hn-models in control
TRANSCRIPT
SOME MODERN TRENDSIN CONTROL THEORY
Dmitry A. [email protected], www.ipu.ru, www.mtas.ru
Institute of Control Sciences RASMoscow
Institute of Physics and Technology Moscow
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
STRUCTURE OF CONTROL SYSTEM
…
Control is “the process of checking to make certain that rules or standards being applied” (Macmillan Dictionary).
Control is “the act or activity of looking after and making decisions about something” (Merriam-Webster Dictionary).
Control is “an influence on a controlled system with the aim of providing the required behavior of the latter” (Theory of Organizations Control)
CONTROL SYSTEM
CONTROLLED OBJECT
Stateof
controlledobject
Control
External disturbances
OBJECTS, METHODS AND MEANS OF CONTROL
CONTROL
OBJECTS METHODS
MEANS
Measuring, transformative,
actuating Informational, computational
…
Control is “the process of checking to make certain that rules or standards being applied” (Macmillan Dictionary).
Control is “the act or activity of looking after and making decisions about something” (Merriam-Webster Dictionary).
Control is “an influence on a controlled system with the aim of providing the required behavior of the latter” (Theory of Organizations Control)
Scientific knowledge
about the object of control
Control theory
Applications
Technologies of control …
…
t
Typical time of development
CYCLE OF THEORY DEVELOPMENT
t
1860-е 1900 1930 1940 1950 1960 1970 1980 1990 2000 2010КонецXVIII века
150 YEARS OF CONTROL THEORY (Technics)
Scientific knowledge
about the object of control
Control theory
Applications
Technologies of control …
…
t
Typical time of development
Institute of Control Sciences (ICS RAS, Moscow):- System Theory and General Control Theory;- Techniques of Control in Complicated Engineering and Man-Machine Systems;- Theory of Control in Inter-Disciplinary Models of Organizational, Social, Economic, Medical and Biological and Environment Protection Systems;- Theory and Techniques in Development of Software-and-Hardware and Engineering Tools of Control and Complicated Data Processing and Control Systems;- Scientific Fundamentals of Technologies in Vehicle Control and Navigation;- Scientific Fundamentals of Integrated Control Systems and Automation of Technological Industrial Processes.
USA: CalTech, Harvard Univ., MIT, Stanford Univ., Univ. of California, etc.UK: Imperial College, Oxford Univ., Cambridge Univ., etc.Germany: Univ. of Stutgard, Techn. Univ. ofDarmstadt , etc.Italy: University of Rome, Politecnico di Torino, Politecnico di Milano, etc.
France: SUPELEC Paris, Univ. of Grenoble, LAAS-SNRS Toulose,etc.Australia: Univ. of New-castle, Australian Nat. Univ., etc.Sweden: University of Linkoping, Lund University.Japan: Japan Advanced Inst. of Science and Techn., Kyoto Univ., etc.
CONTROL THEORY IN RUSSIA
… and in the World
…
SCIENTIFIC RESEARCHESAND APPLIED PROJECTS OF ICS RAS. I
Classical linear control theory
Control of mechanical
systems
Nonlinear systems
Control in quantum systems
Control of systems with distributed parameters
Theory of stability and stabilization
Control of moving objects
Robust and adaptive control
General control theory
)(),,(.
twtuxfx +=
Study of control problems for a multi-mode submarine in the case of
emergency. The intelligent control level is based on production rules for current and predicted situations
Full-scale multi-mode computer training complex for Russian Navy
General view of the training complex
SCIENTIFIC RESEARCHESAND APPLIED PROJECTS OF ICS RAS. II
Integrated solutions for upper level control systems in nuclear power engineering with application in Russia and abroadØProgramming languageØOperating systemØEngineering equipment (in cooperation with the Nizhny Novgorod Scientific Research Institute for Engineering Systems)
ØApplied software
SCIENTIFIC RESEARCHESAND APPLIED PROJECTS OF ICS RAS. III
Equipment for automation devices and computers
New effects in nanostructures
Modern jet devices(operable up to 500˚ C, in vibration and radiation environment)
Multi-channel sensors
SCIENTIFIC RESEARCHESAND APPLIED PROJECTS OF ICS RAS. IV
Aviation maps
Navigation maps
Digital topographic mapsLarge-scale plans
Digital relief matrix
3D-modeling
3D-modeling
Map making by field surveyMap updating by air survey
Map updating by space survey
Geographic Information Systems
SCIENTIFIC RESEARCHESAND APPLIED PROJECTS OF ICS RAS. V
3D-Modeling in control
SCIENTIFIC RESEARCHESAND APPLIED PROJECTS OF ICS RAS. VI
APPLICATIONS OF CONTROL THEORY
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
USA (NSF) priorities:Group control. Combat control. Control in financial and economic systems. Control in biological and ecological systems. Man and team in a control loop.Unified theory of control, computation and communication (С3). …
European priorities:Man-machine symbiosis (modeling a man in a control loop and as a controlled subject). Distributed and networked systems. Production, safety and strategies of heterogeneous control. New principles of interdisciplinary coordination and control. …
Russian Academy of Sciences:Methods and means of communicational and networkedcontrol of multi-level and distributed dynamic systems under uncertainty; intellectual control.
HETEROGENEOUS = DIVERSIVE(NETWORKED)
+ DISTRIBUTED+ HIERARCHICAL
(controlled object, control system and communications).
HETEROGENEOUS CONTROL MODELS
HIERARCHIES AND NETWORKES:CONTROLLED OBLECT, CONTROL SYSTEM
AND COMMUNICATIONS(example: Project Management)
FROM NETWORKES AND HIERARCHIES
TO HIERARCHIES OF NETWORKESAND NETWORKES OF HIERARCHIES
Пакет работ
УровеньWBS
УровеньWBS
УровеньWBS
Пакетработ
Пакетработ
Пакетработ
Пакетработ
Пакетработ
Пакетработ
. . . . . .
Центр
Центр
Центр
OBS
Руководство проекта
Пакетработ
Планy * A'∈
Результатдеятельности
z A0∈
Действие)y А'∈
ЦентрЦентр Центр
АЭ АЭ АЭ АЭ АЭ АЭ
RBS
Функциональная структура предприятия
Стимулирующее воздействие
Группапроектов
ПроектПроектПроектУровень проектов
NETWORK SHEDULE
WBS OBS
RBS
Responsibility allocation
Resources allocation
Authority allocation
Степень
достижения
поставленны
х це
лей
Время реализации
Исходное состояние
Целевое состояние
Оценка состояния
Планируемое состояние
Реальное состояние
Плановая траектория
Реализуемая траектория
Прогноз реализуемой траектория
Возможная траектория в рамках
существующей стратегии (работа над
ошибками)
Оценка затрат для возврата на плановую траекторию развития
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
2000 1950 1940 1960 1970 1980 1990
Theory of automatic control
Game theory, operations research
Cybernetics
Decision-making, ;public choice theory
Mechanism design (MD)
Multiagent systems(DAI)
Discrete optimization, optimal control
2010
Network structures (С3)
Organizational systems (MAN)
Social systems
(SOCIETY)
Ecological systems
(NATURE)
Technico-organizational and man-machine
systems Ecologo- economical
systems
Economical systems
PRODUCTION
Regulatory-axiological
systems
Noospheric systems
Socio-ecological systems
Technical systems
Socio- economical
systems
INTERDISCIPLINARY SYSTEMS
Optimization of hierarchicaland network structures
Territory
Government
Industry
Enterprise
Coordination of interaction and decision-making in multiagent systems
Interests concordance in ecologo-economical systems
Reflexive control
Mechanisms of organizations control
Temporal network organization
А
Б В
Г
2 12 121 А В ВА ВАВ
EXAMPLES OF INTERDISCIPLINARY SYSTEMS
Confrontation Hierarchies Collective Decision-making Cooperative Control
Interaction. Distributed Optimization (e.g. Task Assignment) Mission Planning
“Implementation”. Formation Control
Stabilization. Consensus Problem
Operational level
Action
Tactical level
Strategic level (decision-making,
adaptation, learning, reflexion)
Level of goal-setting and choice of functioning
mechanisms
Exte
rnal
inf
orm
atio
n
Dyn
amic
syst
ems
Art
ifici
al
Inte
llige
nce
Mod
els o
f C
olle
ctiv
e B
ehav
ior
Gam
e T
heor
y
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
GENERAL TOPICS
ACC-2011***
*CDC – Conference on Decision and ControlECC – European Control Conference12-15 December 2011
Organized by IEEE, Orlando, Florida> 1500 papers
**IFAC World Congress28 Aug – 2 Sept., 2011> 1500 papers
***ACC – American Control Conference29-30 June 2011Organized by IEEE, San Francisco, USA > 900 papers
CDC-ECC-2011*
Математи-ческая теория
9%
Приложения19%
"Cредства"4%
"Классика"49%
"Сетевизм"19%
Applications 19%
Mathematical theory
9%
Technical means 4% “Networks”
19%
Classic 49%
"Сетевизм"11%
"Классика"32%
Cредства5%
Приложения44%
Математи-ческая теория
9%
IFAC-2011**
Applications 44%
“Networks” 11%
Classic 32%
Mathematical theory
9%
Technical means 5%
Математи-ческая теория
5%
Приложения33%
Cредства5%
"Классика"42%
"Сетевизм"14%
Technical means 5%
Applications 33%
Mathematical theory
5%
Classic 42%
“Networks” 14%
МАС и консенсус
35%
Кооператив-ное
управление21%
Коммуникации в МАС35%
Верхние уровни
управления4%
Другое4%Highest
control levels 4%
Communi- cations in MAS
35%
Cooperative control
21%
Consensus problems
35%
Others 4%
Другое8%Верхние
уровни управления
10%
Коммуникации в МАС26%
Кооператив-ное
управление10%
МАС и консенсус
46%
Highest control levels
10%
Communi- cations in MAS
26%
Cooperative control
10%
Consensus problems
46%
Others 8%
МАС и консенсус
33%
Кооперативное управление
15%
Коммуникации в МАС31%
Верхние уровни управления
13%
Другое8%
Highest control levels
13%
Communications in MAS
31% Cooperative
control 15%
Consensus problems
31%
Others 8%
«NETWORKS»
II
I III
IV
ACC-2011
СDС-ECCC-2011 IFAC-2011
II
IIII
IV
I
II
III
IV
Другие19%
Морские подвижные
объекты2%
Автомобили и автотрафик
11%
Биология и медицина
13%Мехатроника и
роботы11%
Производство2%
Авиация и космос
9%
Энергетика33%
Others 19%
Energetics 33%
Aero-Space 9%
Production 2% Mechatronics
and robots 11%
Maritime mobile objects
2%
Avto-vehicles and traffic
11%
Bio-Med 13%
Энергетика17%
Авиация и космос
7%
Производство18%
Мехатроника и роботы
13%
Биология и медицина
13%
Автомобили и автотрафик
13%
Морские подвижные
объекты3%
Другие16%
Others 16%
Energetics 17%
Aero-Space 7%
Production 18%
Mechatronics and robots
13%
Bio-Med 13%
Avto-vehicles and traffic
13%
Maritime mobile objects
3%
Другие8%Морские
подвижные объекты
11%
Автомобили и автотрафик
8%
Биология и медицина
17%Мехатроника и
роботы18%
Производство8%
Авиация и космос
8%
Энергетика22%
Others 8% Energetics
22%
Aero-Space 8%
Production 8%
Mechatronics and robots
18%
Bio-Med 17%
Avto-vehicles and traffic
8%
Maritime mobile objects
11%
APPLICATIONS
ACC-2011
СDС-ECCC-2011 IFAC-2011
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
MULTIAGENT SYSTEMS (MAS)
«… a rat or individual locusts are not too clever, and almost harmless. However, flocks of rats or swarms of locusts can have a devastating impact».
Modern trends:- decentralization- miniaturization- intellectualization
Centralized control
Decentralized (group) control
NETWORK
MAS
Material MAS
Virtual MAS (softbots)
Wheeled UAVs AUVs …
Specificity of MAS:üMultiple components;üDistributed, networked
communications;üHierarchy;ü Intelligence (autonomy):• rationality (decision-making under
uncertainty and cognitive restrictions);• autonomous goal-setting, goal-
oriented behavior;• reflection;• cooperative and/or competitive
interactions (the formation of coalitions, information, and other confrontation).
MULTIAGENT SYSTEMS: SPECIFICITY
MULTIAGENT SYSTEMS: ARCHITECTURE OF AN AGENT
Confrontation Hierarchies Collective Decision-making Cooperative Control
Interaction. Distributed Optimization (e.g. Task Assignment) Mission Planning
“Implementation”. Formation Control
Stabilization. Consensus Problem
Operational level
Action
Tactical level
Strategic level (decision-making,
adaptation, learning, reflexion)
Level of goal-setting and choice of functioning
mechanisms
Exte
rnal
inf
orm
atio
n
Dyn
amic
syst
ems
Art
ifici
al
Inte
llige
nce
Mod
els o
f C
olle
ctiv
e Be
havi
or
Gam
e T
heor
y
Confrontation Hierarchies Collective Decision-making Cooperative Control
Interaction. Distributed Optimization (e.g. Task Assignment) Mission Planning
“Implementation”. Formation Control
Stabilization. Consensus Problem
Operational level
Action
Tactical level
Strategic level (decision-making,
adaptation, learning, reflexion)
Level of goal-setting and choice of functioning
mechanisms Ex
tern
al
i
nfor
mat
ion
Dyn
amic
syst
ems
Art
ifici
al
Inte
llige
nce
Mod
els o
f C
olle
ctiv
e B
ehav
ior
Gam
e Th
eory
MULTIAGENT SYSTEMS: CONSENSUS
Multi-agent system
( ) nitxtxatxn
jjiiji ..., ,1,)()()(
1=−−= ∑
=& – characteristic of agent i,)(tx i,
),()( txLtx −=& ( ) ,)(..., ),()( T1 txtxtx n=
[ ] ,ij n nL ×= l
=≠−
= ∑≠
.,,,
)( ijaija
tik
ik
ij
ijl
Theorem*Consensus is reachable iff the communication graph is “connected”.
CONSENSUS PROBLEM
*Agaev R., Chebotarev P. (A&RC. 9. 2000)
Confrontation Hierarchies Collective Decision-making Cooperative Control
Interaction. Distributed Optimization (e.g. Task Assignment) Mission Planning
“Implementation”. Formation Control
Stabilization. Consensus Problem
Operational level
Action
Tactical level
Strategic level (decision-making,
adaptation, learning, reflexion)
Level of goal-setting and choice of functioning
mechanisms Ex
tern
al
i
nfor
mat
ion
Dyn
amic
syst
ems
Art
ifici
al
Inte
llige
nce
Mod
els o
f C
olle
ctiv
e B
ehav
ior
Gam
e T
heor
y
MULTIAGENT SYSTEMS: FORMATION CONTROL
С3 & FORMATIONS CONTROL
Basic consensus problem
( )ix t& = 1
( ( ) ( ))n
ij j ij
a x t x t=
−∑
Communications & computations
( )ix t& = 1
( )( ( ) ( ))n
cij j ij i ij
ja t x t x tτ τ
=
− − −∑
Model of the controlled object
Autonomous Underwater Vehicles. Edited by N. Cruz. – Rijeka: InTech, 2011.
+ nonlinearity + observability + adaptivity + switching communication matrix
…
Formation control ( )ix t& = vi(t),
( )iv t& = 1
( ( ) ( ))n
ij j ij
b x t x t=
−∑ +
1( ( ) ( ))
n
ij j ij
b v t v t=
−∑ + ci(V(t) – vi(t))
x(t), v(t) V(t)
Confrontation Hierarchies Collective Decision-making Cooperative Control
Interaction. Distributed Optimization (e.g. Task Assignment) Mission Planning
“Implementation”. Formation Control
Stabilization. Consensus Problem
Operational level
Action
Tactical level
Strategic level (decision-making,
adaptation, learning, reflexion)
Level of goal-setting and choice of functioning
mechanisms Ex
tern
al
i
nfor
mat
ion
Dyn
amic
syst
ems
Art
ifici
al
Inte
llige
nce
Mod
els o
f C
olle
ctiv
e B
ehav
ior
Gam
e Th
eory
MULTIAGENT SYSTEMS: PLANNING
Confrontation Hierarchies Collective Decision-making Cooperative Control
Interaction. Distributed Optimization (e.g. Task Assignment) Mission Planning
“Implementation”. Formation Control
Stabilization. Consensus Problem
Operational level
Action
Tactical level
Strategic level (decision-making,
adaptation, learning, reflexion)
Level of goal-setting and choice of functioning
mechanisms Ex
tern
al
i
nfor
mat
ion
Dyn
amic
syst
ems
Art
ifici
al
Inte
llige
nce
Mod
els o
f C
olle
ctiv
e Be
havi
or
Gam
e T
heor
y
MULTIAGENT SYSTEMS: COOPERATIVE CONTROL
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
).,...,,,,...,,(maxArg *,
*1,
*1,
*1
*niiiiiiiiiXxi xxxxxfx
iiσσσσσσ θ +−
∈∈
Informational equilibrium
Information control problem.max),(min
)(→Φ ℑ∈Ψ∈ IIx
IxX
ΨX (I) ⊆ X' – the set of real agents actions, which are stable under information structure I; Φ(x, I) – control efficiency criterion;ℑ – set of feasible information structures.
Taking into account the nontrivial mutual beliefs of agents allows:
1. (normative point of view) to enlarge the set of game’s outcomes, which in its turn increases the efficiency of information control;
2. (descriptive point of view) to describe many practically observed situations, which can not be interpreted as a Nash equilibrium under common knowledge, as informational equilibriums under proper information structure.
ГI = N, (Xi)i ∈ N, (fi(⋅))i ∈ N, Ω, I – reflexive game; Ω – set of feasible states of nature; I – information structure; N – set of players (agents); fi: Ω × X’ → ℜ1 – goal function of the i-th agent.
Information (beliefs) structure
θ1 θi θn
θi1 θij θin
… …
… …
I1
Ii1
Real agent
Phantom agent
Reflexion
. . . . . .
. . . . . .
INFORMATIONAL REFLEXION AND CONTROL
Reflexive model of group aircraft battle
Strategic behavior in collective decision-making. Suppose, that the principal is interested in the result x0∈[d;D] of expert examination. Let the opinions of n experts ri∈[d;D]i∈N, be known by the principal only, who is able to form second-order beliefs of the experts. Collective decision x = π(s), where s ∈ [d;D]n is the vector of opinions, revealed by experts. Denote: x0i(a,ri) is the solution of the following equation:π(a, …, x0, …, a) = ri , di(ri) = max d; x0i(D, ri), Di(ri) = min D; x0i(d, ri),d(r) = (d1(r1), d2(r2), …, dn(rn)), D(r) = (D1(r1), D2(r2), …, Dn(rn)).Statement 15. Any result x0∈[π(d(r));π(D(r))] may be implemented as the collective decision by means of information control with the second rank of reflexion.
SOME APPLICATIONS OF INFORMATIONAL CONTROL
INFORMATIONAL AND STRATEGIC REFLEXION
«Optimizational models of collective behavior
Models of games
Game theory
Models of collective behavior
Informational structures
Concept of equilibrium Informational equilibrium
Nash equilibrium
Control problems
Informational control
MODELS OF STRATEGIC REFLEXION
Reflexive structures k-level models;
cognitive hierarchies models, etc
Models of reflexion in
bimatrix games
Informational reflexion
Strategic reflexion
Prognostical
Reflexive equilibrium
Reflexive control
“Reflexive” models
Level
Phenomenological (descriptive)
Normative
Models of reflexive decision-making
Game theory
(super-intelligent players)
Theory of collective behavior
(rational agents)
LEVEL OF “INTELLIGENCE”
REFLEXION
«Probability» of target destruction:ü Direct attack of 40 “simple” agents: 0,125ü 8 «investigators» + 32 «reflexive» agents: 0,985ü 40 «investigators»: 0,999
Level of hierarchy
Modelled processes Modelling tools
6 Choice of agents and their characteristics
Discrete and multicriteriaoptimization
5 Choice of agents’trajectories and velocities
Optimal control
4 Forecasting by the agent opponents’behavior
Reflexive games
3 Minimization of the probabilityof detection
Optimization and heuristics of choosing the direction of motion
2 Collision and obstacle avoidance
Algorithms of “local”trajectories’ choice
1 Motion toward the target
Equations of dynamics
DIFFUSIVE BOMB MODEL: STRATEGIC REFLEXION
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
Confrontation Hierarchies Collective Decision-making Cooperative Control
Interaction. Distributed Optimization (e.g. Task Assignment) Mission Planning
“Implementation”. Formation Control
Stabilization. Consensus Problem
Operational level
Action
Tactical level
Strategic level (decision-making,
adaptation, learning, reflexion)
Level of goal-setting and choice of functioning
mechanisms Ex
tern
al
i
nfor
mat
ion
Dyn
amic
syst
ems
Art
ifici
al
Inte
llige
nce
Mod
els o
f C
olle
ctiv
e B
ehav
ior
Gam
e T
heor
y
MULTIAGENT SYSTEMS: CONFRONTATION
MODELS OF SOCIAL NETWORKS
The set M of principals
The set of agentsN
…
Problem – find a control vector u, s.t. Φ(X, u) = H(X) – c(u) →
Uu∈max ,
H(⋅) – “gain”, с(⋅) – control costs.
Each principal (from set M) is able to influence on the initial opinions of the agents uij and is interested in forming the final opinions XM. Problem – find equilibrium of the game:
Г = (M, Ujj ∈ M, Gj(⋅)j ∈ M).
A set N of agents form social network G = (N, E). x – vector of initial opinions, X – final opinions.
aij ≥0 – the trust of i-th agent to j-th, k-th agent indirectly influences on i-th agent: xk+1 = A [xk + B uk]. Problem – find total influence of one agents on another, find the agents, that form the final opinions
3. Informational interaction (dynamics)
4. Informational control
5. Information warfare
Newsvine
Habr
2. Structural analysis
1. «Statistical» analysis
ANALYSIS AND MODELINGOF INFORMATIONAL PROCESSES IN SOCIAL MEDIA
11.11.11 06.12 .11 31.12.11 25.01.12 19.02 .12 15.03 .12 09.04 .12 04.05 .12 29.05.12 23.06 .12 18.07 .12
0
5000
10000
15000
Суточное количество сообщений : Политики A и B
0 10 20 30 40 50 60 70?0.4
?0.2
0.0
0.2
0.4
0.6
0.8
1.0Автокорреляционная функция по сообщениям : Политики A и B
0 20 40 60 80 100 120 140
?0.2
?0.1
0.0
0.1
0.2
0.3
0.4
0.5Парная корреляционная функция по сообщениям
Политик A Политик B
11.11 .11 06.12 .11 31.12 .11 25.01 .12 19.02 .12 15.03 .12 09.04 .12 04.05 .12 29.05 .12 23.06 .12
11
10
9
8
7
6
5
4
3
2
1Вейвлет скалограмма ?объём сообщений ?: Политик A
11.11 .11 06.12 .11 31.12 .11 25.01 .12 19.02 .12 15.03 .12 09.04 .12 04.05 .12 29.05 .12 23.06 .12
11
10
9
8
7
6
5
4
3
2
1Вейвлет скалограмма ?объём сообщений ?: Политик B
Политическая жизнь России
Политик A Политик B
Общество / Блогосфера / СМИ
Активностьблоггеров
Активностьблоггеров
11.11 .11 06.12 .11 31.12 .11 25.01 .12 19.02 .12 15.03 .12 09.04 .12 04.05 .12 29.05 .12 23.06 .12
0
20
40
60
80
100Точки статистической разладки: Политик A
11.11 .11 06.12 .11 31.12 .11 25.01 .12 19.02 .12 15.03 .12 09.04 .12 04.05 .12 29.05 .12 23.06 .12
0
100
200
300
400Точки статистической разладки: Политик B
Political activity Person A Person B
Society / Blogoshere / Media
Blogers’ activity Blogers’ activity
Daily mentioned: persons A and B
Autocorrelation by persons
Autocorrelation by messages
Person B Person A
Statistical discord: person A
Statistical discord: person B
Scalogramma: person B
Scalogramma: person A
Characteristics of maximal component• Number of agents – 362.000, • Number of “connections”– 802.000 (12.000.000 comments)Structure of components1 – Discussing component (25%). 2 - Un-popular component (45%).3 – Popular component (8%).4 - Others (22%).
The structure is time-stable:• May 2011 • November 2011• December 2011
45
The rate of information (average) spread - 5.3 steps to receive information FROM any agent;- 5.3 steps to transmit information TO any agent.
THE STRUCTURE OF MAXIMALWEAKLY CONNECTED COMPONENT
1
2
3
4
Number of removed nodes Si
ze o
f max
imal
clu
ster
(%)
Number of removed nodes
Num
ber o
f cla
ster
s
CONTROL IN SOCIAL NETWORKSCONTROL:- OPINIONS,- BELIEFS (TRUST),- REPUTATION,- …- MEMBERSHIP and STRUCTURE.
The removal of small number of the most significant nodes leads to the appearance of a great number of unconnected groups (А). But this groups are small, and the largest group is still connected (B).
Число
авторов
Число
авторов
64.000 – cumulative estimate of politically active blogers in Russian LiveJournal
A
B
BACKGROUND: CONSENSUS PROBLEM
French J.R. A Formal Theory of Social Power // Psychological Review. 1956. 63. P. 181 – 194.
Harary F. A Criterion for Unanimity in French’s Theory of Social Power / Studies in Social Power. – Michigan: Institute of Sociological Research. 1959. P. 168 – 182.
De Groot M.H. Reaching a Consensus // Journal of American Statistical Association. 1974. 69. P. 118 – 121.
Roberts F. Discrete Mathematical Models with Applications to Social, Biological, and Environmental Problems. – Prentice: Prentice Hall, 1976.
Jackson M. Social and Economic Networks. – Princeton: Princeton University Press, 2008.
BACKGROUND: GAME THEORY
Network games
Network formation games Network-based games
Networking games
Cognitive games
Social networks games
Games over the project network-schedules
Network – is a result of game-theoret ical interaction
Network is f ixed and determines the dependence of players gains from their
actions
SOCIAL NETWORKS AND LINEAR DYNAMIC SYSTEMS
xk+1 = A [xk + B uk], k = 0, 1, …
MULTIAGENT SYSTEMS
SOCIAL NETWORKS COGNITIVE MAPS
PageRank Problem
ui
xi
Φ(X, u)
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
Level of hierarchy
Modeled phenomena and processes
Technique of modeling
5 Allocation of forces and means in space
Game theory(Colonel Blotto game, etc.)
4 Allocation of forces and means in time
Optimal control, repeated games, etc.
3 Number of troops dynamics
Lanchester’s equations and its modifications
2 «Local»interactionsof squads
Markovian and other stochastic models
1 Interactionof separatebattle units
Dynamic systems.Finite state automata.Simulation.
HIERARCHICAL COMBAT MODELS
20 40 60 80 100Номер ИПБ
500
1000
1500
2000МЦРMRRR
REFLEXIVE COLONEL BLOTTO GAME
Average maximal rational rank of reflexion (MRRR) ∼ 230 (nonsense!)
Denote by N = [1, …, n] the set of objects, x = (x1, …, xn) – first player’s action, y = (y1, …, yn) – second player’s action, where xi ≥ 0 (yi ≥ 0) – amount of resources, allocated by the first (second) player to i-th object, i = 1,n . Resources are limited (1) i
i Nx
∈∑ ≤ Rx, i
i Ny
∈∑ ≤ Ry.
In probabilistic Colonel Blotto Model (CBM) the probability px(xi, yi) of the first player win on the i-th objectdoes not depend on other objects and is “proportional” to the amount or resources, allocated on this object.:
(2) px(xi, yi) = ( )
( ) ( )
i
i i
ri i
r ri i i
xx yα
α +, py(xi, yi) = 1 – px(xi, yi), где ri ∈ (0; 1], αi > 0, px(xi = 0, yi = 0) =
1i
i
αα +
.
Let BRx(y) = (u1 y1 + ε, …, un yn + ε) denote the best response of first player, where n-dimentional vector u = (u1,…, un) is a solution of the
following knapsack problem: (3)0;11
1
max ,
,
i
n
i i uin
i i xi
uV
u y R
∈=
=
→
≤
∑
∑, ε =
1
1 ( )n
x i ii
R u yn =
− ∑ , i.e. let’s assume that the player tries to win on the most valuable set of
objects, and the rest of resources are equally divided among other objects. Let BRy(x) = (v1 x1 + δ, …, vn xn + δ) denote the best response of first player, where n-dimentional vector v = (v1,…, vn) is a solution of the
following knapsack problem: (4) 0;11
1
max,
,
i
n
i i vin
i i yi
vV
v x R
∈=
=
→ ≤
∑
∑, δ =
1
1 ( )n
y i ii
R u xn =
− ∑ .
Rank gameexploration
MRRR
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automation
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
MODERN TRENDS IN MAS
1) Strategicdecision-making
2) Increasing roleof game-theory
3) Benchmark scenarios and problems
Confrontation Hierarchies Collective Decision-making Cooperative Control
Interaction. Distributed Optimization (e.g. Task Assignment) Mission Planning
“Implementation”. Formation Control
Stabilization. Consensus Problem
Operational level
Action
Tactical level
Strategic level (decision-making,
adaptation, learning, reflexion)
Level of goal-setting and choice of functioning
mechanisms
Exte
rnal
inf
orm
atio
n
Dyn
amic
syst
ems
Art
ifici
al
Inte
llige
nce
Mod
els o
f C
olle
ctiv
e B
ehav
ior
Gam
e T
heor
y
MAS & STRATEGIC BEHAVIOR: As Is
Game theory
MAS, group control
Distributed optimization
Game theory
MAS, group control
Algorithmic game theory
Distributed optimization
Models of collective behavior, bounded
rationality
MAS & STRATEGIC BEHAVIOR: As Is
Game theory, mechanism design
MAS, group control
Algorithmic game theory
Distributed optimization
Models of collective behavior, bounded
rationality
Experimental economics, Behavioral
Game Theory
MAS & STRATEGIC BEHAVIOR: To Be
«MAXIMAL INTELLECTUALIZATION»
Intention to maximize the “intellectualization” is limited not only by “costs”(computational, cognitive, economical, etc), but by the inefficient “over-complexity”.
«Intellectualization»
«Costs»
«Result»
«Effect»
MAS: SOME QUALITATIVE RESULTS
1) The level of MAS’ “intellectualization” should be adequate to the problem in hand (taking into account various “costs”).
2) The intention to maximize the “intellectualization” at the higher levels of agent’s architecture leads to a centralized scheme.
3) Trend to the integration of networked MAS, game theory and artificial intelligence.
GENERAL TRENDS
1) Inter-disciplinary: control objects, methods and means of control.2) Network/hierarchical structure of controlled object, control system and
communications.3) Intra-paradigmal problems: «linearity» of development, desire to reduce the
problem to well-known, i.e. «internal» problems of any subject field. Self-isolation of different braches of control science. The demand for the creation of new adequate mathematical technique.
4) «Heuristical» applications: the concept of bounded rationality (under the lack of time, ability or necessity) – instead of optimal pseudo-optimal solutions are heuristically searched.
5) Unification:5C = Control + Computation + Communication + Cost + Cycle.
6) Heterogeneous (hierarchical, complex, integrated) modeling. Problems of models «coupling», search for common language. Generating and replicating typical solutions of control problems.
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
K(ρm) = )(
maxmEy ρ∈
f0(y) → ℘∈mρ
max .
Вертикальные связи
Горизонтальные связи
А
Б В
Г
Временная сетевая
организацияC Network
organization
Network organization -a structure, where temporal connections between the elements are actualized for the period of solving certain problem.
The problem of structural synthesis: find the number of levels of hierarchy m and distribution of agents among the levels (feasible structure ρm), which maximize the efficiency f0(y), given that the agents in corresponding hierarchical game choose their equilibrium actions: Examles:
- distributed production;- IT-projects;- group control, etc.
HIERARCHICAL AND NETWORK ORGANIZATIONS
Statement. For any technological structure and any equilibrium x ∈ E1
N(ρm) there exist the set of decision-making strategies
in a game Г2(ρm0), which guarantees to all the
decision-makers the same levels of utilities, as in the initial game.
OPTIMIZATION OF HIERARCHIES
Development of the organization management structure
Data collection and processing algorithms
Design of assembly plant Task management in network structures
…
Optimization of control hierarchy
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
LABOR SUPPLY (theory)
α α0
τ(α)
0
I = 1
α α0
τ(α)
0
I = 2
α α0
τ(α)
0
I = 3
α
α0
τ(α)
0 αmax
I = 4
α α0
τ(α)
0 αmax
I = 5
τ(α) – desired working time (hours/day), α – wage rate
Индекс (1999)
I25%
II53%
III6%
IV16%
Индекс (2009)
I25%
II41%
III16%
IV8%
V10%
Sample volume> 500
Sample volume> 5000
LABOR SUPPLY (“practice”)
α α0
τ(α)
0
I = 1
α α0
τ(α)
0
I = 2
α α0
τ(α)
0
I = 3
α α0
τ(α)
0 αmax
I = 4
α α0
τ(α)
0 αmax
I = 5
τ(α) – desired working time (hours/day), α – wage rate
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
Economic agents (enterprises)
Environment (NATURE)
Control Center (principal)
Interaction with environment
Results of
activity
Control
State of nature
ECOLOGO-ECONOMICAL SYSTEM
Model: u ≥ 0 – production level; y ≥ 0 – security level; z(u), ϕ(y) – enterprise’s costs; H i(u, y) – principal’s income;
σi(u′, u, x, y) =
≠≠==
'
'
или,0,,
uuxyuuxyV i ;
K = 1, 2, …, k – the set of principals. Goal function of i-th principal: Φ i(σi(⋅), u, y) = H i(u, y) – σi(u, y), i ∈ K, Goal function of the enterprise: f(σi(⋅), u, y) = c u + ∑
∈σ
Kii yu ),( – z(u) – ϕ(y).
u* = arg 0
max≥u
[c u – z(u)];
Φ*i =
0,max
≥yu [Hi(u, y) – c (u* – u) + [z(u*) – z(u)] – ϕ(y)], i ∈ K;
S = u ≥ 0, y ≥ 0 | ∃ V ∈ k+ℜ : Hi(u, y) – Vi ≥ Φ*
i, i ∈ K,
∑∈Ki
iV = c (u* – u) – [z(u*) – z(u)] + ϕ(y);
Λ = u ≥ 0, y ≥ 0, V ∈ k+ℜ | Hi(u, y) – Vi ≥ Φ*
i, i ∈ K,
∑∈Ki
iV = c (u* – u) – [z(u*) – z(u)] + ϕ(y).
Φ*0 =
0,max
≥yu [ ∑
∈Kii yuH ),( – c (u* – u) + [z(u*) – z(u)] – ϕ(y)].
Theorem. The compromise set is not empty iff Φ*
0 ≥ ∑∈
ΦKi
i* .
ECOLOGY VS ECONOMY
Territory
Government
Industry
Enterprise
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
The problem of project’s duration reductionModel: ∆ – the required reduction of project duration; yi – the required reduction of i-th operation duration; xi – plan of reduction of i- th operation duration; ri – «efficiency» of i-th agent; N – set of agents (operations); ci (yi, ri) = ri ϕ (yi / ri) – costs function of i-th agent; λ – rate of payment for the reduction of project duration; fi (yi, ri) = λ yi – ci (yi, ri) – agent’s goal function; s = (s1, s2, ..., sn) – agents’ message;
Theorem. The procedure of decision-making xi (∆, s) = Vsi ∆,
λ (∆, s) = ϕ' (∆ / V), where V = ∑∈Ni
is , is straightforward, minimizes
total costs and allows any decentralization.
Procedures of decision-making (examples):manipulable non-manipulable
IS TRUTH-TELLING PROFITABLE?
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
HIERARCHICAL AUTOMATIZATION
INFORMATIONAL SYSTEMS
OLAP, BSC, DSS ERP MES SCM, CRM, PMS MRP2 MRP, CRP SCADA, DCS PLC, MicroPC
Finances,material
resources,personnel
MRP2
MRP, CRP
SCADA,DCS
PLC, MicroPC
INFORMATIONALSYSTEMS
Control ofTS «Optimization » Control of OS
OLAP, BSC, DSS
ERP
MES
SCM, CRM, PMS
? ?…
1. HISTORY AND TRENDS OF CONTROL THEORY2. HIERARCHICAL AND NETWORKED MODELS3. INTERDISCIPLINARY CONTROL MODELS4. EMPHASIS OF RECENT CONFERENCES5. MULTIAGENT SYSTEMS: SPECIFICITY AND ARCHITECTURE OF AN
AGENT6. HIERARCHICAL MODELS (EXAMLES):
- Diffusive bomb- Social networks- Combat modeling
7. SOME TRENDS IN NETWORKED AND HIERARCHICAL MODELING8. CONTROL THEORY: ANTI-INTUITIVE RESULTS (EXAMPLES)
- Optimization of hierarchies- Labor supply- Ecology VS Economy- Is truthtelling profitable?- Hierarchical automatization
9. PAST, PRESENT AND FUTURE OF CONTROL THEORY
PLAN
PAST, PRESENT AND FUTURE OF CONTROL THEORY (objects of control)
1860-е 1900 1930 1940 1950 1960 1970 1980 1990 2000 2010 2020 …
Mechanical systems
Technical systems Organizational and informational systems
Decentralized intellectual systems
t
- Technical systems
- Economical systems
- Ecological systems
- Live systems
- Social systems
???Степень
достижения поставленных целей
«Relevancy»
“Application field”
«UNCERTAINTY» PRINCIPLE
(«Application field») x («Relevancy») ≤ Const
Mathematics
Psyghology, Sociology, Pedagogy
Economics
Biology
Chemesrty Physics
“Weak” sciences
“Strong” sciences
CONTROL THEORY
LIMITS OF SCIENCE
Thank you !