for digital simulation and advanced computation

The Minnesota SupercomputingInstitute for Digital Simulationand Advanced Scientific Com-

putation (or MSI, as it is generallyknown) recently celebrated its twentiethbirthday. MSI was born and put intofull operation during 1984–86. It cameinto being through the timely andunique convergence during the early1980s of a series of national, state, andUniversity initiatives to strengthen su-percomputer manufacturing industriesin the state of Minnesota and to im-prove high-performance computer ac-cess and computational science and en-gineering at American universities. Es-pecially given the tremendous contribu-tions that MSI has made and continuesto make to the University of Minneso-ta’s international competitiveness, wethought it timely to recap some of theevents that led to MSI’s being created,as well as some of the early Institutehistory. The legacy of MSI and its com-putational infrastructure is rich. Forbrevity we concentrate here only on its

formative years, leaving the more com-plete story for another contribution.

The world of high-performancecomputing was very different in theearly 1980s. Much of what we take forgranted today, including such basics asnetworking, “the Internet,” and data vi-sualization, was either brand-new orunknown still. In addition, we feel it isimportant for us all to remember thecentral role that Minnesota and theMSI have played in the ever more im-portant field of high-performance com-puting. In fact, Minnesota, especially byway of the creative genius of SeymourCray, an alumnus of the University, wasthe birthplace of supercomputing andwas for many years the principal centerfor the design and manufacture of su-percomputers. Con-trol Data Corpora-tion, born here inMinnesota in 1957and Cray Research,Inc., formed here in1972, both intro-duced high-impact se-quences of the world’sfastest vectorized com-puters, while pioneer-ing such innovativecomputer technologiesas LSI chips andRISC, vector and par-allel architecturestrategies. As this pio-

neering effort grew, leaders of the localcomputer industry, most notably JohnRollwagen, CEO at Cray, and Bill Nor-ris at CDC, strongly urged the Univer-sity of Minnesota to develop a highprofile on the frontiers of academiccomputing and computer engineering.This message was articulated especiallyeffectively through an industrial adviso-ry council set up by Institute of Tech-nology Dean Roger Staehle when he ar-rived here in 1979.

The University took a bold step inthis direction in 1981, becoming thefirst American university to purchase aclass VI supercomputer (a 100megaflop Cray 1). In order to help

Supercomputing Institute


Twenty Years Ago:

Birth of the Supercomputing Institute

Volume 22, Number 2

Also in This Issue

Self-Consistent Tight-Binding Calculations . . . . . . . . . . . . . . . .4

Dennis Hejhal Awarded GårdingPrize . . . . . . . . . . . . . . . . . . . . . .7

VLab Tutorial . . . . . . . . . . . . . . . .7Research Reports . . . . . . . . . . . . . .9

Summer 2006 Research Bulletin of the Supercomputing Institute

for Digital Simulation and Advanced Computation

continued on page 2

Figure 1. MSC computer room with prototype and produc-tion Cray 2s and a Cyber 205.

Institute of Technology

(In recognition of the Supercomputing Institute’s 20th anniversary, the following article was prepared by Professors Thomas W. Jones, De-partment of Astronomy, and David A. Yuen, Department of Geology and Geophysics, both Fellows of the Supercomputing Institute andlong-time Institute researchers. Photographs are courtesy of Tom Jacobson and Scott Bertilson.)

manage the considerable expense of ac-quiring and operating the Cray 1, whilelimiting direct costs to academic com-puter users, the University decided alsoto sell supercomputing resources to out-side, non-academic users. To maintainits tax-exempt status while selling com-puter resources on the open market, in1982 the University created a for-profitcompany, the Minnesota Supercomput-ing Center, Inc. (originally known asResearch Equipment, Inc., or REI).Ownership of MSCI was shared withthe University of Minnesota Founda-tion, which was actually the majorityowner. Into 1985 supercomputing facil-ities continued to be operated by Uni-versity employees at its Academic Com-puting Center in Lauderdale, just offHighway 280. Both the original Cray 1and its successor, a single-processorCray 2, were initially installed at thatsite. By early 1986 MSCI was operatedas an independent company, and super-computer hardware ownership wastransferred to MSCI. The company’sboard of directors came from the Uni-versity of Minnesota Foundation andfrom the University administration. ItsCEO was John Sell, previously a Uni-versity Academic Computing Centeremployee. Coincidentally, the City ofMinneapolis and its Mayor Don Fraser,together with the State of Minnesotaand Governor Rudy Perpich, were try-ing to revitalize the old milling districtbetween downtown and the University

West BankCampus. Thisled to a so-called“High TechCorridor” initia-tive along Wash-ington Avenue.As a contribu-tion to this ef-fort and as aneffort to helpencourage finan-cial support forsupercomputingfrom the statelegislature,MSCI and its

supercomputers moved into a remod-eled soft drink warehouse at 1200Washington Avenue South in the fall of1986.

During this same period a broaderserious national concern was developingaround the realization that the US wasnot maintaining a competitive edge inhigh performance computing, generally.In 1982 the Department of Defenseand the National Science Foundationsponsored a blue ribbon panel to exam-ine this problem, chaired by Peter Laxof the Courant Institute. The resultingLax Report of 1984 stressed the needfor a more coordinated and aggressivenational academic computing effort,noting in particular that researchers atAmerican universities did not have ade-quate access to high-performance com-puters. In response to theLax Report, the NSF cre-ated an Office for Ad-vanced Scientific Com-puting and subsequentlyissued a call for proposalsto create several nationalacademic supercomput-ing centers.

Back at the Universityof Minnesota, in late1983 an ad hoc group offaculty, led by mathemat-ics professor George Sell,were aware of the likeli-hood of a national aca-demic supercomputing ini-

tiative and began to prepare for a Min-nesota bid. Physics professor TomWalsh agreed to lead that effort. The re-sulting plan involved creation of an aca-demic institute for high-performancecomputing distinct from MSCI. In sup-port, the University, as part of its initia-tive to secure legislative supercomput-ing funding, agreed to create severalnew faculty positions to recruit leadersin supercomputer-based research. TheMinnesota Supercomputer Institute wasthus born. The newly recruited facultywould be identified as Fellows of theMSI. An international and interdiscipli-nary faculty search was carried out inlate 1984 through the Institute of Tech-nology, then headed by Dean Ettore In-fante. The original MSI Fellows hiredthrough that search were Jan Almlöf(Chemistry), Paul Woodward (Astrono-my), David Yuen (Geophysics) andJohn Zabolitzky (Physics). AneesurRahman (Physics) was also recruited alittle later, but, unfortunately, he suc-cumbed to cancer shortly after his ar-rival at the University of Minnesota.Peter Patton, who had been director ofthe University Academic ComputingCenter, became the first MSI director,and Tom Walsh became the MSI scien-tific director. In an effort to broadenthe role of the MSI within the Univer-sity, the Office of Academic Affairs alsotook over its responsibility from IT.

In the meantime the University didsubmit its NSF bid to become one of

Supercomputing Institute Research Bulletin Summer 20062


Figure 2. Thomas Jacobson (right) and Paul Woodward with early Cray 2 application: Mandelbrot set explorer via remote shell from SGI Iris.

Figure 3. Early demo of Cray 2 networked via TCP/IPwith diverse clients.

four national supercomputing centers.The proposal, with Tom Walsh as Prin-cipal Investigator, was very forwardthinking and thus risky. It was centeredon the brand-new Cray 2 supercomput-er, which was the first supercomputerto utilize a flavor of the UNIX operatingsystem and to incorporate both vectorand parallel architectural features. Theplanned University Cray 2, with fourprocessors, also had 8 GB of memory,which was enormous for its day. Whilethe University did obtain what amount-ed to bridge funding from the NSF, itwas not selected to be one of the na-tional supercomputing centers. Fromthat point on the MSI has been operat-ed through state funds, although itspresence here has facilitated very sub-stantial outside funding to the Universi-ty. In the early MSI years the legislatureincluded a “state special” appropriationto purchase computing time fromMSCI.

As it was starting MSI was a real“seat of the pants” operation that re-quired lots of individual initiative.Among the many challenges was devel-oping networking connections betweenthe supercomputers and the main cam-pus. Networking technology was in itsinfancy then, and such modern stan-dards as the TCP/IP protocol werelargely unknown entities. Tom Jacobsonand Scott Bertilson from MSCI, alongwith Tom Walsh and his designatednetwork administrator, Randy Smith,set out to demonstrate TCP/IP viabilityfor academic computing. Very earlytests included setting up a 56 kbps con-nection between Lauderdale and theUniversity of Wisconsin in Madison

through the ARPANETusing a VAX computeras the local hub. Theidea of implementing aTCP/IP network on thecampus for researcheraccess, while central toeffective use of the Cray2, was resisted by askeptical Universitycommunity more famil-iar with an already ar-chaic remote entrypunch card computer ac-cess. It was important todemonstrate the newer technologieseven though they were not yet author-ized nor funded. This led to such unau-thorized tasks as installing optical fibercables on campus inside various subter-ranean steam tunnels under cover ofdarkness. The fibers actually had to beinstalled more than once, since itturned out that the coatings of the orig-inal fibers could not withstand thesteam tunnels’ extreme temperatures. Inthe end these initiatives won the dayand the University became an early par-ticipant in what became the moderncomputer communications paradigm.In 1985 and 1986 MSI hosted its firstsummer schools in supercomputing onthe east bank campus.

Before the new supercomputing fa-cilities on Washington Avenue wereavailable, the MSI was administered outof a series of temporary offices. InitiallyTom Walsh and his administrative assis-tant, Angie Vail, occupied two tiny of-fices in the Shepherd Labs on the EastBank campus. Angie recalls arriving forher first day on the job to find a phone

on the floor, butnothing else in ei-ther office. As theoperations cametogether and PeterPatton came onboard as the newdirector, anotherstaff person, AnnJohns, was hired.The MSI stafftransferred into

another temporary, larger, but very iso-lated location in the Business and Tech-nology Center across from theMetrodome on the West Bank. In Sep-tember 1986, MSI, along with MSCI,moved into its new, permanent facilityon Washington Avenue, where it re-mained until 2002. Once in the newbuilding, MSCI installed the Cray 1and both Cray 2 supercomputers, aswell as a Cyber 205 from CDC, mak-ing it one of the premier supercomput-ing centers in the world.

A few months after the move, PeterPatton left the University to take a cor-porate position in Austin, Texas. TomWalsh resigned his scientific directorposition in November 1986. By thattime, according to MSI records, alreadyover 250 University faculty researcherswere utilizing MSI supercomputers.The MSI was temporarily managed byan executive committee of MSI Fellows,but shortly thereafter, the two directorpositions were merged, and Don Truh-lar from the Department of Chemistrywas appointed to be the director. He re-mained in that position for almost 18years, guiding the MSI through a majorrevolution in the nature and scope ofhigh-performance computing.

Acknowledgements: We are verygrateful to many colleagues who helpedus reconstruct the events of this story.We especially thank Scott Bertilson,Tom Jacobson, Mitch Luskin, RamaMurthy, Skip Scriven, George Sell,Angie Vail, and Tom Walsh for theirrecollections.


Summer 2006 Supercomputing Institute Research Bulletin 3

Figure 4. Stuart Levy doing network testing.

Figure 5. Radical idea of UNIX as a supercomputer OS.


Physics

Though first principles calcula-tions that repeatedly solve thequantum mechanical Schrö-

dinger equation for the electrons ofatomic systems of chemical and materi-als science interest are now possible, thesize of the systems on which such calcu-lations can be done remains significant-ly constrained. Such calculations are re-quired for determining the ground stateatomic configuration (in Born-Oppen-heimer approximation) of a system ofmaterials science or chemical interest,and also for studying the classical atom-ic dynamics of such systems while con-tinually recalculating the forces on theatoms as the electrons rearrange them-selves. For routine, repeated calcula-tions, a first-principles calculationallimit of about 100 atoms is characteris-tic. On the other hand, the systems ofinterest in biochemistry and materialsnanoscience are larger than this by twoor more orders of magnitude, thoughthe details of the electronic structure arestill crucial for understanding of theimportant properties. For this reasonthere is a need for intermediate scalemethods, which take account of elec-tronic structure but are computationally

simpler and faster than full first princi-ples calculations.

The research group of Professor J.Woods Halley, Department of Physicsand Supercomputing Institute Fellow,has been working for several years onone such approach, called self-consis-tent tight binding (SCTB). In thatmethod, one uses a much smaller, local-ized basis of functions for description ofthe electronic wave functions than thebases used in first-principles methods.No integrals are performed. The modelis parametrized by the matrix elementsof the Hamiltonian in the limited basis.The calculations are self-consistent inthe charges at the atomic sites at theHartree level. The matrix elements ofthe model are fitted to more accuratefirst principles calculations on smallclusters or solids.

Recently, Professor Halley and re-search associate Alexander Tchernatin-sky have developed a new algorithm forimplementing calculations in thismethod that has increased the speed ofcalculations in which clusters are re-laxed to their ground state configura-tion by a factor of 10.

In the approach that SCTB and sim-ilar approaches were previously using,

the total energy of the assembly of nu-clei and electrons is expressed (withsome approximation for the electron-electron interaction terms) as a func-tional of positions {Ri} of the atoms andthe density matrix ρ describing theelectrons. (In a basis of position eigen-states, only the diagonal terms in thedensity matrix are required. This is theHohenberg-Kohn theorem. The densitymatrix is expressed in an arbitrary basisat this stage in the discussion, however.)At a given set of {Ri}, one minimizesthe total energy with respect to ρ sub-ject to certain constraints (discussedbelow) and then calculates the forces onthe atoms from the resulting densitymatrix. The atoms are then moved inaccordance with the calculated forces(either relaxationally or inertially) andthen the process is repeated. The slowstep is the minimization with respect toρ. In the earlier implementation (as pi-oneered by Kohn and Sham) one ex-presses the total energy in terms of or-bitals (sometimes called Penrose or-bitals) that are eigenstates of the densitymatrix treated as an integral operatorand then minimize the energy with re-spect to those orbitals. The result is aset of one-electron-like equations that

A New Structural Relaxation Technique for

Self-Consistent Tight-Binding Calculations

Figure 1. Computational results of a series of nanocrystals of anatase titanium dioxide relaxed to the ground state atomicand electronic configuration using SCTB and the old algorithm compared with those calculations using the new algorithmin relaxational mode. The results are the same but calculation time is 10 times faster.


are solved self consistently, much as inthe Hartree-Fock method. The solutionof the equations requires a matrix diag-onalization, which is slow despite excel-lent and highly refined algorithms, andthe cost scales as nearly the cube of thematrix dimension.

In the new algorithm, an idea fromDavid Vanderbilt’s group at Rutgers iscombined with elements of a methoddue to Roberto Car and MigueleParinello. Vanderbilt and coworkerspointed out that the solution to the setof one-electron equations could be re-framed in a way that did not involve amatrix diagonalization. Given an effec-tive one-electron Hamiltonian H1 they

minimized the quantity Tρ((3ρ2 –2ρ3)(H1 – µN)) and showed that theresult was equivalent to the ρ resultingfrom the solution of the Kohn-Shamequations. The peculiar expression 3ρ2

– 2ρ3, when minimized, gives a densitymatrix satisfying ρ2 = ρ, which is a con-dition (called idempotency) equivalentto the requirement that the Kohn-Shamorbitals be orthonormal. The Vanderbiltmethod works and is, in principle, cost-ing a computation time that scales lin-early in the number of electrons. A self-consistency loop, however, is still re-quired because H1 depends on ρ in theapplication to electronic structure. TheVanderbilt method was implemented

for SCTB and though it worked, it wasnot faster than the traditional algo-rithm.

The Car-Parinello (CP) method is away to avoid the matrix diagonalizationby letting the electronic wave functionschange as the atoms move, in such away that the electronic wave functionsfluctuate during the simulation aroundtheir correct values. Car and Parinellochose the coefficients of an expansionof the Kohn-Sham orbitals in a planewave basis as the variables that fluctuatein this way during the simulation. Theycomputed a classical “equation of mo-

Figure 2. Data revealing how rapid relaxation to the ground state configuration occurs for a 105-atom cluster.

Physics

continued on page 6


Physics

tion” for these coefficients, ignoring thecomplicating factor of keeping the or-bital orthogonal. The electron dynam-ics is unphysical, but it achieves thegoal of making the electronic wavefunctions oscillate around the right an-swer. The CP method is quite elegantin that the ionic coordinates and theelectronic ones are all treated on essen-tially the same footing. However theplane wave basis is very inappropriatefor the SCTB method.

In the new algorithm, these twoideas are combined. Consider the totalenergy functional Etotal(ρ{Ri}), where ρis the density matrix. Replace ρ every-where in it by the Vanderbilt functionP = 3ρ2 – 2ρ3. One can show that theminimum of the function Etotal(ρ,{Ri})– µ(N – TrP)with respect to the ele-ments of ρ and µ would solve the elec-tronic structure problem for fixed {Ri}.Now instead of using the coefficients ofa plane wave expansion of Kohn Shamorbitals, implement the Car-Parinelloidea using the elements of the densitymatrix. The result is a set of equationsof the form:

This can be called inertial CP(which is the form the original creatorsof CP used). One can also replace theleft hand sides by terms of the formγdx/dt where x = Ri, ρiµ,jν, or µ (relax-ational CP). Both forms of the algo-rithm were successfully implementedfor SCTB, with the best results whenall the left hand sides are relaxational.

The Halley group previously relaxeda series of nanocrystals of anatase titani-um dioxide to the ground state atomicand electronic configuration usingSCTB and the old algorithm. (Two ofthe smaller ones are shown in the insetto Figure 2.) They recently repeatedthose calculations using the new algo-rithm in relaxational mode. The resultsand compared computational times areshown in the table in Figure 1. The re-sults are identical and the computation-al times (per processor) are about 10times faster. These clusters are actuallystill too small to profit from the orderN features of the Vanderbilt idea.

For the 105-atom cluster, Figure 2shows some data revealing how thisrapid relaxation to the ground state

configuration occurs. The totalnumber of electrons and a

measure of the idempotency of the den-sity matrix are shown as a function ofthe number of simulation steps. Theplot of total electron number showsthat the electron number is very wellpreserved throughout the relaxation.On the other hand, the second plotshows that the density matrix relaxesthe most slowly to idempotency, so thatduring most of the relaxation the densi-ty matrix is not very close to idempo-tency. Thus one way to describe thephenomenon is to say that the systemfinds a faster route to the ground statewhen it is able to explore paths throughthe density matrix space that are notidempotent.

In the future this algorithm will berefined and applied to study clusters 10times larger than those explored before.The inertial forms of the algorithm formolecular dynamics studies is being re-fined and should make more realisticstudies of systems like water-metal in-terfaces possible. At very large systemsizes, the order N features of the algo-rithm will become relevant, making itsadvantages over traditional methodseven greater.


Dennis Hejhal, School of Math-ematics and SupercomputingInstitute Fellow, has been

awarded the prestigious Eva and LarsGårding prize by the Royal Physio-graphic Society in Lund, Sweden. Thisorganization, founded in 1772, sup-ports natural sciences in Sweden withresearch grants and prizes. Besides hisposition at the University of Minnesota,Professor Hejhal is also a professor ofmathematics at Uppsala University inSweden. The award was presented onDecember 2, 2005 at the University ofLund.

Professor Hejhal’s research involvesinvestigations into analytic number the-ory and spectral theory on surfaces ofnegative curvature. The paper that wonthe Gårding prize is concerned with theRiemann hypothesis, which states thatall nontrivial zeros of the Riemann zetafunction lie on the critical line of com-

plex numbers with real part 1/2. Thishypothesis is important to assertionsabout the distribution of prime num-bers. The paper was published in Inter-national Mathematics Research Noticesin 2000.

Professor Hejhal is a long-time prin-cipal investigator at the Supercomput-

ing Institute. Much of his early experi-mental work was performed on the In-stitute’s Cray supercomputers duringthe 1980s and early 1990s. He usescomputer modeling to gain insight intomathematical theorems.

Mathematics

Dennis Hejhal Awarded Gårding Prize

VLab Tutorial, May 21–June 3, 2006

The Virtual Laboratory for Earthand Planetary Materials (VLab)held a Tutorial on Computa-

tional Materials/Mineral Physics onMay 21–June 3, 2006 at the Universityof Minnesota. This two-week tutorialincluded presentations on theoreticalfoundations and hands-on experiencewith the open source package QuantumESPRESSO.

Further information about the tuto-rial, including notes from the lectures,can be found at the VLab Web site:

www.vlab.msi.umn.edu/events /first-Tutorial.shtml

Information about the 2007 tutorialwill be posted on the VLab and Super-computing Institute Web sites when itbecomes available.

The VLab is supported by the Na-tional Science Foundation and involves

collaborators from: the University ofMinnesota; Stony Brook University;Florida State University; Louisiana StateUniversity; Indiana University; the Uni-

continued on page 10

Events, Symposia, and Meetings

continued on page 8

Right: VLab Tutorial attendees talkduring a break.

versity of California–Santa Barbara;DEMOCRITOS Modeling Center(Italy); Daresbury Laboratory (UK);and University College London (UK).

The principal investigator is Profes-sor Renata M. Wentzcovitch, Depart-ment of Chemical Engineering andMaterials Science and SupercomputingInstitute Fellow. The co-principal inves-tigators are: Professor Yousef Saad, De-partment of Computer Science and En-gineering and Supercomputing InstituteFellow; Professor J. Ilja Siepmann, De-partment of Chemistry and Supercom-

puting Institute Fellow; Professor Don-ald G. Truhlar, Department of Chem-istry and Supercomputing Institute Fel-low; Professor David A. Yuen, Depart-ment of Geology and Geophysics andSupercomputing Institute Fellow; Pro-fessor Philip B. Allen, Department ofPhysics, Stony Brook University; Pro-fessor Gordon Erlebacher, Departmentof Information Technology, FloridaState University; Professor Bijaya B.Karki, Department of Computer Sci-ence, Louisiana State University; Profes-sor Marlon Pierce, Department of In-

formation Technology, Indiana Univer-sity; and Professor Frank Spera, Depart-ment of Geology and Geophysics, Uni-versity of California–Santa Barbara.

For more information about theVLab, see:

www.vlab.msi.umn.edu/

Events, Symposia, and Meetings


Clockwise from top:VLab Tutorial participants participate in a hands-on workshop.Matteo Cococcioni, Massachusetts Institute of Technology, instructed some of the workshops.Tao Sun, Stony Brook University (left) and Koichiro Umemoto, University of Minnesota and another VLab Tutorial instructor,at their workstations during a workshop.

Research Reports


Aerospace Engineering

and Mechanics

2006/60, June 2006Variational Characterization of aQuasi-Rigid BodyR. Fosdick and G. Royer-Carfagni

Chemical Engineering

and Materials Science

2006/74, July 2006 Post-Perovskite MgSiO3 Investigat-ed by First PrinciplesT. Tsuchiya, J. Tsuchiya, andR. M. Wentzcovitch

2006/75, July 2006 and VLab 2006-4

MgSiO3 Postperovskite at D” Con-ditionsR. M. Wentzcovitch, T.Tsuchiya, and J. Tsuchiya

2006/76, July 2006 and VLab 2006-5

Dissociation of MgSiO3 in theCores of Gas Giants and Terrestri-al ExoplanetsK. Umemoto, R. M. Wentzcov-itch, and P. B. Allen

2006/77, July 2006 and VLab 2006-6

Spin Transition in Magne-siowüstite in Earth’s Lower MantleT. Tsuchiya, R. M. Wentzcov-itch, C. R. S. da Silva, and S. deGironcoli

2006/78, July 2006 and VLab 2006-7

Elasticity of CaSiO3 Perovskite atHigh Pressure and High Tempera-tureL. Li, D. J. Weidner, J. Brodholt,D. Alfe, G. D. Price, R. Caracas,and R. Wentzcovitch

2006/79, July 2006 and VLab 2006-8

Phase Stability of CaSiO3 Per-ovskite at High Pressure and Tem-perature: Insights From Ab InitioMolecular DynamicsL. Li, D. J. Weidner, J. Brodholt,D. Alfè, G. D. Price, R. Caracas,and R. Wentzcovitch

2006/80, July 2006 and VLab 2006-9

Density Functional Study of Vibra-tional and Thermodynamic Prop-erties of RingwooditeY. G. Yu and R. M. Wentzcov-itch

2006/81, July 2006 and VLab 2006-10

Composition Controlled Spin Po-larization in Co1-xFexS2: Electron-ic, Magnetic, and ThermodynamicPropertiesL. Wang, T. Y. Chen, C. L.Chien, J. G. Checkelsky, J. C.Eckert, E. D. Dahlberg,K. Umemoto, R. M. Wentzcov-itch, and C. Leighton

2006/82, July 2006 and VLab 2006-11

NaMgF3: A Low-Pressure Analogof MgSiO3K. Umemoto, R. M. Wentzcov-itch, D. J. Weidner, and J. B.Parise

2006/83, July 2006 and VLab 2006-12

Vibrational and ThermodynamicProperties of Forsterite at MantleConditionsL. Li, R. M. Wentzcovitch, D. J.Weidner, and C. R. S. Da Silva

2006/84, July 2006 and VLab 2006-13

Electronic Structure of Co1-xFexS2K. Umemoto, R. M. Wentzcov-itch, L. Wang, and C. Leighton

2006/85, July 2006 and VLab 2006-14

Pressure Induced High Spin toLow Spin Transition in Magne-siowüstiteT. Tsuchiya, R. M. Wentzcov-itch, C. R. S. da Silva, S. deGironcoli, and J. Tsuchiya

2006/86, July 2006and VLab 2006-15

Theory of Spintronic MaterialsJ. R. Chelikowsky, E. Kaxiras,and R. M. Wentzcovitch

Chemistry

2006/42, April 2006 and CB 2006-11

QCRNA1.0: A Database ofQuantum Calculations for RNACatalysisT. J. Giese, B. A. Gregersen,Y. Liu, E. Mayaan, A. Moser,K. Range, O. Nieto Faza,C. Silva Lopez, A. Rodriguez deLera, G. Schaftenaar, X. Lopez,T.-S. Lee, G. Karypis, andD. M. York

2006/43, April 2006 and CB 2006-12

CHARMM Force Field Parametersfor Simulation of Reactive Inter-mediates in Native and Thio-Sub-stituted RibozymesE. Mayaan, A. Moser, A. D.Mackerell, Jr., and D. M. York

2006/44 April 2006 and CB 2006-13

Nucleophilic Attack on PhosphateDiesters: A Density FunctionalStudy of In-Line Reactivity in Di-anionic, Monoanionic, and Neu-tral SystemsX. Lopez, A. Dejaegere,F. Leclerc, D. M. York, andM. Karplus

Names of University of Minnesota principal investigators appear in bold type.

Research Reports


2006/45, April 2006 and CB 2006-14

Transesterification Thio Effects ofPhosphate Diesters: Free EnergyBarriers, Kinetic and EquilibriumIsotope Effects From Density-Func-tional TheoryY. Liu, B. A. Gregersen,A. Hengge, and D. M. York

2006/46, April 2006 and CB 2006-15

Multilevel and Density FunctionalElectronic Structure Calculationsof Proton Affinities and Gas-PhaseBasicities Involved in BiologicalPhosphoryl TransferK. Range, C. Silva Lopez,A. Moser, and D. M. York

2006/47, April 2006 and CB 2006-16

Normal Modes of Redox-Active Ty-rosine: Conformation Dependenceand Comparison to ExperimentK. Range, I. Ayala, D. York, andB. A. Barry

2006/51, May 2006 Microscopic Structure and Solva-tion in Dry and Wet OctanolB. Chen and J. I. Siepmann

2006/52, May 2006Structure and Dynamics of the Ac-queous Liquid-Vapor Interface: AComprehensive Particle-BasedSimulation StudyI-F. W. Kuo, C. J. Mundy, B. L.Eggimann, M. J. McGrath, J. I.Siepmann, B. Chen, J. Vieceli,and D. J. Tobias

2006/53, May 2006Critical Properties of AluminumD. Bhatt, A. W. Jasper, N. E.Schultz, J. I. Siepmann, andD. G. Truhlar

2006/54, May 2006Molecular Simulation Approachesto SolubilityK. E. Anderson and J. I. Siep-mann

2006/55, May 2006Synthesis, Molecular and Electron-ic Structure, and TDDFT andTDDFT-PCM Study of the Solva-tochromic Properties of(Me2Pipdt)Mo(CO4) Complex(Me2Pipdt = N,N’-Dimethylpiper-azine-2,3-dithione)V. N. Nemykin, J. G. Olsen,E. Perera, and P. Basu

2006/63, June 2006QM/MM: What Have WeLearned, Where Are We, andWhere Do We Go From HereH. Lin and D. G. Truhlar

2006/64, June 2006Variational Transition State Theo-ry and Multidimensional Tunnel-ing for Simple and Complex Reac-tions in the Gas Phase, Solids, Liq-uids, and EnzymesD. G. Truhlar

2006/66, June 2006Predicting Aqueous Free Energiesof Solvation as Functions of Tem-peratureA. C. Chamberlin, C. J. Cramer,and D. G. Truhlar

2006/67, June 2006Combined Valence Bond-Molecu-lar Mechanics Potential-EnergySurface and Direct DynamicsStudy of Rate Constants and Ki-netic Isotope Effects for the H +C2H6 ReactionA. Chakraborty, Y. Zhao, H. Lin,and D. G. Truhlar

2006/68, June 2006Design of Density Functionals byCombining the Method of Con-straint Satisfaction With Parame-trization for Thermochemistry,Thermochemical Kinetics, andNoncovalent InteractionsY. Zhao, N. E. Schultz, andD. G. Truhlar

2006/69, June 2006Direct Calculation of Coupled Di-abatic Potential-Energy Surfacesfor Ammonia and Mapping of aFour-Dimensional Conical Inter-section SeamS. Nangia and D. G. Truhlar

2006/70, June 2006Comparative DFT Study of vander Waals Complexes: Rare-GasDimers, Alkaline-Earth Dimers,Zinc Dimer, and Zinc-Rare-GasDimersY. Zhao and D. G. Truhlar

2006/71, June 2006Converged Vibrational EnergyLevels and Quantum MechanicalVibrational Partition Function ofEthaneA. Chakraborty and D. G. Truh-lar

2006/72, June 2006Assessment of the Pairwise AdditiveApproximation and Evaluation ofMany-Body Terms for Water Clus-tersE. E. Dahlke and D. G. Truhlar

Civil Engineering

2006/49, May 2006Solute Transfer Into a SedimentLayer Under a Moving WaterWave: Numerical Solution of the2-D Advection/Dispersion Equa-tion and Derivation of 1-D Dis-persion CoefficientQ. Qian, V. R. Voller, and H. G.Stefan

2006/57, May 2006Benchmark of QUICK Scheme forEnvironmenbtal ApplicationQ. Qian, V. R. Voller, and H. G.Stefan

2006/61, June 2006Singular Boundary Elements forThree-Dimensional ElasticityProblemsB. B. Guzina, R. Y. S. Pak, andA. E. Martinez-Castro

Names of University of Minnesota principal investigators appear in bold type.

Research Reports


2006/62, June 2006Streamflow in Minnesota: Indica-tor of Climate ChangeE. V. Novotny and H. G. Stefan

Computer Science

and Engineering

2006/42, April 2006 and CB 2006-11

QCRNA1.0: A Database ofQuantum Calculations for RNACatalysisT. J. Giese, B. A. Gregersen,Y. Liu, E. Mayaan, A. Moser,K. Range, O. Nieto Faza,C. Silva Lopez, A. Rodriguez deLera, G. Schaftenaar, X. Lopez,T.-S. Lee, G. Karypis, andD. M. York

2006/58, May 2006Greedy Coarsening Strategies forNon-Symmetric ProblemsS. MacLachlan and Y. Saad

Geology and Geophysics

2006/56, May 2006Resolution Tests of 3-D ConvectionModels by Travel-Time Tomogra-phy: Effects of Rayleigh Numberand Regular vs. Irregular Parame-terizationM. Behounkova, H. Cizkova,C. Matyska, D. A. Yuen, andM. S. Wang

Mechanical Engineering

2006/50, May 2006Two-Dimensional NumericalStudy of Atmospheric PressureGlows in Helium With ImpuritiesP. Zhang and U. Kortshagen

Medicinal Chemistry

2006/48, April 2006 and CB 2006-17

Spectrophotometric Determinationand Computational Evaluation ofthe Rates of Hydrolysis of 9-Amino-Substituted AcridinesJ. R. Goodell, B. Svensson, andD. M. Ferguson

Plant Pathology

2006/41, April 2006 and CB 2006-10

Towards Efficient Isolation of RGene Orthologs From MultipleGenotypes: Optimization of LongRange-PCRM. J. Sanchez and J. M.Bradeen

Surgery

2006/59, May 2006 and CB 2006-18

A Seven Nanosecond Particle MeshEwald (PME) Periodic BoundarySolvated Molecular DynamnicSimulation of the Phase 434 cl Re-pressor Protein DNA Binding Do-main in Complex With Its Cog-nate Operator (OR2) DNA Se-quenceL. F. Harris, M. R. Sullivan, andP. D. Popkin-Harris

Instructions for obtaining copies of UMSI Research Reports can be found on our Web site:

www.msi.umn.edu/cgi-bin/reports/searchv2.html

The 2006 Annual Research Report of the Supercomputing Institute is now available. You can download a PDFfrom our Web site:

www.msi.umn.edu/general/Reports/annual.html

or request a copy by sending email to: [email protected]

or mail to:

University of MinnesotaSupercomputing Institute

599 Walter117 Pleasant Street SE

Minneapolis, Minnesota 55455

Please help us keep our

mailing list up to date

Check one:

❏ Add name to mailing list

❏ Make corrections/changes as shown

❏ Remove name from mailing list

Name__________________________Institution ______________________Address __________________________________________________________________________________________________________________

Phone _________________________Email__________________________

This information can be mailed to the

Su p e rcomputing Institute, 599 Wa l t e r,

117 Pleasant St reet SE, Minneapolis, MN 55455,

f a xed to (612) 624-8861, or emailed to re q u e s t s @ m s i . u m n . e d u

For more information on the University of Minnesota Su-percomputing Institute, please contact us at the addressgiven below. You may also consult the Supercomputing In-stitute’s Internet pages:

www.msi.umn.edu

This information is available in alternative formats uponrequest by individuals with disabilities. Please send email to [email protected] or call 612-625-1818.

Editor/photographer: Tracey A. Bartlett

The University of Minnesota is committed to the policy that allpersons shall have equal access to its programs, facilities, andemployment without regard to race, color, creed, religion, na-tional origin, sex, age, marital status, disability, public assistancestatus, veteran status, or sexual orientation.

© 2006 Un i versity of Mi n n e s o t a

Contains a minimum of 10% postconsumer waste

Supercomputing Institute 599 Walter117 Pleasant Street SEMinneapolis, MN 55455

Change Service Requested

for digital simulation and advanced computation

Documents