latvia, [email protected]

This work is devoted to solving the problem of AVL systems effec-tiveness estimation. In this work we consider the tasks of definingthe factors which determine the effectiveness of AVL systems, and ofobtaining their numerical values and distribution laws. To reduce theanalysis complexity, we divide AVL systems into subsystems and re-define accordingly the factors defining their effectiveness. The resultsof this work can be used to obtain an effectiveness index.

3 - Chinese Companies Distress Prediction: An Applica-tion of Data Envelopment AnalysisZhiyong Li, Business School, University of Edinburgh, Flat13, 35 Peffer Bank, EH16 4FE, Edinburgh, Scotland, UnitedKingdom, [email protected], Jonathan Crook, GalinaAndreeva

Traditional distress prediction models employ financial ratios to pre-dict distress. We investigate the predictive accuracy of efficiency mea-sures along with ratios in predicting distress in Chinese companies.Unlike previous studies where DEA was used to generate a singleefficiency-TE, we decompose TE into PTE and SE. These variablesare introduced into LR to make prediction. Efficiency of firms for sev-eral industrial sectors is integrated in one industry-specific LR. Theresults show the predictive power is improved by this efficiency. BothCross-sectional and panel models are tested.

227


4 - Analyzing Portfolios of DMUs with Ratio-Based Effi-ciency Analysis (REA)Ahti Salo, Systems Analysis Laboratory, Aalto UniversitySchool of Science and Technology, P.O. Box 11100, Otakaari1 M, 00076, Aalto, Finland, [email protected], Juuso Liesiö,Yongjun LiIn this paper, we extend Ratio-Based Efficiency Analysis [Salo andPunkka, 2011, Mgmt Sci 57/1] for the comparison of portfolios, de-fined as combinations of DMUs where the share of each DMU is re-quired to satisfy linear constraints. In particular, we establish a dom-inance relationship among such feasible portfolios in order to deter-mine which portfolios are more efficient than others. We also examinethe composition of efficient portfolios, which allows us to determineminimum and maximum bounds on the share of any DMU across allefficient portfolios.


OR and Ethics IStream: OR and EthicsInvited sessionChair: Cristobal Miralles, Depto. Organización de Empresas,Universidad Politecnica de Valencia, Cami de Vera s/n, 46022,Valencia, Spain, [email protected]: Fred Wenstøp, Strategy and Logistics, BI NorwegianSchool of Management, Nydalsveien 37, 0483, Oslo, Norway,[email protected]

1 - On Multidimensional Performance Measures — ACritical Reflection and Future DevelopmentsJoão Clímaco, INESC-Coimbra, 3004-512, Coimbra,Portugal, [email protected] this communication we start by summarizing and exemplifying thedifficulties to build quality of life indicators and sustainability indi-cators, aggregating several dimensions. Ethical, methodological andpractical issues are emphasised. Secondly, we discuss how and inwhich cases the notion of capability, proposed by A.Sen, could be use-ful, and we propose a multidimensional visual procedure to explore theinformation provided by adequate dashboards, helpful in those situa-tions. A software demonstration will be done. Finally, future trends ofresearch are outlined.

2 - Responsibility in Decision-SupportSven Diekmann, Philosophy & Ethics of Technology,Eindhoven University of Technology, P.O. Box 513, IPO 1.14,5600 MB, Eindhoven, Netherlands, [email protected] Researchers (ORs) often influence decisions on ethical is-sues, e.g. HIV policies. Although they do not decide themselves, theirwork makes them accountable for consequences of supported deci-sions. However, are ORs also morally blameworthy? Are they ac-countable for unforeseeable consequences? I distinguish responsibil-ity, accountability and blameworthiness. In the following, I investigatecriteria for holding ORs accountable or blameworthy. I conclude thatORs can always be hold accountable for consequences related to theirwork, while blameworthiness depends on specific conditions.

3 - Multicriteria analysis ranking the most developed EUcountriesZeljko Racic, Operational research, Faculty of economics,Majke Jugovica 4, Banjaluka, Bosnia And Herzegovina,[email protected], Matea ZlatkovicAnalysis of socio-economic development shows that it is a multidi-mensional category. That multidimensionality of socio-economic de-velopment will be the subject of this paper; specifically, the group ofthe most developed EU countries. Starting with the selection of devel-opment indicators, executed based on principal components analysisusing the statistical software SPSS, ie. the latest version of PASWStatistics 20, through their determination of satisfactory or "optimal"number and order of importance, using the PROMETHEE method, itwill be up to the value of a synthetic indicator of the development levelderived from the set of selected indicators.

4 - A value structured approach to conflicts in environ-mental managementFred Wenstøp, Strategy and Logistics, BI Norwegian Schoolof Management, Nydalsveien 37, 0483, Oslo, Norway,[email protected]

The paper aims to suggest how OR practitioners ought to approachvalue conflicts in environmental management. The approach is prac-tical and founded on three pillars: ethics, neuro-economics, and deci-sion sciences. A rational approach problems is too often hampered bystrong emotions elicited by value conflicts among stakeholders. Thus,considerable resources are often squandered on ill founded projectswith detrimental effects. This could be mitigated if the OR analyst hasa deeper understanding of ethical reasons for choice, and knowledgeof practical methods to deal with values.



Stream: DEA and Performance Measurement (con-tributed)Contributed sessionChair: Chih-Hai Yang, Economics, National Central University,300 Jhongda Road, 320, Jhongli, Taiwan, [email protected]

1 - Comparing groups of efficiency in Algarve Por-tuguese HotelsIsabel Pedro, CEG-IST, Avenida Rovisco Pais, 1049-001,Lisboa, Portugal, [email protected], Ricardo Oliveira, RuiMarques

This paper uses non-parametric techniques to investigate the efficiencyof Algarve Portuguese hotels. The Algarve is one of the most impor-tant tourism places in Europe and a popular destination for touristsfrom several countries. By using a recent statistical test, this researchinvestigates the influence of star rating, golf courses and location inhotel efficiency. We concluded that star rating does not matter as faras efficiency is concerned and that location and the existence of golfcourses might be relevant. Our study points out hotels without golfcourses as more efficient.

2 - On the need for reform of the Portuguese Judi-cial System — Does Data Envelopment Analysis evi-dence support it?Sérgio Santos, Faculty of Economics, University of Algarve,Campus de Gambelas, 8005-139, Faro, Portugal,[email protected], Carla Amado

The Portuguese judicial system has attracted considerable criticism inrecent years and demands for reform have gained prominence. By us-ing DEA and focusing on first instance courts, we found evidence thatsupports these calls for reform. We found a fairly inefficient sector,with only one fifth of the 206 courts analysed in 2009 being efficient,and a potential to reduce the courts’ caseload by around 30%. Al-though most of this improvement can be achieved with a better man-agement of cases, scale factors also seem to play an important role inexplaining inefficiency.

3 - Evaluation of alternative targets for graduate dis-tance courses in brazil using multiobjetive modelsbased on data envelopment analysisSilvio Gomes Júnior, Coordenação de Engenharia deProdução, UEZO-Fundação Centro Universitário Estadual daZona Oeste, Avenida Manuel Caldeira de Alvarenga, 1203,Campo Grande, 23070-200, Rio de Janeiro, Rio de Janeiro,Brazil, [email protected], Lidia Angulo-Meza, JoãoCarlos Soares de Mello

228


Classic DEA models determine efficiency indexes of DMUs. For aninefficient DMU one target is found. Subsequent works use multiob-jective models, called MORO, that determine alternative targets. Theseprojections are obtained with non proportional variations of the vari-ables. In some real problems, some variables cannot change their lev-els. These variables are called non discretionary variables. In this pa-per, we present a case study that involves the efficiency evaluation ofthe Center for Higher Distance Education in Brazil with an integrationof MORO models and non discretionary variables.

4 - Produce Patents or Journal Articles? A Cross-Country Comparison of R&D Productivity ChangeChih-Hai Yang, Economics, National Central University, 300Jhongda Road, 320, Jhongli, Taiwan,[email protected]

This paper compares R&D productivity change across countries con-sidering two outputs: patent and journal articles. Using directionaldistance function we develop a Luenberger R&D productivity changeindex. We then decompose it into efficiency change and technicalchange. Estimates show that R&D productivity growth is mainly at-tributed to technical changes. However, non-OECD countries performbetter on both efficiency and technical changes than OECD countries.Moreover, the main source of R&D productivity growth is patent-oriented rather than the journal article-oriented.


Sustainable Construction Processes andsupplier selection

Stream: OR for Sustainable DevelopmentInvited sessionChair: Folke Björk, Civil and Architectural Engineering, KTH -Royal Institute of Technology, Brinellv. 23, SE 10044, Stockholm,Sweden, [email protected]

1 - Multi-criteria assessment of the sustainable renova-tion according on the methodology of the buildingcertification systemMilena Medineckiene, Department of Civil and ArchitecturalEngineering, Royal Institute of Technology, Brinellvagen 23,SE 10044, Stockholm, Sweden,[email protected], Folke Björk, ZenonasTurskis

This research deal with how Multi Criteria Decision Making (MCDM)can be applied when building renovation is to be connected to envi-ronmental certification. The presentation will highlight the methodol-ogy of the MCDM methods and explain about how the criteria fromthe certification systems can be adapted to mathematical algorithms ofthe MCDM. The reseach will result in a methodology for the buildingowner in public or private sector, which will help to integrate sustain-able technology with competitiveness. This should help the buildingowners to make decisions for sustainable renovation.

2 - Classification of factors influencing environmentalmanagement of enterpriseTatjana Tambovceva, Faculty of Engineering Economics andManagement, Riga Technical University, Mezha Street 1/7-213, LV-1048, Riga, Latvia, [email protected]

In article the analysis of different factors influencing environmentalmanagement was done. Company’s external and internal environmentelements were clustered. The author distinguishes between the follow-ing groups of factors: production — technical, economic, political —legal, organizational, ecological and risk factors, as well as educationand culture level. The study is based on the theoretical analyses ofliterature and scientific publications, as well as on research based onsurvey results collected from Latvian companies.

3 - Sustainability prioritizing constructing projects ofmunicipalities based on AHP and COPRAS-G: a casestudy about construction of footbridges in IranEdmundas Kazimieras Zavadskas, Department ofConstruction Technology and Management, VilniusGediminas Technical University, Sauletekio ave. 11,LT01001, Vilnius, Lithuania, [email protected],Mohammad Hasan Aghdaie, Sarfaraz Hashemkhani Zolfani

This study gives a framework for municipalities to prioritize theirprojects based on this framework. The footbridge project of Sari Cityin Iran is selected as a case study for this research. Selecting area forconstructing footbridges including quantitative and qualitative factorssuch as the Total cost, Environmental factors, Socio-economic factorscan be viewed as a kind of Multi-Attribute Decision Making (MADM)problem. The aim of this study is to use Analytic Hierarchy Process(AHP) and COPRAS-G methods for evaluation and selection the newfootbridges alternatives.

4 - Sustainability assessment model of supplier selec-tion problemJolanta Tamosaitiene, Department of ConstructionTechnology and Management, Vilnius Gedimimas TechnicalUniversity, Sauletekio ave. 11, LT–10223, Vilnius, Lithuania,[email protected], Edmundas KazimierasZavadskas, Zenonas Turskis

This paper aims to utilize a hybrid multi—criteria decision making(MCDM) model in supplier selection. Eight evaluation criteria —cost, quality, distance, delivery reliability, reputation, technology level,compatibility and development ability are identified. The Analytic Hi-erarchical Process is initially used for calculating the weight of eachcriterion and the COmplex PRoportional ASsessment of alternativeswith Grey relations (COPRAS-G method) is adopted for ranking sup-pliers. In order to illustrate the model’s functional ability, Kalleh Com-pany is chosen as a case study.


Economic modeling and finance

Stream: Cooperative Game TheoryInvited sessionChair: Sirma Zeynep Alparslan Gok, Mathematics, Faculty of Artsand Sciences, Suleyman Demirel University, Faculty of Arts andSciences, Suleyman Demirel University, Department ofMathematics, 32260, Isparta, Turkey,[email protected]

1 - The Choquet Integral as a continuous certaintyequivalentGianni Bosi, Dipartimento di Scienze Economiche, Aziendali,Matematiche e Statistiche, University of Trieste, Italy,[email protected], Magalì E. Zuanon

The concept of a certainty equivalent is relevant in many fields whenpreferences over stochastic situations are considered (for example,stochastic cooperative games or risk attitudes in a multiperiod con-text). We present a characterization of the existence of a continuouscertainty equivalent which can be represented as the Choquet integralwith respect to a concave probability distortion, therefore continuingin some sense previous work by Bosi, Zuanon et al. [MathematicalSocial Sciences, 2003, 2007].

2 - An Adversarial approach to maximize gain due to fi-nite precision errors in basic financial computationsGanesh Perumal M, International Institute of InformationTechnology, Bangalore, Electronics City, Hosur Road,560100, Bangalore, Karnataka, India,[email protected], Abhilasha Aswal, G. N.Srinivasa Prasanna

229


Basic Financial transactions including deposit, withdrawal and cur-rency conversion may be erroneous due to the usage of IEEE 754 bi-nary standards. This cumulative error can be significantly maximizedin ones favour (especially in multicurrency transactions), using a cleveradversary, utilizing information about the error generation and propa-gation process. We give an optimization model to maximize gain dueto such errors.

3 - Handling peak loads in queuing service systemsAbraham Seidmann, University of Rochester, 14627,Rochester, NY, United States,[email protected], Mariya Bondareva

We study the efficient design of queuing service systems with tightservice level constraints and non-stationary periodic arrival rates. Theresearch introduces the concept of ’peaker’ capacity services for han-dling variable peak-load demands. Internal expert teams handle thebase-load of transactions, while the external peaker team is contractedto handle only the peak demand. We developed a two-staged gameto choose the optimal contracting policy with the peaker, and a robustprocedure to determine both the base-load - and the peaker’s capacitylevels.

4 - Pricing Strategies for Gaming-on-DemandSumanta Basu, Operations Management, Indian Institute ofManagement Calcutta, K-302, New Academic Block, IIMCalcutta, D.H. Road, Joka, 700104, Kolkata, West Bengal,India, [email protected], Soumyakanti Chakraborty

Social gaming has pushed the traditional video gaming industry to thebrink of disruption using concepts like cloud gaming or gaming-on-demand. However, the success of the gaming-on-demand model woulddepend to a large extent on pricing, e.g. non-linear usage based fee, afixed fee or a combination of both. In this paper, we develop a pricingschedule for a typical cloud game provider by modeling the alterna-tives available to a consumer and the propensity to engage in gaming.We also determine the conditions that determine the switching behav-ior of customers from usage based to fixed fee plan.


Recent Achievements in Regression andClassificationStream: Computational StatisticsInvited sessionChair: Fatma Yerlikaya Ozkurt, Scientific Computing, Institute ofApplied Mathematics, Industrial Engineering Department, MiddleEast Technical University, 06800, Ankara, Turkey,[email protected]: Semih Kuter, Department of Forest Engineering, CankiriKaratekin University, Cankiri Karatekin Universitesi, OrmanFakultesi, 18200, Cankiri, Turkey, [email protected]

1 - Restructuring the forward step of CMARS by a map-ping approachElcin Kartal Koc, Statistics, Middle East Technical University,Department of Statistics, No:234, 06800, Ankara, Turkey,[email protected], Cem Iyigun, Inci Batmaz,Gerhard-Wilhelm Weber

CMARS is proposed as a new approach to MARS with its model-basedalternative to the backward step. Instead of stepwise elimination step,a penalized residual sum of squares is used as a Tikhonov regulariza-tion problem which is solved with a conic quadratic programming. Thedrawback of CMARS is its time inefficiency. In this study, to increasethe run time of CMARS, a mapping idea is proposed for knot selectionin the forward step. Integration of mapping idea into CMARS-forwardalgorithm can decrease the computing time significantly by consider-ing less number of data points as candidate knots.

2 - Balance Optimization Subset Selection (BOSS) forCausal Inference with Observational DataAlexander Nikolaev, Department of Industrial and SystemsEngineering, University at Buffalo (SUNY), 409 Bell Hall,14260-2050, Buffalo, NY, United States,[email protected], Sheldon Jacobson

To make causal inferences outside the experimental realm, researchersattempt to post-process large observational data sets to mimic experi-mental data. The primary approach used has been to match treated anduntreated units to achieve a measure of balance. We explain BOSS,a novel approach that eliminates the matching stage and directly opti-mizes balance. It identifies sets of statistically indistinguishable treat-ment and control groups, perfectly optimized with respect to one ormore pre-defined balance measures, giving a minimally biased estima-tor of the treatment effect.

3 - Testing of MARS on MODIS Images for Local Atmo-spheric CorrectionSemih Kuter, Department of Forest Engineering, CankiriKaratekin University, Cankiri Karatekin Universitesi, OrmanFakultesi, 18200, Cankiri, Turkey, [email protected],Zuhal Akyurek, Gerhard-Wilhelm Weber, Ayse Özmen

In this study, Multivariate Adaptive Regression Splines (MARS) isused for local atmospheric correction. Being adaptive and flexible,both linear and non-linear problems with large data sets can be han-dled by MARS. Local atmospheric correction is carried out by apply-ing MARS method on real data, sets of MODIS (Moderate-resolutionImaging Spectroradiometer) images over Europe and Turkey. Obtainedresults are compared with the results of SMAC (Simplified Method forAtmospheric Correction — a numerical radiative transfer model) algo-rithm both numerically and visually.

4 - Estimation of Multi-dimensional Stochastic Differen-tial Equations with CMARS MethodFatma Yerlikaya Ozkurt, Scientific Computing, Institute ofApplied Mathematics, Industrial Engineering Department,Middle East Technical University, 06800, Ankara, Turkey,[email protected], Gerhard-Wilhelm Weber

CMARS, an alternative approach to the well-known data mining toolMARS, is based on a penalized residual sum of squares for MARS asa Tikhonov regularization problem. It treats this problem by a ConicQuadratic Programming. Stochastic Differential Equations are widelyused to represent noisy and real world problems. SDEs are usuallyhard to represent and to resolve by a computer. Before, we identified1-dimensional SDE in a simplified manner by using CMARS. Now,we aim to estimate multi-dimensional SDEs with CMARS. The theo-retical results and implementation of this study will be presented.


Recent Advances of OR on SocietalComplexity

Stream: Methodology of Societal ComplexityInvited sessionChair: Dorien DeTombe, Methodology of Societal Complexity,Chair Euro Working Group, P.O.Box 3286, 1001 AB, Amsterdam,Netherlands, [email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - By revitalization of ancient abandoned cemeteries tosocietal and economic strengthening of small sitesEva Kralova, Institute of History and Theory of Architectureand Monument Restoration, Slovak university of technology,Faculty of architecture, Namestie slobody 19, 81245,Bratislava, Slovakia, Slovakia, [email protected]

230


For societal and economic development of small sites, it is important toutilize local potential. Cultural traditions, local monuments and mem-ory places are parts of the potential source for revitalization. Aban-doned cemeteries represent a special group of such socially acceptedand venerated places; without an active local societal life it becomes acomplex societal problem. We present the case study of an old ceme-tery in Banska Stiavnica, Slovakia (UNESCO World Heritage). We in-troduce the approach and processes of revitalization using operationalresearch.

2 - A surprising equivalence between Consolation ofPhilosophy and Methodology for Societal Complex-ityNicolae Bulz, National Defence College, Romania,[email protected]

There is the affirmation on equivalence: two attempts toward difficult"problem solving". 1) Consolation of Philosophy, by Roman philoso-pher Boetius and 2) Methodology for Societal Complexity, by prof.Dorien J. DeTombe. To point on a single occurrence of the conceptproblem/solve across the text of Consolation of Philosophy: Book III.True Happiness and False. "I understand the problem, but I desire tohear how thou wouldst solve it." An inquiry within the entire contentof Boetius’ texts would possible depict significant equivalence betweenhis and our attempts on "problem solving".

� WC-37Wednesday, 12:30-14:00CC-Act

Operations Research in Health Care I

Stream: Health Care ManagementInvited sessionChair: Patrick Hirsch, Institute of Production and Logistics,University of Natural Resources and Life Sciences, Vienna,Feistmantelstrasse 4, 1180, Wien, Austria,[email protected]

1 - Weekly routing and scheduling of home health careservicesAndrea Trautsamwieser, Institute of Production and Logistics,University of Natural Resources and Life Sciences, Vienna,Feistmantelstraße 4, 1180, Vienna, Austria,[email protected], Patrick Hirsch

The task of planning the visits of mobile nurses can be modeled asrich periodic vehicle routing problem. However, several constraintshave to be met. The clients need to be visited regularly during theweek by appropriately skilled nurses who should not change too fre-quently. Moreover, working time regulations, time windows, and inter-dependent visits have to be considered. We propose a new matheuristicmethod that combines a Branch-Price-and-Cut approach with a Vari-able Neighborhood Search based metaheuristic. This new approach istested in extensive numerical studies with real life data.

2 - Scheduling of urban home health care services us-ing time-dependent multimodal transportKlaus-Dieter Rest, Institute of Production and Logistics,University of Natural Resources and Life Sciences, Vienna,Feistmantelstrasse 4, 1180, Vienna, Austria,[email protected], Patrick Hirsch

Optimization of home health care services is a rather young but quicklyevolving research area due to the expected significant increase in de-mand. We present a Tabu Search based solution approach where nursescan use public transport (bus, tram, subway, and train) in addition tocars, bikes, and walks. While the latter are currently considered astime-independent, public transport is based on actual time-tables. Thedeveloped solution approach is tested with real world data from theAustrian Red Cross in Vienna.

3 - Using agent-based modelling and simulation in pa-tient adherence to medicationChristos Vasilakis, Clinical Operational Research Unit,University College London, 4 Taviton Street, WC1H 0BT,London, United Kingdom, [email protected], ChristinaPagel, Steve Gallivan, Nick Barber, Martin Utley

It is known that patient adherence to medication can determine the ef-fectiveness of a therapy. It is less known what the impact on clinicaleffectiveness is of a number of interacting factors: the pharmacokineticfeatures of a drug and drug regimen, the different patterns of patientadherence and the interaction of patients and prescribers in improvingadherence. In this study we used agent-based modelling and simula-tion to explore the impact of hypothetical patterns of behaviour andinteraction between patients and prescribers on drug concentrations.

4 - Uncovering the hospital length of stay distribution: afinite mixture model approachMaria Guzman Castillo, University of Southampton, UnitedKingdom, [email protected], Sally Brailsford,Honora Smith

Most of the research on modelling length of stay (LoS) attempts to finda suitable probabilistic distribution to describe patient LoS. However,single distributions have shown limited success for this task. Lately,it has been suggested that LoS data may in fact represent a mixtureof several distributions. These models are refereed as finite mixturemodels, where a continuous variable consists of 2 or more clusters ofobservations with different parameters. This research extends previ-ous work by exploring different types of mixtures and by integratingpatient attributes into the model.

� WC-38Wednesday, 12:30-14:00HH-Colombus

Applications of MultiobjectiveOptimization

Stream: Multiobjective OptimizationInvited sessionChair: Carlos Henggeler Antunes, DEEC, University of Coimbraand INESC Coimbra, Rua Antero de Quental 199, 3000-033,Coimbra, Portugal, [email protected]

1 - A multi-objective model to determine electrical loadcontrol actionsCarlos Henggeler Antunes, DEEC, University of Coimbra andINESC Coimbra, Rua Antero de Quental 199, 3000-033,Coimbra, Portugal, [email protected]

The exploitation of demand-side resources using load control actions(LCA) in electricity distribution networks has the potential to bringbenefits to several stakeholders (customers, utilities, etc.). LCA con-sist in the application of power curtailment actions (on-off patterns),changing the usual working cycles, to obtain operational and economicbenefits. A multi-objective model for determining those LCA is pre-sented, which takes into account different stakeholder perspectives. Anoverall energy resource management is provided, including loads, lo-cal generation and storage.

2 - A Fuzzy Goal Programming Model for Solving Ag-gregate Production-Planning Problems under Uncer-taintyFernando Marins, Production, UNESP - São Paulo StateUniversity, Av. Ariberto Pereira da Cunha, 333, 12516-410,Guaratinguetá, SP, Brazil, [email protected], AneirsonSilva, José Arnaldo Montevechi, Ubirajara Ferreira

We propose a Fuzzy Goal Programming Model (FGP-APP) for the ag-gregate production-planning problem. It was made an application ina Brazilian company of sugar and ethanol milling. The FGP Modeldepicts the comprehensive production process of sugar, ethanol, mo-lasses and derivatives, and considers the uncertainties involved in the

231


ethanol and sugar production, as well as in decisions related to agricul-tural and logistics phases, on a weekly-basis planning horizon, includ-ing the whole harvesting season and the periods between harvests. Themodel was applied to a real case with excellent results.

3 - A Multi-objective Portfolio Optimization Model withHigher Order MomentsBijaya Krushna Mangaraj, Production & OperationsManagement Area, XLRI School of Business & HumanResources, Jamshedpur-831001, India, C.H. Area(East),Jamshedpur, Jamshedpur, Jharkhand, India,[email protected], Gayatri Biswal

Markowitz’s seminal work on modern portfolio theory is based on astrong normality assumption of the distribution of portfolio returns andproposed a parametric approach using mean and variance. However,due to non-parametric nature of such distributions, study on modernportfolio efficiency measure has been extended to a mean-variance-skewness-kurtosis framework. We discuss a multi-objective approachof portfolio optimization using fuzzy goal programming for generat-ing a non-dominated portfolio in a four-dimensional case. We test themodel with a sample of stocks from Mumbai Stock Exchange.

4 - Trade-off in statistical design processSamuel Bohman, Department of Computer and SystemsSciences, Stockholm University, Forum 100, 164 40, Kista,Kista, Sweden, [email protected], Maria Kalinina, AronLarsson

The design and planning of a statistical survey can be regarded as acomplex decision process involving multiple stakeholders and a di-verse set of users. The process involves finding a balance betweenstatistical quality, burden on respondents and costs. Our research isaimed towards developing a prototype system consisting of computa-tional core based on the combination of multi-objective optimization,multi-criteria analysis, and a user-friendly interface. The overall objec-tive of the research is to maximize statistical quality while minimizingthe burden on respondents and costs.

� WC-39Wednesday, 12:30-14:00HH-Cousteau

Cops and Robber Games

Stream: Graph SearchingInvited sessionChair: Nancy Clarke, Mathematics and Statistics, AcadiaUniversity, 12 University Avenue, B4P2R6, Wolfville, NovaScotia, Canada, [email protected]

1 - Slow Firefighting on GraphsMargaret-Ellen Messinger, Mathematics Department, MountAllison University, New Brunswick, Canada,[email protected]

In firefighting and graph searching problems, a contaminant typicallyspreads very quickly. We introduce a variant in which the contaminantspreads at a slower rate. At each time step, the searchers (or firefight-ers) protect one vertex and the contaminant (or fire) spreads from onevertex to all unprotected, unburned neighbours. On infinite graphs,the goal of the firefighter is to surround the burned vertices with a setof protected vertices, while the goal of the fire is to avoid such con-tainment. We will consider results on both the two-dimensional andthree-dimensional grids.

2 - Cops and a Robber on a graph embedded into a sur-faceDirk Oliver Theis, Mathematics, Otto von Guericke UniversityMagdeburg, Germany, [email protected], Nancy Clarke

Given a graph embedded on a surface genus $g$, what is the minimumnumber of Cops needed to catch a Robber?

We discuss an approach based on covering spaces. For non-orientablesurfaces, this technique improves the formerly known bounds effort-lessly. For orientable surfaces, with considerable effort, we reduce the

question to one concerning a polytope in the homology space of thesurface defined by the cycles of the graph.

Joint work with Nancy Clarke (Acadia, NS) and Samuel Fiorini, Gwe-naël Joret, Mathieu Carette (ULB, Brussels).

3 - Ambush Cops and RobbersNancy Clarke, Mathematics and Statistics, Acadia University,12 University Avenue, B4P2R6, Wolfville, Nova Scotia,Canada, [email protected]

A variation of the Cops and Robber game is introduced in which therobber side consists of two robbers. The cops win by moving onto thesame vertex as one of the robbers after a finite number of moves. Therobbers win by avoiding capture indefinitely or by both moving ontothe same vertex as the cop. (Otherwise, the robbers are on distinct ver-tices.) We present a variety of results including a recognition theoremfor graphs on which a single cop can guarantee a win, as well as astrategy that can be used by the cop to win on such graphs.

� WC-40Wednesday, 12:30-14:00HH-Livingstone

Model quality and modelling sustainableenergy and transport systems

Stream: SD Modelling in Sustainable DevelopmentInvited sessionChair: Pierre Kunsch, MOSI, Vrije Universiteit Brussel, Pleinlaan2, 1050, Brussels, Belgium, [email protected]

1 - Comparing a BRT and a PRT systemTsutomu Mishina, Systems Science and Technology, AkitaPerfectural Univeristy, 84-4 Tsuchia-Ebinokuchi, 015-0055,Akita, Japan, [email protected], Alejandra GomezPadilla

Bus Rapid Transit (BRT) is a proven effective method for mass trans-portation. Developing countries have introduced it big cities. PersonalRapid Transit (PRT) is a relatively new transportation system, with de-veloping applications usually in developed countries. This work com-pares of both systems. The data of an actually operating BRT systemis considered, and this data is used supposing that a PRT system isrunning instead. This is, for the same demand of passengers, smallervehicles are used, which pass more frequently, run faster and will onlystop at the final destination station.

2 - Automatic control of turbo generators setting auto-matic frequencyMeglouli Hocine, Automatique, Université de Boumerdes,boulevard de l’indepandence, cité 392 lgts bt 17 G08, 35000,Boumerdes, Boumerdes, Algeria, [email protected]

The production-consumption balance is ensured, in normal operation,two additional automatic actions: setting primary and secondary con-trol. Their performance is critical to the safety of the system.

The primary control, it is crucial for power system reliability during thelarge amplitude variations, to master quickly the transition frequencyso as not to reach the top levels of shedding frequency.

3 - Phasing-out nuclear energy in Belgium: is it feasi-ble?Pierre Kunsch, MOSI, Vrije Universiteit Brussel, Pleinlaan 2,1050, Brussels, Belgium, [email protected]

The Belgian government has decided to phase-out nuclear power plantsfrom 2015 on till 2025 while they represent today more than half theelectricity production. By means of system dynamics modelling thispresentation investigates different scenarios to check the feasibility ofsuch decisions. It is found that gas and coal would be the winners,rather than renewable energy sources; this is rather the opposite of thesought for environment-friendly evolution.

232


4 - Model Quality: How to Build Better ModelsMarkus Schwaninger, Institut für Betriebswirtschaft,Universität St.Gallen, Dufourstrasse 40a, 9000, St.Gallen,St.Gallen, Switzerland, [email protected]

Building high-quality models is perhaps the primary challenge aheadof us in modeling and simulation. This rationale is grounded in theConant/Ashby-theorem, which will be introduced. Thereafter, strate-gies and methods for the validation of dynamic-simulation models,and other measures for the enhancement of model quality, will be pre-sented.

� WC-41Wednesday, 12:30-14:00HH-Heyerdahl

Modern project for involvement youngpeople in OR community and OReducationStream: Young People for System Theory, Optimizationand EducationInvited sessionChair: Bohdan Pukalskyi, Students Science Association, NationalTechnical University of Ukraine, Kovalskyi provulok 5, 03057,Kyiv, Ukraine, [email protected]: Kateryna Pereverza, Students Science Association,National Technical University of Ukraine, Kyiv, Scherbakova str.,42/44, app. 54, Kyiv, Ukraine, [email protected]

1 - Achieving interdisciplinarity in OR by additional ed-ucational activities. Example of Summer SchoolAACIMPIryna Smolina, Metal Physics, NTUU "KPI", 03056, Kyiv,Ukraine, [email protected]

Interdisciplinarity is one of essential characteristics of OR as well asof all modern science. Additional educational activities like summerschools, on-line courses and internships are effective tools for spread-ing this principle among new generations of OR researchers. Sum-mer school "AACIMP" is good example of such educational activities,which combines different OR approaches to solve various problemsand provide students with better understanding of interdisciplinarity ofcontemporary science. We present the experience of organization ofthe Summer School "AACIMP".

2 - Odss.4SC: A summer school in Optimization and De-cision Support Systems for Supply Chain Manage-mentGiuseppe Bruno, Dipartimento di IngegneriaEconomico-Gestionale, Università Federico II di Napoli,Piazzale Tecchio n.80, I80125, Napoli, IT, Italy,[email protected], Ana Amaro, Miguel Casquilho,Albert Corominas, Andrea Genovese, Juan Manuel GarciaLopez, Amaia Lusa, Johan Magnusson, Henrique Matos, JoaoMiranda, Sergio Rubio

Our Summer School is taking place in Portalegre, Portugal, July, 1-15, within the Erasmus Intensive Programme. The audience is formedby MSc./PhD. students on Engineering and Logistics specialties, frommore than 12 countries. It aims at: (a) describing the industry-basedsupply chain operations with a particular concern to the green logisticaspects; (b) using optimization models to solve specific problems. Thelectures follow a pull approach, from distribution to supply subjects;the lab sessions are supported by IBM/ILOG. We present the programand the future perspectives.

3 - Mathematically formalized methodology for scenariostudy. Approaches for scenario space explorationKateryna Pereverza, Students Science Association, NationalTechnical University of Ukraine, Kyiv, Scherbakova str.,42/44, app. 54, Kyiv, Ukraine, [email protected]

Scenario analysis is well known technique for exploration of future de-velopment of different complex systems as countries, cities, companiesetc. The stage of scenario space exploration is one of the most rele-vant in scenario study. Big amount of built scenarios can be exploredusing different OR methods as associated analysis, Bayes network etc.The mathematically formalized methodology for scenario study will bepresent in the talk. The example of scenario space analysis for countryusing mentioned methods will be shown.

� WC-42Wednesday, 12:30-14:00BW-Amber

Queueing Systems II

Stream: Stochastic Modeling and Simulation in Engi-neering, Management and ScienceInvited sessionChair: Erik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39, 85577,Neubiberg, Germany, [email protected]: Devin Sezer, Middle East Technical University, Institute ofApplied Mathematics, Eskisehir Yolu, 06531, Ankara,[email protected]: Basak Tanyeri, Bilkent University, Turkey,[email protected]

1 - Analyzing Multi-server Queueing Model with no Wait-ing LineHanifi Okan Isguder, Statistics, Dokuz Eylul University,Dokuz Eylul University Faculty of Science, Department ofStatistics Tinaztepe Buca, 35160, Izmir, Turkey,[email protected], Sener Akpinar, Atabak Elmi, AlperHamzadayi, Simge Yelkenci KoseIn this study a queueing system with recurrent arrivals, three hetero-geneous servers, and no waiting line is examined. In the system cus-tomers choose only one server among the empty servers with differentprobabilities. When all servers are busy, customers depart from thesystem without taking any service. These customers are called "lostcustomer’. The probability of losing a customer is computed for thequeueing system.

2 - Distribution Evaluation and Decomposition forQueueing Networks ModellingJean-Sébastien Tancrez, Louvain School of Management,Université catholique de Louvain, UCL - LSM, 151, Chausséede Binche, 7000, Mons, Belgium,[email protected] modeling methodology is presented for queueuing networks withgeneral assumptions : fork/join networks, general service time distri-butions and finite buffers. The general distributions are discretized byaggregation of the probability masses, and the transformed system isanalytically modeled using a Markov chain. First, we show that it al-lows a fine estimation of the cycle time distributions. Second, we intro-duce a decomposition approach to analyze larger queueing networks.Using computational experiments, we show that it leads to accurateperformance evaluation and that the fine distribution estimation seemsto bring an improvement.

3 - Alternative stochastic approaches for condition-based production-maintenance controlMichael Kaluzny, Business Administration, Production andLogistics, University of Dortmund, Germany,[email protected], Ralf GössingerCondition-based maintenance is analysed for a decentralisedproduction-maintenance system where deterioration depends on pro-duction parameter settings. In order to keep up a defined machineavailability an information exchange protocol is developed. On thisbasis alternative stochastic approaches to calculate the machine condi-tion triggering the order release are formulated and numerically anal-ysed. These approaches take waiting time between maintenance orderrelease and order fulfillment as well as possibly modified productionparameter settings during this waiting time into account.

233


4 - Optimal Control Formulation of Query Model for Au-thentication SystemsDevin Sezer, Middle East Technical University, Institute ofApplied Mathematics, Eskisehir Yolu, 06531, Ankara,[email protected], Ferruh Ozbudak, Yildirim UstunThis research studies the N step adaptive attack on an authenticationsystem using an authentication oracle described in (Safavi-Naini, Wild,2008). We represent the maximum success probability of this type ofattack as the value function of an optimal control problem and derivethe dynamic programming equation that the value function satisfies.The same paper cited above proves that randomized queries in such anattack have the same probability of success as that of a deterministicattack. We rederive this fact using our control representation.

� WC-44Wednesday, 12:30-14:00BW-Marble

Information and Intelligent Systems I

Stream: Information and Intelligent SystemsInvited sessionChair: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]: Wuyi Yue, Intelligence and Informatics, Konan University,8-9-1 Okamoto, Higashinada-ku, 658-8501, Kobe, Japan,[email protected]

1 - Malfunction diagnosis on air-conditioning usingrough sets approachHong Tau Lee, Industrial Engineering and Management,National Chin-Yi University of Technology, Taipin, TaichungCounty, 411, ROC, Taiwan, [email protected], Sheu-HuaChenIn order to effectively diagnose faults causing, this paper purposes asimplified approach for diagnosing air conditioning system which isbased on rough sets theory. This approach converts the reasons andsymptoms to condition and decision attributes respectively. Finally,possible reasons can be diagnosed while the specific set of core at-tributes of symptom occurred. This is a simplified way of avoiding thetesting of all the items that are usually used in the traditional methodof determining the fault cause.

2 - Approach for the prolongation of the lifetime of awireless sensor networkAbdelmalek Boudries, University of Bejaia, university ofSétif, Algeria, Boumansour Cne Oued-Ghir W- Béjaia, Béjaia,Algeria, [email protected], Makhlouf AliouatWe propose an approach for the connectivity maintenance in the wire-less sensor networks. The principle of the solution is the following:a node which wants to transmit data chooses the way of the greatestweight. The idea is that if the remaining rate of energy decreases thenthe weight of the way containing this node decreases. The aimed ob-jective is to maintain the connected network for longest time possibleby prolonging the lifetime of the nodes sensors which have consumedmuch energy. The suggested approach was tested on an example ofscenario and showed its effectiveness.

3 - Analysis of real-time traffic with the sleep mode inIEEE 802.16m networksWuyi Yue, Intelligence and Informatics, Konan University,8-9-1 Okamoto, Higashinada-ku, 658-8501, Kobe, Japan,[email protected], Shunfu Jin, Xiaofei ZhangIn this paper, based on the working principle of the sleep mode in IEEE802.16m, a queueing model with multiple heterogeneous vacations ispresented. Considering the packet correlation in real-time traffic, thesystem arrival is supposed to be a discrete-time Markovian process.The formulas for the average latency and energy saving ratio are ob-tained, and the numerical results are given. The time length of thesleep cycle is optimized by constructing a cost function. The authorsare grateful to Prof. M. Kokai for his helpful comments in this paper.

4 - Planning of Map Making Technique by PlanningMethod with Continuously Operating Reference Sta-tion (CORS-TR)Omur Esen, Office of Construction and Infrastructure, SelcukUniversity, Division of Cartography, 42250, Konya, Selcuklu,Turkey, [email protected], I.Bulent Gundogdu

CORS project has started in USA and Turkey as part of the projectCORS-TR approved by Istanbul Culture University, General Direc-torate of Land Registry and Cadastre and General Command of Map-ping cooperation and by The Scientific and Technological Council ofTurkey sponsorship. Aim is to determine data by 7/24 observation inTurkey according to geographical position real time and by post pro-cessing in a fast and economical way and at the same time modeling ofionosphere and troposphere, requiring geological data and determiningtransformation parameters between ED50 and ITRF Dates.

� WC-45Wednesday, 12:30-14:00BW-Water

Advanced Revenue Management

Stream: Advanced Inventory Control and PricingStrategiesInvited sessionChair: Claudius Steinhardt, Department of Analytics &Optimization, University of Augsburg, Universitätsstraße 16,86159, Augsburg, Germany,[email protected]

1 - Pricing-cum-Inventory Decisions in Supply ChainNetworksLambros Pechlivanos, Athens University of Economics andBusiness, Greece, [email protected], Panos Seferlis

An optimization-based control framework that simultaneously deter-mines the optimal inventory and product pricing policies is developedfor multi-product, multi-echelon supply chain networks. Product pricemanipulation can be used to alleviate congested transportation routes,by altering appropriately the demand profile at the end-point nodes ofthe network. A flexible node-level pricing policy can be understood asa substitute instrument to supply chain management, as it succeeds inaltering the flow of orders customers place by redirecting orders to thedesired end-point nodes.

2 - Modeling of customer choice behavior in revenuemanagementPetr Fiala, Dept. of Econometrics, University of EconomicsPrague, W.Churchill Sq. 4, 13067, Prague 3, Czech Republic,[email protected]

Revenue management is the process of understanding, anticipating andinfluencing customer behavior in order to maximize revenue. The com-mon modeling approaches assume that customers are passive. Today’scustomers actively evaluate alternatives and make choices. Revenuemanagement pays increasing attention to modeling the behavior of in-dividual customers. A modeling approach for strategic customer be-havior is proposed. The evaluation of alternatives can be done by DEAbased evaluation methods. The approach is based on instruments ofartificial neural networks and Petri nets.

3 - Incorporating Auction Games in Logistics Optimiza-tion Academic CourseCharis Marentakis, Industrial Management and Technology,University of Piraeus, Karaoli and Dimitriou Street, 18534,Pireaus, Greece, [email protected], Dimitrios Emiris, MariaGiannopoulou

Auctions and Revenue Management methods are currently evolving inthe research and practice of Logistics Management. Several Opera-tions Research techniques are engaged in this direction, aiming to sup-port pricing decisions related to optimal capacity and resource manage-ment. The present article describes experiences from the enrichment

234

EURO 25 - Vilnius 2012 WD-02

of a Logistics Optimization course with an auction-based freight trans-port business game. It explains why and how to engage auctions inconformance with the theoretic operations research part of the courseand demonstrates major outcomes focusing mainly on the behavioralstudy of participants and the convergence between participants’ valua-tions.

4 - Decision tools for online pricing and sourcing fortruckload truckingHani Mahmassani, Transportation Center, NorthwesternUniversity, 600 Foster St, 60208, Evanston, Illinois, UnitedStates, [email protected], Diego Klabjan, HamedBabai, Young Woong Park, Christopher Lindsey

This study develops tools to support online operations of a third-partynon-asset carrier in matching truckload shipments between shippersand carriers. The model considers respective shipper and carrier ac-ceptance likelihood in an optimal pricing engine that maximizes profitfor the matching agent.


� WD-01Wednesday, 14:30-16:00RB-Alfa

Tutorial Lecture: Professor Karla HoffmanStream: Keynote, Tutorial and Plenary LecturesTutorial sessionChair: Mikael Rönnqvist, Département de génie mécanique,Université Laval, G1V 0A6, Québec, Canada,[email protected]

1 - Auctions: Why they proliferating and what you needto know to participateKarla Hoffman, Department of Systems Engineering andOperations Research, George Mason University, Mail Stop4A6, 4400 University Drive, 22030, Fairfax, Virginia, UnitedStates, [email protected]

The advent of the Internet has led to the creation of global market-places in which sales of everything from low-cost used merchandiseto billion dollar government procurements are conducted by auction.While some auction designs have been extremely successful, othershave failed disastrously. This talk will describe a variety of alterna-tive auction designs, and will explore why one design is likely to bemore appropriate than another for a given application. We will beginwith a brief history of auctions, provide reasons for eBay’s spectacu-lar success, and highlight the Internet’s impact on auction theory andpractice. We will then move on to a discussion of the strategies thatare available to bidders in the most-utilized auction formats, therebyproviding each of you with hints on how to be more successful whenparticipating in auctions. Throughout the talk, we will highlight therole that operations research analysts can play in designing, running,and assisting participants in business-to-business and high-stakes gov-ernment auctions.

� WD-02Wednesday, 14:30-16:00RB-Beta

Methods for Resource-ConstrainedProject Scheduling

Stream: Project Management and SchedulingInvited sessionChair: Vincent Van Peteghem, People, markets and humanities,EDHEC Business School, 24, avenue Gustave Delory, 59057,Roubaix, France, [email protected]

1 - Solving resource-constrained project schedulingproblems with new mathematical programming for-mulationsThomas Kyriakidis, Department of Engineering Informatics &Telecommunications, University of Western Macedonia,Karamanli and Lygeris Street, 50100, Kozani, Kozani, Greece,[email protected], Georgios Kopanos, Michael Georgiadis

In this work, 2 binary-integer programming discrete-time models and2 mixed-integer continuous-time formulations are presented for the so-lution of resource-constrained project scheduling problems. These for-mulations are compared with 4 state-of-the-art literature models us-ing a total of 2760 benchmark problem instances (j30 and j60 fromPSPLIB, and 1800 problem instances generated by RanGen2). A de-tailed computational comparison study illustrates the efficiency of theproposed mathematical formulations, with the continuous-time math-ematical formulations featuring the best overall performance.

235

WD-03 EURO 25 - Vilnius 2012

2 - Solving the resource-constrained project schedulingproblem using large neighborhood searchWolfgang Summerauer, Department of BusinessAdministration, University of Vienna, Bruenner Strasse 72,1210, Vienna, Vienna, Austria,[email protected], Christian Almeder,Richard Hartl

We present an algorithm based on large neighborhood search (LNS) tosolve the resource-constrained project scheduling problem (RCPSP),where a project consists of activities that have to be scheduled underresource and precedence constraints. The objective is to find a sched-ule that minimizes the makespan of the project. We developed newtailor-made destroy and repair operators and also adapted existing op-erators from the literature to tackle the RCPSP. Preliminary results forthe benchmark instances of the literature look promising with respectto solution quality and computational time.

3 - An Invasive Weed Optimization algorithm for the re-source availability cost problemVincent Van Peteghem, People, markets and humanities,EDHEC Business School, 24, avenue Gustave Delory, 59057,Roubaix, France, [email protected], MarioVanhoucke

In this study, an Invasive Weed Optimization (IWO) algorithm for theresource availability cost problem is presented, in which the total costof the (unlimited) renewable resources required to complete the projectby a pre-specified project deadline should be minimized. The IWOalgorithm is a new search strategy, which makes use of mechanismsinspired by the natural behavior of weeds in colonizing and finding asuitable place for growth and reproduction. All algorithmic compo-nents are explained in detail and computational results for the RACPare presented.

� WD-03Wednesday, 14:30-16:00RB-L1

Demand Responsiveness andTransportation Network Design

Stream: Optimization in Public TransportInvited sessionChair: Gabriel Gutiérrez-Jarpa, School of Industrial Engineering,Pontificia Universidad Católica de Valparaíso, 2362807,Valparaíso, Chile, Chile, [email protected]

1 - An analytical model to estimate the optimal cyclelength of demand responsive feeder transit servicesLuca Quadrifoglio, Civil Engineering, Texas A&MUniversity, CE/TTI Bldg. Room 301-I, 3136 TAMU,77843-3136, College Station, Texas, United States,[email protected], Shailesh Chandra

An increase of "livability" depends on solutions to the issue of first/lastmile access to transit. We are working to identify the feeder’s opti-mal cycle duration to maximize LOS. At first, a simulation case studywith demand data from El Cenizo, TX has been conducted by sim-ulation. We are now developing a handy analytical model to obtainresults, without relying on extensive simulation analyses. Experimentsappear to validate our model, which would allow planners, decisionmakers to quickly identify the best feeder transit operating design ofany given residential area.

2 - Dynamic Vehicle Routing for Demand ResponsiveTransportation ServicesJorge Pinho de Sousa, Faculdade de Engenharia daUniversidade do Porto / INESC Porto, Campus da FEUP, RuaDr. Roberto Frias, 4200-465, Porto, Portugal,[email protected], Rui Gomes, Teresa Galvão Dias

Providing quality public transportation may be very expensive in lowand unpredictable demand scenarios. Demand Responsive Transporta-tion (DRT) services address these issues with flexible routes and fre-quencies. We present an innovative approach for designing and op-erating these services, considering the perspectives of the differentstakeholders. This multi-objective approach integrates simulation andheuristics in a DSS, aiming at minimizing operating costs and max-imizing service quality. The DSS is also a tool to assess the perfor-mance of different DRT deployments and operating rules.

3 - Cut and Brach Synchronization Bus TimetablingYasmin Rios-solis, Graduate school of systems engineering,Universidad Autonoma de Nuevo Leon, Facultad deIngenieria Mecanica y Electrica, Av. Unviersidad s/n, 65450,Monterrey, Nuevo Leon, Mexico,[email protected], Omar Ibarra-RojasWe focus on bus timetabling problems of typical networks of LatinAmerica: large network where passenger transfers must be favored,almost evenly spaced departures are sought, bus bunching of differentlines must be avoided, timetable generation is needed several times aday, and travel times are irregular. We formulate this timetabling prob-lem with the objective of maximizing the number of synchronizations.We provide valid inequalities to tighten the formulation. Experimen-tation shows that our methods obtains solutions less than 1% from theoptimum in minutes even for large instances.

4 - The Hierarchical Network Design Problem with Traf-fic CaptureGabriel Gutiérrez-Jarpa, School of Industrial Engineering,Pontificia Universidad Católica de Valparaíso, 2362807,Valparaíso, Chile, Chile, [email protected], PabloIbacache, Vladimir MarianovThe hierarchical network design problem with traffic capture (HNDP-TC) finds a main path joining an origin and a destination, and sec-ondary paths that start on the main path and visit some nodes. Totalconstruction costs of the main and secondary paths are minimized, andcaptured traffic maximized. Traffic is captured if the use of the networkis more convenient than competing systems. Applications are found intransportation and distribution systems, and telecommunications. Weuse Branch and Bound for small instances, and an ad hoc heuristic forlarge instances.


Advances for solving Quadratic and OtherDifficult 0-1 Optimization Problems

Stream: Combinatorial OptimizationInvited sessionChair: Monique Guignard-Spielberg, OPIM, University ofPennsylvania, 5th floor, JMHH, 3730 Walnut Street, 191046340,Philadelphia, PA, United States, [email protected]: Peter Hahn, Electrical and Systems Engineering, Universityof Pennsylvania, 200 South 33rd Street, 19104-6314, Philadelphia,PA, United States, [email protected]

1 - Eigenvalues of Voting Power IndexesLee Papayanopoulos, MSIS, Rutgers University, RBS, 1Washington Park, 07102, Newark, NJ, United States,[email protected] explore the eigenvectors and eigenvalues of power indexes used inlegislative representation and their use in integer/mixed integer searchmethods for "fair" weighted votes, namely, votes that render the vot-ing power distribution (a combinatorial vector function) near-collinearwith an associated population vector of constituencies.

2 - Extensive experiments with the improved hybrid ge-netic algorithm for the quadratic assignment problemAlfonsas Misevicius, Multimedia Engineering, KaunasUniversity of Technology, Studentu st 50-400a/416a,LT-51368, Kaunas, Lithuania, [email protected]

236


Hybrid genetic algorithms (HGAs) are among the most efficient opti-mization techniques. In contrast to the standard GAs, which are basedon natural evolution in a straightforward manner, the hybrid GAs canbe understood as modeling of more complex environments, where thelifecycle transformations play the important role. In this work, sev-eral modifications of the improved HGA for the quadratic assignmentproblem (QAP) are tested, in particular: initial burst, incorporating theiterated local search, extra improvement, cohesive crossover, differentpopulation update schemes, using restarts, etc.

3 - The hypergraph assignment problemOlga Heismann, Zuse Institute Berlin, Germany,[email protected], Ralf Borndörfer

The hypergraph assignment problem (HAP) is the generalization ofassignments on directed graphs to directed hypergraphs. It serves, inparticular, as a universal tool to model several train composition rulesin vehicle rotation planning for long distance passenger railways. Evenfor problems with a small hyperarc size and hypergraphs with a spe-cial partitioned structure the HAP is NP-hard. We present interestingpolyhedral results derived from a complete facet classification of smallHAP polytopes. These can be generalized to cut inequalities for theLP relaxation of large-scale problems.

4 - 0-1 quadratic optimization problems: convexificationand solution.Monique Guignard-Spielberg, OPIM, University ofPennsylvania, 5th floor, JMHH, 3730 Walnut Street,191046340, Philadelphia, PA, United States,[email protected], Lucas Létocart, Gérard Plateau

Convexification of the objective function of a 0-1 quadratic optimiza-tion problem (GQAP, QKP,...) is theoretically possible using an SDPmodel and its dual. We will discuss the numerical difficulties encoun-tered in the process, and will show that for small to medium size in-stances, one may be able to obtain an equivalent convex model and areasonable lower bound. One can also try to generate improved boundsand good feasible solutions using the convex hull relaxation (CHR),and/or try to solve the convex model by standard MIQCP software,such as CPLEX.


Topics in integer and mixed integerprogramming II

Stream: Discrete Optimization, Geometry & Graphs(contributed)Contributed sessionChair: Iskander Aliev, School of Mathematics, Cardiff University,Senghennydd Road, CF24 4AG, Cardiff, [email protected]

1 - Optimize the payout probability table in slot ma-chinesXiaoming Liu, FBA, University of Macau, University ofMacau, Macau, NA, MACAU, China, [email protected],Zhaotong Lian, Xin Li

This paper theoretically derives the game values of slot machines fromthe payout probability distribution. It provides explicit formulas for allquantitative measurements that are of interest to both casino managersand players. Optimization of the pay-out distribution is also given tomaximize the pulls per losing player. The optimal payout and proba-bility lists are shown. Some numerical results are given. These resultscan be utilized by slot managers to choose the best trade-off customerexperience and casino gain.

2 - The Master Corner Polyhedron: VerticesVladimir Shlyk, Mathematics 3, Belarusian State TechnicalUniversity, 65 Independence av., 15 Kulman str., Apt. 27,220100, Minsk, Belarus, [email protected]

As shown by R. Gomory, the set of vertices of the master corner poly-hedron over a group is the union of orbits under the stabilizer of acertain group element. We prove that two combinatorial operationstransforming vertices to their neighbors map the orbits onto the others.So any system of representatives for some special orbits forms a poly-hedron vertex basis, often small in size. We geometrically characterizethe irreducible points and establish relations between the coefficientsof the facets through a given vertex. The computational potential ofthe results is discussed.

3 - The K-Separator ProblemMohamed Ahmed Mohamed Sidi, Telecom SudParis, 9 RueCharles Fourier, 91000, Evry, France,[email protected], Walid Ben-ameur,Jose Neto

Let G be a vertex-weighted undirected graph and k be a positive num-ber. We want to compute a minimum-weight subset of vertices whoseremoval leads to a graph where the size of each connected componentis less than or equal to k. Let us call such a set a k-separator. If k=1 weget vertex cover problem. The case k=2 is equivalent to compute thedissociation number of a graph (case of unit weights). We give someformulations of the problem. We point out some cases where the prob-lem is easy to. Some simple approximation algorithms are described.Numerical results are also reported.

4 - s-Fold Feasibility of Integer KnapsacksIskander Aliev, School of Mathematics, Cardiff University,Senghennydd Road, CF24 4AG, Cardiff, [email protected]

The integer knapsack problem is a well-known problem in integer pro-gramming. Since this problem is NP-hard in general, we are interestedin well-solved special cases. In particular, when the right-hand sidessatisfy certain geometric criteria, the existence of a feasible solutioncan be determined immediately, without the need for complex optimi-sation techniques. We focus on the case where the knapsack polytopecontains at least a prescribed number of integer points. We show thatthe feasibility of the problem appears to be closely related to the be-haviour of the s-covering radius.

� WD-06Wednesday, 14:30-16:00RB-Gamma

Game Theory Applications

Stream: Dynamical Systems and Game TheoryInvited sessionChair: Luis Ferreira, Matemática, Escola Superior de EstudosIndustriais e de Gestão, IPP, ESEIG, Rua D Sancho I, número 981,4480-876, Vila do Conde, Porto, Portugal,[email protected]

1 - Analysis of stroke patients EEG signals on the baseof cooperative game theoryHamidreza Navidi, science, Shahed Univercity, 0098, Tehran,Iran, Islamic Republic Of, [email protected], MajidHassanpour-ezatti, Saeid Barjesteh, Hadi Aghayari

Electroencephalogram (EEG) is useful tool for detection of stroke.Here, a model presented on the base of cooperative game theory toanalyses of features extraction of stroke patient EEG signals and thecontribution rate of these features on outbreak of stroke is calculated.This features obtained from EEG by use of detrended fluctuation anal-ysis method. In this research the data set were collected from 27 strokepatients. The result showed that some channels of EEG in stroke pa-tients have more contribution rate in stroke detection. This finding iscorrelated with some practical observation.

2 - A game theoretical approach to the Emergency Med-ical Vehicle - Emergency Department interfaceVincent Knight, School of Mathematics, Cardiff University,CF24 4AG, Cardiff, United Kingdom, [email protected]

237


At the forefront of any healthcare system are two interacting services:the Emergency Medical Vehicle (EMV) service and the EmergencyDepartment (ED). Often EMVs are subject to long delays at a particu-lar ED whilst the ED suffers from upstream congestion due to arrivalsof other patients. In this work, a game theoretical model is given. TheED’s strategies correspond to the allocation of resources throughoutthe hospital whilst the EMV service is modelled as a routing game.Various insights are given that inform as to optimal levels of policyregulation.

3 - Non-conflicting redistribution of optimal income fornon-cycling network systemsSergei Schreider, School of Mathematical and GeospatialSciences, RMIT University, GPO Box 2476V Melbourne,3001, Melbourne, Victoria, Australia,[email protected]

For LP optimized systems whose network structure is represented witha graph without inner cycles (which is a particular case of Berge’s nor-mal hypergraph) the stable optimal solution exists. The stability meansthat the total optimal income of the system can be distributed betweenplayers in the way that no individual player or coalition of players canbe better off playing individually. Such stable income redistributionwas found for two network systems, one of which water allocation andthe second one represents the natural gas allocation system.

4 - Strategy for Land Acquisition: a game-theoreticalmodelS Deman, Economics & Statistics, Centre for Statistics,Economics & Finnace, PO Box 17517, SE9 2ZP, London,United Kingdom, [email protected]

The process by which houses, land, and other forms of real propertyare sold has been the subject of considerable economic research overthe last few decades. However, the recent crash in housing prices inUS and in Western European countries made the study of this subjecteven more important.

� WD-07Wednesday, 14:30-16:00RB-Eta

Applications of Vehicle Routing 2

Stream: Vehicle Routing and Logistics OptimizationInvited sessionChair: Marco Castro, Faculteit Toegepaste EconomischeWetenschappen, Universiteit Antwerpen, Prinsstraat 13, 2000,Antwerpen, Belgium, [email protected]

1 - An Exact Approach for the Clustered Vehicle RoutingProblemDaniele Vigo, DEIS, University of Bologna, Via Venezia 52,47023, Cesena, Italy, [email protected], Maria Battarra,Gunes Erdogan

We consider the Clustered Vehicle Routing Problem (Clu-VRP), whichis a generalization of the well-known Capacitated Vehicle RoutingProblem (CVRP) in which the customers are grouped into clusters.Whenever a vehicle visits one customer in a cluster it must visit all theremaining vertices of the cluster before leaving it. The Clu-VRP is NP-Hard and has many applications in distribution logistics. We presentan integer programming formulation which is used as a base for exactBranch and Cut and Branch and Cut and Price algorithms which aretested on instances adapted from the literature.

2 - A Comparison of Cross Entropy Algorithms for theVehicle Routing ProblemEdgar Possani, Department of Mathematics, InstitutoTecnologico Autonomo de Mexico, Rio Hondo No.1, Col.Progreso Tizapan, C.P. 01080, Mexico City, D.F. - Mexico,Mexico, [email protected], Marta Cabo Nodar

The cross entropy (CE) method was developed to estimate rare eventprobabilities through simulation, and has been applied to some opti-mization problems. To solve a VRP we try to estimate the proba-bility that the total distance traveled is less than a given value, overall feasible routes. We have developed new CE heuristics based on acluster-first route-second scheme which do not limit the search overthe solution space as the existing CE algorithm did. We also tacklethe SVRP where the demands follow a given distribution. We presentcomputational results to evaluate these approaches.

3 - Orienteering Problem with Hotel Selection: A Vari-able Neighborhood Search MethodAli Divsalar, Center for Industrial Management/Traffic &Infrastructure, KULeuven, Belgium,[email protected], Pieter Vansteenwegen, DirkCattrysse

In the orienteering problem with hotel selection (OPHS) a set of ver-tices with a score and a set of hotels are given. The goal is to determinea fixed number of connected trips that visits some vertices and maxi-mizes the sum of the collected scores. Each trip is limited in lengthand should start and end in one of the hotels. A variable neighborhoodsearch algorithm is developed to solve the OPHS based on a neighbor-hood search operator designed for the hotel selection and some neigh-borhoods for the regular OP. Two appropriate sets of instances withknown optimal solutions are designed.

4 - Improved Methods for the Travelling SalespersonProblem with Hotel SelectionMarco Castro, Faculteit Toegepaste EconomischeWetenschappen, Universiteit Antwerpen, Prinsstraat 13, 2000,Antwerpen, Belgium, [email protected], KennethSörensen, Pieter Vansteenwegen, Peter Goos

In this talk, a new formulation and a new metaheuristic solution pro-cedure for the travelling salesperson problem with hotel selection(TSPHS) is presented. The metaheuristic is a multi-start procedurethat outperforms existing heuristics on all benchmark instances. Wealso provide a number of new optimal solutions found by a commer-cial solver extended with a dedicated cutting plane procedure, as wellas new best known solutions for most benchmark instances.

� WD-09Wednesday, 14:30-16:00RB-Zeta


Stream: EURO/ROADEF ChallengeInvited sessionChair: H. Murat Afsar, Industrial Systems, University ofTechnology of Troyes, 12 rue Marie Curie, BP 2060, 10010,Troyes, France, [email protected]: Eric Bourreau, COCONUT, LIRMM, 161 Rue Ada, 34000,Montpellier, France, [email protected]: Emmanuel Guere, Google, 38 Avenue de l’Opera, 75002,Paris, France, [email protected]: Ender Özcan, Computer Science, University ofNottingham, Jubilee Campus, Wollaton Road, NG8 1BB,Nottingham, United Kingdom, [email protected]: Kedad-Sidhoum Safia, LIP6 - UPMC, 4 Place Jussieu,75005, Paris, France, [email protected]: Christian Artigues, LAAS, CNRS, 7 avenue du ColonelRoche, 31077, Toulouse Cedex 4, France, [email protected]: Marc Sevaux, Lab-STICC - CNRS UMR 3192, Universitéde Bretagne Sud - UEB, Centre de Recherche - BP 92116, 2 rue deSaint Maude, 56321, Lorient, France, [email protected]

1 - A late acceptance metaheuristic for the machine re-assignment problemWim Vancroonenburg, CODeS, KAHO Sint-Lieven, KAHOSint-Lieven, Gebroeders De Smetstraat 1, B-9000, GENT,Belgium, [email protected], Tony Wauters

238


A local search method with a Late Acceptance strategy for the machinereassignment problem is proposed. Two neighborhoods have been de-veloped: probabilistic single reassignment and swap assignment. Bothhave been equipped with an efficient delta evaluation procedure. Eachiteration, one of the two is chosen at random, and only one neighbor-ing solution is evaluated. The results show great performance on thesmall dataset, and managed to obtain the first place in the qualifica-tion round. Further tuning will be performed on the medium and largeinstances for the final phase of the challenge.

2 - Basic VNS for Machine reassignmentNenad Mladenovic, School of Mathematics, BrunelUniversity, Kingston Lane, UB8 3PH, Uxbridge, Middlesex,United Kingdom, [email protected], BassemJarboui, Dragan UrosevicWe propose Basic VNS for Challenging machine reassignment. Twoneighborhood structures are used for Shaking: reassignment and swap.Local search uses re-assignment moves. We divide the set of con-straints into hard and soft. Hard constraints are always taken into ac-count in any move, while soft are included into objective function withincreasing penalty function. In other words, our objective function isdynamic and includes some constraints, if they are violated.

3 - ROADEF/EURO Challenge 2012 : Final Results an-nouncementEric Bourreau, COCONUT, LIRMM, 161 Rue Ada, 34000,Montpellier, France, [email protected], H. Murat Afsar,Christian Artigues, Emmanuel Guere, Kedad-Sidhoum Safia,Ender Özcan, Marc SevauxWe present the results of the ROADEF/EURO 2102 challenge, an in-ternational optimization contest propose jointly by EURO, the FrenchOR society (ROADEF) and an industrial partner (GOOGLE). Manyprizes are offered. Google propose 5000 euros in the junior category,5000 euros in the senior category and 10 000 euros for a specific OpenSource category. Intermediate qualification results (available sincefebruary 2012 on http://challenge.roadef.org/) have already shown thatthe competition is very tight, but after this presentation, the suspensewill be over as the winners will be revealed.

� WD-10Wednesday, 14:30-16:00RB-Theta

Integrated Planning of Operations

Stream: Supply Chain ManagementInvited sessionChair: Thomas Makuschewitz, BIBA - Bremer Institut fürProduktion und Logistik GmbH at the University of Bremen,Hochschulring 20, 28359, Bremen, Germany,[email protected]: Marc-André Isenberg, BIBA - Bremer Institut fürProduktion und Logistik GmbH at the University of Bremen,Germany, [email protected]

1 - A graph theory based heuristic for capacity and costoriented integrated production and maritime trans-port schedulingJens Hartmann, BIBA - Bremer Institut für Produktion undLogistik GmbH at the University of Bremen, Hochschulring20, 28359, Bremen, Bremen, Germany,[email protected], Thomas Makuschewitz, EnzoFrazzon, Bernd Scholz-ReiterInterdependencies between production scheduling and maritime trans-port determine costs and lead times in global supply chains. An in-tegrated scheduling combines flexible production and land transportschemes with overseas transport running a given timetable. In our talkwe present a heuristic scheduling method based on the constructionof a cost-weighted graph for each shipment, containing only feasiblepaths regarding time and capacity. In an iterative approach an over-all schedule is obtained by finding shortest paths for shipments. Theresults are obtained using a real world test case.

2 - Solution of facility layout problem through discreteparticle swarm optimization algorithm and an appli-cationAbdullah Oktay Dundar, Akoren Vocational School, SelcukUniversity, Akoren, KONYA, Turkey,[email protected], Mehmet Akif Sahman, MuhammetBezirci, Ali Alagoz

It is aimed in this study to demonstrate the difference in business coststhat will emerge after situating the units of the manufacturing plantin the best way and thus reducing the movement within the plant. Inthis context, the facility layout problem of a manufacturing businesslocated in the city of Konya was re-designed through Discrete ParticleSwarm Optimization algorithm, and the cost differences between thecurrent and the designed plant location were demonstrated.

3 - Scheduling a flowshop of batch processors contain-ing individual job familiesMarc-André Isenberg, BIBA - Bremer Institut für Produktionund Logistik GmbH at the University of Bremen, Germany,[email protected], Bernd Scholz-Reiter

We introduce a scheduling problem consisting of a batch processorflow shop. Each stage has one batch processor working by the batchavailability model. The problem contains stage specific incompatiblejob families. This induces non-permutation schedules, inconsistentbatches and stage across interdependencies of jobs. Additionally, itincludes unequal processing times, distinct due dates, time periods andbatch size proportional setup times. Batch sizes underlie minimal andmaximal bounds. We present the corresponding ILP, a greedy heuris-tic, a genetic algorithm and experimental results.

� WD-12Wednesday, 14:30-16:00RB-Omicron

Port and Liner Management

Stream: Container Terminal OperationsInvited sessionChair: Haakon Lindstad, Marintek AS, Otto Nielsensvei 10, 7450,Trondheim, Norway, [email protected]

1 - Integrated and robust planning of bulk port opera-tionsNitish Umang, Transport and Mobility Laboratory(Transp-OR), Ecole Polytechnique Federale De Lausanne,Avenue Tir Federale 92, 1024, Ecublens, Vaud, Switzerland,[email protected], Michel Bierlaire

Bulk ports have received considerably less attention than container ter-minals in the field of large scale optimization. We present two exactsolution approaches based on mixed integer programming and gener-alized set partitioning, and a heuristic approach to model the dynamic,hybrid berth allocation problem in bulk ports. We make the model ro-bust to uncertainties and present a dynamic recovery algorithm to solvethe berth allocation problem in real time. We also formulate an MIPto consider the integrated planning of berth and yard space allocation,and implement it using column generation.

2 - Port dynamics - A challenge for network optimizationin liner shippingNguyen Khoi Tran, Logistics System, Institute of ShippingEconomics and Logistics, Universitaetsalle 11-13, 28359,Bremen, Germany, [email protected], Hans-Dietrich Haasis

The presentation aims to review port development as well as optimiza-tion study in liner network.Ports have integrated into global supplychain. Their scope has surpassed traditional sea side and included alsoland side connection.Optimization review concentrates on route opti-mization,fleet operation,container planning with various problems andmethodologies. Naturally, network optimization should be consideredboth sea and land factors. Our simulation illustrates the impact of var-ious logistics costs on total door-to-door cost,in particular inland andinventory carrying costs.

239


3 - Knowledge Management of Macro and Micro SeaportRepositories using Data Mining TechniquesAna Halabi Echeverry, Computing, Macquarie University, 17aGilda St, North Ryde, 2113, Sydney, NSW, Australia,[email protected], Deborah Richards, AyseBilgin, Jairo Montoya-TorresWe propose that repository data can be connected into a systemicframework for decision-making of port authorities (PAs). We proposea systemic model we call the Port Decision System Approach (PDSA).Initially we test it, using cluster analysis to identify patterns with theavailable data. Results help to understand the similarities and differ-ences between subsystems of PDSA. Secondly, we use neural networksas a method to validate the associations by using combinations of in-puts and outputs. Thus, we provide a proof of concept of a strategicdecision making tool for PAs in emerging regions

4 - Multi objective scheduling and environmental rout-ing of maritime vesselsHaakon Lindstad, Marintek AS, Otto Nielsensvei 10, 7450,Trondheim, Norway, [email protected] scheduling and routing which concerns the optimal assignment ofavailable cargoes to maximize profit is a complex problem. Increasedenvironmental concern due to climate change adds another dimensionto the scheduling and makes it multi objective. This study has fo-cused on developing a methodology for calculating fuel consumptionand emissions as function of sea conditions and vessel characteristics.And to use it for scheduling in a multi objective environment wherevoyage requirement will vary between minimizing voyage emissions,minimize voyage cost or maximizing profit.

� WD-13Wednesday, 14:30-16:00RB-Tau

Emerging Issues in New ProductDevelopment

Stream: OR Applications in the Automotive IndustryInvited sessionChair: Thomas Volling, Institute of Automotive Management andIndustrial Production, Technische Universität Braunschweig,Katharinenstr. 3, 38106, Braunschweig, Germany,[email protected]

1 - A Proposal of Diagnosis Procedure for BTO ProductsYuji Sato, Graduate School of Policy Science, Mie ChukyoUniversity, 1846 Kubo, 515-8511, Matsusaka, Mie, Japan,[email protected] objective of this paper is to propose a diagnosis procedure of user’spreference for a BTO product, and to subsequently optimize the spec-ification of the product for a manufacturing company. By combin-ing cost-benefit analysis which evaluates potential products, and AHPwhich quantifies subjective judgment in evaluations, user’s preferenceis diagnosed. The benefit of this approach is to be able to provideboth the decision support and corporate strategy necessary to ensuresuccess in the future. A case study is carried out to demonstrate theapplicability of the proposed approach.

2 - Algorithms for Extending Battery’s lifespan in Elec-tric VehiclesRon Adany, Computer Science, Bar-Ilan University, 52900,Ramat-Gan, Israel, [email protected], Doron Aurbach, SaritKrausIn this paper we propose algorithms to improve the battery’s lifespan.The battery, which is actually a pack of batteries, is designed to be dis-charged in a specific current whereby other currents, higher or lower,have negative effects on its lifespan. The common discharge methodsupplies the current demand using all the batteries in the pack. Sincethe discharge current demands are not constant, there is room for im-provement. We propose a switching algorithm which, for each dis-charge demand, selects a subset of batteries to be discharged as well asthe discharge current for each.

3 - Coordination by contracts in decentralized productdevelopment processes with uncertain developmentresultsKerstin Schmidt, Institute of Automotive Management andIndustrial Production, Technische Universität Braunschweig,Katharinenstraße 3, 38106, Braunschweig, Germany,[email protected], Thomas Volling, Thomas SpenglerProduct development processes in the automotive industry are dis-tributed over various companies. If inappropriate contracts are usedwithin these collaborations, inefficiencies can occur due to existinguncertainties with regard to the development results and differing ob-jectives of the independent companies. To this end, we develop ananalytical optimization model for the representation of decentralizedproduct development processes, taking into account uncertainties andsubstitution possibilities. Based on this model, the coordination abilityof different kinds of contracts is analyzed.

� WD-15Wednesday, 14:30-16:00RB-2101

Theory and algorithms of bilevelprogramming II

Stream: Variational Inequalities and Bi-Level ProblemsInvited sessionChair: Alain B. Zemkoho, Mathematics and Computer Sciences,Technical University Bergakademie Freiberg, Akademiestr. 6,09599, Freiberg, Saxonia, Germany,[email protected]

1 - Solution algorithm for linear bilevel programmingproblemStephan Dempe, Mathematics and Computer Sciences,Technische Universitaet Freiberg, 09596, Freiberg, Germany,[email protected], Shili PengLet S(x) be the solution set mapping of a parametric linear optimiza-tion problem having the parameter in the right-hand side of the linearconstraints. Then, the linear bilevel programming problem consists inminimizing a linear objective function subject to the variable x beingan element of a certain polyhedron and y being an element of S(x). Anoptimal solution can be found at a vertex of the underlying polyhedronand a simplex-like algorithm is formulated which moves along the fea-sible vertices. This algorithm can be globalized using a tunneling idea.

2 - Some results in semivectorial bilevel optimal controlproblemHenri Bonnel, ERIM, University of New Caledonia, B.P. R4,F98851, Noumea, New Caledonia, [email protected],Jacqueline MorganI will present optimality conditions for the bilevel optimal controlproblems where the upper level is a scalar optimal control problemto be solved by the leader, and the lower level is a multiobjective con-vex optimal control problem to be solved by several followers actingin a cooperative way inside the greatest coalition. I will consider theso-called optimistic case as well as the pessimistic one.

3 - On an enumerative algorithm for solving eigenvaluecomplementarity problemsJoaquim Judice, Instituto de Telecomunicações, Polo II, 3000,Coimbra, Coimbra, Portugal, [email protected], LuísFernandes, Hanif Sherali, Maria ForjazIn this talk, we discuss the solution of linear and quadratic eigenvaluecomplementarity problems (EiCPs) using an enumerative method. Thealgorithm solves these EiCPs by finding global minima of special non-linear programs (NLPs). Procedures for computing the interval thatcontains all the eigenvalues are introduced. Necessary and sufficientconditions for stationary points of these NLPs to be solutions of theEiCPs are established. Some computational experience is reported tohighlight the efficiency and efficacy of the enumerative algorithm forsolving the EiCPs.

240


� WD-16Wednesday, 14:30-16:00RB-2103

Recent Advances in Linear Programmingand Related Subjects

Stream: Linear and Conic ProgrammingInvited sessionChair: Tibor Illés, Differential Equations, Budapest University ofTechnology and Economics, Egry J. u. 1., 1111, Budapest,Hungary, [email protected]

1 - Advantages and disadvantages of successive linearprogramming methods for solving non-linear prob-lemsZsolt Csizmadia, Xpress, FICO, FICO House, Starley Way,Birmingham, United Kingdom, [email protected]

For solving general non-linear problems using out of the box solu-tion methods, choosing second order algorithms is often the rule ofthumb entrusting in their general robustness and their ability to solvemany problems without extensive tuning. However, there are sev-eral problem classes and applications where a first order method isbetter suited for efficiently solving the right formulation. The talkwill present examples and applications where an integrated successivelinear programming method outperforms second order approaches bymagnitudes.

2 - The numerical behavior of s-monotone index selec-tion rules for the simplex algorithmAdrienn Nagy, FICO, FICO House, Starley Way,Birmingham, United Kingdom, [email protected], TiborIllés

Several variations of index selection rules for Simplex algorithms likethe Last-In-First-Out or the Most-Often-Selected-Variable rules notonly are theoretically finite, but also provide significant flexibility inchoosing a pivot element. Based on an implementation of the primalSimplex and the Monotone Build-Up (MBU) Simplex method usingthe callable library of FICO Xpress, the practical advantage of thesemethods are evaluated using public benchmark LP test sets. Our resultsalso provide numerical evidence that the MBU-Simplex algorithm is aviable alternative to the Simplex algorithm.

3 - Cutting Plane Methods for Solving Inexact ProblemsFariba Pidani, Department of Mathematics, Mahshahr Branch,islamic Azad University, Mahshahr, Iran, Iran, IslamicRepublic Of, [email protected]

In this paper we study some well-known cases of nonlinear program-ming problems, presenting them as instances of Inexact Linear Pro-gramming. The class of problems considered contains, semidefiniteprogramming, second order cone programming and special cases of in-exact semidefinite programming. Strong duality results for the nonlin-ear problems studied are obtained via Lagrangian duality. Using theseresults we propose some dual algorithms. The proposed algorithms canbe interpreted as cutting plane (also non polyhedral) or discretizationalgorithms.

4 - The s-Monotone Index Selection Rules for Pivot Al-gorithms of Linear ProgrammingTibor Illés, Differential Equations, Budapest University ofTechnology and Economics, Egry J. u. 1., 1111, Budapest,Hungary, [email protected], Zsolt Csizmadia, Adrienn Nagy

We introduce the concept of s-monotone index selection rule for lin-ear programming problems. We show that several known anti-cyclingpivot rules like the minimal index, last-in-first-out and the most-often-selected-variable pivot rules are s-monotone index selection rules. Fur-thermore, we show a possible way to define new s-monotone pivotrules. We prove that several known algorithms like the primal (dual)simplex, MBU-simplex algorithms and criss-cross algorithm with s-monotone pivot rules are finite methods.

� WD-17Wednesday, 14:30-16:00RB-2105

DC programming and DCA 2

Stream: Nonconvex Programming: Local and GlobalApproachesInvited sessionChair: Tao Pham Dinh, INSA Rouen, 76131, Rouen, France,[email protected]

1 - DC Programming via BMI constraintsYi-Shuai Niu, CORIA UMR 6614, French National Centre forScientific Research, Site Universitaire du Madrillet BP 12,76801, SAINT ETIENNE DU ROUVRAY, France,[email protected], Tao Pham DinhWe propose efficient method for DC programming via BMI (BilinearMatrix Inequality) constraints which has important applications in ro-bust control. This is a very difficult NP-hard problem, the difficultylies in the nonconvexity of the objective function and BMI constraints.We reformulate the problem as a DC program with convex constraintsvia penalization techniques and propose DC Algorithm (DCA) for nu-merical solution. DCA requires solving a convex program via LMIconstraints in each iteration. Numerical simulations and comparisonswith PENBMI show good performance of our method.

2 - Combining multiobjective constraint satisfaction andsequential control for solving the sensors manage-ment of the intelligence processAli Khenchaf, Lab-Sticc UMR CNRS 6285, ENSTABretagne, 2 Rue François Verny, 29806, Brest, France,[email protected], Manh Nguyen Duc,Dambreville Frédéric, Christophe Osswald, AbdelamalekToumi, Jean-Christophe CexusThis paper deals with problems of information collection in the intelli-gence process. The intelligence process takes place within hierarchicalmilitary organizations. That leads to bi-level sensor management opti-mization problems, consisting of both a coordination level and sensorscontrol level. We will consider problems derived from actual oper-ational scenarios, combining a multiobjective constraint satisfactionproblem at the coordination level with sequential optimal controls atthe sensors level. The modeling of the problem and solution methodswill be presented and compared.

3 - A new efficient deterministic approach for mixed 0-1 multiobjective programming. Application to shiftscheduling problems.Viet Nga Pham, Mathematics, National Institute for AppliedSciences, Rouen, France, Avenue de l’Université BP8, 76801,Saint-Étienne-du-Rouvray, Haute Normandie, France,[email protected], Hoai An Le Thi, Tao Pham DinhIn this paper, we propose a new deterministic approach for mixed zero-one linear multiobjective programming. It consists in solving only onemixed zero-one linear program by either CPLEX solver or DC algo-rithm (DCA). Computational results applying to shift scheduling prob-lems demonstrate the efficiency of our method with respect to existingalgorithms, especially in the large-scale setting.Keywords: mixed zero-one multiobjective linear optimization, integerprogramming, DC algorithm (DCA), shift scheduling problems.

� WD-18Wednesday, 14:30-16:00RB-2107

Location Analysis and Disruptions

Stream: Network and Discrete LocationInvited sessionChair: Mozart Menezes, MIT-Zaragoza International LogisticsProgram, Zaragoza Logistics Center, 50197, Zaragoza, Zaragoza,Spain, [email protected]

241


Chair: Serigne Gueye, Computer Science, University of Avignon,88212, Avignon, France, [email protected]

1 - An exact cooperative method for the simple plant lo-cation problemPhilippe Michelon, LIA, Université d’Avignon et des Pays deVaucluse, 339 Ch. des Meinajaries, Agroparc BP 91228,84911, Avignon Cedex 9, France,[email protected] this paper, we present a cooperative primal-dual method to solve theSPLP exactly. It consists of a primal process, and effective tabu search,and a dual process, which performs a LaGrange branch-and-boundsearch. Both processes cooperate by exchanging information whichhelps them find the optimal solution. Further contributions includenew techniques for improving the evaluation of the branch-and-boundnodes: decision-variable bound tightening rules applied at each node,and a sub gradient caching strategy to improve the bundle method ap-plied at each node.

2 - Critical Facility Capacitated r-Interdiction LocationProblem: an example of grocery stores with limiteddemandYuran Choi, Systems and Information Engineering, Universityof Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki, 305-8537,Tsukuba, Japan, [email protected], TsutomuSuzukiOur focus is on maximal covering problem which considers two inter-acting players: interdiction and prevention. This study aims to mini-mize the maximum increase of cost caused by disruption so the systemcan retain the highest coverage on the next interdiction step. We spec-ify the facilities as grocery stores with limited demand. The interdic-tion is capacitated according to the size of stores and total interdictionis bounded. With this problem, it is possible to discuss the necessityof participation of governments for preventing critical stores from dis-ruption on a supposition of stagnation.

3 - A Bilinear Programming Formulation for the p-Median Problem with Unreliable Facilities and Dis-ruption Probability ClassesSerigne Gueye, Computer Science, University of Avignon,88212, Avignon, France, [email protected], MozartMenezesThe p-Median problem with unreliable facilities is the problem of lo-cating unreliable facilities on a network that minimize the expectedcost of serving the customers. The facility availability depends on adisruption probability which is independent of its location. We assumethat a finite number of different probabilities exist. The approach cor-responds to aggregating different designs into risk classes. A BilinearProgramming Formulation is proposed to solve exactly the problem.Some numerical results are given using randomly generated instances.

� WD-19Wednesday, 14:30-16:00RB-2111

Survival Analysis and Simulation

Stream: Simulation Methods in FinanceInvited sessionChair: Coskun Kus, Statistics, Selcuk University, 42225, Facultyof Science, Campus, Konya, Turkey, [email protected]: Aysegul Iscanoglu Cekic, Statistics, Selcuk University,KONYA, Turkey, [email protected]

1 - A new lifetime distribution with increasing failure rateIsmail Kınacı, Statistics, Selçuk University, Selçuk University,Science Faculty„ Konya, Turkey, [email protected],Demet SezerIn this study, a new two-parameter lifetime distribution with increas-ing failure rate is introduced. Various properties of this distribution arediscussed. The Fisher information matrix is derived analitically in or-der to obtaining the asymptotic covariance matrix. Simulation studiesare performed and experimental results are illustrated based on a realdata set.

2 - Optimal Warranty Policy for a Pareto DistributedProducts based on Progressive Censored SampleNeriman Karadayı, Statistics Department, Selcuk UniversityFaculty of Science, Selcuk University, 42225, Konya, Turkey,[email protected], Yunus Akdogan, Coskun Kus,Shuo-Jye WuIt is important to attract consumers in highly competitive markets. Oneof the effective ways of doing this is through warranties granted to theproducts. Consumers would like to buy high-priced products so long asthey are supplied a reliability for the products in question. A decisionproblem related with the warranty involving a combination of free re-placement and pro-rata policies is examined. Bayesian approach basedon progressive censored product failure time data having Pareto distri-bution is used to determine the optimal warranty lengths. An numericalexample is also given.

3 - Optimal Progressive Group Censoring Scheme un-der Cost Considerations for Pareto DistributionCoskun Kus, Statistics, Selcuk University, 42225, Faculty ofScience, Campus, Konya, Turkey, [email protected],Yunus Akdogan, Shuo-Jye WuIn this article, optimal design under the restriction of pre-determinedbudget of experiment is developed for the Pareto distribution when thelife test is progressively group censored. We use the maximum likeli-hood method to obtain the point estimator of the Pareto parameter. Wepropose two approaches to decide the number of test units, the numberof inspections, and the length of inspection interval under limited bud-get such that the asymptotic variance of estimator of Pareto parameteris minimum. A numerical example is given to illustrate the proposedmethod. Some sensitivity analysis is also studied.

4 - Generalized Linear Mixed Models (GLMMs) with a Fi-nancial ApplicationNeslihan Iyit, Statistics Department, Selcuk University,Faculty of Science Faculty, Konya, Turkey,[email protected] generalized linear model (GLM) is used to determine the relation-ship between a response variable and fixed effects. A generalized linearmixed model (GLMM) is a statistical model that extends the class ofGLMs by incorporating random effects.

GLMMs are important in the analysis of finance data. When dealingwith the finance data, the distribution of the response variable can bechosen from the exponential family. In this study, GLMM with a fixedand random effect will be implemented on a financial data set.

� WD-20Wednesday, 14:30-16:00RB-2113


Stream: Financial OptimizationInvited sessionChair: Ayse Arik, actuarial science, hacettepe university, BeytepeCampus, Cankaya, Ankara, Turkey, [email protected]

1 - Performance measurement of online algorithms us-ing geometric Brownian motion and quantile regres-sionEsther Mohr, Information and Technology Management,Saarland University, P.O. Box 151150, 66041, Saarbrücken,Germany, [email protected], Robert Dochow, GünterSchmidtWe investigate the problem of estimating the independent risk and re-turn represented by standard deviation and expected return of a knownonline conversion algorithm. The question is whether these param-eters should be estimated by a theoretical price model or by an em-pirical procedure through experiments. To avoid data-snooping biaswe present a hybrid of the theoretical and the empirical approach with3 steps. The problem is formulated as a linear function of parame-ters, and solved by linear programming. Our results generalize formerworks.

242


2 - Optimal Dynamic Tax Evasion: A Portfolio ApproachFrancesco Menoncin, Economics, Brescia University, Via S.Faustino, 74/B, 25122, Brescia, Italy, [email protected],Rosella Levaggi

An agent invests on both a risk-less and a risky asset (following aGBM) and chooses how much of the latter to declare. A fine (Pois-son jump process) must be paid on evasion if caught. Main results are:(i) inter-temporal consumption is higher with evasion, (ii) evasion isaffected neither by the return nor by the volatility of the risky asset,(iii) evasion increases the investment in the risk-less asset, (iv) againstevasion high fine is more effective than frequent controls, (v) for a suf-ficiently high level of taxation on the risk-less asset evasion is zero.

3 - Optimal Cash Management Using Impulse ControlPeter Lakner, IOMS, New York University, 44 West 4th St.,Suite 8-61, 10012, New York, NY, United States,[email protected], Joshua Reed

We solve the optimization problem of a manager who controls theamount of cash on hand by occasionally moving funds back and forthbetween his/her reserves and the cash on hand. The cash on hand pro-cess without any managerial intervention is assumed to follow a Levyprocess. The manager is charged fixed plus proportional costs for anyspecific adding to or withdrawing from the reserves and there are op-portunity costs for keeping too little or too much cash on hand. Themanager’s objective is to minimize the expected total discounted costsover an infinite horizon.

4 - The Lee Carter Method and Poisson Log-BilinearModel: An Application To Turkish Census DataAyse Arik, actuarial science, hacettepe university, BeytepeCampus, Cankaya, Ankara, Turkey,[email protected], Basak Bulut, Erengul Ozkok,Meral Sucu

We present Lee-Carter (LC) and Poisson log-bilinear model, suggestedby Brouhns et al. (2002),for modelling Turkish mortality rates. Af-ter estimating the LC parameters then we re-estimated the parametersunder the Poisson log-bilinear model using bootstrap method. Projec-tions are also made by using the estimated parameters obtained fromthese two methods and are compared with the actual mortality rates.References: Brouhns et al. (2002), A Poisson log-bilinear regressionapproach to the construction of projected life tables. Insr Math Econ31:373-393.

� WD-21Wednesday, 14:30-16:00RB-2115

Capital Asset Pricing Models

Stream: Numerical Methods in FinanceInvited sessionChair: Fabian Lutzenberger, FIM Research Center Finance &Information Management, University of Augsburg, Universitätsstr.12, 86159, Augsburg, Germany,[email protected]

1 - Testing the Capital Asset Pricing Model (CAPM) onthe Central and South-East European Emerging Se-curities MarketsZdravka Aljinovic, Faculty of Economics, University of Split,Matice hrvatske 31, 21000, Split, Croatia,[email protected], Josipa Dzaja

The paper examines if the Capital Asset Pricing Model (CAPM) is ad-equate for capital asset valuation on the Central and South-East Euro-pean emerging securities markets using monthly stock returns for ninecountries for the period of January 2006 to December 2010. Precisely,it is tested if beta, as the systematic risk measure, is valid on observedmarkets by analysing are high expected returns associated with highlevels of risk, i.e. beta. Also, the efficiency of market indices of ob-served countries is examined.

2 - Risk Sensitive Approach to Inventory ManagementMuge Tekin, Industrial Engineering, Koc University,Rumelifeneri Yolu, Koc Universitesi, 34450, Istanbul, – Select–, Turkey, [email protected], Suleyman Ozekici

Newsboy models are one of the core models in inventory managementliterature. The classical newsboy model aims to maximize the expectedprofit and the risk caused by demand uncertainty is disregarded. How-ever, the inventory managers may prefer less gain but more stable out-comes. We follow a mean-variance approach to the newsboy problemand take care of risk. We also aim to incorporate supply randomnessinto our model, as well. In the second part, supposing the randomnessis correlated with financial markets the newsboy invests in a portfolioof financial instruments and hedges the risk.

3 - Valuation — Liquidity (Risk) and Cost of EquityMichael Ludwig, FIM Research Center Finance &Information Management, University of Augsburg, Germany,[email protected], Stefan Stöckl, DennisDiepold

While several empirical studies could find a significant impact of liq-uidity (risk) on asset returns, its effect on valuation has not been in-vestigated yet. Therefore, we examine whether the market value of acompany can be estimated more exactly when cost of equity are usedthat do explicitly account for liquidity (risk) and all other variables(e.g. cash flows) being equal. We find that estimated company val-ues are significantly closer to the actually observed market capitaliza-tion, when the cost of equity are calculated with the liquidity-adjustedCAPM instead of the CAPM.

4 - Multiperiod valuation and the CAPMFabian Lutzenberger, FIM Research Center Finance &Information Management, University of Augsburg,Universitätsstr. 12, 86159, Augsburg, Germany,[email protected], Stefan Stöckl, VaskoIsakovic

This paper compares different applications of the Sharpe-Lintner-Mossin Capital Asset Pricing Model (CAPM) to the valuation of as-sets with stochastic cashflows in more than one periods ahead usingSTOXX Europe 600 data. We show that the cross-section of mar-ket capitalizations can be explained best by taking into account time-varying CAPM-based discount rates with time-varying beta coeffi-cients, market risk premiums and risk free rates. Since discount ratesare usually assumed to be constant over time, we make an importantcontribution to making better multiperiod investment decisions.

� WD-22Wednesday, 14:30-16:00RB-2117

Fuzzy Systems in Geopolitics andDisaster Management

Stream: Fuzzy Optimization - Systems, Networks andApplicationsInvited sessionChair: Erik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39, 85577,Neubiberg, Germany, [email protected]: Alexis Pasichny, Students’ Science Association, NationalTechnical University of Ukraine, 37, Peremohy av., 03056, Kiev,Ukraine, [email protected]

1 - Preparation of Complex Landslide Simulation Re-sults with Spatial Data Mining Methods for DecisionSupportErik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39,85577, Neubiberg, Germany, [email protected], EvaNuhn, Wolfgang Reinhardt, Stefan Pickl

243


Recent catastrophic landslide events caused large human and materialdamages. In order to improve protection against such kinds of naturalhazards a new approach for early warning systems for landslide eventsis pursued where the results of complex finite element simulations oflandslides are coupled with geoinformation systems. This allows adetailed investigation of unstable slopes with sophisticated crisp andfuzzy spatial data mining methods as well as a user friendly prepara-tion of the complex simulation results for decision support.

2 - Fuzzy civilization-state model for system analysis ofworld geopoliticsAlexis Pasichny, Students’ Science Association, NationalTechnical University of Ukraine, 37, Peremohy av., 03056,Kiev, Ukraine, [email protected]

Proposed study is based on development of fuzzy representation forthe global cultures’ distribution across world countries’. Several origi-nal methods for construction of fuzzy world civilization structure usingdifferent types of expert data are proposed. Given model allows to de-fine formal procedures for identification of world civilizations’ corestates and fault lines for testing of S. Huntington’s ’clash of civiliza-tions’ concept.

3 - A concurrent neuro-fuzzy inference system for auto-motive service industryFehmi Evren Dinç, Industrial Engineering, KocaeliUniversity, 41000, Kocaeli, Turkey, [email protected],Yildiz Yulugkural, Zerrin Aladag

Automotive post purchase service network and its quality can be effec-tive on consumers’ purchasing consideration as much as product andservice themselves. In this article, an intelligent model is developedto determine the motor, transmission and power unit problems of spe-cific vehicle model. A Concurrent Neuro-Fuzzy Inference System hasbeen generated with Self-Organized Map (SOM) and Adaptive NeuralFuzzy Inference System (ANFIS) for the detection and estimation ofthe vehicles’ malfunctions. SOM which takes part in model structure,has been positioned as ANFIS’s antecedent input feeder.

� WD-23Wednesday, 14:30-16:00RB-Delta

Recent Advances in Convex Optimization

Stream: Convex OptimizationInvited sessionChair: Aharon Ben-Tal, Industrial Engineering and Mangement,Technion-Israel Institute of Technology, Technion City, 32000,Haifa, Israel, [email protected]: Attila Gilanyi, Faculty of Informatics, University ofDebrecen, Pf. 12, 4010, Debrecen, Hungary, [email protected]

1 - Iterative solution methods for a canonical dc pro-gramming problemSyuuji Yamada, Graduate School of Science and Technology,Niigata University, 8050 IKarashi-2nocho, 9502181, Niigata,Japan, [email protected], Tamaki Tanaka,Tetsuzo Tanino

In this talk, we consider a canonical dc programming problem (CDC)to minimize a linear function over the difference between a compactconvex set and an open bounded convex set. For (CDC), many globaloptimization algorithms have been proposed. However, since the vol-ume of data necessary for executing such algorithms increases in pro-portion to the number of iterations, such algorithms are not effectivefor large scale problems. Hence, to calculate an approximate solutionof a large scale (CDC), we propose new iterative solution methods byutilizing polar coordinates.

2 - Revisiting the construction of gap functions for vari-ational inequalities via conjugate dualityLiana Cioban, Babes-Bolyai University, Cluj Napoca,Romania, Mihail Kogalniceanu, nr. 1, 400084, Cluj Napoca,Romania, [email protected], Ernö Robert Csetnek

Based on conjugate duality we construct gap functions for general vari-ational inequalities formulated by a perturbation function. In convexsetting and if a regularity condition is fulfilled, these become gap func-tions. By particularizing the perturbation function we rediscover sev-eral gap functions from literature. We characterize the solutions of thegeneral variational inequalities by the properties of the convex subdif-ferential. If no regularity condition is fulfilled, we give necessary andsufficient sequential characterizations for the solutions.

3 - Database for Convex OptimizationShriguru Nayak, IT, IIIT-B, IIIT-B, 26/C, Electronic City,560100, Bangalore, Karnataka, India,[email protected], Neha Bhatia, G. N. SrinivasaPrasanna

Convex optimization methods rely on efficient representation of highdimensional convex constraints, both linear and non-linear. Vertexonly and facet only representations of the feasible region are subop-timal in terms of storage. Extending work on 3D polytope represen-tations, we propose a hybrid of vertex and facet representation to ef-ficiently represent convex polytopes and general convex bodies. Effi-cient query algorithms are presented, including membership of a pointin the convex body, its volume, and generation of information equiva-lent convex bodies.

4 - An iterative algorithm for the reflexive opti-mal approximation solutions of matrix equationsAXB+CYD=EHeming Sun, College of Science, Hohai University, 1 Xi KangRoad, 210098, Nan Jing, China, [email protected]

In this paper, we consider the following 3-steps optimization problem.First step, define matrices X and Y as reflexive (anti-reflexive) ma-trices. Second step, minimize the Frobenius norm of AXB+CYD-E.Third step, find the nearest matrices to two given matrices among thesolutions of second step. General optimization problem is two-stepsoptimization problem. We present an iterative algorithm to solve theabove 3-steps optimization problem by making use of the hybrid steep-est descent method (HSDM).

� WD-24Wednesday, 14:30-16:00CC-A11

OR and the ArtsStream: OR and the ArtsInvited sessionChair: Vitaly Podobedov, Computational Mathematics andCybernetics, Moscow State University, Vorobievy Gory, MSU,119992, Moscow, Russian Federation, [email protected]

1 - OR and Abstract PaintingVitaly Podobedov, Computational Mathematics andCybernetics, Moscow State University, Vorobievy Gory,MSU, 119992, Moscow, Russian Federation, [email protected]

Local optimization algorithms and deterministic chaos are the maincreative subjects of a new style of mathematical art — the abstractdeterminism painting. Compared to fractal art, abstract determinismexpands the creative possibilities of chaos by obscuring a mathemati-cal nature of its paintings; even a discerning spectator can hardly saywho made it — a computer or an artist-person? Software for generat-ing the test functions and the domains of attraction of local optima arepresented. Numerous illustrations, allowing the aesthetic and semanticinterpretations are provided.

2 - Optimization of inner operation of a museum to en-hance the educational impact on visitorsKatarina Simoncicova, Institute of History and Theory ofArchitecture and Monument Restoration, Slovak university oftechnology, Faculty of architecture, Slovakia,[email protected], Eva Kralova

244


This paper develops the process of creation of interior operation, whichis important part of securing the fluency of education. It will be showedon example of conversion of a former heat plant to a poly-functionalmuseum of technology that will mediate the history of industry. Theethic approach is in the sensitization of the relations between visitorsand architectural heritage and education of ecology and sustainability.The aim of this project is to evaluate efficient interior operation andnecessity of changing perceptions throughout the visitor’s path.

3 - Composing counterpoint musical scores with vari-able neighborhood searchDorien Herremans, Operations Research Group ANT/OR,University of Antwerp, Dolfijnstraat 82, 2018, Antwerpen,Belgium, [email protected], Kenneth Sörensen

In this research a variable neighborhood search (VNS) algorithm is de-veloped that can generate musical fragments of arbitrary length. Theobjective function of the algorithm is based on a quantification of ex-isting fifth species counterpoint rules. A thorough parametric analysisof the VNS reveals the significance of the algorithm’s parameters onthe quality of the composed fragment, as well as their optimal settings.The VNS algorithm has been implemented as a user-friendly softwareenvironment for composition, called Optimuse.

4 - Ball motion inside symmetric surfaces, described byusing special diagramsLina Otradnova, Department of Mechanics and Mathematics,Lomonosov MSU, GSP-1, Leninskie Gory, 119991, Moscow,Russian Federation, [email protected]

Problems of ball motion between two parallel planes, inside a sphereor a circular cylinder, and also disk motion between two parallel linesand inside the angle, by inertia in the absence of gravity are considered.During the impact, an instant constraint is imposed, and a condition ofrolling without slipping is satisfied. Two cases of impacts are given:absolutely elastic and not absolutely elastic. Diagrams illustrating thiseffect are presented. Such model with impacts can be found in every-day life, namely in sport: pole vault, or table tennis.


Simulation Based Decision Support

Stream: Simulation Based Decision SupportInvited sessionChair: Guido Siestrup, Business Information Systems, HochschuleFurtwangen University, Robert-Gerwig-Platz 1, 78120,Furtwangen, Germany, [email protected]

1 - Simulation approach for measuring benefits of cut-ting stock optimization with usable leftoversMihael Cesar, Business Informatics, Faculty of Economics,Kardeljeva pl. 17, 1000, Ljubljana, [email protected], LukaTomat, Miro Gradisar

We present a method for one-dimensional cutting stock optimizationwith usable leftovers (UL). All leftovers longer than t are returned backto stock and reused in future orders. The question is how to set t. Lowert produces more UL in stock and therefore higher trim-loss in futureorders. If it is too high, too many leftovers become trim-loss. Withhundreds of simulations we found the optimal number of leftovers instock for various stock-to-order size ratios. The proposed method min-imizes trim-loss and at the same time maintains optimal number of ULin stock.

2 - A risk management approach for logistics agglomer-ationsGuido Siestrup, Business Information Systems, HochschuleFurtwangen University, Robert-Gerwig-Platz 1, 78120,Furtwangen, Germany, [email protected], ClaudiaBreuer, Hendrik Wildebrand, Hans-Dietrich Haasis

Logistics agglomerations play an important role within today’s logis-tics networks. At the same time, they represent critical infrastructures.This means that a serious disturbance may lead to severe consequences.Consequently, there is a need for short-term decision support to iden-tify possibilities of compensation for e.g. capacity failure in case ofdisruption. Methodically, we focus on decision matrices. These areintegrated in a multi-agent-based distributed simulation environment.The empirical background is based on two freight villages representinga type of logistics agglomerations.


Multi-Criteria Decision Making andApplications 2

Stream: Multi-Criteria Decision Making and Environ-mental ManagementInvited sessionChair: King-Wah Pang, Department of Logistics and MaritimeStudies, The Hong Kong Polytechnic University, Hung Hom,Kowloon, Hong Kong, Hong Kong, [email protected]: Mikhail Kuznetsov, Moscow Institute of Physics andTechnology, Russian Federation, [email protected]

1 - Multi-criteria Internet Shopping Optimization Prob-lemJedrzej Musial, Computer Science and CommunicationsResearch Unit, University of Luxembourg, Luxembourg,[email protected], Jacek Blazewicz, Pascal Bouvry

Internet Shopping Optimization Problem answers the question of howto manage a multiple-item shopping list over several shopping loca-tions. ISOP problem is also very interesting from the practical pointof view since it concerns shopping optimization for all the users of theInternet.

Multi-criteria ISOP introduces a model within a wider class of prob-lems including additional variables and characteristics (e.g. expirationdates, product priorities). Computational experiments with greedy al-gorithms were conducted to check their applicability.

2 - A hybrid decision making model for labor flexibilityErdem Aksakal, Industrial Engineering, Gazi University, GaziUniversity Faculty of Engineering, Maltepe, 06570, Ankara,Turkey, [email protected], Metin Dagdeviren, IhsanYüksel

Globalization forces firms to consider the flexibility in human re-sources. Labor flexibility plays an important role in organizations inall parts. As a result of including the labor requirements in market re-quirements, labor flexibility and its factors become a considered costvalue to the firms. Without labor flexibility, firms cannot take efficientand competent decisions. Under labor flexibility, 4 criteria will be ex-amined and tried to find out that which one is the most affecting onlabor flexibility.

3 - A knowledge based decision support system forwarehousing management to improve operating ef-ficiencyKing-Wah Pang, Department of Logistics and MaritimeStudies, The Hong Kong Polytechnic University, Hung Hom,Kowloon, Hong Kong, Hong Kong, [email protected]

Data mining approaches have been applied in information extractionfor applications such as customers relationship management. In retail-ing industry, this technique is used to extract consumers buying behav-ior when customers frequently order similar products together, thus itis beneficial to store these correlated products nearby to reduce the or-der picking operating time and cost. We present a knowledge baseddecision support system for the storage location assignment that min-imizes the total travel distances for both put-away and order pickingoperations.

245



Applications in Decision Making &Decision Analysis

Stream: Decision Support SystemsInvited sessionChair: Pascale Zaraté, Institut de Recherche en Informatique deToulouse, Toulouse University, 118 route de NarBonne, 31062,Toulouse, France, [email protected]: Fatima Dargam, SimTech Simulation Technology, RiesStrasse 120, 8010, Graz, Austria, [email protected]

1 - A decision making framework for a Colombian busi-ness groupJulián Benavides, Finance, Universidad Icesi, Cra 89 10-80,760032, Cali, Valle, Colombia, [email protected], FelipeHenaoWe develop a 7-steps framework to facilitate decision-making. Itemerges from an ongoing project where different sources of data weregathered in order to identify the group’s needs in decision-making. Itcombines approaches of strategic management, Soft OR and decisionanalysis. Activities such as problem formulation, definition of goalsand SWOT add to define relevant criteria. Alternatives are ranked us-ing scenario planning. 3 major corporate decisions have been success-fully undertaken: Portfolio prioritization, partner selection for a newventure, and disposal of a failed initiative.

2 - Development of a territorial vulnerability indicatorframework for cross-border risk analysisTina Comes, Institute for Industrial Production (IIP),Karlsruhe Institute of Technology (KIT), Hertzstr. 16, 76187,Karlsruhe, Germany, [email protected], MarjorieVannieuwenhuyse, Frank SchultmannTo reduce vulnerability against natural disasters, decision makers mustimplement adequate risk management strategies. Since risks ignorenational borders, the need for a transnational risk management arises:cooperation in cross-border areas must be reinforced, resources shouldbe cumulated. By identifying key drivers and weaknesses, vulnera-bility analysis is the first step to an efficient risk management. Wepresent an indicator framework for regional vulnerability analysis atcross-border scale and show how it is used as the basis for cross-borderrisk management.

3 - Requirements analysis and conceptual design ofdata warehouseDragana Becejski-Vujaklija, Faculty of OrganizationalSciences, University of Belgrade, Jove Ilica 154, 11000,Belgrade, Serbia, [email protected], Milena PanovicThe subject of paper is requirement analysis and conceptual design ofdata warehouses and proposal of general method for requirement anal-ysis and conceptual design, which should be a methodological frame-work in implementing these phases within the overall data warehousedesign. The general method for requirement analysis and conceptualdesign that combines current approaches is presents. Based on the re-search, the general steps that should be applied to specific projects inorder to ensure a higher level of longevity as well as acceptance of theusers are identified and described.


Seismic Modeling with OR

Stream: Recent Advances in Earthquake StudiesInvited sessionChair: Aysegul Askan, Civil Engineering, Middle East TechnicalUniversity, Inonu Bulvari, 06531, Ankara, Turkey,[email protected]

Chair: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - Ensemble methods for classification of volcano -seismic signalsCindy Mora-Stock, Institut für Geowissenschaften,Christian-Albrechts-Universität, Otto-Hahn-Platz 1, 24118,Kiel, Schleswig-Holstein, Germany,[email protected], Cristian Bravo

Volcanoes present different types of seismic activity depending onthe source origin, such as tremors, long-period, and volcano-tectonicevents. In Chile, Villarrica and Llaima are two of the most active volca-noes, constantly presenting seismicity that can be classified into thesedifferent types. To obtain a faster and reliable classification of futureactivity, pattern recognition ensemble methods were constructed usingneural networks and support vector machines. The usefulness of theprocedure applied to a real-life data set with 18 months of data fromthese volcanoes is presented.

2 - An integrated model for disaster mitigation and re-sponse decisionsAlper Döyen, Industrial Engineering, Bogazici University,Department of Industrial Engineering, Bogazici University,Istanbul, Turkey, [email protected], Necati Aras, GülayBarbarosoglu

We propose a two-stage stochastic integer programming model for pre-disaster planning of mitigation and response decisions in an integratedmanner. The mitigation decisions involve both building and transporta-tion infrastructure retrofitting. The objective is to minimize the totalcost of retrofitting, transportation and shortage of relief item demand.The deterministic equivalent of the model is formulated as an MILPand solved by Lagrangean heuristics. Test results show that the pro-posed heuristics show good performance for a wide range of probleminstances.

3 - A data mining approach for modeling high-frequencyspectral decay of ground motions for NorthwesternTurkeyFatma Nurten Sisman, Civil Engineering, Middle EastTechnical University, METU Üniversiteler Mah. DumlupınarBlv. No:1,06800 Çankaya, Ankara, Turkey,[email protected], Onur Pekcan, Aysegul Askan

In this study, we use data mining methods on 114 strong ground mo-tion records with different magnitudes, source-to-site distances, andsite classes. We use both traditional regression techniques and datamining approaches to describe the relationships between kappa valuesand the aforementioned independent variables. We then compare theoutcomes of data mining techniques with those of traditional methodsto better highlight the important characteristics of the ground motiondataset. We express a regional Kappa model for Northwestern Turkey.

4 - Waveform tomography with numerical optimizationtechniquesAysegul Askan, Civil Engineering, Middle East TechnicalUniversity, Inonu Bulvari, 06531, Ankara, Turkey,[email protected], Volkan Akcelik, Jacobo Bielak, OmarGhattas

In this study, we present a recent waveform inversion method, basedon least-squares optimization for solving the nonlinear full waveforminverse problem of determining the crustal velocity and intrinsic atten-uation properties of sedimentary valleys in seismically active regions.The inverse problem of retrieving the crustal velocity is formulated as aconstrained optimization problem. We present the method with inver-sions for sedimentary models of the San Fernando Valley in California,under SH-wave propagation. Finally, we discuss the sensitivity of themethod to various model parameters.

246



Optimization and Data Mining (II)

Stream: Optimization and Data MiningInvited sessionChair: Minyoung Cho, Korea University, Korea, Republic Of,[email protected]

1 - The application of neural networks and linear pro-gramming discriminant models on the file fragmentclassification problemErich Wilgenbus, School of Computer, Statistical andMathematical Sciences, North West University PotchfestroomCampus, Internal box 107, Private Bag X6001, North-WestUniversity Potchefstroom Campus, 2520, Potchefstroom,South Africa, [email protected], Hennie Kruger, Tiny DuToitThe file fragment classification problem is of importance in networkand computer forensics. This classification problem requires predic-tion of the correct file type to which a given file fragment belongs.Several approaches have been proposed in literature. This paper com-pares the use of neural networks and linear programming discriminantmodels to solve this binary classification problem. This multi-classproblem is solved using a sequence of binary neural network and lin-ear programming classification models. The results of empirical ex-perimentation will be presented.

2 - Clustering anthropometric data addressed to theclothing industryTeresa Leon, Statistics and OR, Universitat de Valencia, Dr.Moliner 50, 46100, Burjassot, Spain, [email protected],Guillermo Vinue, Guillermo Ayala, Sandra Alemany, JuanDomingoA 3D anthropometric survey of the female population was conductedin Spain by the Spanish Ministry of Health. Its aim was to generateanthropometric data addressed to the clothing industry. A sample of10,415 Spanish females from 12 to 70 years old randomly selectedwas measured using a 3D body scanner and 95 anthropometric mea-sures were obtained. We explore this dataset using cluster analysistechniques. We use a dissimilarity measure based on McCulloch etal’s. We modify the original measure, which is merely based on thesum of squared discrepancies over each individual feature, by takinginto account the user’s opinion, using an OWA operator.

3 - Fault Classification using Data Mining Technique forSemiconductor Process MonitoringMinyoung Cho, Korea University, Korea, Republic Of,[email protected], Jonghyuck Park, Jun Seok Kim,Sung-Shick Kim, Jun-Geol BaekProcess control is essential to operating in the semiconductor processefficiently. This paper considers fault classification of semiconductorcyclic signal for process control and monitoring. In general, processsignal usually take the different pattern depending on different causesof fault. Fault classification by causes of faults could improve the pro-cess control through a rapid diagnosis. This paper proposes the methodthat can classify fault causes as each one class. Data-Mining Techniquesuch as Neural-network, SVM, Average distance, k-NNDD, HotellingT2 are used to perform the method.


OR and Ethics IIStream: OR and EthicsInvited sessionChair: Cristobal Miralles, Depto. Organización de Empresas,Universidad Politecnica de Valencia, Cami de Vera s/n, 46022,Valencia, Spain, [email protected]

1 - Identifying the upper age limit to extend screeningfor breast cancer in england and walesRachid Rafia, HEDS, University of Sheffield, ScHARR,Regent court, 30 Regent street, S14DA, Sheffield, UnitedKingdom, [email protected], Alan Brennan, JasonMadan, Lynda Wyld, Karen Collins

A mathematical model that allows the impact of screening policies onbreast cancer diagnosis and subsequent management to be assessedwas constructed to evaluate the optimal upper age for a breast cancerscreening policy. The model has two parts - a natural history model ofthe progression of breast cancer up to discovery, and a post-diagnosismodel of treatment, recurrence and survival. This study suggests thatan extension of the current NHSBSP upper age limit for invitationsfrom 69 to 78 would represent a cost-effective use of NHS resources.

2 - Multi-criteria decision making with criteria values es-timated from survival data: a case study in heart fail-ureJing Zhao, Department of Epidemiology, University MedicalCenter Groningen, PO Box 30.001, 9700 RB, Groningen,Netherlands, [email protected], Douwe Postmus, HansHillege

Clinical decision making is often based on assessing the patient’s abso-lute risk of experiencing a disease event. When more than one event isconsidered, the problem is complicated by dependencies between dif-ferent events and criteria value trade-offs. The stochastic multi-criteriaacceptability analysis method enables to take into account the inter-criteria dependencies and identify the value trade-offs through MonteCarlo simulation. In this presentation, we show how a multi-criteriadecision model can be constructed with the measurements estimatedfrom a survival analysis.

3 - Complexity of decision making with multiple stake-holders: a case study.Cristobal Miralles, Depto. Organización de Empresas,Universidad Politecnica de Valencia, Cami de Vera s/n,46022, Valencia, Spain, [email protected], AlyssonCosta, Maria Antónia Carravilla

Complexity theory has been used extensively in the field of strategicmanagement to understand how organizations or firms adapt to theirenvironments. Complementary, the CSR paradigm brings a wider per-spective of the decision making process where different stakeholdersshould be considered. Starting from a typical optimization problem,this case study aggregates pieces of information that consecutively in-crease the complexity of decision making; analyzing carefully thosevariables that influence the decision and the stakeholders and values inconflict. Thanks to: CORSARI MAGIC DPI2010-18243.



Stream: DEA and Performance Measurement (con-tributed)Contributed sessionChair: Hanns de la Fuente, Facultad de Ciencias Económicas yAdministrativas, Pontificia Universidad Católica de Valparaíso,Avenida Brasil 2950, 6 Piso, 2340025, Valparaíso, Valparaíso,Chile, [email protected]

1 - Measuring the efficiency and productivity change inGlobal Telecom OperatorsChrysovaladis Prachalias, Economics Department, AthensUniversity, Proairesiou 8, 11636, Athens, Greece,[email protected], Rajiv Banker

247


The Liberalization of Telecommunications Markets in combinationwith the globalization of economies created new operational condi-tions of former state owned telecommunications operators, challengesand enterprising occasions. The aim of this paper is to assess the ef-ficiency and productivity change in Global Telecom Operators aftertheir transformation. The input oriented Data Envelopment Analysis isapplied to the data of former public telecommunication operators forthe period of 2002-2007.

2 - Returns to Scale and Incentive Regulation in BrazilElectricity Distribution IndustryAna Lopes, CEPEAD, Federal University of Minas Gerais -UFMG, Brazil, av. Antonio Carlos, 6627, Face Sala 4042,31270901, Belo Horizonte, Minas Gerais, Brazil,[email protected], Rajiv Banker

We address the empirical question of maintaining returns to scale as-sumption when using DEA models in incentive regulation. Althoughthe complete specification of a production relationship between all in-puts, outputs and environmental variables may exhibit NDRS or CRSfor a local monopoly, we prove that the estimated production func-tion will necessarily exhibit VRS with a region of decreasing returnsin the simpler empirical abstraction. Data from 28 Brazilian electricitydistribution companies also statistically rejects the NDRS assumptionproposed by the regulator.

3 - Performance measurement in the marketing produc-tivity in services. An application to the U.S. financialsector using stochastic frontier modelsHanns de la Fuente, Facultad de Ciencias Económicas yAdministrativas, Pontificia Universidad Católica deValparaíso, Avenida Brasil 2950, 6 Piso, 2340025, Valparaíso,Valparaíso, Chile, [email protected]

The literature on marketing productivity shows the absence of mea-surements, especially in the case of quantitative measures and in themeasurement of the influence of marketing assets. This paper appliesthe Theoretical Model of Marketing Productivity (MTPM) in the caseof U.S. financial sector. The results show decreasing technical effi-ciency levels, and demonstrate the positive influence of total assets,administrative staff and sales growth on the productivity of marketingand a negative influence of investments in marketing and labor costson the productivity of marketing.

4 - Efficiency evaluation and analysis of Third Party Lo-gistics in BrazilMariana Almeida, Production engennig, Escola de Engenhariade Sao carlos, Avenida Trabalhador Sãocarlense, 400 -Centro, 13566-590, Sao Carlos, Sao Paulo, Brazil,[email protected], Luis Oliveira

This paper deals with the efficiency evaluation and analysis of 44 ThirdParty Logistics working in Brazil, using Data Envelopment Analysis— DEA for this purpose. The DEA model used considers variablereturns to scale, since the analyzed companies have diversified sizes,and oriented to maximize their outputs. The use of variable selectiontechniques was critical to reach a subset of variables with greater rep-resentativeness for the system. As the main practical result, it was pos-sible to identify the best players, regarding the efficiency of productionprocesses and the returns to scale.


Optimization Modeling II

Stream: Software for OR/MSInvited sessionChair: Bjarni Kristjansson, Maximal Software, Ltd., BoundaryHouse, Boston Road, W7 2QE, London, United Kingdom,[email protected]: Robert Fourer, Industrial Engineering and ManagementSciences, 2145 Sheridan Road, 60208-3119, Evanston, IL, UnitedStates, [email protected]

1 - Alternatives for Scripting in Conjunction with an Al-gebraic Modeling Language for OptimizationRobert Fourer, AMPL Optimization, 2521 Asbury Avenue,60201-2308, Evanston, IL, United States, [email protected]

Optimization modeling languages are fundamentally declarative, yetsuccessful languages also offer ways to write scripts or programs.What can scripting in a modeling language offer in comparison tomodeling in a general-purpose scripting language? Some answers willbe suggested through diverse examples in which the AMPL modelinglanguage is applied to parametric analysis, solution generation, heuris-tic optimization, pattern enumeration, and decomposition. Concludingcomments will touch on the complexity of scripts seen in practical ap-plications, and on prospects for further improvements.

2 - CMPL - Coliop|Coin Mathematical Programming Lan-guageMike Steglich, University of Applied Sciences Wildau,Germany, [email protected]

CMPL is a mathematical programming language and a system formathematical programming and optimization of linear optimizationproblems. CMPL is a COIN-OR project initiated by the TechnicalUniversity of Applied Sciences Wildau and the Institute for Opera-tions Research and Business Management at the Martin Luther Uni-versity Halle-Wittenberg. This talk will begin with an overview aboutthe language, followed by the description of some advanced modellingtechniques. Finally, we will present the interaction between CMPL andselected solvers.


Recent Advances in StatisticsStream: Computational StatisticsInvited sessionChair: Pakize Taylan, Mathematics, Dicle University, 21280,Diyarbakır, Turkey, [email protected]: Elcin Kartal Koc, Statistics, Middle East TechnicalUniversity, Department of Statistics, No:234, 06800, Ankara,Turkey, [email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - A simulation study to compare conditional and nor-malized residuals in diagnostic checkingMehmet Güray Ünsal, Statistics, Gazi University, GaziUniversity Science Faculty, Statistics Department, 06100,Ankara, Turkey, [email protected], Resat Kasap

In this study, a simulation study has been done for comparing theconditional and normalized residuals in diagnostic checking in Box-Jenkins Modelling Process. Firstly, it was given basic informationabout different types of residuals then a simulation study has beengiven to show different decision cases when the conditional and nor-malized residuals were considered in diagnostic chencking.

2 - Statistical Analysis of Narrative SequencesFatma Selcen Phelps, ABEMIS, Westminster College, 501Westminster Ave, 65251, Fulton, MO, United States,[email protected], William Guinee, Eugene Johnsen

This cross-disciplinary study extends the benefits of analytics to therealm of research in the social sciences, specifically the detection ofan underlying "narrative grammar’ in Alcoholics Anonymous meet-ings. Binomial distributions, and distributions estimated by simulatinga large number of meetings with the same parameters as those thatwere observed, allow us to test the hypothesis that there is an under-lying narrative sequence in AA speeches, which allows the audienceto vicariously experience giving up alcohol repeatedly, as a turbulentstory with a satisfactory resolution.

248



Technological change and productivity:permanent features of population health

Stream: Occupational Heath & WellbeingInvited sessionChair: M. Harvey Brenner, School of Public Health, Departmentof Behavioral and Community Health, University of North TexasHealth Science Center, 3500 Camp Bowie Blvd., EAD-711A,76107, Fort Worth, TX, United States, [email protected]: Elena Andreeva, Berlin University of Technology,ACK-185, Ackerstr. 76, 13355, Berlin, Germany,[email protected]

1 - Large-scale organisational downsizing and workers’health during the global crisis of the last decadeElena Andreeva, Berlin University of Technology, ACK-185,Ackerstr. 76, 13355, Berlin, Germany, [email protected] the recent global economic crisis, enterprises have introduced avariety of strategies for managing the "surplus’ employees. This paperaddresses the issues of identifying and understanding the consequencesof such strategies, first of all, with respect to health responses of work-ers affected by the large scale downsizing. We consider the currentinformation needs in occupational health research and discuss factorswhich can potentially modify the patterns of health responses in theEuropean context.

2 - Macroeconomic influences on life expectancy in ad-vanced industrialized countriesM. Harvey Brenner, School of Public Health, Department ofBehavioral and Community Health, University of North TexasHealth Science Center, 3500 Camp Bowie Blvd., EAD-711A,76107, Fort Worth, TX, United States,[email protected] policy is a crucial component in the health of industri-alized countries. Key findings are: GDP per capita in purchasing powerparity is the most important predictor of increased life expectancy inindustrialized countries. The damaging effect of income inequality canbe seen only in age groups under 45. The replacement of manufactur-ing employment by services employment is the second most importantfactor beneficially influencing life expectancy. Decreased labor forceparticipation is a significant source of increased mortality in middleage.

3 - Health policy amidst financial austerity — the case ofHungaryPéter Mihályi, University of Pannonia (Veszprém) andInstitute of Economics, Hungarian Academy of Sciences,8200, Budapest, Hungary, [email protected] early 2009, WHO advised governments not to cut health expendi-tures during the global financial crisis. When the present Hungariangovernment took office in mid-2010, its programme didn’t envisageany major change. However, in the context of the Bismarckian socialinsurance reform, the policies aimed at cutting pension expendituresat the end resulted in totally unexpected cuts in health expenditures aswell as in drastic health policy shifts. The paper will describe in de-tails how the model implanted in 1992 was replaced by the NHS modelwithin a few months without any policy discussion.

� WD-37Wednesday, 14:30-16:00CC-Act

Operations Research in Health Care II

Stream: Health Care ManagementInvited sessionChair: Sally Brailsford, University of Southampton, School ofManagement, SO17 1BJ, Southampton, United Kingdom,[email protected]

1 - Combining Discrete Event, Agent Based and SystemDynamics Simulation to explore the Health and So-cial Care implications of Age Related Macular Degen-erationJoe Viana, School of Social Science, University ofSouthampton, University Road, Highfield, SO17 1BJ,Southampton, Hampshire, United Kingdom,[email protected], Andrew Amos Channon, SallyBrailsford

Age Related Macular degeneration (ARMD) causes sight loss, a newtreatment requiring monthly injections has increased the demand onOphthalmology departments (OD), which is compounded by the UK’sageing population. A Discrete Event Simulation (DES) model of anOD has been produced to evaluate performance. An Agent Basedmodel has been combined with the DES model, representing the needand receipt of social care of individuals. An agent’s sight is capturedby a System Dynamics (SD) model of each eye. The purpose of themodels is to explore how social and health care interact.

2 - Modelling the dental workforce in Sri LankaSally Brailsford, University of Southampton, School ofManagement, SO17 1BJ, Southampton, United Kingdom,[email protected], Dileep de Silva

This talk presents a system dynamics simulation model of the sup-ply and demand for dental care in Sri Lanka. The model representsthe flows of dentists through the complex recruitment and career pro-gression phases through to retirement. This is augmented by a de-mand model, which combines empirical data on the time required tocarry out various dental procedures with secondary clinical and demo-graphic data to project the future incidence of different dental condi-tions. The overall supply-demand model enables policy-makers to testvarious policies for staffing, training and recruitment.

3 - Managing limited bed capacity at hospitalsBurhaneddin Sandikci, Booth School of Business, Universityof Chicago, 5807 South Woodlawn Avenue, 60637, Chicago,IL, United States, [email protected], Tom Best, DonEisenstein, David Meltzer

To address adverse effects of limited inpatient bed capacity, an aca-demic medical center received government dispensation to partition itsinpatient beds into mini-hospitals or wings. Each wing has a specificdesignation of the types of patients it can admit, and the number ofbeds it is allocated. A patient requesting hospital services can be ad-mitted only if a bed is available in the appropriate wing. We present amodel to help hospital administrators make wing formation decisionsand a novel approach to solve the hard problem of forming wings alongwith a rich set of numerical results.

� WD-38Wednesday, 14:30-16:00HH-Colombus

Multiobjective Optimization and WaterManagement

Stream: Multiobjective OptimizationInvited sessionChair: Javier Cano, Rey Juan Carlos University, Spain,[email protected]: Angel Udias, Rey Juan Carlos University, Spain,[email protected]

1 - Pareto Data Mining based decision for Global WaterRestoration Decision ProcessAngel Udias, Rey Juan Carlos University, Spain,[email protected], Andrés Redchuk, Javier Cano,Lorenzo [email protected]

249


Sustainability of water resources has become a challenging problemworldwide. We present a hydroinformatics management tool designedto find an optimal Program of Measures (PoMs) to achieve WaterFramework Directive objectives. It combines 1) a water quality model(to simulate effects of PoMs) 2) a Multi-Objective Evolutionary Al-gorithm to identify efficient trade-offs between PoMs’ costs and waterquality 3) data-mining of the Pareto optimal set to extract knowledgefrom optimal decisions in a usable form. We have applied our method-ology in inner Catalan watersheds with promising results.

2 - A decision support system for the multiobjectivemanagement of the Kwanza riverKiombo Jean Marie, Universidad Rey Juan Carlos, 28943,madrid, Spain, [email protected], Javier Cano,David Rios-Insua

Water management has become a challenging problem worldwide, es-pecially in developing countries, due to the growing scarcity of naturalwater resources; a demographic explosion in many urban settlements;and the inherent increase in the demand of energy and natural freshwater. We provide a model for the multiobjective management of theKwanza River, taking into account uncertainty in various involved pro-cesses, and the need to plan over a long period of time, and the presenceof several conflicting interests. We describe also a decision supportsystem implementing our model.

3 - Role of carbon capture technologies in the Spanishindustry in 2030 under a CO2 reduction scenario us-ing the TIMES-Spain energy optimisation modelDiego García, Energy System Analysis Unit, CIEMAT, Av.Complutense 40, E28040, Madrid, Madrid, Spain,[email protected], Helena Cabal, Machteld Van DenBroek, Yolanda Lechón, Antonio Alonso-Ayuso

In this work, a techno-economic assessment of CO2 capture technolo-gies for relevant industrial sectors (cement, steel, and oil refineries)is presented. Main capture techniques are described, such as pre-combustion, oxyfuel combustion, and postcombustion. Introducingthese technologies in the TIMES-Spain energy optimization model re-sults show that, under a CO2-reduction scenario (20% by 2020 and30% by 2030 below 1990 levels), emissions in the cement sector de-crease 73% comparing to the BaU scenario at the end of the period.Nevertheless, there is no CO2 capture in the other sectors.

� WD-39Wednesday, 14:30-16:00HH-Cousteau

Graph and Digraph Searching Games

Stream: Graph SearchingInvited sessionChair: Nancy Clarke, Mathematics and Statistics, AcadiaUniversity, 12 University Avenue, B4P2R6, Wolfville, NovaScotia, Canada, [email protected]

1 - On Constructive Characterization of Forbidden Mi-nors for k-Searchable GraphsOznur Yasar Diner, Informations Technology, Kadir HasUniversity, Cibali Kampusu, Balat, 34083, Istanbul, Turkey,[email protected]

Edge searching is a pursuit evasion game played on graphs. The aimis to find a search strategy which ensures that all edges of the graphare secure. The least number of searchers needed is the edge searchnumber of the graph, denoted s(G). A major problem is to character-ize the k-searchable graphs, those graphs G such that s(G) is at mostk. We can characterize the k-searchable graphs by giving the completelist of forbidden minors. Such sets are only known when k is less thanfour. In this talk we discuss how clique sum operations give rise toconstructively obtain some forbidden minors.

2 - LIFO-search on digraphsPaul Hunter, Department of Computer Science, University ofOxford, Wolfson Building, Parks Road, OX1 3QD, Oxford,United Kingdom, [email protected] searching games on digraphs are notoriously badly behaved.Different notions of connectivity lead to different parameters, mono-tonicity is rare and tight min-max theorems are even rarer. We presenta searching game that characterizes the cycle-rank of a digraph - ameasure that has been around for several decades - and show thatall standard variations of the game (invisible/visible, monotone/non-monotone, strongly-connected/directed paths) coincide, yielding anexact min-max theorem for cycle-rank.

3 - Eternal Domination- being ever watchfulStephen Finbow, St Francis Xavier University, Canada,[email protected] a network under constant surveillance thru a noisy channel.Searchers are stationed throughout the network so that they may bequickly dispersed to investigate any suspicious activity. To investigatea searcher is presumably moved out of position and hence the othersearchers may also have to move to ensure any additional activity onthe network can be inspected. When the searchers are able to respondindefinitely, the network is said to be eternally dominated. We willfocus on recent results and conjectures regarding grid graphs.

� WD-40Wednesday, 14:30-16:00HH-Livingstone

Modelling sustainable productionsystems

Stream: SD Modelling in Sustainable DevelopmentInvited sessionChair: Francois Guerrin, UPR Recyclage & Risque, Inra & Cirad,Station de la Bretagne - BP 20, 97408, Saint-Denis, France,[email protected]

1 - Modeling and prioritizing requirements for designingsustainable supply chainsAnjali Awasthi, CIISE, Concordia University, H3G 1MB,Montreal, Quebec, Canada, [email protected],ZhongHua ZhangWe present a technique called Sustainable Function Deployment de-veloped on the concept of Quality Function Deployment to model cus-tomer and technical requirements, establish relationship between themand prioritize them for developing sustainable supply chains. The cus-tomer requirements for sustainable supply chains are obtained throughVoice of the Customer surveys and the technical requirements usingsystematic literature review. In addition, priority matrix and cause-and-effect diagram techniques are applied to screen the requirements.Numerical application is provided.

2 - A fuzzy farming efficiency estimation model for agri-cultural developmentTomas Balezentis, Lithuanian Institute of AgrarianEconomics, Lithuania, [email protected], Alvydas BalezentisReaching an economic efficiency is the foremost objective of agricul-tural development. The paper focuses on development of the multi—criteria benchmarking of farming efficiency across different farmingtypes in Lithuania. In this study time series data from the FADNwere analyzed. The fuzzy MULTIMOORA method was applied onthe selected indicators identifying farming efficiency in terms of land,labour, and capital exploitation. The most efficient as well as least effi-cient farming types were identified. Such analysis provides a rationalefor strategic decision making.

3 - Integrating multidisciplinary knowledge into simula-tion models to organize organic wastes recycling inagricultureFrancois Guerrin, UPR Recyclage & Risque, Inra & Cirad,Station de la Bretagne - BP 20, 97408, Saint-Denis, France,[email protected], Moussa NDienor, HeriniainaRamahefarison, Virginie Parnaudeau, Jean-Marie Paillat

250


Substituting chemical fertilizers by organic wastes is a way to im-prove agriculture sustainability. It requires designing new managementscenarios of agricultural production systems based on the knowledgeavailable on their biophysical and managerial components. Simulationmodels representing those systems as productive units spread over aterritory together with data from agronomical experiments are jointlyused to assess the performances of the scenarios against agronomicaland environmental criteria. This approach is illustrated on two exam-ples in periurban areas of Senegal and Madagascar.

� WD-41Wednesday, 14:30-16:00HH-Heyerdahl

Applications of Control Theory inDynamical Systems

Stream: Control Theory & System Dynamics (con-tributed)Contributed sessionChair: Ioan Radu Morar, Automation Dept, Technical Universityof Cluj-Napoca, Ghe Baritiu Street 26-28, room 350, 400027,Cluj-Napoca, Cluj, Romania, [email protected]

1 - A model to particles flow composed by twometastable potentialsElso Drigo Filho, Department of Physics, Sao Paulo StateUniversity, Rua Cristovao Colombo 2265, 15054-000, SaoJose do Rio Preto, Sao Paulo, Brazil, [email protected],Marcelo Tozo de Araujo

A fluid flow through a tube can promote a sedimentation of particleswhich reduces the diameter of the tube. This phenomenon is particu-larly important to oil transport which induces the deposition of paraffinin the inner wall of the tube. In order to simulate this process, we sug-gest a model that involves two time dependent metastable potentials.The solution of a Fokker-Planck equation for each potential is the inputto describe the time dependence of the total flow. Numerical results fordifferent values of parameters are presented.

2 - The mechatronic system optimization with use of ge-netic programmingMarian Klucik, Institute of Control and Industrial Informatics,Slovak University of Technology, Ilkovicova 3, 812 19,Bratislava, Slovakia, [email protected], Ladislav Jurisica,Anton Vitko, Peter Paszto, Jaroslav Hanzel

The aim of this article are the desired properties of mechatronic sys-tems, and the desired properties of mechatronic systems design. In thedesign process is essential to develop the method for search the opti-mal system structure and parameters, which ensures sufficient synergybetween system components. One means of solving this difficult prob-lem is to use genetic programming. With this method it is possible tosolve the optimization problem with consideration of the large numberof criteria. In the conclusion is proposed one case of control systemoptimization of mechatronic system.

3 - Fault-tolerant Control Systems in Stability and Re-covery control Of Damaged AircraftM. Navabi, Dynamics and Control, Shahid BeheshtiUniversity, GC, 123, thr, Iran, Islamic Republic Of,[email protected], M. Radaei

In the emergency cases where conventional flight controls are lost, analternative means of control is needed to maintain a certain level of sta-bility and performance to avoid potential crashes and accidents. In thispaper two classified inter-control-disciplinary of fault-tolerant controlsystems are introduced: 1) passive fault-tolerant control based on ro-bust control techniques, and 2) active fault-tolerant control using directand indirect adaptive control, reconfigurable control, and eigenstruc-ture assignment technique for different damage scenarios. Simulationresults are discussed.

4 - Modeling an Unmanned Aerial Vehicle (Quadrotor)Ioan Radu Morar, Automation Dept, Technical University ofCluj-Napoca, Ghe Baritiu Street 26-28, room 350, 400027,Cluj-Napoca, Cluj, Romania, [email protected], IoanNascu

In this paper the authors describe the process of obtaining a modelfor an Unmanned Aerial Vehicle (UAV) designed by the Parrot, calledAR.Drone. The aim of this process is to find a simple model thatcan approximate as good as possible the non-linear dynamics of thereal model. A simulator was designed in computational software andtested, with a purpose of developing control techniques and strategies.Simulation results are presented at the end of the paper

ACKNOWLEDGMENT: TUCN, research supported by the projectPOSDRU/107/1.5/S/78534

� WD-42Wednesday, 14:30-16:00BW-Amber

Stochastic Modelling in ComputationalBiology and Environmental Sciences

Stream: Stochastic Modeling and Simulation in Engi-neering, Management and ScienceInvited sessionChair: Erik Kropat, Department of Computer Science, Universitätder Bundeswehr München, Werner-Heisenberg-Weg 39, 85577,Neubiberg, Germany, [email protected]: Zeev (Vladimir) Volkovich, Ort Braude Academic College,Yehiam 6, 21823, Karmiel, Israel, [email protected]: Gerhard-Wilhelm Weber, Institute of Applied Mathematics,Middle East Technical University, ODTÜ, 06531, Ankara, Turkey,[email protected]

1 - Earthquake Scenario Reduction by Symmetry Rea-soningSteven Prestwich, Computer Science, Cork ConstraintComputation Centre, University College, Cork, Ireland,[email protected]

A recently identified problem is that of finding an optimal investmentplan for a transportation network, under the assumption that an earth-quake or other disaster may destroy each link in the network with agiven probability. A network based on the Istanbul highway systemhas a billion scenarios, but it has been found that sampling a millionscenarios gives reasonable results. In this work we use symmetry rea-soning to reduce the number of scenarios to a few thousand, and Underan additional but realistic assumption we further reduce this number toa few hundred.

2 - Modelling the meteorological effects on air temper-ature for Konya city in Turkey: the approaches ofquantile regression and quantile regression neuralnetworksIlkay Altindag, Statistics, Selçuk University, SelcukUniversity Science Faculty Depertmant of Statistics KampusSelcuklu/KONYA, Konya, Turkey, [email protected],Nimet Yapici Pehlivan

Globalization forces firms to consider the flexibility in human re-sources. Labor flexibility plays an important role in organizations inall parts. As a result of including the labor requirements in market re-quirements, labor flexibility and its factors become a considered costvalue to the firms. Without labor flexibility, firms cannot take efficientand competent decisions. Under labor flexibility, 4 criteria will be ex-amined and tried to find out that which one is the most affecting onlabor flexibility.

3 - Stochastic optimization methods in change-point de-tectionTatiana Polushina, NTNU, Norway, [email protected],Georgy Sofronov

251


The change-point problem arises in a wide variety of fields, includingspeech and image processing, seismology, industry and financial math-ematics. Multiple change-point models are also important in many bi-ological applications and, particularly, in analysis of DNA sequences.In this talk, we consider various stochastic optimization methods tochange-point detection, using Monte Carlo simulation to find estimatesof change-points as well as parameters of the process on each seg-ment. We also demonstrate the methods for a realistic problem arisingin computational biology.

4 - A randomness test based on the Minimal SpanningTree approachZeev (Vladimir) Volkovich, Ort Braude Academic College,Yehiam 6, 21823, Karmiel, Israel, [email protected]

A stochastic model intended to evaluate the relative randomness of twogiven texts is offered. Pairs of samples, drawn from different texts, arecompared. The relative randomness is measured via the resemblanceof the samples in the pooled one when the mixture quality is repre-sented by the total number of edges connecting points from differentsamples in the constructed minimal spanning tree. Under the homo-geneity hypothesis, this amount is normally distributed. Numericalexperiments exhibit the ability of the method to distinguish texts gen-erated by different random sources.

� WD-43Wednesday, 14:30-16:00BW-Granite

Emerging applications in finance andinsuranceStream: Stochastic ProgrammingInvited sessionChair: Giorgio Consigli, Mathematics, Statistics and ComputerSciences, University of Bergamo, Via dei Caniana 2, 24127,Bergamo, Italy, [email protected]

1 - An application of institutional asset-liability manage-ment to a large P&C insurance companyGiorgio Consigli, Mathematics, Statistics and ComputerSciences, University of Bergamo, Via dei Caniana 2, 24127,Bergamo, Italy, [email protected], Vittorio Moriggia,Gaetano Iaquinta, Angelo Uristani

Recent trends in the insurance sector have highlighted expansion oflarge insurance firms into asset management. In addition to their histor-ical liability exposure, the growth of investment divisions has causedincreasing exposure to market fluctuations. This has led to stricter riskmanagement requirements as reported in the Solvency II 2010 studyby the European Commission. We present an ALM model tested on areal-world case study, combining in a dynamic framework an optimalstrategic asset allocation problem with risk capital and P&C businessconstraints for a large insurer.

2 - Multistage strategic asset allocation with alternativeinvestmentsAngelo Uristani, Mathematics, University of Bergamo, 20149,Bergamo, Italy, [email protected], Giorgio Consigli,Vittorio Moriggia

In recent years, due to increasing market risk and low expeced returns,investors are expanding their investment universe. The so-called al-ternative investments have become increasingly interesting since theyprovide good diversification benefits, despite their poor liquidity. Inthis work we present an ALM multi-stage problem which includes pri-vate equity and real estate along with more common investment classessuch as stocks and bonds. The complex model specification and its nu-merical implications are analysed in detail.

3 - An Alternative Approach for Exponential SmoothingMethodAli Sabri Taylan, Market Oversight, Turkish DerivativesExchange, Vadeli Islem ve Opsiyon Borsasi A.S., AkdenizCd. Birsel Is Mrk. No:14 Kat:6, 35210, Izmir, Turkey,[email protected], Güçkan Yapar

The orginal exponential smoothing has performed well in numerousempirical studies, and it is well established as an accurate and op-timal forecasting method. However, there are two main problemsabout choosing the smoothing constant and starting value in exponen-tial smoothing theory. In this study, we suggest an alternative methodfor smoothing constant and starting value. The modified method givesmore weights than the classical method to most recent observations.The forecasting error is compared to the error in forecasts obtained bythe original.

4 - Quadrangle Concept in Classification and OnlineTrading AlgorithmsStan Uryasev, Director of Risk Management and FinancialEngineering Lab, University of Florida, FL 32611-6595,Gainesville, 303 Weil Hall, Florida, United States,[email protected], Peter Tsyurmasto

Quadrangle Concept in Risk Management provides a convenient toolfor building new numerical approaches involving decision making inuncertain environment. We use Quadrangle for the development ofnew classification algorithms with application to the intraday tradingstrategies. We are interested in a simple trading strategy: "buy low andsell high" on daily basis. The position is closed in the end of the day.We are planning to use for solving the classification problem the CVaR-based Support Vector Machine implemented with Portfolio Safeguard(PSG) software.

� WD-44Wednesday, 14:30-16:00BW-Marble

Information and Intelligent Systems II

Stream: Information and Intelligent SystemsInvited sessionChair: Igor Kabashkin, Transport and TelecommunicationInstitute, Lomonosova iela 1, Riga, Latvia, [email protected]: Hans Georg Zimmermann, Corporate Technology CT T,Siemens AG, Otto-Hahn-Ring 6, 81730, München, Germany,[email protected]

1 - E-Commerce & m-Commerce: A Specific IntegratedPlatformDavid L. La Red Martínez, Informática, Universidad Nacionaldel Nordeste, 9 de julio 1449, (3400), Corrientes, Corrientes,Argentina, [email protected], Valeria E. Uribe,Sofía J. Vallejos

In recent years, the evolution of technology has given great leap. Inter-net, combined with telephone, becomes a means of social interactionthat is currently present in all areas of daily life. One of the manyapplications of this integration occurs between the e-commerce andm-commerce, even with little development in the Argentina (exceptmobile banking sector). The developed platform supports e-commerceand m-commerce for a pharmaceutical entity, manages informationabout the client and the marketing of products and over-the-countermedications.

2 - Text mining and visualization with self-organizingmapsPavel Stefanovic, Vilnius University, Institute of Mathematicsand Informatics, Akademijos str. 4, LT-08663, Vilnius,Lithuania, [email protected], Olga Kurasova

The aim of the research is text mining and visualization with self-organizing maps. Text to Matrix Generator toolbox was used to con-struct term-document matrices from text collections. The influenceof different options has been investigated. A self-organizing map hasbeen used for text clustering and graphical representation (visualiza-tion). A comparative analysis has been made with 40 different articlesfrom four areas: artificial neural network, bioinformatics, optimiza-tion and self-organizing maps. The quantization and topographic er-rors have been measured to estimate the SOM quality.

252


3 - Design of architecture of embedded system for opti-mal reliability of complex systemIgor Kabashkin, Transport and Telecommunication Institute,Lomonosova iela 1, Riga, Latvia, [email protected]

The digital systems provide higher productivity and greater flexibility,but it is also accepted that they cannot be fault-free. The role of built intest embedded system in electronics has grown in prominence with theadvances in system complexity. The architecture of an embedded sys-tem is an abstraction, that typically doesn’t show detailed informationsuch as software source code or hardware circuit design. Our paperdescribes the approach for design of optimal architecture of embeddedsystem with minimal cost of reliability identification during life cycleof complex systems.

4 - Selecting Ore for an Iron Ore MineJim Everett, Centre for Exploration Targeting, University ofWestern Australia, 49 Goldsmith Rd, 6009, Nedlands, WA,Australia, [email protected]

Interpolated drilling data provide assay estimates for each block in themodeled mine, identifying it as ore or waste, depending on its iron andcontaminants (such as silica, alumina and phosphorus). Industry prac-tice accepts a block as ore if its grade lies within a quadrant, above acut-off in iron and below cut-off in each contaminant. A compositecut-off function is shown preferable to quadrant selection. The methodmaximizes the yield of ore of required product grade and matchesmarginal cost to marginal value, optimizing the resource net value. Itcan apply to other mine types.

� WD-45Wednesday, 14:30-16:00BW-Water

IBM Research Applications

Stream: IBM Research ApplicationsInvited sessionChair: Marco Laumanns, IBM Research - Zurich, 8803,Rueschlikon, Switzerland, [email protected]

1 - Planning under Uncertainty — Case Study: EnergyCost Minimization for a Water Treatment and Distri-bution Network.Susara van den Heever, Research, IBM, Ireland,[email protected], Niall Brady, Olivier Verscheure, KateLehane

We address the problem of planning under uncertainty where the plan-ning decisions are represented by binary variables. As a case study, weconsider the minimization of energy costs due to pumping for a watertreatment and distribution network, where energy prices vary duringthe day due to factors such as wind. The actual prices are only knownseveral days after the planning decisions have been made. We present aMixed Integer Linear Programming model, as well as two approachesto capture the uncertainty, namely a stochastic programming approachand a heuristic scenario hedging approach.

2 - Nonlinear optimization for decision problems in wa-ter distribution networks under uncertaintyMartin Mevissen, IBM Research Dublin, IBM TechnologyCampus, Damastown Industrial Park, Mulhuddart, 15, Dublin,Ireland, [email protected]

Decision problems in water distribution networks are challenging dueto both, nonlinearity in the hydraulic models and demand uncertaintyover the network. We propose an approach for pressure optimizationand optimal design decisions over the network, which is based on anovel quadratic model for the nonlinearity and constructs demand un-certainty sets from given sampled data. We use convex optimizationand MINLP methods to obtain an optimal solution, which is robustwith respect to demand uncertainty and based on an accurate approxi-mation for a real world water distribution network.

3 - An efficient way to find an optimal power series forcharging electric vehiclesJacint Szabo, Business Optimization, IBM Research Lab,Zurich, Säumerstrasse 4, 8803, Rüschlikon, Switzerland,[email protected], Olivier Gallay, Olle Sundstroem

The charging of electric vehicles (EV) is usually managed by chargingservice providers, who buy electric power from the retailer to satisfythe demand of the EV pool in such a way that the forecasted consump-tion profiles of the EV’s are respected, and power grid load constraintsare taken into account. In this contribution we present an efficient wayto calculate an optimal power series over the considered time horizon.

4 - Multi-Period Production Planning Under Non-Compliance RiskMarco Laumanns, IBM Research - Zurich, 8803,Rueschlikon, Switzerland, [email protected], AlwinHaensel, Ban Kawas, Eleni Pratsini, Steven Prestwich, CatalinStefan Tiseanu

We address a production planning setting for pharmaceutical compa-nies under the risk of failing quality inspections undertaken by the reg-ulatory authorities. After reviewing the single-period problem of max-imizing the worst-case revenue and the worst-case expectation, differ-ent stochastic programming and dynamic programming formulationsfor the multi-period version are discussed. We show how decision-dependent probabilities, which are due to risk transfer between prod-ucts of the same site, can be handled by solving a MINLP for eachstate in the multi-period dynamic program.

253

WE-01 EURO 25 - Vilnius 2012


� WE-01Wednesday, 16:30-17:30RB-Alfa

IFORS Distinguished Lecture: ProfessorRalph E. Gomory

Stream: Keynote, Tutorial and Plenary LecturesPlenary sessionChair: Dominique de Werra, EPFL, 1015, Lausanne,[email protected]

1 - Forty Years of Corner PolyhedraRalph E. Gomory, Stern School, New York University, 260Douglas Road, 10514, Chappaqua, New York, United States,[email protected]

This talk will discuss the evolution of corner polyhedra from their be-ginnings in the study of Stock Cutting problems. We will discuss boththeir practical and theoretical aspects. We will see that corner poly-hedra are by themselves the simplest integer programs and thereforemight be expected to be more amenable to analysis than the more com-plex I.P. s of which they are always a part. This expectation is fulfilledby some theoretical insights gained from corner polyhedra. The practi-cal linkage stems from the fact that cutting planes for corner polyhedraare cutting planes for the complex practical problems of which they area part. No knowledge of corner polyhedra is assumed in this talk andhistorical and personal aspects will be included throughout.

254

STREAMS

Actuarial Sciences andStochastic CalculusInvited

Ricardo Josa-FombellidaUniversidad de [email protected]

Juan Pablo Rincon-ZapateroUniversidad Carlos III de [email protected](s): 193 sessions

Advanced Inventory Controland Pricing StrategiesInvited

Alf KimmsUniversity of [email protected]

Robert KleinUniversität [email protected](s): 453 sessions

Analytic Hierarchy Process,Analytic Network ProcessInvited

Josef JablonskyUniversity of Economics [email protected]

Y. Ilker TopcuIstanbul Technical [email protected](s): 416 sessions

Applications of LocationAnalysisInvited

Sibel A. AlumurTOBB University of Economics [email protected]

Ioannis GiannikosUniversity of [email protected]

Mercedes LandeteUniversity Miguel Hernández [email protected](s): 175 sessions

Arc RoutingInvited

Stefan IrnichJohannes Gutenberg [email protected]

Claudia BodeJohannes Gutenberg [email protected](s): 112 sessions

Boolean and Pseudo-BooleanOptimizationInvited

Endre BorosRutgers [email protected](s): 84 sessions

Business Excellence inLogisticsInvited

Martin Josef [email protected](s): 33 sessions

Combinatorial OptimizationInvited

Silvano MartelloUniversity of [email protected]

Paolo TothUniversity of [email protected](s): 47 sessions

Commodities and FinancialModelingInvited

Rita DecclesiaSapienza University of [email protected]

Ronald HochreiterWU Vienna University of Economicsand [email protected]

Yeliz Yolcu OkurMiddle East Technical [email protected](s): 192 sessions

Computational Biology,Bioinformatics and MedicineInvited

Jacek BlazewiczPoznan University of [email protected]

Metin TurkayKoc [email protected]

Giovanni FeliciConsiglio Nazionale delle [email protected](s): 374 sessions

Computational StatisticsInvited

Pakize TaylanDicle [email protected]

Gerhard-Wilhelm WeberMiddle East Technical [email protected](s): 352 sessions

Container Terminal OperationsInvited

Iris F.A. VisUniversity of [email protected]

Christian BierwirthMartin-Luther-UniversityHalle-Wittenbergchristian.bierwirth@wiwi.uni-halle.de

Ceyda OguzKoc [email protected](s): 122 sessions

Continuous LocationInvited

Emilio CarrizosaUniversidad de [email protected](s): 183 sessions

255

STREAMS EURO 25 - Vilnius 2012

Control Theory & SystemDynamics (contributed)Contributed

Richard HartlUniversity of [email protected](s): 411 session

Convex OptimizationInvited

Aharon Ben-TalTechnion-Israel Institute [email protected]

Marc TeboulleTel Aviv [email protected]

Attila GilanyiUniversity of [email protected](s): 234 sessions

Cooperative Game TheoryInvited

Mariana Rodica Branzei"Alexandru Ioan Cuza” [email protected]

Sirma Zeynep Alparslan GokFaculty of Arts and Sciences,Suleyman Demirel [email protected](s): 347 sessions

Cutting and PackingInvited

José Fernando OliveiraUniversity of [email protected]

A. Miguel GomesINESC TEC, Faculdade deEngenharia, Universidade do [email protected](s): 9 107 sessions

Data ConfidentialityInvited

Jordi [email protected]

Juan José Salazar GonzálezUniversidad de La Laguna (Tenerife)[email protected](s): 302 sessions

Data Mining and DecisionMakingInvited

Lai-Soon LeeUniversiti Putra [email protected]

Hsin-Vonn SeowUniversity of Nottingham- [email protected](s): 241 session

Data Mining in Early WarningSystemsInvited

Inci BatmazMiddle East Technical [email protected]

Gulser KoksalMiddle East Technical [email protected]

Antonio RodriguesUniversity of [email protected](s): 241 session

Data Mining in the FinancialSectorInvited

Vadim StrijovRussian Academy of Sciences,Computing [email protected](s): 204 sessions

Data Mining, KnowledgeDiscovery and ArtificialIntelligenceContributed

Vincent MousseauEcole Centrale [email protected](s): 314 sessions

DEA and PerformanceMeasurementInvited

Dimitris DespotisUniversity of [email protected]

Ozren DespicAston [email protected]

Meryem Duygun FethiUniversity of [email protected]

Ana CamanhoUniversidade do [email protected]

Vania SenaAston [email protected](s): 366 sessions

DEA and PerformanceMeasurement (contributed)Contributed

Elena FernandezTechnical University of [email protected](s): 29 30 32 358 sessions

Decision Making Modeling andRisk Assessment in theFinancial SectorInvited

Cristinca FulgaGheorghe Mihoc-Caius IacobInstitute of Mathematical Statisticsand Applied Mathematics ofRomanian [email protected](s): 205 sessions

256

EURO 25 - Vilnius 2012 STREAMS

Decision ProcessesInvited

Ahti SaloAalto University School of Scienceand [email protected]

Alec MortonLondon School of [email protected]

Jeffrey KeislerUniversity of Massachusetts [email protected](s): 278 sessions

Decision Support SystemsInvited

Pascale ZaratéToulouse [email protected]

Fatima DargamSimTech Simulation [email protected]

Rita RibeiroUninova - [email protected](s): 274 sessions

Demand and Supply inConsumer Goods and RetailingInvited

Heinrich KuhnCatholic University [email protected]

Winfried SteinerClausthal University of Technology,Institute of Management [email protected]

Rob BroekmeulenTU [email protected](s): 351 session

Discrete Optimization,Geometry & Graphs(contributed)Contributed

Silvano MartelloUniversity of [email protected](s): 57 sessions

Dynamic ProgrammingInvited

Lidija Zadnik StirnUniversity of [email protected](s): 392 sessions

Dynamical Systems and GameTheoryInvited

Alberto A. PintoUniversity of [email protected](s): 63 sessions

Dynamical Systems andMathematical Modelling in ORInvited

Selma BelenCAG [email protected]

Gerhard-Wilhelm WeberMiddle East Technical [email protected](s): 411 session

Emerging Applications inFinance and LogisticsInvited

Norbert TrautmannUniversity of [email protected](s): 303 sessions

Emerging Applications of FatTail Models in FinancialModeling and EngineeringInvited

Audrius KabasinskasKaunas University of [email protected](s): 301 session

Emerging Aspects ofProduction Planning inContinuous Process Industries:Theory, Optimization, andPracticeInvited

Krystsina BakhrankovaSINTEF - Technology and [email protected](s): 303 sessions

Emissions Trading and EnergyInvited

Magnus FröhlingKarlsruhe Institute of Technology(KIT)[email protected](s): 322 sessions

Energy, Environment andClimate 1Contributed

Peter LetmatheRWTH Aachen [email protected](s): 252 sessions

Energy, Environment andClimate 2Invited

Wolf [email protected]

Costas PappisUniversity of [email protected](s): 25 353 sessions

Engineering OptimizationInvited

Wolfgang AchtzigerUniversity of [email protected]

Helder RodriguesTechnical University of Lisbon, [email protected](s): 351 session

Environmental Issues inOperations ManagementInvited

Grit WaltherBergische Universität [email protected](s): 252 sessions

EURO Doctoral DissertationAward (EDDA 2012)InvitedTrack(s): 421 session

257


EURO Excellence in PracticeAward 2012Invited

Michel BierlaireÉcole Polytechnique Fédérale deLausanne (EPFL)[email protected](s): 422 sessions

EURO/ROADEF ChallengeInvited

H. Murat AfsarUniversity of Technology of [email protected]

Christian [email protected]

Eric [email protected]

Emmanuel [email protected]

Ender ÖzcanUniversity of [email protected]

Marc SevauxUniversité de Bretagne Sud - [email protected](s): 3 95 sessions

Experimental Economics andGame TheoryInvited

Ulrike Leopold-WildburgerKarl-Franzens-Universityulrike.leopold@uni-graz.at

Stefan PicklUniversität der Bundeswehr Mü[email protected]


Financial Mathematics and ORInvited

Gerhard-Wilhelm WeberMiddle East Technical [email protected]

Mustafa PinarBilkent [email protected](s): 216 sessions

Financial Modelling & RiskManagement (contributed)Contributed

Stein W. WallaceLancaster University [email protected](s): 184 sessions

Financial OptimizationInvited

Gautam MitraBrunel [email protected]

J. E. BeasleyBrunel University & JB [email protected](s): 203 sessions

ForecastingInvited

Aris SyntetosUniversity of [email protected]

Antonio RodriguesUniversity of [email protected]

Sven F. CroneLancaster University [email protected](s): 244 sessions

Fuzzy Optimization - Systems,Networks and ApplicationsInvited

Erik KropatUniversität der Bundeswehr Mü[email protected](s): 223 sessions

Fuzzy Sets and Soft ComputingInvited

Mikael CollanLappeenranta University [email protected](s): 12 224 sessions

Fuzzy Systems, NeuralNetworks & ArtificialIntelligenceInvited

Heinrich RommelfangerJ. W. Goethe [email protected](s): 122 sessions

Game Theory and LogisticsInvited

Ignacio García-JuradoCoruna [email protected]

Ana MecaUniversidad Miguel Herná[email protected](s): 242 sessions

Game Theory and SocialNetworksInvited

Juan TejadaComplutense Unversity of [email protected](s): 312 sessions

Generalized Differentation andOptimizationInvited

Boris MordukhovichWayne State University, [email protected]

Alexander KrugerUniversity of [email protected]


Juan Enrique Martínez-LegazUniversitat Autònoma de [email protected](s): 235 sessions

258


Geometric ClusteringInvited

Andreas BriedenUniversität der Bundeswehr Mü[email protected]

Peter GritzmannTU Mü[email protected](s): 131 session

Global OptimizationInvited

Herman MawengkangThe University of Sumatera [email protected](s): 172 sessions

Graph SearchingInvited

Boting YangUniversity of [email protected]

Anthony BonatoRyerson [email protected]

Nancy ClarkeAcadia [email protected](s): 392 sessions

Graphs and NetworksInvited

Dominique de [email protected](s): 93 sessions

Health Care ManagementInvited

Sally BrailsfordUniversity of [email protected]

Teresa MeloSaarland University of [email protected]

Stefan NickelKarlsruhe Institute of Technology(KIT)[email protected]

Marion RaunerUniversity of [email protected](s): 372 sessions

Hub LocationInvited

James CampbellUniversity of Missouri-St. [email protected]

Ivan ContrerasConcordia [email protected](s): 172 sessions

IBM Research ApplicationsInvited

Eleni PratsiniIBM Zurich Research [email protected](s): 451 session

Information and IntelligentSystemsInvited

Zuzana OplatkovaTomas Bata University in [email protected]


Innovative Software Tools forMCDAInvited

Patrick MeyerTelecom [email protected]

Vincent MousseauEcole Centrale [email protected](s): 392 sessions

Inventory ManagementInvited

Dolores Romero MoralesUniversity of [email protected](s): 132 sessions

Iterative Methods for EconomicModelsInvited

Adriana GnudiUniversity of [email protected]

Elisabetta AlleviUniversity of [email protected]

Igor KonnovUniversity of [email protected](s): 311 session

Keynote, Tutorial and PlenaryLecturesInvited

Marielle ChristiansenNorwegian University of Science [email protected]

Leonidas SakalauskasInstitute of Mathematics &[email protected](s): 115 sessions

259


Linear and Conic ProgrammingInvited

Miguel AnjosEcole Polytechnique de [email protected]

Tibor IllésBudapest University of Technologyand [email protected](s): 164 sessions

Logistics, Transportation,TrafficContributed

Elena FernandezUniversitat Politecnica de [email protected](s): 33 sessions

Long Term Financial DecisionsInvited

Thomas BurkhardtUniversitaet [email protected](s): 213 sessions

Long Term Planning in Energy,Environment and ClimateInvited

Nadia MaïziMINES [email protected](s): 253 sessions

Lot-sizing and Related TopicsInvited

Christian AlmederEuropean University [email protected]

Bernardo Almada-LoboFaculty of Engineering of [email protected]

Alistair ClarkUniversity of the West of [email protected](s): 134 sessions

Machine Learning and itsApplicationsInvited

Vadim StrijovRussian Academy of Sciences,Computing [email protected]

Sureyya Ozogur-AkyuzBahcesehir [email protected](s): 446 sessions

Manufacturing andWarehousingInvited

Martin GrunowTechnische Universität Mü[email protected](s): 132 sessions

Maritime TransportationInvited

Henrik AnderssonNorwegian University of Science [email protected](s): 5 96 sessions

Mathematical Models in Macro-and MicroeconomicsInvited

Murat KirdarMiddle East Technical [email protected]

Ludmilla KoshlaiInstitute of [email protected]


Mathematical ProgrammingInvited

Florian PotraUniversity of [email protected]

Tamas TerlakyLehigh [email protected]


Goran LesajaGeorgia Southern [email protected]

Sandor Zoltan NemethThe University of [email protected](s): 233 sessions

MatheuristicsInvited

Vittorio ManiezzoUniversity of [email protected]

Stefan VossUniversity of [email protected](s): 61 session

MCDA: New Approaches andApplicationsInvited

Gerhard-Wilhelm WeberMiddle East Technical [email protected](s): 39 4010 sessions

MetaheuristicsInvited

Marc SevauxUniversité de Bretagne Sud - [email protected]

Kenneth SörensenUniversity of [email protected]

Andreas ReinholzGerman Aerospace Center (DLR)[email protected](s): 67 sessions

260


Methodology of SocietalComplexityInvited

Dorien DeTombeChair Euro Working [email protected](s): 362 sessions

Mixed-Integer Non-LinearProgrammingInvited

Sonia CafieriEcole Nationale d’Aviation [email protected]

Claudia D’AmbrosioCNRS - Ecole [email protected](s): 297 sessions

Models of Embodied CognitionInvited

Susie VrobelThe Institute for Fractal [email protected](s): 282 sessions

Multi-Criteria Decision Makingand Environmental ManagementInvited

Vadim StrijovRussian Academy of Sciences,Computing [email protected](s): 263 sessions

Multiobjective OptimizationInvited

José Rui FigueiraTechnical University of [email protected]

Kaisa MiettinenUniversity of [email protected]

Christiane TammerMartin-Luther-UniversityHalle-Wittenbergchristiane.tammer@mathematik.uni-halle.deTrack(s): 38 4113 sessions

Network and Discrete LocationInvited

Francisco Saldanha-da-GamaUniversity of [email protected]

Alfredo MarínUniversity of [email protected]

Stefan NickelKarlsruhe Institute of Technology(KIT)[email protected](s): 184 sessions

Network OptimizationInvited

Bernard FortzUniversité Libre de [email protected]

Luis GouveiaUniversity of [email protected]

Walid Ben-ameurTelecom [email protected](s): 84 sessions

New Trends in Vehicle RoutingInvited

Jorge E. MendozaUniversité Catholique de l’[email protected](s): 73 sessions

Nonconvex Programming:Local and Global ApproachesInvited

Tao Pham DinhINSA [email protected]

Hoai An Le ThiUniversity of [email protected](s): 173 sessions

Nonlinear ProgrammingInvited

Edite M.G.P. FernandesUniversity of [email protected]

M. Fernanda P. CostaUniversity of [email protected]

Lino CostaUniversity of [email protected]

Ana Maria A.C. RochaUniversity of [email protected]

Isabel Espírito SantoMinho [email protected]

Simone GöttlichUniversity of [email protected](s): 15 194 sessions

Nonsmooth OptimizationInvited

Adil BagirovUniversity of [email protected]

Albert FerrerTechnological University of [email protected]

Antonio FuduliUniversita’ della [email protected]

Peter RechtTU [email protected](s): 163 sessions

Numerical Methods in FinanceInvited

Ömür UgurMiddle East Technical [email protected](s): 213 sessions

Occupational Heath & WellbeingInvited

Elena AndreevaBerlin University of [email protected](s): 361 session

261


Operational Research andQuantitative Models in BankingInvited

Constantin ZopounidisTechnical University of [email protected](s): 192 sessions

Operations/Marketing InterfaceInvited

Kathryn E. SteckeUniversity of Texas at [email protected]

Xuying ZhaoUniversity of Notre [email protected](s): 351 session

Optimal ControlInvited

Ekaterina KostinaUniversity of [email protected]

Gernot TraglerVienna University of [email protected]

Erik KropatUniversität der Bundeswehr Mü[email protected]


Optimization and Data MiningInvited

Emilio CarrizosaUniversidad de [email protected]

Theodore TrafalisUniversity of [email protected]

Renato De LeoneUniversità di [email protected](s): 294 sessions

Optimization for SustainableDevelopmentInvited

Herman MawengkangThe University of Sumatera [email protected]


Optimization in PublicTransportInvited

Leo KroonErasmus University [email protected]

Anita SchöbelGeorg-August Universiy [email protected](s): 3 4 129 sessions

OR and EnvironmentalManagementInvited

Marion StevenRuhr-University [email protected](s): 25 262 sessions

OR and EthicsInvited

Cristobal MirallesUniversidad Politecnica de [email protected]

Fred WenstøpBI Norwegian School of [email protected](s): 312 sessions

OR and Real ImplementationsInvited

Ben LevDrexel [email protected]

Belarmino Adenso-DiazUniversidad de [email protected]

Istvan MarosImperial College [email protected](s): 351 session

OR and the ArtsInvited

Vitaly PodobedovMoscow State [email protected](s): 241 session

OR Applications in IndustryInvited

Geir HasleSintef [email protected]

Jens WollenweberDr. Staedtler Transport [email protected](s): 351 session

OR Applications in theAutomotive IndustryInvited

Thomas SpenglerTechnische Universität [email protected]

Thomas VollingTechnische Universität [email protected](s): 133 sessions

262


OR for Development andDeveloping CountriesInvited

Elise del [email protected]

Honora SmithUniversity of [email protected]


Subhash DattaInstitute of Management [email protected]

Leroy WhiteUniversity of [email protected]

Alexander MakarenkoNational Technical University ofUkraine "KPI"[email protected]

Francoise SummersMiddle East Technical [email protected]

Youssef MasmoudiHight School of Commerce of [email protected](s): 338 sessions

OR for SustainableDevelopmentInvited

Leonidas SakalauskasInstitute of Mathematics &[email protected]

Edmundas Kazimieras ZavadskasVilnius Gediminas [email protected]

Tatjana VilutieneVilnius Gediminas [email protected]

Vida MalieneSchool of the Built Environment,Liverpool John Moores [email protected](s): 333 sessions

OR in Agriculture, Forestry andFisheriesInvited

LluisM PlaUniversity of [email protected]

Victor M. AlbornozUniversidad Tecnica Federico [email protected](s): 264 sessions

OR in Health & Life Sciences(contributed)Contributed


OR in Human ResourcesInvited

Bo HuUniversität der Bundeswehr Mü[email protected](s): 241 session

OR in Industry and Software forOR (contributed)Contributed

Mikael RönnqvistDépartement de génie mé[email protected](s): 351 session

OR in Military, Safety andSecurity ApplicationsInvited

Ana Isabel [email protected](s): 283 sessions

OR in Quality ManagementInvited

Ipek Deveci KocakoçDokuz Eylul University Faculty ofEconomics and [email protected]

Gulser KoksalMiddle East Technical [email protected](s): 282 sessions

OR in SportsInvited

Sigrid KnustUniversity of Osnabrü[email protected]

Dirk BriskornUniversity of [email protected]

Alexandra NewmanColorado School of [email protected](s): 351 session

OR in the Oil and Gas SectorsInvited

Irina DolgopolovaMiddle East Technical [email protected](s): 262 sessions

OR in Water ManagementInvited

Halil ÖnderMiddle East Technical [email protected]

Elcin [email protected](s): 261 session

Pioneers in OperationsResearchInvited

Graham RandLancaster [email protected]

Jakob KrarupUniversity of [email protected](s): 241 session

263


Preference LearningInvited

Krzysztof DembczynskiPoznan University of [email protected]

Willem WaegemanGhent [email protected]

Roman SlowinskiPoznan University of [email protected](s): 405 sessions

Production and the Link withSupply ChainsInvited

Lionel AmodeoUniversity of Technology of [email protected]

Farouk YalaouiUniversity of Technology of [email protected](s): 103 sessions

Production Management &Supply Chain Management(contributed)Contributed

Elena FernandezUniversitat Politecnica de [email protected](s): 106 sessions

Project Management andSchedulingInvited

Rainer KolischTechnische Universitaet [email protected](s): 24 sessions

Quantitative Approaches inManagerial and FinancialAccountingInvited

Matthias AmenUniversity of [email protected](s): 451 session

Quantitative Models forPerformance and DependabilityInvited

Markus SiegleUniversitaet der [email protected](s): 362 sessions

Realistic Production SchedulingInvited

Ruben RuizUniversidad Politecnica de [email protected](s): 113 sessions

Recent Advances in EarthquakeStudiesInvited

Aysegul AskanMiddle East Technical [email protected]


Revenue ManagementInvited

Ayse KocabiyikogluBilkent [email protected]

Yuri LevinQueen’s [email protected]

Tatsiana LevinaQueen’s [email protected]

Joern MeissnerKuehne Logistics [email protected](s): 40 44 456 sessions

SchedulingInvited

Erwin PeschUniversity of [email protected]

Vitaly StrusevichUniversity of [email protected](s): 28 sessions

Scheduling under ResourceConstraintsInvited

Joanna JozefowskaPoznañ University of [email protected]

Jan WeglarzPoznan University of [email protected](s): 146 sessions

SD Modelling in SustainableDevelopmentInvited

Pierre KunschVrije Universiteit [email protected]

Erik PruytDelft University of [email protected](s): 402 sessions

Semi-Infinite OptimizationInvited

Vladimir ShikhmanRWTH Aachen [email protected]


Shunsuke HayashiKyoto [email protected](s): 152 sessions

SimulationInvitedTrack(s): 421 session

Simulation Based DecisionSupportInvited

Miroljub KljajicUniversity of [email protected]

Viveca AsprothMid Sweden [email protected](s): 251 session

264


Simulation in ManagementAccounting and ManagementControlInvited

Friederike WallAlpen-Adria-Universitaet [email protected]

Stephan LeitnerAlpen-Adria-Universität [email protected](s): 452 sessions

Simulation Methods in FinanceInvited

Aysegul Iscanoglu CekicSelcuk [email protected]


Soft OR and ProblemStructuring MethodsInvited

Giles HindleHull University Business [email protected](s): 326 sessions

Software for OR/MSInvited

Robert [email protected]

Bjarni KristjanssonMaximal Software, [email protected](s): 33 352 sessions

Stochastic Modeling andSimulation in Engineering,Management and ScienceInvited

Erik KropatUniversität der Bundeswehr Mü[email protected]

Devin SezerMiddle East Technical University,Institute of Applied [email protected]

Basak TanyeriBilkent [email protected]

Zeev (Vladimir) VolkovichOrt Braude Academic [email protected]


Stochastic Modelling inLogistical NetworksInvited

El-Houssaine AghezzafGhent [email protected](s): 453 sessions

Stochastic ProgrammingInvited

Jitka DupacovaCharles Univ, Faculty of Math. [email protected]

Georg PflugUniversity of [email protected]

Andras PrekopaRutgers [email protected]

Alexander ShapiroGeorgia Institute of [email protected]

Stein W. WallaceLancaster University [email protected]

Giorgio ConsigliUniversity of [email protected]

Abdel LisserUniversite de Paris [email protected]

Asgeir TomasgardSintef Technology and [email protected](s): 43 4412 sessions

Structured Optimization andApplicationsInvited

Vladimir ShikhmanRWTH Aachen [email protected](s): 284 sessions

Supply Chain ManagementInvited

Steef van de VeldeRSM Erasmus [email protected](s): 102 sessions

265


Supply Chain PlanningInvited

Moritz FleischmannUniversity of [email protected]

Herbert MeyrUniversity of [email protected](s): 7 84 sessions

Sustainable Living: Cognitive,Social, Economical, Ecologicaland World ViewInvited

Annette HohenbergerMiddle East Technical University(METU)[email protected]

Pedamallu Chandra SekharDana-Farber Cancer [email protected]


Claudia RaveNational University of [email protected](s): 341 session

Sustainable Transport PlanningInvited

Bülent ÇataySabanci [email protected]

Tolga BektasSchool of [email protected](s): 32 sessions

System Dynamics Modeling andSimulationInvited

Markus SchwaningerUniversität [email protected]

Markus SchwenkeUniversity of St. [email protected](s): 411 session

Timetabling and RosteringInvited

Dario Landa-SilvaUniversity of [email protected]

Nysret MusliuVienna University of [email protected]

Timothy CurtoisUniversity of [email protected](s): 145 sessions

Transportation and LogisticsInvited

Maurizio BielliInstitute of Systems Analysis [email protected](s): 115 sessions

Transportation PlanningInvited

Herbert KopferUniversity of [email protected]

Frank SchultmannKarlsruhe Institute of Technology(KIT)[email protected]

Track(s): 111 session

Uncertainty and Perturbationsin Optimization and inEnvironmental and EnergyModellingInvited

Jerzy FilarFlinders [email protected]

Julia PiantadosiUniversity of South [email protected](s): 165 sessions

Variable Neighborhood SearchInvited

José A. Moreno-PérezUniversity of La [email protected]

Nenad MladenovicBrunel [email protected](s): 61 session

Variational Inequalities andBi-Level ProblemsInvited

Stephan DempeTechnische Universitaet [email protected]

Yury KochetovNovosibirsk State [email protected]

Oleg KhamisovInstitute of Energy [email protected](s): 153 sessions

Vector and Set-ValuedOptimizationInvited

Vicente NovoUniversidad Nacional de Educacion [email protected]

Bienvenido Jimé[email protected]

César GutiérrezUniversidad de [email protected](s): 153 sessions

Vehicle Routing and LogisticsOptimizationInvited

Daniele VigoUniversity of [email protected](s): 76 sessions

266


Young People for SystemTheory, Optimization andEducationInvited

Alexis PasichnyNational Technical University [email protected]

Kateryna PereverzaNational Technical University [email protected]

Liza KorotchenkoNtuu "kpi"[email protected]

Alexej OrlovInstitute for applied system analysisof the National Technical Universityof [email protected]

Bohdan PukalskyiNational Technical University [email protected](s): 412 sessions

267

SESSION CHAIR INDEX

AAbreu, Nair TB-05Achtziger, Wolfgang MC-35Adenso-Diaz, Belarmino MA-35,

TC-35Önder, Halil TA-26Özçetin, Erdener TB-07Özcan, Ender WA-03, WB-03,

WC-03, TD-09, WD-09Özdemir, Rifat Gürcan WB-10Özmen, Ayse WA-42Çalık, Ahmet WC-19Afsar, H. Murat WA-03, WB-03,

WC-03, TD-09, WD-09Aghezzaf, El-Houssaine TA-45Airoldi, Mara TB-27Akyer, Hasan TC-07Albareda Sambola, Maria WC-18Algaba, Encarnación MA-34Aliev, Iskander WD-05Allevi, Elisabetta MC-31Almada-Lobo, Bernardo MA-13Almeder, Christian MB-13Alparslan Gok, Sirma ZeynepWB-34,

WC-34Alumur, Sibel A. MC-17Alvarez, Fernando MC-05Alvarez-Valdes, Ramon MC-09Amen, Matthias TD-45Amodeo, Lionel MA-10Andersson, Henrik MD-05Andreeva, Elena WD-36Anido, Carmen MA-25, MB-25Antunes, Carlos Henggeler WC-38Archibald, Thomas TA-16Argyris, Nikolaos TA-27Arik, Ayse WD-20Armañanzas, Rubén WC-29Artigues, Christian WA-02, 03, WB-

03, WC-03, TD-09, WD-09Askan, Aysegul WD-28Assoumou, Edi WB-25Avkar, Tansel MA-21, MB-

21, MC-21, MD-21, TA-21,TB-21

Avrachenkov, Konstantin TC-16

BBabonneau, Frédéric TD-16Baidya, Tara Keshar Nanda TC-20Bakhrankova, Krystsina MA-26,

WA-30Ballestero, Enrique MC-19Barcelo, Jaume MB-11Barketau, Maksim TA-02Barros, Ana Isabel WA-28, WB-28,

WC-28Bartl, David WB-16Baskaran, Geetha TC-14Bastic, Majda MC-24Bautzer, Ana MA-08Beasley, J. E. WB-20Belen, Selma TD-41Ben-Tal, Aharon WD-23

Benedito, Ernest WA-10Benoist, Thierry MD-03Berndt, Rüdiger TC-13Bertazzi, Luca TD-13Bertocchi, Marida MC-43Besiou, Maria WC-10Bielza, Concha WC-29Bierlaire, Michel MC-42, MD-42Björk, Folke WC-33Bjork, Kaj-Mikael MA-22Blanco, Víctor TB-18Blazewicz, Jacek MA-37, MC-37Bode, Claudia TC-11Bohlin, Markus TB-03Boland, John MD-16Bolshoy, Alexander WC-26Boros, Endre WC-08Bortfeldt, Andreas TB-09Bourgeois, Nicolas WA-09Bourreau, Eric WA-03, WB-03, WC-

03, WA-05, TD-09, WD-09Boyacı, Burak TD-18Bozkaya, Burcin MB-17Brabazon, Philip WC-13Brailsford, Sally WD-37Brandao, Jose TC-06Branke, Juergen MA-41Bravo, Cristian MC-20Bravo, Mila MC-19Brenner, M. Harvey WD-36Brieden, Andreas TB-13Briskorn, Dirk MB-02, MD-02Brotcorne, Luce MB-45Bruni, Renato TD-31

CCafieri, Sonia MC-29, TA-29Caimi, Gabrio Curzio WA-12Canca, David WB-12Cangalovic, Mirjana WB-09Cano, Javier MD-33, WD-38Capek, Roman WA-11Captivo, Maria Eugénia MC-06Carravilla, Maria Antónia MB-09Carreras, Ashley MC-32Castro, Marco WD-07Chaabane, Amin MD-10Chabchoub, Habib TA-33, TB-33Chandra Sekhar, Pedamallu TD-34Charousset-brignol, Sandrine MC-44Chehade, Hicham MA-10, TB-10Chen, Chie-beinMA-39, MB-39, MC-

39, MD-39, MA-40, MB-40, MC-40, MD-40

Chen, Yw TC-37Cho, Minyoung WD-29Chowdhury, S.M.Atiqur RahmanWC-

16Christiansen, Marielle ME-01Chub, Olga WC-11Clarke, Nancy WC-39, WD-39Cleophas, Catherine WB-45Collan, Mikael MC-12Comes, Tina TD-20

Consigli, Giorgio WD-43Contreras, Ivan TB-17Cordoba-Pachon, Jose-Rodrigo MD-

32Correia, Gonçalo MA-04Couzoudis, Eleftherios TA-28

DD’Ambrosio, Claudia MC-29, TA-29Daduna, Joachim R. TC-03Dargam, Fatima WB-27, WD-27Datta, Subhash MB-33Dauzere-peres, Stéphane MB-42de Araujo, Silvio TD-10De Cosmis, Sonia TA-42De Cuypere, Eline TC-45de la Fuente, Hanns WD-32De Smet, Yves TB-39de Werra, Dominique WE-01deCampos, Teofilo TA-44, TB-44Decclesia, Rita MA-19Dembczynski, Krzysztof TB-40,

TC-40, TD-40Demenkov, Max MD-15Demir, Alparslan Serhat TA-08Dempe, Stephan WB-15Despotis, Dimitris MB-36DeTombe, Dorien WA-36, WC-36Deveci Kocakoç, Ipek MA-28Dias, Luis C. MD-36Djeffal, Lakhdar WA-14Dolgopolova, Irina MB-26Doll, Claus MB-04Dorsch, Dominik MC-28, MD-28Druart, Frederik MD-18Dugardin, Frédéric MB-10, WA-10Duhamel, Christophe MB-10Dullaert, Wout MB-07, TB-45

EEguia, Ignacio TD-07Ejov, Vladimir TD-04Emmenegger, Jean-Francois TA-22,

TC-22Ennouri, Wissem WC-21Escudero, Laureano Fernando WA-43Eskelinen, Juha WB-30Euler, Reinhardt TC-09Everett, Jim TB-30

FFabian, Csaba I. TC-43Fanghaenel, Diana WC-15Fedrizzi, Mario MD-12Fernández Barberis, Gabriela TA-39Fernandes, Edite M.G.P. MC-15,

TD-19Fernandez, Elena MB-01, WB-01Ferreira, LuisWB-06, WC-06, WD-06Ferrer, Albert MB-16, MC-16Fiems, Dieter TC-45Figueira, José Rui WA-38Fischer, Frank WC-05Forsell, Nicklas WA-25Fourer, Robert WD-33

268

EURO 25 - Vilnius 2012 SESSION CHAIR INDEX

Fröhling, Magnus TD-32Framinan, Jose M. TD-02Franco, L. Alberto MB-32Freixas, Josep MC-34Fullér, Robert MC-12

GGüler, M. Güray TC-10Gönsch, Jochen WA-45Gössinger, Ralf WA-26Göttlich, Simone MB-15Gamboa, Dorabela TD-06García Alvarado, Marthy StívalizMD-

10García-Jurado, Ignacio MA-24Garfinkel, Robert TC-30Gavalec, Martin MB-12Gawiejnowicz, Stanislaw TA-14Geiger, Martin Josef MA-03, MB-03,

MC-03, MD-38Geldermann, Jutta TC-25Giannikos, Ioannis TA-17Gilanyi, Attila WD-23Gnudi, Adriana MC-31Gomes, A. Miguel MA-09Gomes, Luiz F. Autran M. TA-20Gomes, Marta Castilho WB-26Gonçalves, José Fernando TA-09González Alastrué, José Antonio TD-

30Gonzalez-Araya, Marcela TD-26Gorobetz, Mikhail WB-38Gotthardt, Benjamin WA-24Goverde, Rob TC-12Gregory, Amanda TB-32Gribkovskaia, Irina TA-05Griebsch, Susanne MB-18Gritzmann, Peter TB-13Groenevelt, Harry TB-10Grunewald, Martin WB-13Guajardo, Mario WA-13Gucumengil, Alper WB-11Guere, Emmanuel WA-03, WB-03,

WC-03, TD-09, WD-09Guerrin, Francois WD-40Gueye, Serigne WD-18Guignard-Spielberg, Monique WD-04Guillen, Montserrat TB-19Guimarães, Luis MD-13Guimaraes, Paula MD-35Gutiérrez-Jarpa, Gabriel WD-03Gwiggner, Claus TA-04

HHaase, Knut WA-16Habenicht, Walter MA-38Hadjiconstantinou, Eleni TD-04Hahn, Peter WD-04Hakanen, Jussi TA-38Hanafi, Saïd TA-06Harper, Paul TB-37Hartl, Richard WC-01Haurie, Alain TD-16Hernández-Jiménez, Beatriz TC-15Hernández-Pérez, Hipólito WA-07Herranz, Patricia WB-21

Hesse, Martina WB-39Hezer, Seda TC-02Hildmann, Marcus MA-20, MB-20Hiller, Benjamin TB-29Hindle, Giles TA-32Hirsch, Patrick TC-26, WC-37Hofmann, Ulla TD-21Humpola, Jesco TC-29Hunjak, Tihomir MC-41

IIftimie, Bogdan MA-19Ilani, Hagai WB-14Illés, Tibor WD-16Iscanoglu Cekic, Aysegul WB-19,

WC-19, WD-19Isenberg, Marc-André WD-10

JJablonsky, Josef MB-41Janiak, Adam MD-14Jimenez-Lopez, Mariano MA-12Jochem, Patrick TB-35Johnsen, Trond A. V. MA-05Juan, Angel A. MB-16

KKabashkin, Igor WD-44Kabasinskas, Audrius MA-30Kaluzny, Michael WA-26Kana Zeumo, Vivien TC-27Karray, Salma MA-45Kartal Koc, Elcin WD-35Katsev, Ilya WA-34Kazina, Irina MC-24Keisler, Jeffrey MC-27Kentel, Elcin TA-26Keskin, M. Emre MC-08Khachay, Michael TD-44, WA-44,

WB-44Kinoshita, Eizo WA-36, TB-41Klingelhöfer, Heinz Eckart WA-21Koberstein, Achim WB-13, TC-43Koksal, Gulser MB-28Koksalan, Murat MA-03Kolisch, Rainer WC-02Konnov, Igor MC-31Kopa, Milos TB-43Korhonen, Pekka TC-41Koshlai, Ludmilla TA-22, TC-22Kostina, Ekaterina MC-04Kostoglou, Vassilis WB-37Kostrzewska, Marta TB-22Kovalyov, Mikhail Y. MD-14, TB-14Kozan, Erhan WC-12Krarup, Jakob MC-30Krass, Dmitry TA-18Krawczyk, Jacek TB-16Kristjansson, Bjarni WD-33, WB-35Kroon, Leo TD-12Kropat, Erik MC-04, WB-

22, WC-22, WD-22, TB-23,WA-29, TA-42, TB-42, TC-

42, WA-42, WB-42, WC-42, WD-42

Kruger, Alexander MA-23, MB-23,MC-23, MD-23, TA-23

Krushinsky, Dmitry WA-17Kubiak, Wieslaw MA-14Kuefer, Karl-Heinz MC-38Kulak, Ozgur MD-09Kunsch, Pierre WC-40Kuosmanen, Timo MA-36Kus, Coskun WD-19Kusunoki, Yoshifumi TC-31Kuter, Semih WC-35Kuznetsov, Mikhail WC-26, WD-26

LLabbé, Martine TB-04Lachout, Petr MB-43Löhne, Andreas MB-38López-Cerdá, Marco A. TA-23Landete, Mercedes MD-17Lange, Jan TD-14Larsen, Christian TA-10Laumanns, Marco WA-12, WD-45Le Thi, Hoai An WC-17Lee, Gue Myung TC-38Lehrer, Ehud MB-34Leitner, Stephan MC-45, MD-45Leopold-Wildburger, Ulrike TD-22Lepin, Victor TC-05Letchford, Adam MA-29Letmathe, Peter TC-01Lev, Ben TC-35Levchenkov, Anatoly WB-38Levin, Yuri MA-44Levina, Tatsiana MA-44Li, Qing MB-44Liefooghe, Arnaud MA-41Lin, Chin-Tsai MA-39, MB-39, MC-

39, MD-39, MA-40, MB-40, MC-40, MD-40

Lindstad, Haakon WD-12Lisser, Abdel TD-43Liu, Shaofeng WC-27Ljubic, Ivana TC-04Loeffler, Andreas TC-21Lukasiak, Piotr MD-37Luss, Ronny WA-23Lutzenberger, Fabian WD-21

MMacCarthy, Bart WC-13Möst, Dominik TC-32Mahmassani, Hani MC-11Makarenko, Alexander TC-33, TD-33Makulska, Joanna M. TB-23Makuschewitz, Thomas WD-10Maliene, Vida WB-33Manger, Robert WB-05Manitz, Michael MA-42Mannini, Livia MD-11Mar-molinero, Cecilio WA-32Mareschal, Bertrand TC-39Maroto, Concepcion MD-26Martello, Silvano MB-42Masmoudi, Youssef TA-33, TB-33Mauricio, De Souza MC-10Mawengkang, Herman TD-17,

WA-17, TB-34, TC-34McComb, Sara WC-14

269

SESSION CHAIR INDEX EURO 25 - Vilnius 2012

Meca, Ana MB-24Melo, Teresa MA-17Mendoza, Jorge E. MA-07, TA-07,

WC-07Menezes, Mozart WD-18Mercadante, Luciano WA-35Meyer-Nieberg, Silja TA-42Michaels, Dennis MD-29Miettinen, Kaisa TA-38Miglierina, Enrico TD-15Milanic, Martin WB-08Milioni, Armando Zeferino MC-36Miralles, Cristobal WC-31, WD-31Moench, Lars MC-02, TA-13Molho, Elena WA-15Molina, Elisenda MB-31Monsuur, Herman WA-28Montibeller, Gilberto MA-27Morar, Ioan Radu WD-41Morton, Alec TA-27Motrenko, Anastasia TA-31, TC-31Mousseau, Vincent MA-01

NNasution, Azizah Hanim TA-34Nazarenko, Olga MB-10Nediak, Mikhail TA-40Nemeth, Sandor Zoltan TC-23Nikitina, Mariya WC-20Niroomand, Sadegh TB-11Nishide, Katsumasa MD-19Novo, Vicente TC-15, TD-15, WA-15Nunes, Ana Catarina TD-11Nunez, Marina MD-34

OOdegaard, Fredrik MD-44Onel, Selis TB-42Onoda, Takashi TA-31Oppen, Johan MC-07Orlov, Alexej WC-22Oulamara, Ammar TB-14Ozceylan, Eren TD-23Ozogur-Akyuz, Sureyya TC-44

PPaksoy, Turan TD-23Panella, Massimo MA-19Pang, King-Wah WD-26Papadimitriou, Dimitri MB-08Papamichail, K. Nadia MB-27Pasichny, Alexis WD-22, WA-41Paucar-Caceres, Alberto MA-32Pelegrin, Blas TC-18Pereira, Ana I. TA-15Pereverza, Kateryna WB-22, WC-41Perez-Marin, Ana Maria TB-19Pham Dinh, Tao WB-17, WD-17Pichler, Alois MA-43Pillac, Victor TA-07, WC-07Plociennik, Kai MC-38Podobedov, Vitaly WD-24Porschen, Stefan WA-08Prekopa, Andras MD-43Psaraftis, Harilaos MB-05Pukalskyi, Bohdan WC-41

Pulido Cayuela, Manuel A. MC-34

QQuintanilla, Sacramento MC-14

RRaa, Birger TB-45Radulescu, Marius WA-20Rönnqvist, Mikael WD-01, TC-08Ragwitz, Mario MD-25Ramirez, Hector WB-23, WC-23Rand, Graham MD-24Rauscher, Sandra TB-42Recht, Peter TD-05Reddy, Brian TD-39Reyer, Ivan TD-31, TA-44, TB-44Rodríguez Álvarez, Margarita TB-15Rodrigues, Antonio WB-24Rodriguez-Chia, Antonio Manuel

WB-18Rodriguez-Sanchez, Sara Veronica

MC-26

SSafia, Kedad-Sidhoum WA-

03, WB-03, WC-03, TD-09,WD-09

Sakalauskas, Leonidas WA-01Salazar González, Juan José TB-04,

WA-07Salo, Ahti TD-27, WC-30, MB-42Schöbel, Anita TD-12Schönberger, Jörn MA-11Schmidt, Marie TB-12Schultz, Rüdiger WB-43Schwindt, Christoph WB-02, TA-30Sciomachen, Anna TA-11Seidmann, Abraham TA-37Selosse, Sandrine TD-25Sevaux, Marc MD-

01, TB-01, WA-03, WB-03,WC-03, TD-09, WD-09

Sezer, Devin TC-42, WC-42Shafransky, Yakov TA-14Siegle, Markus TC-36, TD-36Siestrup, Guido WD-25Sindhya, Karthik TB-38Singh, Sanjeet WB-32Sinske, Marcel TB-28Sinuany-Stern, Zilla TB-36Slijepcevic-Manger, Tatjana MC-18Smith, Honora MA-33Smith, Leanne TB-37Soler-Dominguez, Amparo MB-30Soumis, Francois WB-04Soysal, Mehmet TB-25Sperandio, Fabricio TD-37Speranza, M. Grazia TE-01Stadtler, Hartmut MB-42Stålhane, Magnus WC-09Staube, Tatjana MA-33Stecke, Kathryn E. TA-35Stefani, Silvana MB-19Steinhardt, Claudius WC-45Sterna, Malgorzata MA-02Sternbeck, Michael MB-35

Stewart, Theodor TD-34Strijov, VadimWA-23, TD-44, WB-44Strusevich, Vitaly TB-02Stummer, Christian MD-27Swiercz, Aleksandra MB-37Syntetos, Aris TA-24, TD-24Szachniuk, Marta MC-37Szantai, Tamas MD-43

TTalasova, Jana MB-22Tammer, Christiane WA-38Tanaka, Tamaki TC-38Tansel, Barbaros TC-17Tanyeri, Basak TD-42, WC-42Tatlidil, Hüseyin TB-20Taylan, Pakize WD-35Teghem, Jacques MB-42Tejada, Juan MA-31Temocin, Büsra TC-42Thanassoulis, Emmanuel TA-36Theocharis, Zoe TC-24Tofallis, Chris TD-29Toledo, Franklina MC-13Tomasgard, Asgeir TA-43Topcu, Y. Ilker MC-10, TA-41Trafalis, Theodore WB-29Tragler, Gernot MD-04Traversi, Emiliano MB-29Tricoire, Fabien MA-06Trodden, Paul MD-30Tsoukiàs, Alexis TC-27Turkay, Metin MA-37Turkensteen, Marcel TB-08

UUdias, Angel MD-33, WD-38Uemura, Yoshiki WB-22

Vvan Ackooij, Wim MC-44van der Laan, Erwin TA-03Van Peteghem, Vincent WD-02van Vuuren, Jan WC-04Verbeeck, Cédric MD-06Vercamer, Dauwe WA-06Verleye, Derek TA-45Vespucci, Maria Teresa MC-43Viana, Ana WA-37Vieira, Cristina C. TB-06Vigna, Elena TA-19Vilutiene, Tatjana WA-33Visagie, Stephan WA-04Vitoriano, Begoña MC-33Volkovich, Zeev (Vladimir) WD-42Volling, Thomas WD-13Vrobel, Susie TC-28, TD-28

WWaegeman, Willem TB-40, WA-40,

WB-40Wäscher, Gerhard TA-01, TE-01Wallace, Stein W. TD-01Wang, Song TC-23Weber, Christoph TA-25

270

EURO 25 - Vilnius 2012 SESSION CHAIR INDEX

Weber, Gerhard-Wilhelm MC-01, WA-17, MA-21, MC-21, TC-22, TD-22, MA-23, MB-23, MC-23, MD-23, TB-23, TD-23, WD-28,TB-33, TD-33, TC-34, TD-34, MC-35, WD-35, WC-36, TB-39, TD-41, WA-42,WD-42, WC-44

Weber, Richard MD-20Wedley, William MD-41Weglarz, Jan MB-14Weistroffer, Heinz Roland TD-38Wenstøp, Fred WC-31Werner, Axel MD-08

Wheat, David WB-41Wiedenmann, Susanne WC-25Windeatt, Terry TC-44Winkelkotte, Tobias TD-35Wojciechowski, Pawel MB-37Wozabal, David TC-19Wright, Mike MB-06

XXavier, Adilson Elias MA-16Xufre, Patricia TB-24

YYalaoui, Farouk MC-10, TA-10Yan, Fei TA-44, TB-44Yang, Chih-Hai WC-32

Yatskiv, Irina TB-31Yerlikaya Ozkurt, Fatma WC-35Yildirim, U.Mahir TD-03Yue, Wuyi WC-44

ZZabih, Ramin TD-08Zadnik Stirn, Lidija WA-39Zaraté, Pascale WA-27, WD-27Zemkoho, Alain B. WD-15Zhang, Yi TC-16Zimmermann, Hans Georg WD-44Zolotova, Tatiana TB-21

Škrabic Peric, Blanka MA-18

271

AUTHOR INDEX

AA, Srinivas TB-37A. S. Castro, Ricardo MC-36A.C. Rocha, Ana Maria TD-19Aïssani, Djamil MD-08Abaev, Pavel TD-36Abastante, Francesca TC-27Abbaszadeh, Shahin TA-40Abdelouahab, Zaghrouti WB-04Abdul Rahim, Siti Khatijah Nor WC-

14Ablanedo Rosas, José Humberto TD-

06Abreu, Nair TB-05Absi, Nabil MD-06Achtziger, Wolfgang WB-15Adali, Yelda Cansu TC-10Adany, Ron WD-13Adekoya, Adebola MB-41Adenso-Diaz, Belarmino TC-35Adjiman, Claire WC-15Adler, Nicole TB-36Álvarez, David TB-09Álvarez-López, Alberto A. TC-10Álvarez-Miranda, Eduardo TC-04Öktem, Sükran MC-24, TD-34Öllinger, Michael TB-13Önder, Halil TA-26Önkal-Atay, Dilek TC-24Özçetin, Erdener TB-07Özcan, Ender TA-06, TD-09, WD-09,

TC-14Özdemir, Rifat Gürcan TC-10, TC-33Özmen, Ayse TB-23, WC-35Çalık, Ahmet WC-19Çatay, Bülent TD-03Çelebi, Dilay TC-44Çimen, Emre WA-29Çodur, Muhammed Yasin WA-24Ülkü, Tolga TB-36Ümit, Hakan WA-26Ünsal, Mehmet Güray WD-35Affenzeller, Michael MC-03Aflaki, Arian WC-10Afordakos, Orestis MB-32Afsar, H. Murat TD-09, WD-09Agarwal, Shweta TD-42Agatz, Niels MC-11Aggarwal, Remica WB-32Aggogeri, Francesco TA-42Aghayari, Hadi WD-06Aghdaie, Mohammad Hasan WC-33Aghezzaf, El-Houssaine TC-34Agra, Agostinho MA-06, MB-37Agustín, Alba WA-07, MB-16Ahi, Emrah TB-20Ahmed-Nacer, Mohamed TA-31Ahn, Heinz MC-36, TA-36Ailon, Nir TB-40Airola, Antti TB-40Airoldi, Mara TB-27Akal, Tülay WB-24Akartunali, Kerem TD-14Akcelik, Volkan WD-28

Akdogan, Yunus WB-19, WC-19,WD-19

Akehurst, Ron TA-32Akgün, Ibrahim TC-17Akpinar, Sener MC-06, MD-39,

WC-42Akrour, Riad WB-40Aksakal, Erdem WD-26, MD-41Aksen, Deniz MC-03Aktas, Emel MB-10, WB-10Aktin, Tülin MB-28Akyer, Hasan TC-07Akyurek, Zuhal WC-35Aladag, Zerrin WD-22, MA-40Alagoz, Ali WD-10Alain, Faye MA-08Alalawin, Abdallah TD-13Albareda Sambola, Maria WC-18Albayrak, Y. Esra MB-40Alberg, Dima MA-20Alcañiz, Manuela TB-19Alcaide Lopez de Pablo, DavidMA-38Aldasoro, Unai WA-43Aldea, Anamaria TD-20Alefragis, Panayiotis TD-09Alemany, Mareva TC-08Alemany, Sandra WD-29Alfandari, Laurent WA-03Alfano, Marina TC-28Alfred, Mbairadjim Moussa MB-12Algaba, Encarnación WB-12, MA-34,

WA-34Aliev, Iskander WD-05Alimohammadi, Amene MD-37Alimohammadi, Shohreh MD-37Aliouat, Makhlouf WC-44Alirezaee, Mohammad-Reza TA-36Alirezaee, Nassrin TA-36Aliyeva, Nigar WC-22Aljinovic, Zdravka WB-20, WD-21Allevi, Elisabetta MC-31, MA-43Allmendinger, Richard TB-30Almada-Lobo, Bernardo MA-

13, MC-13, MD-13, TA-30,TB-30, TD-37

Almagro, Jose MA-35Almaraz Luengo, Eduardo MD-21Almaraz Luengo, Elena MD-21Almeder, Christian WD-02, MB-13,

TB-30Almeida, Mariana WD-32Alonso Martínez, Maria Teresa MA-

09, MC-09Alonso, Federico MB-20Alonso-Ayuso, Antonio TA-29,

WD-38Alp Bilisik, Özge Nalan TB-41Alparslan Gok, Sirma Zeynep WB-34Alpern, Steve TB-18Alps, Ivars MD-38Alptekinoglu, Aydin TA-35Altefrohne, Maximilian WB-13Altindag, Ilkay WD-42Altinel, I. Kuban MC-08

Altiparmak, Fulya WC-18Altun, Adem Alpaslan WA-30Alumur, Sibel A. MC-17, TB-17Alvarez, Felipe WB-23, WC-23Alvarez, Fernando WC-09Alvarez, María Jesús TD-07Alvarez-Valdes, Ramon MA-09,

MB-09, MC-09Alvelos, Filipe MB-08, MC-08,

MA-13, WA-37Alves, Cláudio WC-07Alves, Diogo MA-13Alves, Luciene MC-36Alves, Maria João MA-38Alzate, Carlos WC-29Alzbutas, Robertas WA-33, TD-37Alzbutien, Giedr TD-37Alzorba, Shaghaf WA-38Amado, Carla WC-32, MD-35Amand, Guillaume TA-45Amann, Katharina MC-13, TC-25Amaro, Ana TB-30, WC-41Amaya, Ciro Alberto TC-11Ambrosino, Daniela TA-11Amen, Matthias TD-45Amini, Mehdi MD-27Amini, Mohammad TB-18Amorim, Pedro MA-13, TA-30,

TB-30Anderson, Chris MD-44Andersson, Henrik MD-05, WC-09Andrade, Jose luis TC-37Andreeva, Elena WD-36Andreeva, Galina WC-21, WC-30Andres, Carlos TC-35Andreu, Laura MD-18Angelelli, Enrico TB-04, TC-19Angelis, Vassilis MB-21Angulo-Meza, Lidia WC-32Anholcer, Marcin MC-11Anido, Carmen WA-44Anily, Shoshana MB-24Anjos, Miguel WA-16Ansell, Jake TA-16, WC-21Antczak, Maciej MD-37Antczak, Tadeusz TC-15Antunes, Carlos Henggeler MD-36,

WC-38Aoudia-rahmoune, Fazia WC-12Apanaviciene, Rasa WB-33Appa, Gautam TC-09ApSimon, Helen TD-32Apt, Jay TC-43Arana-Jiménez, Manuel TC-15Aranburu, Larraitz WA-43Aranda Almansa, Joaquin TB-09Arantes, Amílcar TA-08, TA-42Arapoglu, R. Aykut MD-06Aras, Necati MC-08, WD-28Araujo, Marcelo Tozo de WD-41Araz, CeyhunMD-02, MB-03, MC-35Archibald, Thomas TA-16, TB-42Arda, Yasemin TA-45Arellano-Garcia, Harvey MB-15

272

EURO 25 - Vilnius 2012 AUTHOR INDEX

Arenas-Parra, Mar MA-12Argyris, Nikolaos TA-27Arias, Alicia MC-36Arik, Ayse WD-20Arin, Arif MB-06Arin, Javier WB-06Arkhipov, Dmitry TD-02Armañanzas, Rubén WC-29Arneric, Josip MA-18, MB-21Aronsson, Martin WA-12Aros-Vera, Felipe TB-17Arthanari, Tiru WC-17Artigues, Christian WA-02, TD-09,

WD-09Asbjørnslett, Bjørn Egil MA-05Asgari, Nasrin MA-07Ashley, Mary V. MA-37Ashrafi, Mahdi MD-37Asif, Muhammad WA-35Askan, Aysegul WD-28Asmuss, Julija TD-36Asmuss, Svetlana MD-12Assad, Arjang MD-24Assis, Laura MA-25Assoumou, Edi TD-25, WA-25,

WB-25Aswal, Abhilasha WB-25, WC-34Atak, Erman MC-41Atalik, Gultekin WC-22Atasever, Ilknur MA-23Atasoy, Bilge MB-11Atkins, Jonathan TB-32Attia, Ahmed TA-10Attux, Romis TD-24Atygayev, Damir TB-22Aurbach, Doron WD-13Aussel, Didier WC-15Avanzi, Claudia MC-31Aviv, Yossi TA-35Avkar, Tansel TB-21Awasthi, Anjali MB-28, WD-40Ayala, Guillermo WD-29Ayayi, Ayi Gavirel TB-28Ayag, Zeki TC-33Aykan, Fulya WB-24Aytaç, Esra MC-41Azad, Md. Abul Kalam TD-19Azadegan, Aida TB-32Azevedo, Anibal TA-09, TB-33Azizi, Majid TC-41

BBabai, Hamed WC-45Babic, Zoran MC-45Babonneau, Frédéric TD-07, TD-16Bachmat, Eitan MC-28Badin, Luiza TD-20Baek, Jun-Geol WD-29Baena Mirabete, Daniel TD-30Bürgi, Markus MD-28Bürgy, Reinhard TC-02Büsing, Christina TB-12Baes, Michel MD-29Baesens, Bart TB-31Baesler, Felipe WC-11

Baghban Asghari nejad, HoomanWC-27

Bagheri, Masoud TB-06Bagirov, Adil MC-16, TB-31Bahbouh, Radvan MD-12Bahn, Olivier TD-16Bahovec, Vlasta MB-33Baidya, Tara Keshar Nanda TC-20Bailly-Bechet, Marc WA-09Bajbar, Tomas MC-28Bakhrankova, Krystsina TA-30Balaisyte, Jurgita WC-10Baldo, Tamara MD-13Balestrassi, Pedro Paulo TD-29Balezentis, Alvydas WD-40Balezentis, Tomas WD-40Balin, Abit TB-41Ballerstein, Martin MD-29Ballestero, Enrique MC-19Ballestin, Francisco WB-02, MC-14Banal Estanol, Albert TA-25Banciu, Mihai MD-44Banker, Rajiv WD-32Bao, Truong Q. MB-38Baptiste, Pierre TB-14Baraçlı, Hayri TB-41Baranov, Oleksii WC-05Barbósa-Póvoa, Ana Paula TA-10,

TA-30, TB-30Barbarosoglu, Gülay WD-28Barbati, Maria WC-18Barber, Nick WC-37Barceló, Jaume MB-11Barcus, Ana MA-27Bardadym, Tamara TC-22Bareisis, Arnas TA-21Bargiela, Andrzej TC-14, WC-14Barjesteh, Saeid WD-06Barkalov, Konstantin WA-23Barketau, Maksim MA-02, TA-02Barnhart, Cynthia MA-04Barrena, Eva WB-12, MD-17Barros, Ana Isabel WA-28, WB-28Barthelemy, Thibaut MC-03Bartke, Philipp WB-45Bartl, David WB-16Barton, Paul I. TC-29Barysavets, Tatsiana TA-05Barz, Tilman MB-15Bashiri, Mahdi TB-06, TB-39Baskaran, Geetha TC-14Basoglu, Ismail MC-18Bastert, Oliver MA-30Bastic, Majda MC-24Bastings, Ingrid WC-28Basu, Sumanta WC-34, MB-45Batista dos Santos, Lucelina TC-15Batmaz, Inci WB-24, WC-35Batta, Rajan TA-18Battaïa, Olga TD-13, TB-45Battarra, Maria WD-07Battiti, Roberto WA-40Baucells, Manel MC-27Baumann, Philipp TA-30Baumbach, Birk MC-07Baumung, Martin TC-17

Bautzer, Ana MA-08Bawa, Daljit Singh WC-14Baykasoglu, Adil WA-11Bayoglu, Levent MD-02, MB-03,

MC-35Bayraktar, Demet TC-44Baysal, Gokce MA-28Bazan, Peter TC-13Bastürk, Nalan WC-26Beasley, J. E. WB-20Becejski-Vujaklija, Dragana WD-27Becker, Jörg TD-24Beek, Onne MB-07Bekiaris, Michail MA-18Bektas, Tolga TD-03Bektas, Zeynep MD-30Beleggia, Beatrice MD-42Belin, Bruno WB-25Bell, Michael TA-33Bell, Peter MD-44Bellanger, Adrien TB-14Bellenguez-Morineau, Odile WB-02Bello Rivas, Juan M. MC-15Belo Filho, Márcio Antônio Ferreira

TB-30Belov, Gleb TA-09Belovas, Igoris MA-30Belton, Valerie TC-27Beltran-Royo, Cesar MA-35Ben Ameur, Hachmi WA-20Ben-ameur, Walid WD-05Benati, Stefano WB-18, WB-34Benavides, Alexander J. WC-03Benavides, Julián WD-27Bendotti, Pascale MC-44Benediktovich, Vladimir TB-05Benedito, Ernest WA-10, MD-30Benezech, Vincent WB-12Benito, Antonio MC-19Benmansour, Rachid MB-02Benoist, Thierry MD-03, WB-03,

WB-35, MC-42Bentaha, Mohand Lounes TB-45Berbig, Dominik TC-07, WB-13Beresnev, Vladimir WB-15Berezovskyi, Oleg TC-22Bergantinos, Gustavo MA-24Berger-Wolf, Tanya MA-37Berghman, Lotte WC-16Bergiante, Nissia MD-35Berk, Erhan TC-41Berlinska, Joanna MA-02Berman, Oded TA-18Bernaert, Korneel TD-24Bernardo, Giovanni TD-32Berndt, Rüdiger TC-13Bernstein, Fernando MB-24Berrachedi, Abdelhafid TD-05Bertazzi, Luca TD-13Bertocchi, Marida MA-43, MC-43,

TA-43Berzins, Lorena MB-41Besikci, Umut WC-02Besiou, Maria WC-10Best, Tom WD-37Betts, Gavin TB-32

273

AUTHOR INDEX EURO 25 - Vilnius 2012

Bezirci, Muhammet WD-10Bhatia, Neha WD-23Bianchessi, Nicola TB-04Bianchi, Monica TD-15Bianchi-Aguiar, Teresa MC-09Bielak, Jacobo WD-28Bielza, Concha WC-29Bierlaire, Michel MB-11, WD-12Bierwirth, Christian MC-02Bijak, Katarzyna MC-20Bilbao, Jesus-Mario MA-34Bilbao-Terol, Amelia MA-12Bilge, Ümit WC-02Bilgen, Bilge TC-08Bilgic, Taner TC-10Bilgin, Ayse WD-12Billaut, Jean-Charles TB-14, WA-30Billio, Monica MC-21Billionnet, Alain MC-29Bing, Xiaoyun MC-17Bisdorff, Raymond TC-39Biswal, Gayatri WC-38Björk, Folke WC-33Bjork, Kaj-Mikael MA-22Blair, Graeme TA-16Blanco, Luis MA-35Blanco, Víctor TB-18Blanquero, Rafael TC-18Blazewicz, Jacek WD-26, MB-37,

MD-37Blockmans, Tom WB-34Bloemhof, Jacqueline MC-17, TB-25Bludau, Bastian MC-38Boaventura-Netto, Paulo OswaldoTB-

06Boccato, Levy TD-24Bock, Stefan MB-02Bode, Claudia TC-11Bohlin, Markus TB-03Bohman, Samuel WC-38Bolance, Catalina TB-19Boland, John MD-16, TD-16Boley, Mario TD-31Bolshakov, Vitaly MC-03Bolshoy, Alexander WC-26, MC-37Bondareva, Mariya WC-34Bonenti, Francesca MC-31Bonnel, Henri WD-15Bont, Leo MD-26Borbely, Antal TC-28Borges, José TD-37Borgwardt, Steffen TB-13Borndörfer, Ralf WD-04, WA-12Borodin, Dmitry WB-04Boros, Endre WB-08, WC-08Bortfeldt, Andreas MA-09Bortz, Michael MC-38Bosco, Adamo TC-11, TD-13Bosi, Gianni WC-34Bosso, Sandro MD-42Bottero, Marta TC-27Bouckaert, Stephanie WB-25Boudjeloud-Assala, Lydia WB-17Boudries, Abdelmalek WC-44Bouhtou, Mustapha MB-45Bouman, Paul TC-12

Bounkhel, Messaoud TA-23Bourgeois, Nicolas WA-09Bourlakis, Michael MB-10Bouroubi, Sadek TA-28Bourreau, Eric WA-05, TD-09,

WD-09Boutsinas, Basilis TA-17Bouvry, Pascal WD-26Boyacı, Burak TD-18Boylan, John TA-24Boysen, Nils MD-02Bozkaya, Burcin MB-17Bozkurt, Levent MC-35Brabazon, Philip WC-13Bracke, Martin WA-35Brady, Niall WD-45Brînzei, Nicolae TD-36Bräysy, Olli MB-07Brailsford, Sally WC-37, WD-37Branco Costa, Antonio MB-28Branco, Antonio C. S. WC-27Branda, Martin TB-43Brandao, Jose TC-06Brandt, Felix WA-03, WC-07Brauner, Nadia TB-02Bravo, Cristian MC-20, WD-28Bravo, Mila MC-19Bravo, Raquel WA-05Brazis, Viesturs TA-03Bregar, Andrej MC-45Brekelmans, Ruud MD-42Brennan, Alan WD-31Brenner, M. Harvey WD-36Breschan, Jochen MD-26Breuer, Claudia WD-25Breun, Patrick TD-32Brezina, Ivan MD-03Brieden, Andreas TB-13Briens, François TD-25Briskorn, Dirk MB-02, MD-02Brison, Valérie TA-20Brodnik, Andrej TD-14, MD-15Broekmeulen, Rob MB-35Bronshtein, Efim WC-20Brotcorne, Luce MB-45Brozova, Helena MB-22Brugha, Cathal MD-32, TA-32Brunelli, Matteo MD-12Bruni, Renato MB-20, WB-20Bruno, Giuseppe TA-18, WC-18,

WC-41Buccheri, Rosolino TC-28Buchheim, Christoph MB-29Buchheit, Grégory TD-36Buer, Tobias MA-11Bukauskas, Feliksas MA-37Bukhsh, Waqquas Ahmed MD-30Buljubasic, Mirsad WB-03Bulut, Basak WD-20Bulz, Nicolae WC-36Bunte, Stefan WA-02Buoni, Alessandro MB-22Burachik, Regina MD-23Burger, Alewyn WC-04Buriol, Luciana S. WC-03

Burkhardt, Thomas TC-21, TD-21,WA-21

Burnak, Nimetullah MA-28Burnaz, Sebnem MC-10Burr, Jennifer TA-32Bursa, Nurbanu MA-33Buscher, Udo MC-02Buskens, Erik TD-37Butelle, Franck WA-03Button, Kenneth TD-29

CCañal, Verónica MA-12Cañavate Bernal, Roberto Javier MD-

17Cabal, Helena WD-38Caballero, Rafael TC-06Cabo Nodar, Marta WD-07, WB-39Caceres Cruz, Jose MB-16Cadarso, Luis TD-12Caetano, Sabrina WB-30Cáceres, Ma Teresa TD-18Cafieri, Sonia TA-29Cailloux, Olivier TD-39, TC-40Calafiore, Giuseppe WC-05Cales, Ludovic MC-21Camanho, Ana TD-29, TD-31,

MC-36, MD-36Camargo, Ricardo MB-08, TC-17Camargo, Victor MC-13Cambini, Alberto MC-23Camiz, Sergio MC-37Campigotto, Paolo WA-40Canca, David WB-12Cangalovic, Mirjana WB-09Cano, Javier MD-33, WD-38Cantarella, Giulio Erberto MD-11Cao, Qi TD-37Cao, Wenhua MC-37Capek, Roman WA-11Captivo, Maria Eugénia MC-06Cardoso Dias, Bruno WC-11Cardoso, Domingos TB-05, WB-16Caremi, Claudio MD-42Cariou, Pierre MA-05Carmo, Jose Luis TB-24Caro, Felipe TB-25, MC-42Carosi, Laura MC-23Carotenuto, Pasquale TB-11Carpente, Ma Luisa MA-35Carrasco, Miguel WB-23Carrasco-Gallego, Ruth WC-10Carravilla, Maria Antónia MB-09,

MC-09, WD-31Carreras, Ashley MC-32Carreras, Francesc MC-34Carrese, Stefano WB-14Carrión, Miguel TB-35, MC-43Carrizosa, Emilio TC-18Carroll, Paula MA-08Carvalho, Lilian MD-25Casey, Brad WC-12Casquilho, Miguel WC-41Castaño Cabrales, Jose Luis MD-33Castanier, Bruno WB-21Castro, Alberto MB-08

274


Castro, Javier MA-31Castro, Jordi TC-30, TD-30Castro, Marco WD-07, WA-14Cattrysse, Dirk WD-07, TB-12Caulkins, Jonathan WC-01Cavique, Luís TA-31Cayla, Jean-Michel TD-25Cerdeira-Pena, Ana MA-35Cerqueti, Roy MA-19Cesar, Mihael WD-25Cesarone, Francesco MB-20, WB-20Cevik Onar, Sezi WB-10, TA-41Cexus, Jean-Christophe WD-17Chaabane, Amin MD-10Chaichiratikul, Pairoj TD-04Chaing, Hui-Min MD-40Chakraborty, Abhishek WB-10Chakraborty, Soumyakanti WC-34Champion, Thierry WB-23Chandra, Shailesh WD-03Chang, Chiao-Chen TA-31, TD-31Chang, Dong Shang TC-25Chang, Jen-Chia MD-39Chang, Kuochung MA-39Chang, Pao-Long MC-40Chang, Tsung-Sheng TB-03Channon, Andrew Amos WD-37Chanpum, Montathip WC-12Chaovalitwongse, W. Art MA-37Chardy, Matthieu MA-08, MC-08Charousset-brignol, Sandrine MC-44Chassein, André MB-02Chatterjee, Ashis WB-10Chauhan, Satyaveer S. MB-28Chaves, Antônio TA-09, TB-33Cheaitou, Ali MA-05Chelouah, Rachid TA-06, TA-31Chen, An TA-19Chen, Bo MA-34Chen, Chie-bein MC-39Chen, Frank Y TB-10Chen, JongChen MC-15, TB-31Chen, Pei-Ting MD-39Chen, Sheu-Hua WB-27, WC-44Chen, Vivien Y.C. MA-40, MD-40Chen, Yuh-Wen TC-37Cheng, Chun-Yuan TC-45Cheng, Weiwei WB-40Chernichovsky, Dov TB-36Chiang, Naiyuan TD-43Chiang, Yu-Min MB-40, MC-40Chien, Hung-ju MA-40Chilcott, Jim TA-32Chin, Yang-Chieh TA-31, TD-31Chiwiacowsky, Leonardo TC-06Cho, Minyoung WD-29Choi, Yuran WD-18Choo, Eng MD-41Chou, Chun An MA-37Chou, Ying-Chyi MC-40Choudhary, Rekha TA-28Chowdhury, S.M.Atiqur RahmanWC-

16Christensen, Tue TB-04Christiansen, MarielleMA-06, WC-09Christie, Marc WB-25

Christofides, Nicos TC-21Chuang, Yen-Ching MB-41Chub, Igor MD-09Chub, Olga WC-11Chung, Chia-Shin MD-10Church, Richard MD-26Churilov, Leonid WA-27Chutani, Anshuman MA-45Ciaffi, Francesco TA-11Cichenski, Mateusz MA-02Cileg, Marija WB-10, MD-18Cimler, Richard MB-12Cincyte, Rasa WB-41Cioban, Liana WD-23Cipriani, Ernesto MD-11, TA-11Ciric, Zoran WB-10Civinova, Jitka MA-36Cizmesija, Mirjana MB-33Claassen, G.D.H. (Frits) MB-13,

MC-26Clímaco, João WC-31Clark, Alistair TD-30, MC-33Clarke, Nancy WC-39Clausen, Uwe TB-03, WB-12Clautiaux, François WC-03, WC-07Cleophas, Catherine WB-45Cobre, Juliana WB-30Coelho Silva, Ricardo MA-12Coffman, Ed MA-14Cogurcu, M.Tolga TC-31Collan, Mikael MC-12Collins, Karen WD-31Colombani, Yves MA-30Comellas, Jaume MB-08Comes, Tina TD-20, WD-27Conejo, Antonio J. MC-43Consigli, Giorgio WD-43Consiglio, Andrea MC-18Constantino, Miguel WA-37Contardo, Claudio WC-18Contreras, Ivan TB-17Contreras, Javier TB-16Cookson, Emily MA-44Corbett, Charles TB-25Cordeau, Jean-François MD-05Cordeiro, Clara TD-24Cordeiro, Isabel TC-20Cornejo, Oscar WC-11Corominas, Albert WC-41Coronado-Hernandez, Jairo MC-45Correia, Gonçalo MA-04Correia, Isabel MA-17, TB-17Cortazar, Gonzalo MB-20Cortinhal, Maria TD-11, TA-17Cosgun, Ozlem MB-26Coskun, Semih TC-03Costa, Alysson WD-31Costa, Anabela MA-21, TB-21Costa, Fabricio TD-10Costa, João Paulo MA-38Costa, Julian MA-24Costa, M. Fernanda P. MD-15, TA-15Costa, Marie-Christine MA-08Costa, Sueli MD-09Coti, Camille WA-03Couellan, Nicolas WB-29

Coutinho-Rodrigues, Joao WB-27Couzoudis, Eleftherios MD-28Coves, Anna M. TD-02Crainic, Teodor Gabriel MB-06,

WC-07, TC-12, WC-18Crama, Yves MA-34Creput, Jean-Charles TA-06Crevits, Igor WA-06Crispim, José TD-07Crook, Jonathan MC-20, WC-30Cruz Corona, Carlos MA-12Cruz, F. R. B. TC-36Csetnek, Ernö Robert WD-23Csizmadia, Zsolt WD-16Cuesta, Jordi TC-30Cunha, João TD-31Cyrino Oliveira, Fernando LuizTB-24

DD’Acierno, Luca WB-12D’Aleo, Giuseppe MD-42D’Alessandro, Simone TD-32D’Ambrosio, Claudia MC-29D. Nasiri, Saeideh WC-16Dadenne, David WC-27Daduna, Joachim R. TC-03Díaz-Madroñero, Manuel MA-12Döyen, Alper WD-28Dagdeviren, Metin WD-26, TD-33,

MD-41Dahlberg, Hans WA-12Dahms, Florian TB-03Damay, Jean WA-12Dambreville, Frédéric WC-17Daneshvar, Sahand WA-32Dangaard Brouer, Berit MC-05,

WC-09Dangerfield, Brian WB-41Danilkina, Alexandra TD-36DanisOncul, Selmin WB-14Dao Thi, Thu Ha TC-36Daou Pulido, Amir José MD-25Dargam, Fatima WC-27Darlay, Julien WB-03, WB-35Dasaklis, Tom TC-25, TB-37Dasgupta, Bhaskar MA-37Daskalaki, Sophia TB-20, WA-24Datta, Subhash MB-33de Araujo, Silvio TD-10De Asmundis, Roberta WC-29De Baets, Bernard WA-40de Carvalho, J. M. Valério WC-07De Causmaecker, Patrick TC-14De Cosmis, Sonia WA-29De Cuypere, Eline TC-45de Freitas Rodrigues, Rosiane WC-11de Graaf, Gimon TA-37de Joode, Jeroen MA-26de Kok, Ton WA-43de la Fuente, Hanns WD-32de la Torre, Rocio MD-30De Leone, Renato MD-16, WA-29de los Santos, M. Antonia TD-02,

MD-11de Miranda Borba, Leonardo WC-03De Rosa, Vincenzo TD-35

275


de Ruijter, Tom WA-40de Silva, Dileep WD-37De Smet, Yves TB-39de Sousa, Amaro TC-04, MA-08,

MB-08, MC-08De Turck, Koen TC-45de Villiers, Anton WC-04De Wolf, Daniel TC-29Deb, Kalyanmoy TB-38, MA-41Debnath, Subhrajit MA-10deCampos, Teofilo TA-44Decclesia, Rita MA-19Degel, Dirk TA-17Dehouche, Nassim WC-26Dejaeger, Karel TB-31Dejax, Pierre TB-45Dekker, Rommert WA-07Del Pia, Alberto MD-29Delahaye, Daniel TA-04Delgado Pineda, Miguel TB-09,

TD-15Delgado, Alberto WC-03, MD-05Delgado, Alexandrino MA-06Delgado-Gomez, David MD-17Della Croce, Federico WA-30Dellaert, Nico WA-43Deman, S WD-06Dembczynski, Krzysztof WB-40Demenkov, Max MD-15Demir, Alparslan Serhat TA-08Demir, Emrah TD-03Demircan, Ece Arzu MA-42Demirel, Neslihan MC-17Demirovic, Emir WB-03, WA-05Dempe, Stephan WB-15, WD-15den Hertog, Dick MD-42Densing, Martin MA-43Dereli, Turkay WA-11Dereniowski, Dariusz MA-14Desaulniers, Guy TB-01Deshmukh, Abhijit MB-30Despotis, Dimitris MB-36DeTombe, Dorien WA-36Deveci Kocakoç, Ipek MA-28Dewilde, Thijs TB-12Dhara, Anulekha TC-23Dhungana, Meghana MA-44Di Maggio, Andrea TC-27Di Pace, Roberta MD-11Diarrassouba, Ibrahima TD-13Dias, Luis C.MB-27, MC-27, MD-36,

TC-40Diaz, Javier MC-02, MC-15Diekmann, Sven WC-31Diepen, Guido WB-35Diepold, Dennis WD-21Dierkes, Stefan TD-45Dietz, Chris WA-34Dimaki, Katerina MB-21Dinç, Fehmi Evren WD-22Dindarian, Azadeh TB-25Diner, Oznur Yasar WD-39Dios-Palomares, Rafaela MA-38Dirksen, Jakob MC-05Divnic, Tomica TC-05Divsalar, Ali WD-07

Djeffal, Lakhdar WA-14Djordjevic, Milan MD-15Dmitruk, Andrei TA-23Doan, Phi TA-04Dochow, Robert WD-20Doerner, Karl WA-07Dogu, Eralp MA-28Dokka, Trivikram TC-09Dolgui, Alexandre TA-02, TD-13,

TB-45Dolgun, Leman Esra MA-28Doll, Claus MB-04Domínguez, Oscar WA-07Domingo, Juan WD-29Domingo-Ferrer, Josep TC-30Dominguez, Enrique WB-18Dominguez, Ruth MC-43Domonkos, Tomas MB-25Donduren, M. Sami TC-31Doorman, Gerard TA-43, TC-43Doran, Des WB-10Dorsch, Dominik MD-28Dorta-González, Pablo TC-18Dragan, Irinel MA-24Drapkin, Dimitri WB-43Driessel, Rene TA-13Drigo Filho, Elso WD-41Drinias, Theodor WA-24Drozdowski, Maciej MA-02Du Toit, Tiny WD-29, TA-31Dubedout, Hugues WB-03Dubois-Lacoste, Jérémie MA-41Dudin, Alexander TC-42Dudouit, Renaud WA-25Duenas, Alejandra TC-14Duer, Mirjam WA-16Dufour, Francois TC-16Duginov, Oleg TC-05Dullaert, Wout MB-07, TB-45Duman, Ekrem TC-19Dumicic, Ksenija WC-21Dundar, Abdullah Oktay WD-10,

WA-30Dundulis, Gintautas WA-33Dunford, Charlotte MD-32Dupacova, Jitka TD-01, TB-43Dupin, Nicolas MC-44Duque, Joaquim R. WB-26Durán, Guillermo WB-08Durbach, Ian MA-27Durkan, Mehmet WA-28Durneac, Cristina-Petronela WB-32Dytczak, Miroslaw TA-14Dzaja, Josipa WD-21Debczynski, Marek TA-14

EEckhardt, Alan WB-26Edinger Munk Plum, Christian MC-

05, WC-09Edwards, Neil TD-16Eglese, Richard TD-03Eguia, Ignacio TD-07Ehrgott, Matthias WC-14, WA-38Eijgenraam, Carel MD-42Eiselt, H.a. TA-17

Eisenstein, Don WD-37Ejov, Vladimir TD-04Ekenberg, Love TD-27Ekici, Ali MC-03Ekinci, Cumhur TD-31Ekinci, Yeliz TB-34Ekmekci, Esin MC-10El Faouzi, Nour-Eddin TA-29EL Haj Ben Ali, Safae TB-18El-Borai, Mahmoud TD-41El-Ga’aly, Ashraf MA-10El-Said El-Nadi, Khairia TD-41Elhallaoui, Issmail WB-04Elloumi, Sourour MC-29Elmi, AtabakMC-06, WA-11, MD-39,

WC-42Emde, Simon MD-02Emel, Erdal TB-06Emet, Stefan WA-05Emiris, Dimitrios MA-11, WC-45Emmenegger, Jean-Francois TA-22Emrouznejad, Ali TA-36Engell, Sebastian WA-30Epstein, Leonardo TB-10Ercan, Secil MD-30Erdin Gundogdu, Ceren MB-17Erdogan, Gunes WD-07Erdogan, Sevgi TD-23Erjavec, Natasa MB-33Erkut, Erhan MB-01Ernst, Andreas MA-17Erol, ISil WC-21Ersoy, Cem MC-08Escamilla, Carmen G. TB-18Escribano Ródenas, María del Carmen

TA-39Escudero, Laureano FernandoMD-17,

TA-29, MC-43, WA-43Esen, Omur WC-44Eshragh Jahromi, Ali WA-26, WB-37Eskelinen, Juha TD-29, WB-30Eski, Ozgur MD-02, MB-03, MC-35Esmaeeli, Amin WC-27Espejo, Inmaculada WB-18Espinoza, Daniel WA-42Espuña, Antonio TB-02Esquirol, Patrick WA-30Estellon, Bertrand WB-03, WB-35Euler, Reinhardt TC-09Everett, Jim WD-44Evers, Lanah WA-28Ewe, Hendrik MC-38

FFacchi, Giancarlo MD-04Faco’, Joao Lauro D. TD-26Faezipour, Mehdi TC-41Fagerholt, Kjetil MA-05, MD-05Fagiano, Lorenzo WC-05Fahimnia, Behnam WA-26Fahmi, Fidan WB-24Faias, Jose TB-24Fakharzadeh, Alireza MC-04, TB-15Falbo, Paolo MA-19, MB-19Falke, Lutz WA-08Fanghaenel, Diana WC-15

276


FarajiDavar, Nazli TA-44Farid, Suzanne TB-30Fasano, Giorgio MB-09Fasth, Tobias TD-27Fathi, Masood TD-07Fattahi, Ali TC-03, TB-11Faulin, Javier WA-07, MB-16Fayzur Rahman, Dewan MA-25Fedotova, Kristine WA-33Fedrizzi, Mario MB-22Fedrizzi, Michele MD-12Feillet, Dominique MD-06Felici, Giovanni WC-08Felix, Bastian TC-32Fellag, Hocine MD-33Felletti, Daniele MB-19Fenerci, Kübra TC-19Ferhat, Arezki MD-34Fernández Barberis, Gabriela TA-39Fernandes, Edite M.G.P. MD-15,

TA-15, TD-19Fernandes, Luís WD-15Fernandez, Elena TB-17Fernandez, Jose TB-18Fernandez, Pascual TC-18, TD-18Fernandez-Viagas Escudero, Victor

MC-14Ferreira Filho, Virgilio Jose Martins

WB-42Ferreira, Carlos TD-11Ferreira, Luis WB-06, WC-06Ferreira, Pedro TB-24Ferreira, Ubirajara WC-38Ferrer, Albert MB-16, MC-16Ferretti, Valentina TC-27Fertig, Emily TC-43Fescioglu-Unver, Nilgun MA-42,

TD-42Fetta, Angelico TB-42Fiala, Petr WC-45Fichtner, Wolf TB-35Fidelis, Krzysztof MD-37Fiems, Dieter TC-45Fiestras-Janeiro, Ma Gloria MB-24Figueira, Gonçalo MA-13Figueiredo, Rosa Maria MB-37Filar, Jerzy TC-16Filippi, Carlo TB-04, MA-06Finbow, Stephen WD-39Finke, Gerd TB-02Finta, Lucian WA-03Fioole, Pieter TD-12Firat, Murat MB-02Firmansyah, Firmansyah TC-34Fischer, Frank WC-05Fischetti, Matteo WB-01Flatberg, Truls TC-12, TA-30Fliedner, Malte MD-02Fliedner, Thomas WC-02Flier, Holger TB-03Flisberg, Patrik WA-13, MD-42Flor, Guerrero WB-21Florentino, Helenice TD-10Flores, Ramon MB-31Floudas, Christodoulos MC-29Flynn, James MD-10

Fodstad, Marte MA-26Fondacci, Rémy TA-29Fonseca, Carlos M. TB-06Forjaz, Maria WD-15Forsgren, Malin WA-12Fortemps, Philippe TD-40Fortz, Bernard MA-08Fotopoulos, Christos MB-10Fourer, Robert WD-33Fourneau, Jean-Michel TC-36Frédéric, Dambreville WD-17Fröhling, Magnus TD-32Framinan, Jose M. TD-02, WB-11,

MC-14França, Felipe MA-16Franc, Vojtech WA-23Francfort, Stanislas MA-08, MC-08Franco, L. Alberto MA-27, MA-32,

MB-32Frangioni, Antonio TC-30Frankowicz, Marek TD-33Frankowska, Helene MD-04Frazzon, Enzo WD-10Frei, Oleksandr WB-44Freixas, Josep MC-34Friedman, Lea TA-36, TB-36Fries, Carlos Ernani WB-30Frisk, Mikael WA-13, MD-42Fukuda, Yutaka TA-04Fukushima, Masao TB-15, 16Fulga, Cristinca TB-20Furems, Eugenia TA-20Furió, Salvador TC-35Furmans, Kai TC-07, WB-13

GG. Hernandez-Diaz, Alfredo TC-06G. Nejad, Mohammad MD-27Gabay, Michaël TB-02, WA-03Gabor, Adriana F. WA-07Gabriel, Steven TA-25, WA-26Güler, M. Güray TC-10Günther, Elisabeth TB-12Günther, Markus MD-27Güran, Celal Barkan WA-20Gürbüz, Tuncay MB-40Gaertner, Thomas TD-31Gökçen, Hadi MC-17Gönsch, Jochen WA-45Gössinger, Ralf WC-42Göttlich, Simone MB-15Gaidamaka, Yuliya TD-36Galafassi, Cristiano TC-06Galindo, Gina TA-18Gallay, Olivier WD-45Gallegos, David MD-11Galli, Laura MA-29Gallien, Jérémie MC-42Gallivan, Steve WC-37Gallo, Giorgio TD-32Gallo, Mariano WB-12Galvão Dias, Teresa WD-03, MD-36Gamboa, Dorabela TD-06Gamrath, Gerald WC-05Ganas, Ioannis MC-06Ganobjak, Michal MB-25

Garín, María Araceli WA-43García Alvarado, Marthy StívalizMD-

10García Quiles, Sergio WB-18García Sipols, Ana Elizabeth TD-42García, Diego WD-38García, María D. TD-18García-Jurado, Ignacio MB-24Garcia Centeno, Maria Del Carmen

TA-39Garcia Lopez, Juan Manuel WC-41Garcia-Flores, Rodolfo MA-17Garcia-Laguna, Juan MD-10Garcia-Sabater, José P. MC-45Garcia_Perez, Fernando TB-09Gardi, Frédéric WB-03, MA-35,

WB-35, MC-42Garfinkel, Robert TC-30Gargallo, Pilar MD-18Gass, Saul MD-24Gastaldi, Massimiliano MB-11Gaudutis, Ernestas WA-33Gavalec, Martin MB-12, TC-13,

MA-22Gavranovic, Haris WB-03, WA-05Gawiejnowicz, Stanislaw MD-14,

TA-14Gawron, Piotr WC-03, MB-37Gebhard, Marina TD-35Gee, Deborah TA-16Gehring, Hermann MA-11Geiger, Martin Josef MB-03, MC-03Geipele, Ineta MA-33, WA-33,

WB-33Geipele, Sanda WA-33, WB-33Geissler, Bjoern TB-29Gel, Yulia MD-16Geldermann, Jutta MC-13, TC-25,

WC-25, WB-39GenÇ, Asır WB-19Gendreau, Michel WC-07, WA-43Gendron, Bernard TB-04Genova, Krassimira WB-27Genovese, Andrea WC-41Gentile, Claudio TC-30Georgiadis, Michael WD-02, TA-25Geraldo, Diego MC-36Gerdessen, J.c. MB-13Geroliminis, Nikolas TD-18Gesteira, Claudio MA-16Gestrelius, Sara WA-12Gevezes, Theodoros MB-37Ghattas, Omar WD-28Ghiani, Gianpaolo TD-13Ghodsi, Reza WB-13Ghosh, Diptesh MC-06Giandomenico, Monia MA-29Gianfreda, Angelica MB-19Giannikos, Ioannis TA-17Giannopoulou, Maria WC-45Giffhorn, Edilson TB-32Gil-Lafuente, Ana Maria TB-19Gila Arrondo, Aranzazu TB-18Ginda, Grzegorz TA-14Ginestar, Concepción MD-26Glass, Celia TA-02

277


Glazebrook, Kevin TB-42, MA-44Glorie, Kristiaan WA-28Gnudi, Adriana MC-31Goberna, Miguel TA-15, TB-15Goddard, Wayne TD-05Godfrey, Patrick MD-32Godinho, Maria Teresa TC-04Goel, Asvin MA-07Goeppel, Mathias MA-04Goerdt, Andreas WA-08Goerigk, Marc TB-12Gogos, Christos TD-09Goldengorin, Boris WA-17Golikova, Victoria TA-22Gollmer, Ralf MD-29Gollowitzer, Stefan TB-04Gomes Júnior, Silvio WC-32Gomes, A. Miguel MA-09, MB-09,

MD-09Gomes, Christina MC-37Gomes, Gastão MC-37Gomes, Luiz F. Autran M. TA-20Gomes, Marta Castilho WB-26Gomes, Rui WD-03Gomez Gonzalez, Daniel MA-31Gomez Padilla, Alejandra WB-32,

WC-40Gomez, Arthur WB-04, TC-06,

WA-06Gomory, Ralph E. WE-01Gonçalves, Graça TA-26Gonçalves, José Fernando TA-09Gondzio, Jacek TB-07, WC-16Gontis, Vygintas MD-18Gonul, Mustafa Sinan TC-24González Alastrué, José Antonio TD-

30González-Arangüena, Enrique MA-

31, MB-31González-Császár, Eduardo TB-10Gonzaga, Clovis WB-16Gonzalez-Araya, Marcela TD-26Gonzalez-Brevis, Pablo WC-16Gonzalez-R, Pedro L. TC-37Goos, Peter WD-07, WA-14Goossens, Dries MB-18Gorczyca, Mateusz TC-02Gordini, Angelo MD-42Gorelik, Viktor WB-04Gori, Stefano MD-11Gorobetz, Mikhail MD-38Gorokhovik, Valentin MB-23Goto, Yusuke MC-45Gotthardt, Benjamin WA-24Goumoto, Kensuke TB-15Gouveia, Luis TC-04, MA-08, TA-26Gouveia, Maria MD-36Goverde, Rob TC-12Grace, Adam TA-42Grad, Sorin-Mihai MB-38Gradisar, Miro WD-25Gröflin, Heinz TC-02Gram, Markus WA-10Granado, Bertrand TA-06Grandjean, Arnaud MC-44Granichin, Oleg WA-42

Granot, Daniel MA-24Grasas, Alex MB-16, TA-35Grass, Dieter MC-04Grebennik, Igor WC-05Gregory, Amanda TB-32Gribkovskaia, Irina TA-05, WC-09Griebsch, Susanne MB-18Griffiths, Jeff WA-37Grigorenko, Olga WB-22Grimaud, Frédéric TB-45Grinshpoun, Tal WB-14Grisar, Catherine MD-45Gritzmann, Peter TB-13Grmusa, Jeca WA-37Groenevelt, Harry TB-10Groissböck, Markus MA-25Gromicho, Joaquim MA-09Gronalt, Manfred TC-26Grossmann, Ignacio MA-01Grothey, Andreas MD-30Grothmann, Ralph WA-24Gruber, Aritanan WC-08Grunewald, Martin WB-13, WC-13Grunow, Martin TA-13Grzybowski, Jerzy MA-23Guajardo, Mario WA-13, MB-26Guarracino, Mario WC-29Gucumengil, Alper WB-11Guegan, Dominique MC-21Guenther, Hans-Otto MA-05, MA-13Guere, Emmanuel TD-09, WD-09Gueret, Christelle MA-07Guerrero, Fernando TD-07Guerrin, Francois WD-40Gueye, Serigne WD-18Guignard-Spielberg, Monique WD-04Guillen, Montserrat TB-19Guimarães, Luis MD-13Guimaraes, Paula MD-35Guimaraes, Thiago TB-03, TB-09Guinee, William WD-35Guinzbourg, Anna MB-28Guivarch, Céline TD-25Guleria, Sanjeev Kumar WC-14Guliashki, Vassil WB-27Gulserin, Meltem TC-31Gultekin, Hakan WB-11Gunalay, Yavuz MD-26Gunay, Elif Elcin MB-28Gundogdu, I.Bulent WC-44Gungor, Askiner TC-03Gunkel, David TC-32Guo, Xuezhen MC-26Guo, Yike TB-43Gupta, Ashish TA-44Gurevich, Gregory WB-24Gurrieri, Massimo TD-40Gurvich, Vladimir WB-08Guschinsky, Nikolai TD-13Gutiérrez Fernández, Milagros TA-39Gutiérrez, César TD-15, WA-15Gutiérrez-Jarpa, Gabriel WD-03Gutjahr, Walter WC-02Guzman Castillo, Maria WC-37Gwiggner, Claus TA-04, TC-42

HHaase, Knut WA-16Haasis, Hans-Dietrich WD-12,

WD-25Habenicht, Walter MA-38Hadzi-Puric, Jelena WA-37Hadad, Yossi WB-24, TA-36, TD-37Hadjiconstantinou, Eleni TD-04,

TC-21Haensel, Alwin WD-45Hübner, Ruth MB-29Hüllermeier, Eyke WB-40Hörmann, Wolfgang MC-18Höschl, Cyril MD-12Haftendorn, Clemens MA-26Hagiwara, Motohiro MA-18Hain, Vladimir TD-33Hakanen, Jussi TD-38Halabi Echeverry, Ana WD-12Halfoune, Nadia TA-31Hall, Nicholas WA-35Halme, Merja TD-29Halskau, Oyvind WC-09Hamann, Joern TD-26Hamouma, Moumen MD-08Hamzadayi, Alper MC-06, MD-39,

WC-42Han, Do Hee MA-44Hanácek, Tomás MD-33Hanafi, Saïd MB-02, WA-06, WC-07,

TD-09Hanczar, Pawel TC-08Hans, Erwin WB-37Hansen, Klaus Reinholdt Nyhuus TA-

13Hansmann, Ronny MA-06Hanya, Kherchi WA-42Hanzel, Jaroslav TB-09, WD-41Harmand, Jérôme MC-15Harper, Paul TB-37, TB-42Hart, Diane MA-32Hartikainen, Markus TA-38Hartl, Richard WA-02, WD-02,

WA-07, TA-45Hartmann, Jens WD-10Harvey, Nigel TC-24Haselmann, Tobias WB-02Hashemkhani Zolfani, Sarfaraz WC-

33Hashimoto, Hideki TD-09Haspeslagh, Stefaan TC-14Hassan, Mehedi MC-37Hassanababdi, Morteza TD-17Hassanpour-ezatti, Majid WD-06Hasuike, Takashi WA-10, TA-45Hauge, Kristian Milo TA-07Haurie, Alain TD-16Hayashi, Shunsuke TB-15Haythorpe, Michael TC-16Heggedal, Ane Marte TC-43Heinimann, Hans Rudorf MD-26Heipcke, Susanne MA-30Heismann, Olga WD-04Hellingrath, Bernd TA-07Helmberg, Christoph WC-05, WA-16Helmedag, Fritz WA-21

278


Hemmelmayr, Vera WC-18Henao, Felipe WD-27, MA-32Henkel, Charlotte WB-43Herczeg, Gábor TC-25Hermosilla, Cristopher WB-23Hernández-Jiménez, Beatriz TC-15Hernández-Pérez, Hipólito WA-07Herranz, Patricia WB-21Herremans, Dorien WD-24Herrmann, Frank TA-24Herron, Rebecca MA-32Hertz, Alain MD-06Hervet, Cédric MA-08, MC-08Herzog, Florian MB-20Heskes, Tom WA-40Hesse, Martina WB-39Heuberger, Daniel TC-29Heydari, Somayeh MA-25Hezarkhani, Behzad MA-14Hezer, Seda TC-02Hidayati, Hidayati TA-34Hielscher, Kai-Steffen TC-13Higgins, Andrew MA-17Hilbert, Andreas WB-45Hildenbrandt, Achim MD-03Hildmann, Marcus MB-20Hillege, Hans WD-31, TD-37Hiller, Benjamin TC-29Hilliard, Casey MB-11Hindle, Giles MD-32, TB-32Hinojosa, Yolanda WC-18Hirsch, Patrick TC-26, WC-37Hiziroglu, Abdulkadir TC-22Hnich, Brahim TA-24, TC-45Ho, Chia-Huei MA-39Hoang, Le WB-04Hobbesland, Kirsti MD-05Hochreiter, Ronald MA-19Hocine, Meglouli WC-40Hoehener, Daniel MD-04Hoff, Arild MC-07Hoffman, Karla WD-01Hofmann, Ulla TD-21Hohenberger, Annette TC-28Hohzaki, Ryusuke WB-28Hoji, Eduardo TB-32Holden, Phil TD-16Holecek, Pavel MB-22Holler, Justus TD-24Holnicki, Piotr TD-32Holschemacher, Klaus WA-33Hombach, Laura TB-25Hon, Cheng-Chuang MC-39Hong, Seongtae TD-34Hooker, John N. TA-27Horbach, Andrei MB-02Horiguchi, Masayuki TC-16Horta, Isabel MD-36Hossein Mirzaei, Ali TB-08Hotta, Keisuke TC-20Hou, Jiang-Liang MB-40, MC-40Houda, Michal MB-43Housos, Efthymios TD-09Howard, John WB-06Hrd, Venkata MA-36Hsiao, Yuan-Du MD-39

Hsiung, Liang-Yuan MD-39Hsu, Chin-Min MD-35Hsu, Chiung-Wen MC-40Hsu, Chung-Chian TB-31Hsu, Pi-Fang MB-39, 40Hu, Yike TC-10Huang, Kwei-Long TC-37Huerga, Lidia TD-15Hugues, Paul TD-25Huisman, Dennis TD-12Humpola, Jesco TB-29Hung, Rose MB-40Hungerländer, Philipp WA-16Hunjak, Tihomir MC-41Hunjet, Dubravko MD-36Hunter, Paul WD-39Huppmann, Daniel MA-26Hurink, Johann WB-37Hurkens, Cor MB-02Huynh, Nha-Nghi WC-10Hvattum, Lars Magnus MD-05Hwu, Tian-Jong MB-27, MA-33Hyde, Matthew TA-06

IIaquinta, Gaetano WD-43Ibacache, Pablo WD-03Ibarra-Rojas, Omar WD-03Ide, Jonas WA-38Ignatavicius, Ceslovas WA-33Iimura, Takuya MD-28IJzerman, Hans TC-28Ikem, Precious TA-33Ilani, Hagai WB-14Illés, Tibor TD-14, WD-16Ilsen, Rebecca TD-32Imai, Haruo WB-34Imanirad, Raha WA-27Inci, A. Can WB-21Inkaya, Alper MB-19Inoue, Ayaka WB-29Ioannou, George TD-04Irnich, Stefan TC-11Irzhavski, Pavel TC-05Isakovic, Vasko TD-21, WD-21Iscanoglu Cekic, Aysegul WB-19Isenberg, Marc-André WD-10Isguder, Hanifi Okan MC-06, MD-39,

WC-42Ishikawa, Hiroshi TD-08Isik, Mine TC-31Ismayilov, Elviz WC-22Isomaki, Hannakaisa TD-38Ivakhnenko, Andrey WB-44Ivanov, Dmitry TA-07Ivorra, Benjamin MC-15Iyigun, Cem WB-24, WC-35Izbirak, Gokhan WA-32

JJablonsky, Josef MB-41Jacobs, Neil WA-04, WC-04Jacobson, Sheldon WC-35Jaehn, Florian TA-02Jäkel, Manuel TA-28Jakubiak, Michal TC-07

Jamall, Mohammad TD-05James, Tabitha TD-06Jancauskas, Vytautas WA-17Janiak, Adam TC-02, MD-14, TB-14Janiak, Wladyslaw MD-14Janilionis, Vytautas TD-37Jarboui, Bassem WD-09Jarus, Mateusz MA-02Jazbec, Anamarija MC-41Jaskowski, Wojciech WC-03Jean Marie, Kiombo MD-33, WD-38Jeanjean, Antoine MD-03, MC-42Jelinek, Herbert TB-31Jensen, Rune MD-05Jeon, Jeasu MC-35Jerman Blazic, Borka TA-21Jiang, Hao-Ping TB-03Jimenez-Lopez, Mariano MA-12Jin, Jianyong MB-06Jin, Shunfu WC-44Jochem, Patrick TB-35Jofre, Alejandro MD-08Johnsen, Eugene WD-35Johnsen, Trond A. V. MA-05Johnson, Andrew MA-36Joncour, Cédric TD-13Jones, Dylan MA-07Jongen, Geert WC-25Jongen, Hubertus Th. TA-15, MD-28Jorgensen, Bjorn TC-01José Nogueira Medeiros, Wesley TB-

03Jost, Vincent MD-03Jozefowska, Joanna MA-02, MA-14Juan, Angel A. MB-16Juan, Angel, A. WA-07, MB-16,

MC-16Juan, Pin-Ju MB-39Judice, Joaquim WD-15Jules, Sadefo Kamdem MB-12Jung Grant, Susan TC-21Junqueira, Leonardo MA-09Jurisica, Ladislav TB-09, WD-41Jurun, Elza TC-33Justel, Claudia TB-05Juutinen, Artti MD-38

KKabak, Özgür MB-03Kabashkin, Igor WD-44Kabasinskas, Audrius MA-30Kabyl, Kamal TD-05Kaczmarczyk, Waldemar TB-08Kadima, Hubert TA-06, TA-31Küfer, Karl-Heinz MC-38Köbis, Elisabeth WA-38Köker, Michaela WB-13Kämpf, Uwe TD-28Körez, Muslu Kazım WB-19, WC-19Kahl, Fredrik TD-08Kahraman, Cengiz WC-22Kakeneno, Joe TA-32Kakouris, Iakovos MD-19Kaleta, Mariusz MD-04, MA-20Kalinina, Maria WC-38Kalsyte, Zivile MA-30

279


Kaluzny, Michael WC-42Kamanli, Mehmet TC-31Kamp, Michael TD-31Kan, Tang-Chung MD-39, MA-40,

MD-40Kana Zeumo, Vivien TC-27Kang, Chao-Chung WC-30Kanudia, Amit TD-16Kaparis, KonstantinosMA-29, MB-29Kara, Bahar Yetis MB-17Kara, Yakup TC-02Karacapilidis, Nikos TC-25Karadayı, Neriman WD-19Karakoca, Aydın WC-19Karaman, Kayhan TC-10Karaoglan, Ismail WC-18, TD-23Karelkina, Olga WA-15, TA-38Karimov, Azar TB-21Karray, Salma MA-45Karsu, Ozlem TA-27Kartal Koc, Elcin WB-24, WC-35Kasap, Resat WD-35Kaschub, Thomas TB-35Kaspar, Ralf WB-39, TA-41Kaspi, Moshe MD-14Kasprzak, Marta MB-37Kassay, Gabor TA-23Katin, Igor WC-06Katina, Joana WC-06Katris, Chris TB-20Katsev, Ilya WA-34Katsifou, Argyro MB-35Kaulakys, Bronislovas MD-18Kaur, Parmjit MC-32Kaut, Michal TC-12, MA-26Kawamura, Márcio MC-13Kawano, Hiroyuki TD-03Kawas, Ban WA-42, WD-45Kaya, C Yalcin MD-23Kaya, Onur MC-03Kayaci Çodur, Merve MB-12Kayakutlu, Gulgun MA-03, MD-30,

TC-31Kazancoglu, Yigit TC-40Kazina, Irina MC-24Kanková, Vlasta MB-43Keisler, Jeffrey MC-27Kellerer, Hans TB-02Kenne, Jean-Pierre TB-45Kennedy, Deanna WC-14Kennedy, Jonathan MC-35Kentel, Elcin TA-26Kentrait, Ausra TA-21Keren, Baruch TA-36, TD-37Kergosien, Yannick TB-14Kern, Walter WB-38Keropyan, Aras TB-19Keshvari, Abolfazl WA-32Keskin, M. Emre MC-08Khachay, Michael WB-44Khadraoui, Djamel WB-17Khenchaf, Ali WD-17Khmelnitskaya, Anna MB-31Khokhar, Ashfaq MA-37Khoshgam, Zahra MC-04Khosravi, Somayeh TB-15

Kibzun, Andrey TD-43Kidd, Martin WA-04, WC-04Kiesling, Elmar MD-27Kijima, Masaaki MD-19Kilby, Philip TB-07Kilci, Firat MB-17Kilic, Erdem TB-21Kim, Jun Seok WD-29Kim, Sang Won MD-44Kim, Sung-Shick WD-29Kinoshita, Eizo WA-36, TB-41Kirilov, Leoneed WB-27Kirytopoulos, Konstantinos MC-14Kis, Tibor WB-10, MD-18Kittler, Josef TA-44, TB-44Klabjan, Diego MD-13, WC-45Klamroth, Kathrin WA-38Klein, Sulamita WA-05Kleindorfer, Paul WC-07Kleniati, Polyxeni-Margarita WC-15Klibi, Walid TA-45Klimentova, Xenia WA-37Klingelhöfer, Heinz Eckart WA-21Klucik, Marian TB-09, WD-41Klucik, Miroslav TD-44Klumbyt, Egl WB-33Kneib, Thomas TB-29Knight, Vincent WD-06, TB-37,

WA-37, TB-42Knippel, Arnaud TD-13Koç, Çagrı TD-23Koberstein, Achim TC-43Kobylarczyk, Justyna TD-34Kobylkin, Konstantin WB-44Koch, Marco K. MD-25Koch, Sebastian WA-45Koch, Thorsten TB-29Koilias, Christos WB-37Kojic, Jovana TC-34Kok, Gurhan MB-24Koksal, Gulser MA-28Koksalan, Murat MA-03Kolisch, Rainer WC-02Komenda, Izabela WA-37Komiya, Toru WC-28Konarzewska-Gubala, Ewa TD-20Kong, Jiehong TC-08Konnov, Igor MC-31Kononovicius, Aleksejus MD-18Konstantaras, Ioannis MC-06Kopa, Milos TB-43Kopanos, Georgios WD-02, TA-25Kopfer, Heiko MA-11Kopfer, Herbert MA-11Kopytov, Rostislav MC-24Korhonen, Pekka TD-29, WA-32,

TB-39, MA-41, TC-41Kornafel, Marta MD-23Koronakos, Gregory MB-36Korviny, Petr MA-22Koshlai, Ludmilla TA-22Kostina, Ekaterina MC-04Kostoglou, Vassilis WB-37Kostrzewska, Marta TB-22Kostyukova, Olga TA-15Kothari, Ravi MC-06

Kotlowski, Wojciech WB-40Kotsialos, Apostolos MA-16Kouedeu, Annie Francie TB-45Koukoumialos, Stelios MB-10Kouvela, Anastasia TC-09Kovacevic Vujcic, Vera WB-09Kovalev, Sergey TA-02Kovalyov, Mikhail Y. MA-02, TA-02,

MD-14, TB-14Kowalski, Dawid MA-02KoyuncuoGlu, Mehmet Ulas WC-12Kozan, Erhan WC-12, MD-14Kraemer, Christian TA-28Kralova, Eva WD-24, MB-25, TD-33,

WC-36Krarup, Jakob MD-24, MC-30Krass, Dmitry TA-18Kratica, Jozef WB-09Kraus, Sarit WD-13Krawczyk, Jacek TB-16Kriett, Phillip TA-13Krislock, Nathan MA-29Kristensen, Anders Ringgaard MC-26Kristianto, Yohanes WA-10Kristjansson, Bjarni WB-35Kritikos, Manolis TD-04Kroeller, Alexander WB-17Kronborg, Morten Tolver TA-19Kroon, Leo TD-12Kropat, Erik WD-22, TB-23Kruger, Alexander TA-23Kruger, Hennie WD-29, TA-31Krumke, Sven MB-02Krushinsky, Dmitry WA-17Krysiak, Tomasz TC-02, MD-14Kubat, Gözde MC-24, TD-34Kubiak, Wieslaw MA-14Kubicka, Ewa TD-05Kubicki, Grzegorz TD-05Kucher, Vlad WC-22Kucuk, Mahide MA-23Kucuk, Yalcin MA-23Kuefer, Karl-Heinz MC-38Kuehlwein, Daniel WA-40Kuhn, Daniel WB-42, TB-43Kuhn, Heinrich MB-35Kuhn, Kenneth TB-12Kuhpfahl, Jens MC-02Kuik, Roelof MD-10Kuimova, Anastasia WA-39Kula, Erhun MD-26Kula, Ufuk MB-28, TC-33, TB-45Kulak, OsmanMA-05, TC-07, WC-12Kulak, Ozgur MD-09Kumar, Kundan MA-10Kundakcı, Nilsen MC-41Kunsch, Pierre WC-40Kunter, Marcus MA-45Kunz, Friedrich TC-32Kuosmanen, Timo WB-30, MA-36Kurasova, Olga MB-07, WD-44Kurc, Wieslaw MD-14Kuroiwa, Daishi TC-38Kurz, Sascha MC-34Kus, Coskun WB-19, WC-19, WD-19Kusanovic, Tino TC-33

280


Kusunoki, Yoshifumi WA-44Kuter, Semih WC-35Kuyzu, Gultekin MC-03Kuznetsov, Mikhail WB-26Kuznetsov, Nikita TD-28Kuzyutin, Denis WC-06Kydland, Finn ME-01Kylinger, Martin MB-26Kyriakidis, Thomas WD-02Kınacı, Ismail WC-19, WD-19Kıran, Mustafa Servet TD-23Kıvrak, Abdulkadir MB-26

LLa Red Martínez, David L. WD-44La Spada, Simone MD-11Labbé, Martine TB-04, MA-08Labriet, Maryse TD-16Lachout, Petr MB-43Lackes, Richard TD-10Løkketangen, Arne MB-06, MC-07Lê, Minh TC-36Löhne, Andreas MB-38López Sánchez, Ana Dolores TC-06López, Diana MB-15López, Julio WB-23, WC-23López-Cerdá, Marco A. TA-23López-Ibáñez, Manuel MA-41López-Redondo, Juana TB-18Létocart, Lucas WD-04Laganà, Demetrio TC-11, TD-13Lai, Mu-Hui MD-39Lai, Yi-Chiuan MD-35Lakhdar, Sais TD-10Lakner, Peter WD-20Lal, A.k. WC-14Lambert, Amélie MC-29Lami, Isabella TC-27Lamorgese, Leonardo WA-12Landete, Mercedes MD-17Landsman, Zinoviy TA-19Lange, Jan TD-14Langevin, André TC-11Langton, Sebastian MB-03Lansink, Alfons Oude MC-26Lantz, Frederic WA-39Lancinskas, Algirdas TC-18Lapegue, Tanguy WB-02Laporte, Gilbert TD-03, TA-05Larrañaga, Pedro WC-29Larsen, Christian TB-08, TA-10Larsen, Jesper TA-07Larsson, Aron TD-27, WC-38Laskova, Maya TA-33, TD-35Laugier, Alexandre MB-02Lauks, Gunars TD-36Laumanns, Marco WA-42, WD-45Lavor, Carlile MB-37Layter Xavier, Vinicius MA-16Lazarev, Alexander TD-02, WB-14,

TA-33, TD-35Le Thi, Hoai An WB-17, WC-17,

WD-17Le, Hoai Minh WC-17Leal, Nuno TA-29Lechón, Yolanda WD-38

Lechevalier, Steve WB-25Lee, Dong Myong MA-44Lee, Gregory MC-24Lee, Gue Myung TC-38Lee, Hansuk TD-34Lee, Hong Tau WB-27, WC-44Lee, Hwi Young MD-45Lee, In-Young TD-34Lee, Jinwook MB-21Lee, Mooyeob MB-21Leggate, Alexander TD-14Legillon, François WC-03Lehane, Kate WD-45Lehmann, Thomas TB-29Lehuédé, Fabien MA-17Leitner, Markus TC-04Leitner, Stephan MD-45Lemarié, Bernard WC-07Leme, Rafael C TD-29Lenté, Christophe TB-14Leon, Teresa WD-29, MA-36Leonhardt, Alexander TA-21Leopold, Ulrich MA-03Leopold-Wildburger, Ulrike TD-22Lepin, Victor TC-05Lerida, Josep Lluis MD-43Leroux, Christian MA-17Leskovar-Špacapan, Gabrijela MC-24Letchford, Adam WC-16, MA-29,

MB-29Leurent, Fabien WB-12Levêque, Christèle WC-07Levaggi, Rosella WD-20Levchenkov, Anatoly MD-38, WB-38Levin, Genrikh MA-10, TD-13Levin, Yuri TA-40, MA-44Levina, Tatsiana TA-40Levy, Marc MC-21gencat.cat, Lorenzo WD-38Li, Olga MD-35Li, Qing MB-44Li, Weiqi TB-07, MD-21, WA-39Li, Xiang TC-29Li, Xin WD-05Li, Yan TD-38Li, Yongjun WC-30Li, Zhiyong WC-30Lian, Zhaotong WD-05Liang, Chih-Chin MC-39Liberatore, Federico MC-33Lichner, Ivan WC-21, MB-25Lichtenstein, Maciej TC-02Lidbetter, Thomas TB-18Lidestam, Helene TA-08Lieder, Alexander MD-02Liefooghe, Arnaud WC-03, MA-41Lienert, Judit TB-27Liesa, Francisco MD-11Liesiö, Juuso TD-27, WC-30Lim, Gino MC-37Lima, Priscila MA-16Lin, Chin-Tsai MA-39Lin, Dong MB-11Lin, Hsuan-Tien TD-40Lin, Hung Lung MB-40Lin, Jenn-rong MA-04

Lin, Shin-Yi MB-39Lin, Shu-Yu MB-39Lin, Ya-Han MB-39Lin, Yao-Chang MC-40Linck, Iris WB-04Linden, Isabelle WC-27Lindsey, Christopher WC-45Lindstad, Haakon WD-12Lino, M.Pilar MC-14Liou, James MB-41Liour, Yuanchau MC-39Lipara, Carmen TD-20Lisboa, Adriano WB-21, MA-40Lisser, Abdel TD-43Listes, Ovidiu TD-07Liu, Fuh-Hwa MB-36Liu, Jiyin MB-05Liu, Mei-Ling TC-45Liu, Shaofeng WC-27Liu, Wenrong TC-25Liu, Xiaoming WD-05Liu, Yu-Cheng MB-36Lium, Arnt-Gunnar TC-12, MA-26Liuzzi, Giampaolo MB-29Ljubic, Ivana TB-04, TC-04Llamazares, Bonifacio WB-26Llerena, Francesc MD-34Llorente, Miguel Angel MB-16Lo, Mei-Chen TA-20, MB-27, MA-33Lobato, Marcus WB-21, MA-40Lodi, Andrea MC-29Loeffler, Andreas TC-21Lonc, Zbigniew WA-08Lopes, Ana WD-32Lopes, Carlos MC-08Lopes, Maria João MA-08Lopes, Rui Borges TD-11Lopez, Pierre WA-02Lopez, Ruben WA-15Lopez, Susana MB-31Lopez-Pascual, Joaquin MD-17Lorenzo, Leticia MA-24Lorenzo-Freire, SilviaMA-24, MA-35Lourenço, Helena Ramalhinho MB-

16, MC-16Love, Ernie TC-45Lovison, Alberto TA-38Lozano, Sebastián TC-35Lozovanu, Dmitrii MD-04Lu, Hua-An TB-11Lu, Jung-Ho MA-39, MC-39Lu, Ming-Tsang MC-39Lubis, Asrin TB-34Lucheroni, Carlo MD-16Ludwig, Michael WD-21Luhandjula, Monga K MA-12Lukas, Elmar TC-43Lukasiak, Piotr MD-37Luna, Henrique MB-08, TC-17Luna, Mônica M. M. WB-30Luque, Mariano TB-38Lusa, Amaia MD-30, WC-41Luss, Ronny WA-23Lutzenberger, Fabian WD-21Luz, Carlos J. WB-16Lydia, Maya Silvi TD-17

281


Lyons, Daniel WC-05Lyra, Christiano TD-24, MA-25

MM, Ganesh Perumal WC-34M. Costa, Alysson TC-13, MC-26Ma, Meng WC-21Maç, Elif MA-40Maïzi, Nadia TD-25, WA-25, WB-25MacCarthy, Bart WC-13Macedo, Rita MB-02, WC-07Machado Cardoso Junior, Moacyr

WA-29Machado, Maria Augusta TA-20Machowiak, Maciej MA-02Maculan, Nelson MB-37, TC-44Madan, Jason WD-31Maenhout, Broos WA-02Müller, Sven WA-16Mäkelä, Marko M. WA-15Mészáros, András TC-36Maggioni, Francesca MA-43, TC-43Magirou, Evangelos MB-05Maglaras, George MB-10Magnusson, Johan WC-41Magos, Dimitris TC-09Maheut, Julien TC-35Mahmassani, Hani MC-11, WC-45Maia, Fábio TD-06Maisiuk, Yauhen TA-05Maiza, Mohamed TD-10Majewska, Justyna TD-43Makajic-Nikolic, Dragana WB-09Makarenko, Alexander TC-33Makulska, Joanna M. TB-23, MC-26Makuschewitz, Thomas WD-10Malapert, Arnaud MC-08Malassé, Olaf TD-36Maldonado, Michelli MB-13Maldonado, Sebastian WB-29Malhamé, Roland TD-16Malick, Jerome MA-29Maliene, Vida WB-33Malitsky, Yuri WC-03Malo, Pekka TB-38Mammeri, Mohamed, Lounes TD-39Maneia, Gian Mauro TA-42Mangaraj, Bijaya Krushna WC-38Manger, Robert WB-05Manitz, Michael MA-42Manni, Emanuele TD-13Mannini, Livia MD-11Manon, Yvan MC-08Manuel, Conrado M. MA-31, MB-31Mar-molinero, CecilioMA-18, WA-32Marín, Ángel TD-12Marín, Alfredo MD-17, WB-18,

WC-18Marangoni, Giacomo MC-31Marasovic, Branka WB-20, MA-21,

MB-33Marentakis, Charis MA-11, WC-45Mareschal, Bertrand TC-39Marianov, Vladimir WD-03, TA-17,

TB-17Marinakos, George WA-24

Marins, Fernando WC-38Marmaras, Nicolas MA-32Marmolejo, Antonio MD-30Maroti, Gabor TD-12Maroto, Concepcion MD-26, TD-26Marques, Inês MC-06Marques, Rui WC-32Martein, Laura MC-23Martel, Alain TA-45Martello, Silvano MC-29Martin Caraballo, Ana M. WB-21Martin, Alexander TB-29Martin, Simon TC-14Martin-Campo, F. Javier TA-29Martinez Sykora, Antonio MB-09Martinez, Carme MD-30Martinez, Luis MA-04Martinez-de-Albeniz, Javier MD-34Martinez-Martin, Pablo WC-29Martins de Sá, Elisangela TC-17Mas, Marta MD-30Masala, Elena TC-27Mashira, Hildah TA-33Massol, Olivier TA-25, MD-33Masson, Renaud WB-03Masui, Michiharu MA-31Matallin-Saez, Juan Carlos MB-30Matano, Yoshinori TB-28Mateo, Jordi MD-43Mateo, Manuel MD-30Mathiesen, Lars TB-16Matos Dias, Joana TD-19Matos, Henrique WC-41Matos, Telmo TD-06Matsushita, Yasuhiro MA-18Matthews, Jason WA-04Mattik, Imke MA-13Maturana, Sergio TC-26Matzke, Andreas WC-13Mavri, Maria MA-18Mawengkang, Herman TD-17, TA-34,

TB-34, TC-34Maya Duque, Pablo Andrés WA-14Mayag, Brice TC-40Mayston, David TB-36Mazalov, Vladimir TD-22Mazauric, Vincent WA-25Mazzola, Marco TA-42Mbiydzenyuy, Gideon MB-03McBride, Tom TB-27McComb, Sara WC-14McDowall, Richard WB-41McGarraghy, Seán MA-08McGill, Jeff TA-40McHale, Ian WA-35McKinnon, Ken MD-30Meca, Ana MB-24Medaglia, Andres MA-07Medineckiene, Milena WC-33Medori, Jean-robin MB-45Meeyai, Sutthipong TD-41Megel, Romain WB-03, WB-35Meherrem, Shahlar WB-16Mehmet, Kabak TD-33, MD-41Mehta, Deepak WA-03Meier, Arne WA-08

Meissner, Joern MA-44Melchiors, Philipp WC-02Melnikov, Andrey WB-15Melo, Teresa MA-17Melo, Valdir TB-06Meltzer, David WD-37Memmedli, Memmedaga WB-22Mendes, Armando TA-31Mendoza, Jorge E. MA-07Meneguzzer, Claudio MB-11Menezes, Mozart WD-18Menoncin, Francesco TA-19, WD-20Mera Vazquez, Ana MA-25Merchant, Sue MC-32Merino, María WA-43Merkuryeva, Galina MC-03Merlo, Angelo TA-42Mesa, Juan A. TD-12Meshkinfam, E. MD-15Messinger, Margaret-Ellen WC-39Mesyagutov, Marat TA-09Meterelliyoz, Emre MB-26Metla, Nataliya WC-17Mevissen, Martin WD-45Meyer, Anne WC-07Meyer, Matthias MD-45Meyer, Patrick TC-39, TD-39Meyerov, Iosif WA-23Meyers, Kris MB-45Meza, Sergio TB-22Mezei, József MA-22Mianabadi, Hojjat MC-40Michaels, Dennis MD-29Michallet, Julien WB-03Michel, Sophie TD-13Michel, Terraza MB-12Michelon, Philippe WD-18Michnik, Jerzy MB-27, MA-33Midthun, Kjetil MA-26Mietek, Arleta TD-22Miettinen, Kaisa MD-38, TA-38,

TB-38, TD-38Miglierina, Enrico TD-15, WA-15Miguéis, Vera TD-31Miguez, Geraldo TC-44Mihályi, Péter WD-36Mihalak, Matus TB-03Mika, Marek MB-14Mikhailau, Mikalai TA-05Milanic, Martin WB-08Milioni, Armando Zeferino MC-36Milostan, Maciej MD-37Minh Thuy, Ta WB-17Minnetti, Valentina TC-40Minoux, Michel MB-45Minty, R John TB-37Mirabi, Mehrdad MC-40Miralles, Cristobal TC-25, WD-31Miranda, Gilberto MB-08, TC-17Miranda, Joao WC-41Mirchandani, Prakash MD-44MirHassani, Ali TD-17Mirkov, Radoslava TB-29Misener, Ruth MC-29Misevicius, Alfonsas WD-04Mishina, Tsutomu WB-32, WC-40

282


Miskiewicz, Michal MA-02Missbauer, Hubert TA-13Miyamoto, Yuichiro WB-29, WB-38Miyashiro, Ryuhei WB-38Miyazaki, Kenji TD-33Mladenovic, Marko WA-06Mladenovic, Nenad WA-06, WD-09Mockus, Jonas WC-06Modlinski, Piotr MA-14Moeini, Mahdi WB-17Moench, Lars MC-02, MA-09, TA-13Mohamad Doost, Faeghe TA-08Mohamed Sidi, Mohamed Ahmed

WD-05Mohammadipour, Mona TA-24Mohaupt, Michael WB-45Mohr, Esther WD-20Molho, Elena TD-15, WA-15Molina, Elisenda MA-31, MB-31Molina, Julian TC-06Monacelli, Daniele TB-11Monaci, Michele WB-01Monkkonen, Mikko MD-38Monroy Licht, Ingrid Marcela TC-11Monsuur, Herman WA-28, WC-28Montella, Bruno WB-12Montero, Javier MC-33Montevechi, José Arnaldo WC-38Montibeller, Gilberto MA-27, TD-42Montoya-Torres, Jairo WD-12Montvida, Olga MA-38Mor, Baruch WB-11Mor-Yaroslavtsev, Andrew WB-38Mora-Stock, Cindy WD-28Morabito, Reinaldo MA-09, MD-13Moraes, Leonardo WB-42Morais, Paulo TD-29Morar, Ioan Radu WD-41Moreira, Attus MD-35Moreno, Eduardo WA-42Moreno, Luis MC-02, MC-15Moretti, Antônio TA-09, TB-33Morgado, Miguel MB-27Morgan, Jacqueline WD-15, TB-16Mori, Ryota TA-04Moriggia, Vittorio WD-43Morozkov, Mikhail WA-42Morsi, Antonio TB-29Morton, Alec TA-27Morze, Michal MA-02Moser, Albert TA-28, TC-29Moser, Elke MC-04Mosheiov, Gur WB-11Mosquera Rodríguez, Manuel Alfredo

MB-24Mostert, Erik MC-40Motahhari, Seyyed Mahdia TD-17Motrenko, Anastasia TD-44Mottl, Vadim TB-22Moulis, Frederic MC-08Moura, Pedro TC-04Mourad, Amad MD-08Mourão, Cândida TD-11Mourtos, Yiannis TC-09Mousseau, Vincent TD-39, TC-40Moustapha, Diaby MB-45

Muñoz, Jose WB-18Mueller, Christian TA-42Mula, Josefa MA-12Munari, Pedro TB-07, WC-16Munavalli, Jyoti R. TB-37Murray, Robert TA-16Musatova, Elena WB-14, TA-33,

TD-35Musial, Jedrzej WD-26Musmanno, Roberto TC-11Mustajoki, Jyri WC-26Myndyuk, Olga MD-43

NNabavi, Vahid TC-41Nadal, Esteve MC-26Nadri Pari, Mahdi TD-17Nørstebø, Vibeke MA-26Nafi, Amir WA-27Nagamochi, Hiroshi TC-05Nagaoka, Sakae TA-04, TC-42Nagel, Andrea MA-11Nagurney, Anna TC-23Nagy, Adrienn WD-16Nahorski, Zbigniew TD-32Naibaho, Tutiarny WA-17Naima, Hamadouche TC-42Nait Belkacem, Salima WA-04Nakhai Kamalabadi, Isa TB-08Naroski, Pawel WA-08Nascimento, Leonardo MD-23,

WB-42Nascu, Ioan WA-37, WD-41Nasrollahi, Mahdi MD-40Nasrollahi, Meisam WB-13Nasution, Azizah Hanim TA-34Natividade-Jesus, Eduardo Manuel

WB-27Naujokaitis, Darius TD-37Naumann, Marc TC-43Naumov, Andrey TD-43Navabi, M. MC-15, MD-15, TB-23,

WD-41Navidi, Hamidreza WD-06Nayak, Shriguru WD-23Nazarenko, Olga MB-10NDienor, Moussa WD-40Nediak, Mikhail TA-40, MA-44Neiva, Rui TD-29Nel, Hannelie WA-04Nenes, George MA-28Neralic, Luka MD-36Nesterov, Yurii MD-29Neto, Jose WD-05Neufeld, Janis Sebastian MC-02Neumann, Ludmila MC-36Neves, Manuela TD-24Ng, Chi To MC-10Nguyen Duc, Manh WC-17, WD-17Nguyen, Khanh WB-21Nickel, Stefan MC-17, TB-17Nicoloso, Sara WA-09Nielsen, Jens Perch TB-19Nielsen, Lars Relund TA-10Niemann, André MD-25Nieto-Gallardo, Gracia M. TC-15

Nigro, Marialisa MD-11Nikitina, Mariya WC-20Niknejad, Ali TA-36Nikolaev, Alexander WC-35Nikolakopoulos, Athanassios MC-06Nikolic, Nebojsa WB-09Nikulin, Yury WA-15Niroomand, Sadegh TB-11Nishide, Katsumasa MD-19Nishihara, Michi WC-20Nishimura, Nobuhiko WA-26Niu, Yi-Shuai WD-17Nogales, Amaya TC-18Nogueira, Loana WA-05Noll, Torsten TD-10Nolz, Pamela MD-06Nonås, Lars Magne MA-05Nonobe, Koji TD-09Noordkamp, Wouter WA-28Norikumo, Shunei TB-41Norkin, Vladimir TD-43Norlund, Ellen Karoline TA-05Nossack, Jenny TA-02Nouioua, Karim WB-03, WB-35Novaes, Antonio G.N. WB-30Novo, Vicente TD-15, WA-15Novozhilova, Marina MD-09Nowak, Maciej MB-27, MA-33Nowak, Thomas TC-23Nowak, Uwe MC-38Nuhn, Eva WD-22Nunes, Ana Catarina TD-11Nunez, Marina MD-34Nykamp, Stefan MA-25Nemcová, Zuzana TC-13

OOberscheider, Marco TC-26Ocaktan, Beyazıt TB-45Ochoa, Gabriela TA-06Odegaard, Fredrik MD-44Oggioni, Giorgia MC-31Oghojafor, B.e.a. MB-41Ogiwara, Kei WB-32Oguz, Ceyda MC-01Ohazulike, Anthony WB-38Ohst, Jan Peter WB-05Ohya, Takao TB-41Ojalehto, Vesa TD-38Ojrzynska, Anna TA-37Okhrin, Irena MD-13Oktay Fırat, Seniye Ümit MB-26Okuhara, Koji WA-10, TA-45Okuno, Takayuki TB-15Olabode, Adewoye MD-21Oliveira Moreira, Mayron César TC-

13Oliveira, Aurelio MD-25Oliveira, José Fernando MB-09,

MC-09Oliveira, Luis WD-32Oliveira, Marisa MA-09Oliveira, Ricardo WC-32Oliveira, Sandra Cristina MC-21Olsson, Leif TD-27Olteanu, Alexandru-Liviu TC-39

283


Onel, Selis TC-43Onn, Shmuel MD-29Onoda, Takashi WA-44Onsel, Sule MB-03Onucyildiz, Mustafa TA-26Oorni, Anssi TB-39Opolchenov, Daniil WC-30Oppen, Johan MC-07, MC-11Oprime, Pedro MB-28Or, Ilhan TC-19Oral, Evrim WB-19, WC-19Orlov, Alexej WA-41Orlovich, Yury TC-05Orlovs, Pavels MA-38Ortega Riejos, Francisco A. TD-12,

MD-17Ortigosa, Pilar M. TB-18Ortobelli, Sergio TC-19Ortuño, M. Teresa MC-33Ossadnik, Wolfgang WB-39Osswald, Christophe WD-17OSullivan, Barry WA-03Osuna-Gómez, Rafaela TC-15Otradnova, Lina WD-24Otto, Alena TC-13Oubraham, Aichouche WC-12Ouelhadj, Djamila TC-14Oulamara, Ammar TB-14Ouyang, Liang-yuh MA-39Ovchinnikov, Anton TA-40Owadally, Iqbal TA-19Owen, Guillermo MB-31Oxley, Tim TD-32Oyatoye, Emmanuel O. MB-41Oyola, Jorge MC-07Ozaydin, Ozay MB-03Ozbudak, Ferruh WC-42Ozceylan, Eren TD-23Ozdemir, Ozer WB-22Ozdemir, Ozge MA-26Ozekici, Suleyman WD-21Ozener, Okan MC-03Ozgur, Leyla TC-03Ozkok, Erengul WD-20Ozogur-Akyuz, Sureyya TC-44Ozpeynirci, Ozgur WA-14, TC-40Ozpeynirci, Selin TC-40Ozturk, Gurkan TB-07, MC-16,

WA-29Ozturkkal, Belma TD-42Ogcu, Gamze MB-26

PP. Fernandes, Florbela MD-15, TA-15Pacheco, Maria F WB-16Pacino, Dario WC-03, MD-05Pacqueau, Rémi WB-04Püster, Johannes TD-24Paetz, Alexandra-Gwyn TB-35Pérez Perales, David TC-08Pérez, M. Angeles MC-14Péton, Olivier MA-17Pagel, Christina WC-37Pahikkala, Tapio TD-40Paias, Ana MA-08Paillat, Jean-Marie WD-40

Paiva, Anderson Paulo TD-29Paixão, José MA-21, TB-21Paksoy, Turan TD-23Palacio, Antonio TC-35Palka, Piotr WB-06Pallas, Sabine MD-25Palomo Zurdo, Ricardo TA-39Pan, Jason Chao-Hsien MA-39Pan, Quan-Ke WA-11Panagiotidou, Sofia MA-28Pando, Valentín MD-10Pang, King-Wah MB-05, WD-26Pankratov, Alexander MB-09Pankratova, Yaroslavna WC-06Pankratz, Giselher MA-11Panovic, Milena WD-27Panteli, Antiopi TA-17Papadimitriou, Dimitri MB-08Papadopoulos, Thanos MB-32Papamichail, K. Nadia MB-27Papayanopoulos, Lee WD-04Pappis, Costas TC-25, TB-37Paquet, Marc MD-10Paradi, Joseph MC-18Pardalos, Panos WA-17Paredes, Fernando TA-21Pareja, Paloma MC-19Park, Jonghyuck WD-29Park, Young Woong WC-45Parkes, Andrew J. TA-06Parnaudeau, Virginie WD-40Parpaillon, Jean MC-08Parreno, Francisco MA-09, MC-09Pasenchenko, Yuriy MB-10Pash, David TD-14Pasichny, Alexis WD-22Paslawski, Jerzy MB-33Passarella Freire, Wilhelm MD-23Passerini, Andrea WA-40Paszto, Peter TB-09, WD-41Patek, M.t. MB-13Paterson, Colin TB-42Pato, Margarida MC-06, TA-26Paucar-Caceres, Alberto MA-32Paul, Anand WC-02Paul, Rahul MA-10Pavlov, Alexander TA-07Pavlovic, Ljiljana TC-05Pawlak, Grzegorz MA-02Peña, Gloria MC-02Pecaric, Mario TC-33Pechinkin, Alexandr TD-36Pechlivanos, Lambros WC-45Pedraza Martinez, Alfonso WC-10Pedraza-Martinez, Alfonso WC-10Pedro, Isabel WC-32Pedroso, Joao Pedro MA-25, WA-37Peeters, An TD-24Pekár, Juraj MD-03Pekcan, Onur WD-28Pekgör, Ahmet WC-19Pelegrin, Blas TC-18, TD-18Pelizzari, Cristian MA-19Pelot, Ronald MB-11Peng, Shili WD-15Penna, Puca WB-03

Pennings, Clint TA-24Pensa, Stefano TC-27Perau, Eugen MD-25Perea Olabarria, Eugenio MA-25Pereira, Ana I. MD-15, TA-15Pereira, António MB-37Pereira, Javier TA-21Pereira, Luis Gustavo TB-03Pereverza, Kateryna WC-41Perez Gonzalez, Paz TD-02, WB-11Perez Valdes, Gerardo MA-26Perez, Gloria WA-43Perez, Juan MD-08Perez-Marin, Ana Maria TB-19Peric, Tunjo MC-45Perraudin, Xavier MA-17Pervukhina, Elena TA-22Pesch, Erwin MA-02, TA-02, MD-14Peska, Ladislav TD-41Pesneau, Pierre TC-04Pessoa, Luciana MB-16Peters, Georg WB-22Petersen, Hanne TC-06Petrelli, Marco TA-11Petrovic, Slavica P. WA-36Petrucci, Vinicius WB-03Peypouquet, Juan WC-23Pfeiffer, Dominik TA-07Pflug, Georg MA-43Pfosser, Stefan TD-21Pham Dinh, Tao WB-17, WC-17,

WD-17Pham, Viet Nga WD-17Phelps, Fatma Selcen WD-35Phillips, Antony WC-14Piantadosi, Julia MD-16Piccolo, Carmela TA-18Pichl, Lukas MA-21Pichler, Alois MA-43Pickl, Stefan WD-22Pidani, Fariba WD-16Pietropaoli, Ugo WA-09Pigden, Tim MA-07Pillac, Victor MA-07Pillon Torralba Fernandes, Jéssica

WC-25Pilvar, Rahman MC-11Pilz, Kay MB-18Pimentel, Carina MA-13Pindoria, Sandip WB-35Pinheiro, Sofia TB-05Pinho de Sousa, Jorge WD-03, TD-07Pini, Rita TD-15Pino, José L. TD-18Pinson, Eric TA-07Pinto Ferreira, Ma Eduarda MA-09Pinto_Varela, Tânia TA-30Pires, José MA-08Pirlot, Marc TA-20Pisciella, Paolo TA-43Pisinger, David MC-05, WC-09Pistikopoulos, Efstratios TA-25,

MC-29Pitzer, Erik MC-03Piunovskiy, Alexei TC-16Pivac, Snjezana MD-21, TC-33

284


Pizarro Romero, Celeste MC-33Pizarro, Jose MC-20Pla, LluisM MC-26, MD-43Placido, Antonio WB-12Plastria, Frank TC-18, WB-34Plateau, Gérard WA-03, WD-04Plaza Leiva, Victoria TB-18Plazola Zamora, Laura MC-33Pliskin, Joseph TD-37Plociennik, Kai MC-38Poberiy, Maria WB-44Poddig, Thorsten WA-21Podkopaev, Dmitry MD-38Podobedov, Vitaly WD-24Poesz, Attila TC-41Poklepovic, Tea WB-20Polat, Olcay TC-03, MA-05, WC-12Polat, Refet WB-16Poldi, Kelly TD-10Polovinkin, Alexey WA-23Polushina, Tatiana WD-42Polyak, Boris WA-01Pomarico, Silvia TC-27Pons, Montserrat MC-34Poole, Travis TA-16Pop, Emilia-Loredana MB-38Pope, Brandon MB-30Porcheron, Marc MC-44Porschen, Stefan WA-08Portal, Gabriel WC-03Portela, Maria MB-36, MC-36Portillo, Fabiola WA-32Possani Espinosa, Andre WB-39Possani, Edgar WD-07, WB-39Post, Gerhard F. MA-09Postmus, Douwe WC-26, WD-31,

TA-37, TD-37Potapovs, Andrejs WB-38Potthoff, Daniel TD-12Potts, Chris WA-35Pourouchottamin, Prabodh WA-25Pozo, Miguel Angel TD-12, MD-17Prachalias, Chrysovaladis WD-32Pranevicius, Henrikas MA-37Pranevicius, Mindaugas MA-37Pranevicius, Osvaldas MA-37Prasanna, G. N. Srinivasa MA-10,

WD-23, WB-25, WC-34Pratsini, Eleni WA-42, WD-45Prekopa, Andras MB-21, MD-43Prestwich, Steven TA-24, WD-42,

TC-45, WD-45Prieto, Angel M. MA-38Prieur, Fabien TB-16Prigent, Jean-luc WA-20Prins, Christian WC-07Prior, Diego WA-32Prochorskaite, Agne WB-33Prodan, Radu WA-37Prot, Damien WB-02Protti, Fabio WA-05Prskawetz, Alexia MC-04Psaraftis, Harilaos MB-05Pticina, Irina TD-02Puchinger, Jakob TA-45Puechmorel, Stephane TA-04

Puente, M. Albina MC-34Puerto, Justo TD-12, TB-18, WB-18,

WC-18Puetz, Markus TD-45Pukalskyi, Bohdan WA-41Purba, Ronsen WA-17Pureza, Vitoria MB-28Purutcuoglu Gazi, Vilda WB-24Pylypovskyi, Oleksandr TA-22Pytlak, Radoslaw MD-04

QQi, Zhiquan TD-38Qian, Fubin WC-09Qu, Rong TC-14, WC-14Quadrifoglio, Luca WD-03Quariguasi, Joao TB-25Quessette, Franck TC-36Quintanilla, Sacramento MC-14

RRaa, Birger WA-06, MB-07, TB-45Rabadi, Ghaith MB-06Racer, Michael MD-27Racero, Jesus TD-07Rachaniotis, Nikolaos TB-37Racic, Zeljko WC-31Radaei, M. WD-41Rademaker, Alexandre WC-27Rademaker, Michael WA-40Radjef, Mohammed Said TD-10,

MC-23, MD-34Radulescu, Constanta Zoie WA-20Radulescu, Marius WA-20Radulescu, Sorin WA-20Régin, Jean-Charles MC-08Römisch, Werner TB-29Rönnqvist, Mikael TC-08, WA-13,

MB-26, MD-42Rafels, Carles MD-34Rafia, Rachid WD-31Raghavan, S. TC-04Rahman, Farzana MC-37Rahmo, El-Desouky TD-15Rais, Abdur WA-37Raith, Andrea WC-11, TB-12Rajabloo, Ghasem MC-10Rajkovic, Vladislav TA-21Rakicevic, Zoran TC-34Ram, Camelia TD-27Ramón, Nuria MA-36Ramahefarison, Heriniaina WD-40Ramik, Jaroslav MA-22Ramirez, Hector WC-23Ramjee, Shivani MA-33Ramos Jorge, Diana Rita MA-04Ramos, Angel Manuel MC-15Ramos-Gallego, Ana Bel WB-18Rana, Rupal WA-45Rand, Graham MD-24Rangel, Luis MB-41Rangel, Socorro MB-13Rangraz, Hamed TB-23Rapaport, Alain MC-15Rapine, Christophe TA-29Rashidi Bajgan, Hannaneh TB-33

Rasoolinezhad, Zeinab TB-06Rathgeber, Andreas TA-21Ratli, Mustapha MB-02Rauch, Jonas WB-45Rauscher, Sandra TB-42Raz, David MD-08Razmi, Jafar WB-13Reade, Andy TB-25Rebaine, Djamal TB-14Recht, Peter TD-05Redchuk, Andrés WD-38Reddy, Brian TD-39Reed, Joshua WD-20Rego, Cesar TD-06Reimann, Marc WA-24Reimers, Stian TC-24Reinecke, Philipp TD-36Reinelt, Gerhard MD-03Reinhardt, Wolfgang WD-22Reizes, Erwin TB-28, TD-41Reklaite, Agne WB-34Relvas, Susana TA-10, TA-29Requejo, Cristina MB-37Resat, Giray TB-33Resende, Mauricio MB-16Rest, Klaus-Dieter WC-37Reunanen, Pasi MD-38Reuther, Markus WA-12Revie, Matthew TA-16Rey, David TA-29Reyer, Ivan TB-44Rezapoor, Shabnam MC-10Rezapour, Shabnam TA-08Rezgui, Mohamed MC-08Ribeiro, Cassilda TA-09Ribeiro, Celso MB-06Ribeiro, Cristina MB-09Ribeiro, Patrícia TA-29Ribeiro, Rita WC-27Ricca, Federica MA-19Riccardi, Rossana MC-31Ricci, Olivia WA-25Richards, Deborah WD-12Richards, Russell MB-25Richter, Knut MD-13Riera, Daniel MB-16Rijo, Pedro TA-08Rinaldi, Francesco MB-29Rios-Insua, David MD-33, WD-38Rios-Mercado, Roger Z. MC-26Rios-solis, Yasmin WD-03Ritt, Marcus WC-03Ritzinger, Ulrike TA-45Rivreau, David WA-02Robledo, Jose MC-33Rocha, Humberto TD-19Rocha, Paula WB-42Rocha, Pedro MD-09Rocha, Roger WB-42Rodríguez Álvarez, Margarita TB-15Rodríguez, Victoria TD-07Rodríguez-Puerta, Inmaculada TC-10Rodrigues, Antonio TB-24, WB-24Rodrigues, Rui MD-09Rodriguez Aguilar, Roman MA-30Rodriguez, J. Tinguaro MC-33

285


Rodriguez-Chia, Antonio ManuelWB-18

Rodriguez-Sanchez, Sara VeronicaMC-26

Rodriguez-Ulloa, Ricardo MA-32Roitch, Vladimir TB-43Rojas-Medar, Marko A. TC-15Rokou, Elena MC-14Roland, Julien TB-39Romanin-Jacur, Giorgio MA-06Romanova, Tatiana MB-09Romero, Ruben TB-09Ronconi, Debora MC-13Roos, Kees MD-42Ropke, Stefan TC-06Roset, Alain WC-07Rosseti, Isabel MB-06Rossi, André TA-06, WB-09Rossi, Fabrizio MA-29Rossi, Riccardo MB-11Rossi, Roberto TA-24, TC-45Rossi, Sebastiano MB-20Rossomakhine, SergueiTD-04, TC-16Rotaru, Kristian WA-27Roupin, Frederic WA-03, MA-29Rouwette, Etienne MA-27Rovatti, Riccardo MC-29Rozenknop, Antoine WA-03Rozga, Ante WC-21Rozin, Boris MA-10Rozycki, Rafal MB-14Ruan, Yu-Xun TD-40Rubio, Sergio WC-41Rudec, Tomislav WB-05Rufián-Lizana, Antonio TC-15Ruhland, Johannes TA-37Ruiz, Ana Belen TB-38Ruiz, José L. MA-36Ruiz, Marc MB-08Ruiz, Natalia WC-23Ruiz, Ruben WA-11Ruiz-Garzón, Gabriel TC-15Ruiz-Hernandez, Diego MD-17Rusdiana, Siti TB-34Rustem, Berc MD-19Rustogi, Kabir TB-02Ruuska, Sauli TA-38Ruza, Vecislavs MD-12Ruzika, Stefan WB-05Ruziyeva, Alina WB-15Ryan, David WC-14

SSaatkamp, Helmut MC-26Sabeghi, Narjes TB-18Saborido Infantes, Rubén TB-38Saboya, Martha MB-31Sadeh, Arik WA-27Sadoghi, Amirhossein TA-08Sadyadharma, Hendaru TA-34, TC-34Sadıç, Senay TD-07Sørensen, Michael MA-29Sürmeli, Gökhan TC-44Süss, Philipp MC-38Sörensen, Kenneth MD-01, WD-07,

WA-14, TD-22, WD-24

Safia, Kedad-SidhoumTD-09, WD-09Sagarra, Martí MA-18Sagbansua, Lutfu TD-06Sager, Sebastian WB-45Saha, Rajib TA-37Sahin, Oz MB-25Sahman, Mehmet Akif WD-10,

WA-30Sakalauskas, Leonidas MA-30,

WB-42Salani, Matteo MB-11Salasc, Baptiste WA-26Salazar González, Juan José WA-07Saldanha, Rodney WB-21, MA-40Saldanha-da-Gama, FranciscoMA-17,

MC-17, TB-17Sales, Alessandro WB-14Saliba, Sleman MB-02Salkın, Sultan Ceren MA-42Salles Neto, Luiz TA-09, TB-33Salman, Sibel MC-03Salo, Ahti TD-27, WC-30Salvador, Manuel MD-18Samii, Behzad WA-26, MB-45Samouylov, Konstantin TD-36San-José, Luis A. MD-10Sanchez-rodriguez, Luis TB-35Sander, Onderstal MB-18Sandikci, Burhaneddin WD-37Sano, Marcello MB-25Santos Martin, M. Teresa TD-42Santos, Dorabella MB-08, MC-08Santos, Fellipe WB-21, MA-40Santos, Jorge TA-31Santos, Maristela MA-13, MD-13Santos, Paulo Sergio WC-23Santos, Sérgio WC-32, MD-35Santos-Peñate, Dolores R. TC-18Sarabando, Paula MC-27Saraiva, Erlandson WB-30Sarto, José Luis MD-18Sartık, Yunus MB-26Satiroglu, Sait TB-20Sato, Kimitoshi MA-44Sato, Yuji WD-13Sawada, Kiyoshi WB-05Sawaki, Katsushige MA-44Sbihi, Abdelkader WB-05Scandolo, Giacomo MB-19Scarpel, Rodrigo WA-29Schüle, Ingmar MC-38Schöbel, AnitaTA-01, TB-12, MB-29,

WA-38Schäfer, Andreas TA-28Schäfer, Ulrich TD-45Schönberger, Jörn MA-11, WA-45Schöpperl, Andreas MA-13Schatka, Anne TB-25Schaudt, Oliver WB-08Scheimberg, Susana WC-23Scheithauer, Guntram TA-09Schewe, Lars TB-29Schlechte, Thomas WA-12Schlegel, Patrice WA-40Schmehl, Meike TC-25Schmid, Verena WA-07

Schmidt, Christiane WB-17Schmidt, Günter WD-20Schmidt, Kerstin WD-13Schmidt, Marie TB-12Schnell, Alexander WA-02Schoen, Cornelia TA-35Schoenauer, Marc WB-40Scholten, Lisa TB-27Scholz-Reiter, Bernd WD-10Schoppmeyer, Christian WA-30Schröder, Andreas TC-32Schrödl, Holger MB-14Schramme, Torben WA-02Schreiber, Ulrich MD-25Schreider, Sergei WD-06Schroeder, Werner WA-10Schultmann, Frank TD-20, WD-27,

TD-32Schultz, Rüdiger MD-29, TB-29,

WB-43Schwab, Melanie TC-07Schwaninger, Markus WC-40Schwarze, Silvia MC-05Schweiger, Jonas TB-29, TA-43Schwindt, Christoph WB-02Sciomachen, Anna TA-11Scozzari, Andrea MA-19, MB-20,

WB-20Sebag, Michèle WB-40Sedlak, Otilija WB-10Seferlis, Panos WC-45Segovia-Gonzalez, M. Manuela WB-

21Segura Maroto, Marina MD-26,

TD-26Segura, Baldomero MD-26, TD-26Seidmann, Abraham WC-34, TA-37Seifert, Matthias MC-27Seifert, Ralf W. MB-35Sel, Çagrı TC-08Selih, Jana WB-33Selim, Shokri MA-10Sellami, Khaled TA-31Selosse, Sandrine WA-25, WB-25Sels, Peter TB-12Sena, Galeno TA-09Sender, Julia TB-03Sener, Emrah TB-20Sener, Nazmi MC-12Sengul, Serkan TC-22Senna, Fernanda MC-37Senturk, Sevil WC-22Serafini, Paolo TB-39Serpell, Martin TD-30Sethi, Suresh MA-45Sevaux, Marc MC-03, TA-06, WD-09Seyhun, Omer Kayhan MA-20Sezer, Demet WC-19, WD-19Sezer, Devin WC-42Shabtay, Dvir MD-14Shafransky, Yakov TA-02, TA-14Shahraki, Narges TB-11Sharifi, Mohammad Bagher MC-40Sharma, Megha MB-45Shaw, Duncan TA-32Shen, Hwa-Rong MA-39

286


Sherali, Hanif WD-15Sherker, Shauna MB-25Shevchenko, Yaroslav WA-41Shi, Xiaoning MC-05Shibata, Takashi WC-20Shih, Su-Chuan TA-41Shikhman, Vladimir TA-15, MC-28,

MD-28Shim, Yohan WA-26Shlakhter, Alex TA-18Shlyk, Vladimir WD-05Shorgin, Sergey TD-36Shufan, Elad WB-14Sicilia, Joaquín MD-10Siddiqui, Afzal MA-25Sidorov, Sergey WA-23Siebert, Xavier TD-40Siepermann, Christoph MD-03Siepermann, Markus TD-10Siestrup, Guido WD-25Sigurd, Mikkel M. WC-09Silalahi, Meslin TC-34Silva, Aneirson WC-38Silva, Elsa MC-09Silveira, João MD-35Simaria, Sofia TB-30Simon de Blas, Clara MC-33, MC-36,

TD-42Simon Martin, Jose MC-36Simoncicova, Katarina WD-24Simonetti, Luidi MA-06Simonis, Helmut WA-03Simroth, Axel MC-07Sindhya, Karthik TB-38Singh, Sanjeet WB-32Singh, Shivraj MA-10Singh, Vikram jeet TA-21, MC-30Sinha, Ankur MA-41Sinske, Marcel MC-28Sinuany-Stern, Zilla TB-36Siqueira, Hugo TD-24Sirvent, Inmaculada MA-36Sisman, Fatma Nurten WD-28Skalska, Jolanta MB-04Skar, Christian TA-43Skouri, Konstantina MC-06Sladky, Karel MB-23Sledneu, Dzmitry TA-14Slijepcevic-Manger, Tatjana MC-18Slimani, Hachem MC-23Sliwinski, Tomasz MD-04Slomp, Jannes WA-17Smedley, Jo TA-32Smet, Pieter TC-14Smeulders, Bart TB-22Smid, Martin TB-43Smirnova, Nadezhda WA-34Smith, Chris TA-32Smith, Honora MB-17, WC-37Smith, J. MacGregor TC-36Smith, Jim TD-30Smith, Leanne TB-37Smith, Raymond TC-44Smolina, Iryna WC-41Smriglio, Stefano MA-29Soares de Mello, João Carlos WC-32

Sobeyko, Oleh MC-02Sobotka, Fabian TB-29Sobotka, Olena TD-20Socha, Leslaw TB-22Sofronov, Georgy WD-42Sohn, So Young MB-21, MC-35Sokolov, Boris TA-07Soler-Dominguez, Amparo MB-30Solignac, Manon TD-25Solimanpur, Maghsud MC-10Sologub, Roman TD-44Solsona, Francesc MD-43Song, Xiang MA-07, TA-35Songmene, Victor TB-45Sonnberger, Jana MD-27Sopot, Eugen TA-05Sopp, Karina TD-45Soria-Comas, Jordi TC-30Sosic, Greys MB-24Sotiros, Dimitrios-Georgios MB-36Soto-Silva, Wladimir TD-26Soualah, Sofiane WC-12Soukhal, Ameur TB-14Soumis, Francois WB-04Sourd, Francis WA-12Sousa Santos, Beatriz TD-11Sousa, Luiz Carlos WB-42Souza, Reinaldo MB-06, TB-24Souza, Renato Rocha WC-27Soyertem, Mustafa MA-23Soysal, Mehmet TB-25Spachis, Alexandra TC-21Speck, Jochen WA-03Speckenmeyer, Ewald WA-08Spengler, Thomas WB-13, WC-13,

WD-13, TB-25Sperandio, Fabricio TD-37Speranza, M. Grazia TE-01Spieksma, Frits TC-09, WC-16,

MB-18, TB-22, MA-34Spinler, Stefan WC-07Spliet, Remy WA-07, TC-40Sprengel, Eva-Maria TD-05Squires, Hazel TA-32Srour, Jordan MC-11Stadler, Johannes TA-21Stadler, Michael MA-25Stützle, Thomas TD-19, MA-41Stöcker, Martin MB-15Stöckl, Stefan TA-21, WD-21Stålhane, Magnus WC-09Staino, Alessandro MC-18Stalinski, Piotr MD-10Stamure, Iveta WB-33Stancu, Andreea Madalina MC-23Stancu-Minasian, Ioan MC-23Stangl, Claudia MD-29, TB-29Stanojevic, Bogdana WB-09Stanojevic, Milan WB-09Staube, Tatjana MA-33Staykov, Boris WB-27Steele Santos, Paulo Eduardo TD-29Stefani, Silvana MB-19Stefanovic, Pavel WD-44Steglich, Mike WD-33Steinhardt, Claudius WA-45

Steponavice, Ingrida TA-38Sterna, Malgorzata MA-02Sternbeck, Michael MB-35Stetsyuk, Petro TA-22, TC-22Steuer, Ralph E. WB-20Stewart, Theodor MB-27, TD-34Still, Georg WB-38Stokes, Klara TD-30Stokic, Dejan MB-20Stoklasa, Jan MB-22Stolletz, Raik MD-02, TD-14, MA-42Stoyan, Yuri MB-09, MC-09Strandmark, Petter TD-08Stranieri, Andrew TB-31Stray, Jonas WC-04Strijov, Vadim TD-44Strusevich, Vitaly TB-02Studniarski, Marcin TD-15Stukalin, Dmitriy WB-39Stummer, Christian MD-27Stutzer, Michael TC-21Stygar, Anna TB-23, MC-26Su, Chia-hsien MA-39Suarez-Vega, Rafael TC-18Subbiah, Subanatarajan WA-30Subramanian, Anand WB-03Subrt, Tomas TA-11Sucu, Meral WD-20Suhl, Leena WA-02Summerauer, Wolfgang WD-02Sun, Bin TC-42Sun, Heming WD-23Sun, Li TD-35Sun, Wei WC-27Sundstroem, Olle WD-45Surer, Ozge TA-41Suykens, Johan WC-29Suzuki, Teruyoshi MD-19Suzuki, Tsutomu WD-18Swiercz, Aleksandra MB-37Syafari, Syafari TD-17Syahrini, Intan TB-34Syntetos, Aris TA-24Sys, Christa MC-05Szabo, Jacint WD-45Szafarz, Ariane MC-21Szantai, Tamas MD-43Szołtysek, Jacek TA-37Szubert, Marcin WC-03Szymczak, Jaroslaw MA-02

TTa, Anh Son WB-17Tadic, Ivana MA-21, MB-33Tüncel, Özge MC-03TÜtÜncÜ, G.yazgi TC-14Taheri, Hassan TA-06, TB-18Tainio, Marko TD-32Takahashi, Shingo MC-45Talasova, Jana MB-22Talbi, El-ghazali WC-03, MB-45Talla Nobibon, Fabrice MA-34Tamaki, Mitsushi MB-30Tamarit, Jose MA-09, MB-09, MC-09Tambovceva, Tatjana WC-33Tambunan, Hardi TD-17

287


Tammer, Christiane MB-38, WA-38Tamosaitiene, Jolanta WA-33, WC-33Tan, Tarkan TB-25Tanaka, Katsuaki MA-18, TB-28Tanaka, Ken-ichi WB-38Tanaka, Tamaki WD-23, TC-38Tancrez, Jean-Sébastien MB-35,

WC-42Taner, Mustafa Egemen TC-07,

WC-12Tang, Ling-Lang MC-39Tanino, Tetsuzo WD-23, WA-44Tanrisever, Fehmi WC-25Tansel, Barbaros TC-17Tanyeri, Basak TD-42Tarashnina, Svetlana WA-34Tarczynski, Grzegorz TA-03Tardella, Fabio MB-20, WB-20Tareghian, Hamed Reza TB-18Tarel, Guillaume Jean TD-16Tarim, Armagan TA-24, TC-45Tarkkanen, Suvi TD-38Tarnawski, Tomasz MD-04Tas, Duygu WA-43Taschini, Luca MD-25Tatarinova, Tatiana MC-37Tatlidil, Hüseyin WB-11, TB-20,

MA-33Tavana, Mina MC-15Tavella, Elena MB-32Taylan, Ali Sabri TB-20, TD-31,

WD-43Taylor, Stephen WA-36Tas, Oktay WA-20Taskan, Cüneyt TA-26Tchegho, Geraldine WA-21Tchemisova, Tatiana TA-15Teboulle, Marc WA-23Teghem, Jacques MD-38Teichert, Katrin MC-38Teixeira, Joyce MC-36Tejada, Juan MA-31, MB-31Tejada, Macarena TC-06Tekin, Muge WD-21Telek, Miklós TC-36Temocin, Büsra TB-21Teo, Chee Chong MC-05Teo, Kwong Meng WB-23Ternero, Isabel MD-17Tervonen, Tommi WC-26, TC-40Tezcaner, Diclehan MA-03Thénié, Julien TD-16Thanassoulis, Emmanuel MB-36,

TA-36Theis, Dirk Oliver WC-16, WC-39Theocharis, Zoe TC-24Thiao, Mamadou WB-17Thomas, Lyn MC-20Thomson, Mary TC-24Thuring, Fredrik TB-19Tierney, Kevin WC-03Tiesler, Stephan TA-37Tietz, Christoph WA-24Tikkanen, Olli-Pekka MD-38Tilson, Vera TA-37Tirado, Gregorio MD-05

Tiseanu, Catalin Stefan WD-45Tlili, Youssef WA-27Tobin, Patrick WA-28Todosijevic, Raca WA-06Tofallis, Chris WB-30Toh, Kim-Chuan WA-16Tokareva, Julia TD-22Tokmak, Ismail MC-24, TD-34Toledo, Franklina MB-09, MC-13,

TB-30Tolstad Uggen, Kristin MA-26Tomasgard, Asgeir TC-29, MC-43,

TA-43Tomaskova, Hana MB-12Tomat, Luka WD-25Tombak, Mihkel TB-22Tonon, Daniela MD-04Topaloglu, Huseyin MA-44Topaloglu, Seyda WA-11Topcu, Y. IlkerMC-10, WB-10, TA-41Tor, Ali Hakan MC-16Torezzan, Cristiano MD-09Torr, Philip TD-08, TB-44Torra, Vicenc TD-30Tortosa-Ausina, Emili MB-30Tortum, Ahmet WA-24Toth, Paolo TC-04Toumi, Abdelamalek WD-17Tournamille, Jean-François TB-14,

WA-30Tourniaire, Emeric WA-09Toyasaki, Fuminori TC-23Tozkan, Didem MA-23Trafalis, Theodore WB-29Tragler, Gernot MC-04Tralhão, Lino WB-27Tran, Nguyen Khoi WD-12Tran, Trung Hieu WB-23Transchel, Sandra WC-10Trautmann, Norbert TA-30Trautsamwieser, Andrea WC-37Traversi, Emiliano MB-29Traxler, Renate MB-13Tricoire, Benoît TA-07Tricoire, Fabien TB-07Triki, Chefi TD-13Trodden, Paul MD-30Tryphonas, Marinos MC-18Trzaskalik, Tadeusz MB-27, MA-33Trzpiot, Grazyna TA-37, TD-43Tsagkas, Vassilis MA-32Tsai, Hung-Yi TB-31Tsai, Ming-Feng TD-40Tsekouras, Kostas WA-24Tsivtsivadze, Evgeni WA-40Tsoukiàs, Alexis TC-27Tsyurmasto, Peter WD-43Turan, Belma WA-07Turan, Filiz MD-02, MC-35Turan, Hasan Huseyin MC-12Turanoglu, Ebru WC-22Turco, Marco TB-11Turgut, Hakan MC-24, TD-34Turkay, Metin TC-03, TB-11, TB-33Turkel, Gokcen MA-03Turkensteen, Marcel TB-08

Turskis, Zenonas WC-33Tuyttens, Daniel MD-38Tus Isık, Aysegül MC-41Twaróg, Sebastian TA-37Twyman, Matt TC-24Tzanidakis, Konstantinos TD-32Tzeng, Gwo-HshiungMB-27, MA-33,

MD-40, MB-41Tzu-hsuan, Kuo MD-39, MA-40,

MD-40

UUdias, Angel MD-33, WD-38Ueno, Nobuyuki WA-10, TA-45Uldry, Marc MD-06Ulengin, Burc MB-03Ulengin, Fusun MB-03, TB-34Ulusoy, Gündüz WC-02Umang, Nitish WD-12Uney-Yuksektepe, Fadime MA-37Ungureanu, Valeriu MA-34Unzueta, Aitziber WA-43Uratani, Tadashi WC-20Uray, Nimet TB-34Urban, Josef WA-40Urbanski, Ryszard MA-23Uribe, Valeria E. WD-44Uriol, Juan MD-26Uristani, Angelo WD-43Urosevic, Dragan WD-09Uryasev, Stan MC-21, WD-43Usberti, Fábio MA-25Ustun, Yildirim WC-42Utley, Martin WC-37Uusivuori, Jussi TC-26Uzelac, Filip TA-19Uzun Araz, Ozlem MB-03

VVaaben, Bo MC-05Völker, Markus WA-03Vaiceliunas, Saulius MA-37Vaira, Gintaras MB-07Valdes, Teofilo WA-44Valle, Marco TC-27Vallejos, Sofía J. WD-44Valls, Vicente MC-14Valouxis, Christos TD-09van ’t Wout, Joel TC-12van Berkum, Eric WB-38van Dalen, Jan TA-24van de Ven, Rien WC-28Van Delft, Christian TD-07van den Brink, Rene MA-34, WA-34Van den Broeck, Dennis WA-35Van Den Broek, Machteld WD-38van den Heever, Susara WD-45Van den Poel, Dirk WA-06, TD-31van der Laan, Erwin TA-03, TA-24Van der Meer, Robert TD-14van der Mijden, Tom WB-28van der Ster, Suzanne WA-28van der Vecht, Bob WB-28van der Vorst, Jack MC-17, TB-25van Elst, Nicole WA-28Van Essen, Theresia WB-37

288


Van Huele, Christophe WA-14van Laarhoven, Twan WA-40van Merode, Frits TB-37Van Peteghem, Vincent WD-02van Valkenhoef, Gert WC-26van Vuuren, Jan WC-04Van Wassenhove, Luk WC-10Van Woensel, Tom WA-43Vancroonenburg, Wim WD-09Vandaele, Arnaud MB-15Vanden Berghe, Greet TC-14Vanderpooten, Daniel WC-26Vanhoucke, Mario WA-02, WD-02,

WA-14, MB-45Vannieuwenhuyse, Marjorie WD-27Vanovermeire, Christine TD-22Vansteenwegen, Pieter MD-06,

WD-07, TB-12Varas, Mauricio TC-26Vasilakis, Christos WC-37Vasquez, Michel TD-09Vassilakopoulos, Michael WB-37Vassilakopoulou, Polyxeni MA-32Vaz, Clara TB-36Vazquez Novoa, Nadia TA-36Veelenturf, Lucas TD-12Velasco, Luis MB-08Velten, Sebastian MC-38Venceslau, Helder MD-23, WB-42Veneziano, Thomas TC-39Vera de Serio, Virginia N. TB-15Vera, Jorge TC-26Verbas, Ismail Omer MC-11Verbeeck, Cédric MD-06Verbraken, Thomas TB-31Vercamer, Dauwe WA-06Verdegay, Josè L. MA-12Verleye, Derek TC-34Verly, Céline TB-39Vermeulen, Jack WC-28Verscheure, Olivier WD-45Verter, Vedat MC-17Vespucci, Maria TeresaMC-43, TA-43Vetschera, Rudolf TD-20, MD-27Vial, Jean-Philippe TD-07Viana, Ana MA-25, WA-37Viana, Joe WD-37Vianna, William TB-32Viappiani, Paolo TB-40Vicente, Joaquim Jorge TA-10Vickman, Sara TD-27Vicko, Gabriela MD-18Vidal, Thibaut WB-03, WC-07Vidal-Puga, Juan MA-24Vidalis, Michael MB-10Viegas, José MA-04Vieira, Cristina C. TB-06Vieira, Douglas WB-21, MA-40Vieira, Israel TB-37, TB-42Vielle, Marc TD-16Viertl, Reinhard MC-12Vigerske, Stefan MD-29Vigna, Elena TA-19Vigo, Daniele WD-07, MD-42Vilkkumaa, Eeva TD-27Villas-Boas, Sergio B. MA-16

Villegas, Juan G. MA-07Vilutiene, Tatjana WA-33Vimont, Yannick TD-09Vinue, Guillermo WD-29Virtanen, Kai TB-42Visagie, Stephan WA-04, WC-04Visser, Fernando TC-20Vitko, Anton TB-09, WD-41Vitoriano, Begoña MC-33Vitsentiy, Vitaliy WB-43Vittucci Marzetti, Guiseppe WB-34Vizcaino, José Federico MA-25Vizvari, Bela WC-08, TB-11Vocaturo, Francesca TC-11Vogel, Silvia MB-43Vogt-Schilb, Adrien TD-25Vojtas, Peter WB-26, TD-41Volkovich, Zeev (Vladimir) WA-42,

WD-42Voll, Robert WB-12Volling, Thomas WB-13, WC-13,

WD-13von Selasinsky, Alexander TC-32Von Zuben, Fernando J. MA-25Voss, Stefan MC-05Vozdolska, Ralitza WC-14Vrobel, Susie TD-28Vujosevic, Mirko WB-09Vuppala, Sunil Kumar WB-25

WWacziarg, Adrien WA-25Waegeman, Willem TD-40Wäscher, Gerhard TE-01, MC-09Wagelmans, Albert TD-12Wagner, Andrea WA-38Wagner, Hermann-Josef MD-25Wakolbinger, Lea M. MD-27Wakolbinger, Tina TC-23, MD-27Waligora, Grzegorz MB-14Wallace, Stein W. MB-29, TA-43Wallenius, Jyrki TB-39, MA-41Wallot, Sebastian TD-28Walls, Lesley TA-16Walther, Grit TB-25Walther, Ursula TD-21Walukiewicz, Stanislaw MC-30Wan, Yat-wah TC-42Wang, Hsiao-Fan MB-44Wang, Jiang-Shan TB-31Wang, Min TC-45Wang, Pei-Chun TA-09Wang, Shin-Yun MB-39Wang, Song TC-23Wang, Tai-Yue MD-40Wang, Xiaolu MC-12Wang, Xiaoyu MC-05Wang, Yang-Sheng MA-39Wang, Yi-Hsien MA-39Waraich, Rashid TB-35Watanabe, Takahiro MD-28Waterer, Hamish WC-14Wauters, Tony WD-09Weber, Christoph TA-25, TC-32Weber, Gerhard-Wilhelm TB-21,

MD-23, TB-23, WC-35

Weber, Richard MC-20, WB-22Wedel, Woldemar TC-10Wedley, William MD-41Weglarz, Jan MA-02Weiblen, Judith TC-07Weider, Steffen WA-12Weismantel, Robert MD-29Weistroffer, Heinz Roland TD-38Weitzl, Wolfgang TD-20Wen, Liang TD-03Wendell, Richard E. MD-36Wendt, Oliver TC-10Wensing, Thomas WA-13Wenstøp, Fred WC-31Werner, Adrian TC-12, MA-26Werner, Axel MD-08Werners, Brigitte TA-17Westerlund, Magnus MA-22Westphal, Stephan MC-42Wheat, David WB-41White, Amelia WC-11Wickert, Toni WA-06Wickström, Anna-Laura MC-28Widmer, Marino MD-06Wiecek, Margaret MD-38Wiedenmann, Susanne WC-25Wieloch, Bartosz WC-03Wiesemann, Wolfram WB-42Wijnmalen, Diederik J.D. WC-28,

MC-32Wilbaut, Christophe MB-02Wildebrand, Hendrik WD-25Wilgenbus, Erich WD-29Wilkin, Carla WA-27Williams, Hilary Paul TA-27Williams, Janet TB-37, TB-42Wilson, Michel WB-02Windeatt, Terry TC-44Windridge, David TC-44Winkelkotte, Tobias TD-35Winkelmann, Axel TD-24Winkenbach, Matthias WC-07Witteveen, Cees WB-02Wittmann-Hohlbein, Martina MC-29Wojtkiewicz, Tomasz TA-14Wolfsteiner, Elisabeth TD-20Woll, Oliver TA-25, TC-32Wollenberg, Tobias WB-43Wollenweber, Jens TD-35Wolter, Katinka TD-36Won, Daehan MA-37Wong, Hartanto TA-10Wong, Man Hong TC-20Wotzlaw, Andreas WA-08Wozabal, David TC-19Wozny, Günter MB-15Wright, Mike MB-06Wu, MingHung MA-39Wu, Shuo-Jye WD-19Wu, Xiaoli MB-44Wyld, Lynda WD-31Weglarz, Andrzej TB-23

XXavier Candido, Lilian Caroline TB-

09

289


Xavier de Amorim, Rainer WC-11Xavier, Adilson EliasMA-16, MD-23,

WB-42, TC-44Xie, Ying TB-31Xu, Dong-Ling TA-38Xu, Zhaohui WC-15Xufre, Patricia TB-24

YYüksel, Ihsan WD-26, MD-41Yagi, Kyoko MD-19Yagiura, Mutsunori TD-09Yahi, Zahra TA-28Yahyayi, Mohsen TB-39Yamada, Syuuji WD-23Yan, Fei TB-44Yanasse, Horacio TC-02Yang, Chih-Hai WC-32Yang, Jian-Bo MB-27, TA-38Yang, Liu MC-10Yang, Shih-Ting MB-40, MC-40Yang, Ta-hui MA-04Yanovskaya, Elena WA-34Yapar, Güçkan WD-43Yapici Pehlivan, Nimet WC-19,

TD-23, WD-42Yarar, Alpaslan TA-26Yarar, Hülya TA-26Yaskov, Georgiy MC-09Yassa, Sonia TA-06Yasue, Akinori WC-28Yatskiv, Irina TD-02, WA-44Yatsko, Andrew TB-31Yazdan Shenas, Nima MD-10Yazenin, Alexander MA-21Yazhemsky, Ekaterina TB-36Yearworth, Mike MD-32Yedidsion, Liron MD-14Yeh, Li-Ting TC-25Yelkenci Kose, Simge MC-06,

MD-39, WC-42Yen, Hsin-Yi MC-40

Yeomans, Julian Scott WA-27Yerlikaya Ozkurt, Fatma WC-35Yeung, Thomas WB-21Yigit, Vecihi MB-12Yildirak, Kasirga TD-31Yildirim, U.Mahir TD-03Yildiz, Abdurrahman MD-06Yildiz, Munevvere WB-22Yilmaz, Aycan WC-21Yilmaz, Tarik WB-19Yoda, Kunikazu MD-43Yolcu Okur, Yeliz TB-21Yoon, Moon-Gil MD-45York, Darren MD-32Yozgatligil, Ceylan WB-24Yu, Jiun-Yu TC-37Yu, Po-Lung MB-39Yuan, Zhi TD-19Yue, Wuyi WC-44Yulugkural, Yildiz WD-22Yurshevich, Elena TB-42Yurtkuran, Alkin TB-06Yılmaz, Adem WC-19

ZZabih, Ramin TB-44Zachary, Daniel MA-03Zadnik Stirn, Lidija WA-39Zagorskis, Viktors TD-36Zainal Abidin, Norhaslinda WB-41Zaleskis, Genadijs TA-03Zamarripa, Miguel TB-02Zamorano de Acha, Emilio TD-14Zanella, Andreia MD-36Zanjirani Farahani, Reza TB-33Zaourar, Sofia WA-03Zaraté, Pascale WC-27Zarghami, Mahdi WC-27Zarzo, Alejandro WB-12Zavadskas, Edmundas Kazimieras

WC-33Zazgornik, Jan TC-26

Zbaganu, Gheorghita WA-20Zemkoho, Alain B. WB-15Zendehzaban, Sonia MC-19Zerbo, Boureima TA-06Zhang, Cong TC-05Zhang, Huizhen MA-35Zhang, Rachel MB-44Zhang, Shuzhong TC-20Zhang, Wenjuan MC-30Zhang, Xiaodong MC-37Zhang, Xiaofei WC-44Zhang, Yi TC-16Zhang, ZhongHua WD-40Zhao, Jing WD-31Zhukova, Ksenia TB-44Ziegler, Daniel TA-25Zigrino, Stefano MC-43Zilinskas, Antanas WA-17Zilinskas, Julius TC-18Zimmermann, Hans Georg WA-24Zimmermann, Hans-Jürgen TE-01Zimmermann, Karel MA-22Zimmermann, Uwe T. MA-06Zipf, Michael TC-32Zlatkovic, Matea WC-31Zolotova, Tatiana TB-21, TD-21Zolotykh, Nikolai WA-23Zouba, Mohammed TB-14Zuanon, Magalì E. MC-31, WC-34Zuccollo, James TB-16Zuidwijk, Rob TB-25

Cicková, Zuzana MD-03Iyit, Neslihan WD-19

Senyüzlüler, Filiz WA-11Šakalys, Algirdas MB-05, MD-05Škrabic Peric, BlankaMA-18, MB-21,

WC-21

290

SESSION INDEX

Monday, 8:30-10:00

MA-01: Keynote Lecture: Professor Ignacio Grossmann (RB-Alfa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1MA-02: Scheduling Applications (RB-Beta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1MA-03: Business Excellence in Logistics: Intelligent and Interactive Routing (RB-L1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1MA-04: Optimization of shared urban transportation systems (RB-L2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2MA-05: Maritime fleet size and mix (RB-L3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2MA-06: Matheuristics in Transportation (RB-Gamma) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3MA-07: Novel and emerging VRPs (RB-Eta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3MA-08: Network Optimization 1 (RB-Epsilon) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4MA-09: Cutting and Packing 1 (RB-Zeta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4MA-10: Applications (RB-Theta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5MA-11: Inculsion of foreign transportation ressources (RB-Iota) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5MA-12: Fuzzy Goal Programming and Optimization Methods (RB-Omicron) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6MA-13: Lot-sizing and Related Topics 1 (RB-Tau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6MA-14: Scheduling (RB-Omega) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7MA-16: Advances in Nonsmooth Optimization (RB-2103) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7MA-17: Location and Supply Chain Management (RB-2105) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8MA-18: Financial Crisis Modelling (RB-2107) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8MA-19: Risk management in commodities markets (RB-2111) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9MA-20: Financial Time Series Analysis and Forecasting (RB-2113) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9MA-21: Financial Mathematics and OR 1 (RB-2115) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10MA-22: Soft Computing and SCM Applications (RB-2117) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10MA-23: Generalized Differentiation (RB-Delta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11MA-24: Cooperation and Logistics (CC-A11) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11MA-25: Energy Distribution: Planning and Optimization (CC-A23) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12MA-26: OR Applications in Oil and Gas (CC-A24) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12MA-27: Empirical Research on Decision Processes - Methods & Findings (CC-A25) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13MA-28: OR in Quality Management 1 (CC-A27) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13MA-29: Mixed-Integer Quadratic Problems (CC-A29) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14MA-30: Fat Tail Models in Finance (CC-A31) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14MA-31: Game Theory and Social Networks 1 (CC-A33) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15MA-32: Combining Methods & Multimethodology (CC-A34) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15MA-33: OR for Business and Industrial Development (CC-A37) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16MA-34: Game Solutions (CC-A39) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16MA-35: OR Industry; OR Software (CC-A41) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17MA-36: DEA and Performance Measurement: Metodology 1 (CC-A43) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17MA-37: Bioinformatics I (CC-Act) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18MA-38: Integer Linear and Linear Multiobjective Optimization (HH-Colombus) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18MA-39: MADM Applications I (HH-Cousteau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19MA-40: MADM Applications V (HH-Livingstone) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19MA-41: Evolutionary Multiobjective Optimization (HH-Heyerdahl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20MA-42: Simulation 1 (BW-Amber) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20MA-43: Multistage Stochastic Programming (BW-Granite) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21MA-44: Airline Revenue Management (BW-Marble) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21MA-45: Advertising decisions in the supply chain (BW-Water) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

Monday, 10:30-12h00

MB-01: IFORS Invited Tutorial: Professor Erhan Erkut (RB-Alfa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23MB-02: New scheduling models and algorithms (RB-Beta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23MB-03: Business Excellence in Logistics: Future Transport and Production (RB-L1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23MB-04: Public Transport Perspectives within a Changing Mobility Environment (RB-L2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24MB-05: Maritime Transportation 1 (RB-L3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24MB-06: Design and parameter tuning of metaheurstics (RB-Gamma) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25MB-07: Neighborhood reduction strategies for vehicle routing (RB-Eta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25MB-08: Network Optimization 2 (RB-Epsilon) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26MB-09: Cutting and Packing 2 (RB-Zeta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26MB-10: Performance Analysis and Inventory (RB-Theta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

291

SESSION INDEX EURO 25 - Vilnius 2012

MB-11: Transport Modelling (RB-Iota) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27MB-12: Fuzzy Relations & Neural Networks (RB-Omicron) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28MB-13: Lot-sizing and Related Topics 2 (RB-Tau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28MB-14: Discrete-continuous scheduling (RB-Omega) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29MB-15: Nonlinear Optimization and Applications 1 (RB-2101) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29MB-16: Hybrid Algorithms & Nonsmooth Optimization (RB-2103) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30MB-17: Location and GIS (RB-2105) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30MB-18: Risk Analysis and Options (RB-2107) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31MB-19: Energy and Environmental markets (RB-2111) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31MB-20: Data Mining and Risk Analysis (RB-2113) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31MB-21: Financial Mathematics and OR 2 (RB-2115) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32MB-22: Multiple-criteria evaluation and decision-making under fuzziness: theory and applications (RB-2117) . . . . . . . . . . . . . 32MB-23: Optimality Conditions and Stability (RB-Delta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33MB-24: Centralized Inventory and Production (CC-A11) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33MB-25: Practical Issues of Environmental Management (CC-A23) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34MB-26: Recent Research Advances in the Oil Sector (CC-A24) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34MB-27: MCDA and Decision Processes (CC-A25) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34MB-28: OR in Quality Management 2 (CC-A27) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35MB-29: Robust Optimization and Non-Convex Integer Programming (CC-A29) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35MB-30: Emerging Applications in Finance (CC-A31) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36MB-31: Game Theory and Social Networks 2 (CC-A33) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36MB-32: Understanding the Practice of Soft OR Interventions I (CC-A34) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37MB-33: OR and Modelling for Sustainable Development (CC-A37) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37MB-34: Cooperative games: dynamics, robustness and related topics (CC-A39) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38MB-35: Assortment and Replenishment Planning (CC-A41) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38MB-36: DEA and Performance Measurement: Methodology 2 (CC-A43) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38MB-37: Bioinformatics II (CC-Act) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39MB-38: Duality in Multi-objective Optimization (HH-Colombus) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39MB-39: MADM Applications II (HH-Cousteau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40MB-40: MADM Applications VI (HH-Livingstone) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40MB-41: AHP/ANP 1 (HH-Heyerdahl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41MB-42: EURO Doctoral Dissertation Award (EDDA 2012) (BW-Amber) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41MB-43: Stochastic Programming - Stability, Estimation (BW-Granite) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41MB-44: Inventory, Pricing, and Customer Behavior (BW-Marble) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42MB-45: Applications of Pricing and Revenue Management (BW-Water) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

Monday, 12:30-14:00

MC-01: Keynote Lecture: Professor Ceyda Oguz (RB-Alfa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43MC-02: Scheduling Approaches for Complex Manufacturing Systems (RB-Beta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43MC-03: Business Excellence in Logistics: Advanced Vehicle and Inventory Routing (RB-L1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44MC-04: Optimal Control I (RB-L2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44MC-05: Liner shipping optimization (RB-L3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45MC-06: Production, planning and facilty location (RB-Gamma) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45MC-07: Applications of Vehicle Routing (RB-Eta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46MC-08: Network Optimization 3 (RB-Epsilon) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46MC-09: Cutting and Packing 3 (RB-Zeta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47MC-10: Non-deterministic models (RB-Theta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47MC-11: Transport Planning (RB-Iota) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48MC-12: Application of fuzzy logic and soft methodologies in decision-making and financial modelling (RB-Omicron) . . . . . . 48MC-13: Lot-sizing and Related Topics 3 (RB-Tau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49MC-14: Resource allocation and scheduling in practice: further complexity (RB-Omega) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49MC-15: Nonlinear Optimization and Applications 2 (RB-2101) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50MC-16: Numerical methods of nonsmooth optimization (RB-2103) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50MC-17: Reverse Logistics Network Design (RB-2105) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51MC-18: Financial Modelling (RB-2107) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51MC-19: Quantitative approaches to bank performance and asset evaluation (RB-2111) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52MC-20: Risk Analysis and Credit Scoring (RB-2113) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52MC-21: Financial Mathematics and OR 3 (RB-2115) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53MC-23: Generalized Convexity and Optimization (RB-Delta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53MC-24: OR in Human Resources (CC-A11) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54MC-26: OR in Agriculture I (CC-A24) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54MC-27: Decision Analysis and Group Decision Making (CC-A25) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55MC-28: Nonsmooth Optimization I (CC-A27) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

292

EURO 25 - Vilnius 2012 SESSION INDEX

MC-29: Algorithms and Methods for MINLP problems (CC-A29) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55MC-30: Emerging Applications in Production Management (CC-A31) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56MC-31: Mathematical methods for decision support in energy and industrial sectors (CC-A33) . . . . . . . . . . . . . . . . . . . . . . . . . . . 56MC-32: Understanding the Practice of Soft OR Interventions II (CC-A34) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57MC-33: Humanitarian Logistics and Disaster Management (CC-A37) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57MC-34: Topics in Cooperative Game Theory (CC-A39) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58MC-35: Engineering Optimization (CC-A41) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58MC-36: DEA and Performance Measurement: Applications 1 (CC-A43) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59MC-37: Bioinformatics III (CC-Act) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59MC-38: Multicriteria Decision Support Systems in Industry (HH-Colombus) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60MC-39: MADM Applications III (HH-Cousteau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60MC-40: MADM Applications VII (HH-Livingstone) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61MC-41: AHP/ANP 2 (HH-Heyerdahl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62MC-42: EEPA 2012 - 1 (BW-Amber) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62MC-43: Stochastic Programming: Applications to the energy sector (BW-Granite) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62MC-44: Incorporating uncertainty in Energy Management Optimization Problems: A long and winding road (BW-Marble) . . 63MC-45: Management accounting and management control 1 (BW-Water) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Monday, 14:30-16:00

MD-01: Tutorial Lecture: Professor Kenneth Sörensen (RB-Alfa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64MD-02: Scheduling models in Logistics (RB-Beta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64MD-03: Logistics, Transportation, Traffic - 1 (RB-L1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65MD-04: Optimal Control II (RB-L2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65MD-05: Maritime Transportation 2 (RB-L3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66MD-06: Vehicle routing problems (RB-Gamma) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66MD-08: Network Optimization 4 (RB-Epsilon) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67MD-09: Cutting and Packing 4 (RB-Zeta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67MD-10: Inventory Management 1 (RB-Theta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68MD-11: Traffic Management (RB-Iota) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68MD-12: Decision-making and modelling under fuzziness (RB-Omicron) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69MD-13: Lot-sizing and Related Topics 4 (RB-Tau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69MD-14: Scheduling with variable parameters I (RB-Omega) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70MD-15: Nonlinear Optimization and Applications 3 (RB-2101) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70MD-16: Uncertainty Analysis in Energy and Water Resources (RB-2103) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71MD-17: Further Applications of Location Analysis (RB-2105) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71MD-18: Analysis of Price Dynamics and Herd Behavior (RB-2107) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72MD-19: Financial Decision Making: Application of Mathematical Optimization (RB-2111) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72MD-20: Energy consumption and price forecasting (RB-2113) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73MD-21: Financial Mathematics and OR 4 (RB-2115) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73MD-23: Various Advances in Generalized Differentiation and Optimization (RB-Delta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73MD-24: Pioneers in Operations Research (CC-A11) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74MD-25: Renewable Energies (CC-A23) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74MD-26: OR in Forestry I (CC-A24) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75MD-27: Agent-based Modeling of Diffusion Processes (CC-A25) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75MD-28: Nonsmooth Optimization II (CC-A27) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76MD-29: Selected Topics on Mixed-Integer Non-Linear Programming (CC-A29) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76MD-30: Emerging Applications in Logistics (CC-A31) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77MD-32: Theoretical developments in Soft OR (CC-A34) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77MD-33: OR for Energy and Resource Development in Developing Countries (CC-A37) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78MD-34: Assignment and multicriteria games (CC-A39) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78MD-35: DEA and Performance Measurement: Applications 10 (CC-A41) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78MD-36: DEA and Performance Measurement: Applications 2 (CC-A43) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79MD-37: Bioinformatics IV (CC-Act) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79MD-38: Combinatorial Multiobjective Optimization (HH-Colombus) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80MD-39: MADM Applications IV (HH-Cousteau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80MD-40: MADM Applications VIII (HH-Livingstone) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81MD-41: AHP/ANP 3 (HH-Heyerdahl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82MD-42: EEPA 2012 - 2 (BW-Amber) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82MD-43: Probabilistic Programming (BW-Granite) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82MD-44: Topics in Revenue Management (BW-Marble) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83MD-45: Management accounting and management control 2 (BW-Water) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

293


Monday, 16:30-17:30

ME-01: Plenary Lecture: Professor Finn Kydland (Opera) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

Tuesday, 8:30-10:00

TA-01: Tutorial Lecture: Professor Anita Schöbel (RB-Alfa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85TA-02: Graph Models in Logistics and Scheduling (RB-Beta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85TA-03: Logistics, Transportation, Traffic - 2 (RB-L1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86TA-04: Air Transportation and Propagation (RB-L2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86TA-05: Maritime Transportation in Offshore Industry (RB-L3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87TA-06: Engineering and Hybrid search (RB-Gamma) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87TA-07: Vehicle routing in practical settings (RB-Eta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88TA-08: Supply Chain Design (RB-Epsilon) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88TA-09: Cutting and Packing 5 (RB-Zeta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89TA-10: Inventory Management 2 (RB-Theta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89TA-11: Transport Networks (RB-Iota) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90TA-13: Semiconductor and Pharmaceuticals Production (RB-Tau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90TA-14: Scheduling with variable parameters II (RB-Omega) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91TA-15: Semi-infinite methods and applications (RB-2101) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91TA-16: Asset management applied to energy and the environment (RB-2103) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92TA-17: Covering Models (RB-2105) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92TA-18: Advances in Location Modeling (RB-2107) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92TA-19: Pension Funds (RB-2111) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93TA-20: Multicriteria Analysis for Investment Decision Making (RB-2113) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93TA-21: Financial Mathematics and OR 5 (RB-2115) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94TA-22: Institutional and Technological Changes in Market and Transitional Economies (RB-2117) . . . . . . . . . . . . . . . . . . . . . . . . 94TA-23: Regularity in Nonsmooth Analysis (RB-Delta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95TA-24: Demand Forecasting (CC-A11) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95TA-25: Modelling Markets with Imperfections (CC-A23) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96TA-26: Optimization in Design of Water Systems (CC-A24) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96TA-27: Inequality Averse Decisions (CC-A25) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97TA-28: Hierarchical Optimization (CC-A27) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97TA-29: MINLP in Airline and Air Traffic Management applications (CC-A29) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98TA-30: Modeling Approaches and Solution Algorithms (CC-A31) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98TA-31: Data Mining: Web and Social-Oriented Applications (CC-A33) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99TA-32: Soft OR case studies (CC-A34) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100TA-33: OR and Transportation: Applications in Developing Countries (CC-A37) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100TA-34: Environment and Sustainable Development - Green Teachers (CC-A39) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100TA-35: Pricing and Supply Chain Management (CC-A41) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101TA-36: DEA and Performance Measurement: Applications 3 (CC-A43) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101TA-37: OR in Health & Life Sciences 1 (CC-Act) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102TA-38: Interactive Multiobjective Optimization: Methods and Applications (HH-Colombus) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102TA-39: MCDA: New Approaches and Applications in Financial Markets (HH-Cousteau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103TA-40: Dynamic Pricing (HH-Livingstone) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103TA-41: AHP/ANP 4 (HH-Heyerdahl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .104TA-42: Simulation Models and Decision Analysis (BW-Amber) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104TA-43: Stochastic programming in energy (BW-Granite) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105TA-44: Computer vision (1) (BW-Marble) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105TA-45: Stochastic Logistical Networks (BW-Water) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

Tuesday, 10:30-12h00

TB-01: Tutorial Lecture: Professor Guy Desaulniers (RB-Alfa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107TB-02: Scheduling and Logistics (RB-Beta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107TB-03: Logistics, Transportation, Traffic - 3 (RB-L1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107TB-04: Approaches for Integer Programming (RB-L2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108TB-05: Problems on graphs I (RB-L3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108TB-06: Performance evaluation and new metaheuristics (RB-Gamma) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

294


TB-07: Vehicle Routing and Logistics Optimization (RB-Eta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109TB-08: VMI and Delivery (RB-Epsilon) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110TB-09: Cutting and Packing 6 (RB-Zeta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110TB-10: Inventory Management 3 (RB-Theta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111TB-11: Transportation and Logistics (RB-Iota) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111TB-12: Robustness in Public Transportation (RB-Omicron) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112TB-13: Geometric Clustering (RB-Tau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112TB-14: New models in batch scheduling (RB-Omega) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113TB-15: Algorithms for Semi-Infinite Optimization (RB-2101) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113TB-16: Theory and Applications of Constrained Equilibria (RB-2103) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113TB-17: Hub Location I (RB-2105) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114TB-18: Location Analysis: Advanced Optimization Tools (RB-2107) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114TB-19: Quantitative Models for Risk Management in Insurance (RB-2111) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115TB-20: Financial Modeling and Optimization (RB-2113) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115TB-21: Financial Mathematics and OR 6 (RB-2115) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116TB-22: Mathematical Models in Macro- and Microeconomics 1 (RB-2117) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116TB-23: Mathematical Programming and Regulatory - Ecological Systems (RB-Delta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117TB-24: Financial Forecasting and Risk Analysis (CC-A11) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117TB-25: Sustainable Supply Chains (CC-A23) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118TB-27: Decision Analysis in Practice (CC-A25) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118TB-28: Optimization Modeling (CC-A27) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119TB-29: MINLP and Nonlinear Network Flows in Gas Transport I (CC-A29) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119TB-30: Applications and Decision Support Systems (CC-A31) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120TB-31: Data Analysis and Its Applications (CC-A33) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120TB-32: Developing Methodology in Soft OR (CC-A34) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121TB-33: Logistic and Development (CC-A37) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121TB-34: Optimization for Sustainable Development 1 (CC-A39) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122TB-35: Electric Mobility (CC-A41) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122TB-36: DEA and Performance Measurement: Applications 4 (CC-A43) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123TB-37: OR in Health & Life Sciences 2 (CC-Act) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123TB-38: Hybrid Evolutionary Multiobjective Optimization Algorithms (HH-Colombus) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124TB-39: MCDA: New Approaches and Applications 1 (HH-Cousteau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124TB-40: Preference Learning 1 (HH-Livingstone) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125TB-41: AHP/ANP 5 (HH-Heyerdahl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125TB-42: Networks under Uncertainty (BW-Amber) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125TB-43: Recent Applications of Probabilistic Programming (BW-Granite) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126TB-44: Computer Vision (2) and Machine Learning (BW-Marble) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126TB-45: Logistics and stochastics 1 (BW-Water) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

Tuesday, 12:30-14:00

TC-01: Keynote Lecture: Professor Bjorn Jorgensen (RB-Alfa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128TC-02: Scheduling in Robotics and Manufacturing (RB-Beta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128TC-03: Sustainable Network Design and Railway/Sea Transport (RB-L1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128TC-04: Optimization problems on graphs and networks (RB-L2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129TC-05: Problems on graphs II (RB-L3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129TC-06: Rich vehicle routing problems (RB-Gamma) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130TC-07: Warehouse Operations Optimization (RB-Eta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130TC-08: Applications (RB-Epsilon) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131TC-09: Colourings, Independence and Assignment (RB-Zeta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131TC-10: Pricing and Applications (RB-Theta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132TC-11: Exact Methods for Arc Routing Problems (RB-Iota) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132TC-12: Capacity and Infrastructure (RB-Omicron) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133TC-13: Assembly systems and automotive production (RB-Tau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133TC-14: Nurses Scheduling (RB-Omega) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134TC-15: Vector and Set-Valued Optimization I (RB-2101) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134TC-16: Perturbation and sensitivity in Markov control problems and games (RB-2103) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135TC-17: Hub Location II (RB-2105) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136TC-18: Competitive Location. (RB-2107) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136TC-19: Risk Modeling and Risk Measures (RB-2111) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136TC-20: Risk Assessment in the Financial Sector (RB-2113) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137TC-21: Asset Pricing (RB-2115) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137TC-22: Mathematical Models in Macro- and Microeconomics 2 (RB-2117) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138TC-23: Complementarity Problems, Variational Inequalities and Equilibrium (RB-Delta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138

295


TC-24: Judgmental Forecasting (CC-A11) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139TC-25: Measurement of Sustainability (CC-A23) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139TC-26: OR in Forestry II (CC-A24) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140TC-27: Policy Analytics (CC-A25) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140TC-28: Boundaries and Interfaces (CC-A27) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141TC-29: MINLP and Nonlinear Network Flows in Gas Transport II (CC-A29) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141TC-30: Data confidentiality 1 (CC-A31) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142TC-31: Applications of the Neural Networks (CC-A33) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142TC-32: Risk and Uncertainty in Energy Models (CC-A34) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143TC-33: Understanding, Modeling and Implementation of OR for Sustainable Development (CC-A37) . . . . . . . . . . . . . . . . . . . . 143TC-34: Optimization for Sustainable Development 2 (CC-A39) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144TC-35: Innovation in Logistics (CC-A41) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144TC-36: QMPD Session1 (CC-A43) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145TC-37: OR in Health & Life Sciences 3 (CC-Act) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145TC-38: Theory of Set-valued Optimization and Applications (HH-Colombus) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146TC-39: Latest advances on MCDA software (HH-Cousteau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146TC-40: Preference Learning 2 (HH-Livingstone) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146TC-41: AHP/ANP 6 (HH-Heyerdahl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147TC-42: Queueing Systems I (BW-Amber) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147TC-43: Stochastic programming in industry II (BW-Granite) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148TC-44: Ensemble Learning and Artificial Neural Networks (BW-Marble) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148TC-45: Logistics and stochastics 2 (BW-Water) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

Tuesday, 14:30-16:00

TD-01: Tutorial Lecture: Professor Jitka Dupacova (RB-Alfa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150TD-02: Scheduling in Transport (RB-Beta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151TD-03: Sustainable Road Transport and Vehicle Routing (RB-L1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152TD-04: Routing problems (RB-L2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152TD-05: Problems on graphs III (RB-L3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153TD-06: Relaxation Adaptive Memory Programming (RB-Gamma) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153TD-07: Manufacturing Issues (RB-Eta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153TD-08: Graph cut algorithms in Computer Vision (RB-Epsilon) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154TD-09: EURO/ROADEF Challenge Session 1 (RB-Zeta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154TD-10: Cutting and Packing 7 (RB-Theta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155TD-11: Heuristic Methods for Arc Routing Problems (RB-Iota) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155TD-12: Rescheduling (RB-Omicron) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156TD-13: Logistics and Manufacturing & Warehousing (RB-Tau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156TD-14: Workforce Scheduling I (RB-Omega) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157TD-15: Vector and Set-Valued Optimization II (RB-2101) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157TD-16: Statistical, Games and Optimization Analysis in Climate Meta-Models (RB-2103) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158TD-17: Global Optimization 1 (RB-2105) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158TD-18: Location Analysis (RB-2107) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159TD-19: Heuristics (RB-2111) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159TD-20: Decision Theory and Analysis (RB-2113) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160TD-21: Risk Management (RB-2115) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160TD-22: Experimental Economics and Game Theory (RB-2117) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161TD-23: Recent Advances of Mathematical Programming in Supply Chain Management (RB-Delta) . . . . . . . . . . . . . . . . . . . . . . 161TD-24: Forecasting methods (CC-A11) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .162TD-25: Pathways toward low carbon transport (CC-A23) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162TD-26: OR in Agriculture, Forestry and Fisheries (CC-A24) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163TD-27: Uncertainties in Decision Processes (CC-A25) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163TD-28: Anticipation and Synchronization (CC-A27) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164TD-29: DEA and Performance Measurement: Methodology 3 (CC-A29) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164TD-30: Data confidentiality 2 (CC-A31) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165TD-31: Data Mining in Economics and Ecology (CC-A33) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165TD-32: National Integrated Assessment Models (CC-A34) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166TD-33: OR for Education and Sustainable Development (CC-A37) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166TD-34: OR for Sustainable Living and Development (CC-A39) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167TD-35: Robust logistic applications (CC-A41) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167TD-36: QMPD Session2 (CC-A43) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168TD-37: OR in Health & Life Sciences 4 (CC-Act) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168TD-38: Multiobjective Optimization for Decision Support (HH-Colombus) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169TD-39: MCDA software for real-world MCDA problems (HH-Cousteau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169

296


TD-40: Preference Learning 3 (HH-Livingstone) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170TD-41: Dynamical Systems and Mathematical Modelling in OR (HH-Heyerdahl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170TD-42: Advances in Stochastic Modeling and Simulation (BW-Amber) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171TD-43: Stochastic and robust optimization (BW-Granite) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171TD-44: Model Generation and Model Selection (BW-Marble) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172TD-45: Quantitative Approaches in Managerial and Financial Accounting (BW-Water) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172

Tuesday, 16:30-17:30

TE-01: Celebration of EURO and its Presidents (Opera) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173


WA-01: Keynote Lecture: Professor Boris Polyak (RB-Alfa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174WA-02: New Models for Project Scheduling (RB-Beta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174WA-03: EURO/ROADEF Challenge Session 2 (RB-L1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174WA-04: Combinatorial Optimization: Applications (RB-L2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175WA-05: Topics in combinatorial opimization I (RB-L3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175WA-06: Variable Neighborhood Search and hybrid metaheuristics (RB-Gamma) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176WA-07: Logistics and Vehicle Routing (RB-Eta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176WA-08: Satisfiability and Hypergraphs (RB-Epsilon) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177WA-09: Graphs and networks I (RB-Zeta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177WA-10: Non-deterministic models 3 (RB-Theta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178WA-11: Scheduling problems with additional constraints (RB-Iota) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178WA-12: Railway Optimization I (RB-Omicron) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179WA-13: Uncertainty in Inventory Management (RB-Tau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179WA-14: Workforce scheduling II (RB-Omega) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180WA-15: Vector and Set-Valued Optimization III (RB-2101) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180WA-16: Conic Optimization: Algorithms and Applications (RB-2103) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181WA-17: Global Optimization 2 (RB-2105) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181WA-20: Portfolio Decision Analysis (RB-2113) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182WA-21: Financial Service Management (RB-2115) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182WA-23: Optimization in Machine Learning (RB-Delta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183WA-24: Data Mining and Decision Making (CC-A11) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183WA-25: Future role of nuclear (CC-A23) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184WA-26: OR and Environmental Management (CC-A24) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184WA-27: Decision Support Systems (CC-A25) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185WA-28: OR Military Applications (CC-A27) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185WA-29: Data Mining (CC-A29) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186WA-30: Scheduling Applications (CC-A31) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186WA-32: DEA and Performance Measurement: Methodology 4 (CC-A34) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187WA-33: Sustainable Development in Civil Engineering and Multi-attribute Decision Making (CC-A37) . . . . . . . . . . . . . . . . . . 187WA-34: Solution concepts for TU-games (CC-A39) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187WA-35: Sports and OR (CC-A41) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188WA-36: Societal Complexity, Healthcare and Sustainable Development (CC-A43) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188WA-37: OR in Health & Life Sciences 5 (CC-Act) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189WA-38: Multiobjective Optimization in Location Problems (HH-Colombus) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189WA-39: Dynamic Programming 1 (HH-Cousteau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190WA-40: Preference Learning 4 (HH-Livingstone) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190WA-41: Young scientists doing research in OR (HH-Heyerdahl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191WA-42: Robust Optimization and Randomized Methods (BW-Amber) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191WA-43: Stochastic programming in industry I (BW-Granite) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192WA-44: Data Analysis and Its Applications (BW-Marble) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192WA-45: Non-Airline Revenue Management Applications (BW-Water) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

Wednesday, 10:30-12h00

WB-01: Keynote Lecture: Professor Matteo Fischetti (RB-Alfa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194WB-02: Preemption, Lower Bounds and Robustness in Project Scheduling (RB-Beta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194

297


WB-03: EURO/ROADEF Challenge Session 3 (RB-L1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194WB-04: New Ideas in Integer Programming (RB-L2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195WB-05: Topics in combinatorial opimization II (RB-L3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195WB-06: Industrial Organization (RB-Gamma) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196WB-08: Boolean Optimization in Graph Theory (RB-Epsilon) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196WB-09: Optimization Problems on Graphs (RB-Zeta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197WB-10: Supply Chain Management 1 (RB-Theta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197WB-11: Other realistic scheduling problems (RB-Iota) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198WB-12: Railway Optimization II (RB-Omicron) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198WB-13: Transportation and Logistics (RB-Tau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199WB-14: Timetabling and Transport (RB-Omega) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199WB-15: Theory and algorithms of bilevel programming I (RB-2101) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200WB-16: Linear and Quadratic Programming (RB-2103) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200WB-17: DC programming and DCA 1 (RB-2105) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201WB-18: Discrete Location (RB-2107) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201WB-19: Predictive Modelling in Finance and Insurance (RB-2111) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202WB-20: Financial Optimization 1 (RB-2113) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202WB-21: Data Analysis in the Financial Sector (RB-2115) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203WB-22: Fuzzy Optimization (RB-2117) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203WB-23: Recent Advances on Proximal Methods I (RB-Delta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204WB-24: Data Mining in Early Warning Systems (CC-A11) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204WB-25: Smart cities and Smart grids (CC-A23) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205WB-26: Multi-Criteria Decision Making and Applications 1 (CC-A24) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205WB-27: Web-based Spatial-temporal Collaborative Decision Support Systems (CC-A25) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206WB-28: OR Military and Security Applications (CC-A27) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206WB-29: Support Vector Machines. New results (CC-A29) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207WB-30: DEA and Performance Measurement: Applications 8 (CC-A31) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207WB-32: DEA and Performance Measurement: Applications 5 (CC-A34) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208WB-33: OR in Sustainable Urban Development (CC-A37) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208WB-34: Games and decision making (CC-A39) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208WB-35: Optimization Modeling I (CC-A41) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209WB-37: OR in Health & Life Sciences 6 (CC-Act) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209WB-38: Multiobjective Optimization and Transportation (HH-Colombus) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210WB-39: Dynamic Programming 2 (HH-Cousteau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210WB-40: Preference Learning 5 (HH-Livingstone) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211WB-41: System Dynamics: demographics and methodology (HH-Heyerdahl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211WB-42: Energy Systems Planning (BW-Amber) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212WB-43: Extended Two-Stage Linear Stochastic Programs (BW-Granite) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212WB-44: Collective Learning Procedures (BW-Marble) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213WB-45: Customer-Oriented Revenue Management (BW-Water) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213


WC-01: Keynote Lecture: Professor Jonathan Caulkins (RB-Alfa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214WC-02: Multi-Project Scheduling and Stochastic Project Scheduling (RB-Beta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214WC-03: EURO/ROADEF Challenge Session 4 (RB-L1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214WC-04: Combinatorial Optimization: Applications (RB-L2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215WC-05: Topics in integer and mixed integer programming I (RB-L3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215WC-06: Mathematical Economics (RB-Gamma) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216WC-07: Rich vehicle routing problems (RB-Eta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216WC-08: Boolean Functions and Their Applications (RB-Epsilon) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217WC-09: Maritime Transportation 3 (RB-Zeta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217WC-10: Humanitarian Logistics (RB-Theta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218WC-11: Realistic parallel machines problems (RB-Iota) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218WC-12: Solution Methods at Seaport Container Terminals (RB-Omicron) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219WC-13: Emerging Issues in Order Fulfillment (RB-Tau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219WC-14: Timetabling and People (RB-Omega) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220WC-15: Discrete bilevel optimization and Variational inequalities (RB-2101) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220WC-16: Linear programming: Algorithms and Applications (RB-2103) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221WC-17: Nonconvex Programming (RB-2105) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221WC-18: Location Routing (RB-2107) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222WC-19: Simulation in Estimation (RB-2111) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222WC-20: Financial Optimization 2 (RB-2113) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223WC-21: Ratings of Banks and Companies (RB-2115) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

298


WC-22: Fuzzy Scheduling and Fuzzy Quality Management (RB-2117) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224WC-23: Recent Advances on Proximal Methods II (RB-Delta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224WC-25: OR and Environmental Management - Decision Processes (CC-A23) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225WC-26: Criteria Weighting in Decision Support (CC-A24) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225WC-27: Knowledge Management & Decision Systems (CC-A25) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226WC-28: OR Military Applications II (CC-A27) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226WC-29: Optimization and Data Mining (I) (CC-A29) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227WC-30: DEA and Performance Measurement: Applications 9 (CC-A31) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227WC-31: OR and Ethics I (CC-A33) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228WC-32: DEA and Performance Measurement: Applications 6 (CC-A34) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228WC-33: Sustainable Construction Processes and supplier selection (CC-A37) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229WC-34: Economic modeling and finance (CC-A39) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229WC-35: Recent Achievements in Regression and Classification (CC-A41) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230WC-36: Recent Advances of OR on Societal Complexity (CC-A43) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230WC-37: Operations Research in Health Care I (CC-Act) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231WC-38: Applications of Multiobjective Optimization (HH-Colombus) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231WC-39: Cops and Robber Games (HH-Cousteau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232WC-40: Model quality and modelling sustainable energy and transport systems (HH-Livingstone) . . . . . . . . . . . . . . . . . . . . . . . . 232WC-41: Modern project for involvement young people in OR community and OR education (HH-Heyerdahl) . . . . . . . . . . . . . . 233WC-42: Queueing Systems II (BW-Amber) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233WC-44: Information and Intelligent Systems I (BW-Marble) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234WC-45: Advanced Revenue Management (BW-Water) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234


WD-01: Tutorial Lecture: Professor Karla Hoffman (RB-Alfa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235WD-02: Methods for Resource-Constrained Project Scheduling (RB-Beta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235WD-03: Demand Responsiveness and Transportation Network Design (RB-L1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236WD-04: Advances for solving Quadratic and Other Difficult 0-1 Optimization Problems (RB-L2) . . . . . . . . . . . . . . . . . . . . . . . 236WD-05: Topics in integer and mixed integer programming II (RB-L3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237WD-06: Game Theory Applications (RB-Gamma) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237WD-07: Applications of Vehicle Routing 2 (RB-Eta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238WD-09: EURO/ROADEF Challenge Session 5 (RB-Zeta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238WD-10: Integrated Planning of Operations (RB-Theta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239WD-12: Port and Liner Management (RB-Omicron) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239WD-13: Emerging Issues in New Product Development (RB-Tau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240WD-15: Theory and algorithms of bilevel programming II (RB-2101) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240WD-16: Recent Advances in Linear Programming and Related Subjects (RB-2103) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241WD-17: DC programming and DCA 2 (RB-2105) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241WD-18: Location Analysis and Disruptions (RB-2107) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241WD-19: Survival Analysis and Simulation (RB-2111) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242WD-20: Financial Optimization 3 (RB-2113) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242WD-21: Capital Asset Pricing Models (RB-2115) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243WD-22: Fuzzy Systems in Geopolitics and Disaster Management (RB-2117) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243WD-23: Recent Advances in Convex Optimization (RB-Delta) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244WD-24: OR and the Arts (CC-A11) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244WD-25: Simulation Based Decision Support (CC-A23) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245WD-26: Multi-Criteria Decision Making and Applications 2 (CC-A24) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245WD-27: Applications in Decision Making & Decision Analysis (CC-A25) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246WD-28: Seismic Modeling with OR (CC-A27) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246WD-29: Optimization and Data Mining (II) (CC-A29) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247WD-31: OR and Ethics II (CC-A33) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247WD-32: DEA and Performance Measurement: Applications 7 (CC-A34) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247WD-33: Optimization Modeling II (CC-A37) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248WD-35: Recent Advances in Statistics (CC-A41) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248WD-36: Technological change and productivity: permanent features of population health (CC-A43) . . . . . . . . . . . . . . . . . . . . . 249WD-37: Operations Research in Health Care II (CC-Act) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249WD-38: Multiobjective Optimization and Water Management (HH-Colombus) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249WD-39: Graph and Digraph Searching Games (HH-Cousteau) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250WD-40: Modelling sustainable production systems (HH-Livingstone) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250WD-41: Applications of Control Theory in Dynamical Systems (HH-Heyerdahl) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251WD-42: Stochastic Modelling in Computational Biology and Environmental Sciences (BW-Amber) . . . . . . . . . . . . . . . . . . . . . . 251WD-43: Emerging applications in finance and insurance (BW-Granite) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252WD-44: Information and Intelligent Systems II (BW-Marble) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252

299


WD-45: IBM Research Applications (BW-Water) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253


WE-01: IFORS Distinguished Lecture: Professor Ralph E. Gomory (RB-Alfa) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254

300

technical sessions - ipu.ru

Documents