Embed Size (px)
Citation preview
BNL 66299
Proceedings of RIKEN BNL Research Center Workshop Volume 13
PHYSICS OF THE 1 TERAFLOP
RIKEN-BNL-COLUMBIA QCD PROJECT
October 16, 1998
Organizer
Robert Mawhinney
FIRST ANNIVERSARY CELEBRATION
October 16, 1998
RIKEN BNL Research CenterBuilding 510, Brookhaven National Laboratory, Upton, NY 11973, USA
Other RIKEN BNL Research Center Proceedings Volumes:
Volume 12- Quarkonium Production in Relativistic Nuclear Collisions - BNL-September 28–October 2, 1998- Organizer: Dmitri Kharzeev
Volume 11- Event Generator for RHIC Spin Physics - BNL-66116September 21–23, 1998- Organizers: Naohito Saito and Andreas Schaefer
Volume 10- Physics of Polarimetry at RHIC - BNL-65926August 4–7, 1998- Organizers: Ken Imai and Doug Fields
Volume 9- High Density Matter in AGS, SPS and RHIC Collisions - BNL-65762July 11, 1998- Organizers: Klaus Kinder-Geiger and Yang Pang
Volume 8- Fermion Frontiers in Vector Lattice Gauge Theories - BNL-65634May 6–9, 1998- Organizers: Robert Mawhinney and Shigemi Ohta
Volume 7- RHIC Spin Physics - BNL-65615April 27–29, 1998- Organizers: Gerry Bunce,Yousef Makdisi, Naohito Saito, Mike Tannenbaum and Aki Yokosawa
Volume 6- Quarks and Gluons in the Nucleon - BNL-65234November 28–29, 1997, Saitama, JapanOrganizers: Tashiaki Shibata and Koichi Yazaki
Volume 5- Color Superconductivity, Instantons and Parity (Non? )-Conservation atHigh Baryon Density - BNL-65105November 11, 1997- Organizer - Miklos Gyulassy
Volume 4- Inauguration Ceremony, September 22 andNon-Equilibrium Many Body Dynamics - BNL- 64912September 23–25, 1997- Organizer - Miklos Gyulassy
Volume 3- Hadron Spin-Flip at RHIC Energies - BNL-64724July 21- August 22, 1997- Organizers: T.L. Trueman and Elliot Leader
Volume 2- Perturbative QCD as a Probe of Hadron Structure - BNL-64723July 14–25, 1997- Organizers: Robert Jaffe and George Sterman
Volume 1- Open Standards for Cascade Models for RHIC - BNL-64722June 23–27, 1997- Organizer - Miklos Gyulassy
Photographs by Joseph Rubino, Minoru Yanokura, BNL Video and the Information Technology Division
Dr. Nicholas P. Samios, Professor T. D. Lee,Dr. Akira Masaike, and Dr. Satoshi Ozaki
Dr. Satoshi Ozaki (BNL, RHIC Project Director)Chairperson of the RBRCFit Anniversary Symposium
Tip Photo. (Clockwise, from left)
T. D. Lee, RIKBN BNL Research Center
(RBRC) D~ctor, Columbia University (CU}
Shun-ichi Kobayashi, President ofRIKEN,
The Institute of Physical&Chemical Research,
Japan; Kouichi Yazaki, University of Tokyo,
Japam Minoru Yanokura, RIKEM
Shigemi Ohta, RBRC and KEK,Japan;
Edward McFadden, BW,
Robert Mawhinney, CTJ
John Marburger, BNL Dwecto~ Satoshi Osaki,
RHIC Project Director, Bm
Norman Cbristj CU;
Fujio Sakauchi, RIKBN Akira Ukawa,
University of Tsukuba, Japan;
Raph Kasper, CU; and (center)
Nicholas Samios, RBRC Deputy Director.
Photo L@
Shun-ichi Kobayashl,
President of RIKEN, Japan
Akiko Arima, MichaelJ. Tannenbaum (BNL), George Sterman (SUNY,SB), Peter Bond, Nicholas P.Samios, YousefMakdisi, Richard Melucci, and Satoshi Ozaki (BNL)
Shigemi Ohta, Doris Rueger, Rae Greenberg, Fern Simes, and Akiko Mmawith the RIKEN-BNLSupercomputer
B@RIKEN BNL Research CenterBldg.570,Brookhaven National Laboratory Upton, NY 11973, USA
FIRST ANNIVERSARY CELEBRATION
Friday, October 16,1998Brookhaven National Laboratory
Physics Department, Large Seminar Room
PROGRAM
09:00AM Registration Desk Opens(Physics Department, Seminar Lounge)
10:00 AM SYMPOSIUM
Chairperson: Doctor Satoshi Ozaki
Addresses byProfessor Shun-ichi Kobayashi, President of RIKENDoctor John H. Marburger, III, Director of BNLProfessor T. D. Lee, Director of RIKEN BNL Research CenterProfessor Shigemi Ohta, RIKEN BNL Research Center & KEK
11:15 AM Tour of 0.6 Teraflop Computer, Building 515
12:00Noon Luncheon (Berktzer Hall, Room B)
02:00 PM Parallel Prorram
Tour of BNL Facilities: RHIC, NSLS, and IMAGING CENTER(Meet at Berkner Hall)
Workshop: Physics of 1 TeraflopRIKEN-BNL-Columbia QCD Project(Physics Department, Small Seminar Room)Organizec Robert Mawhinney
I
Remarks by BNL Director John Marburger on the occasion of the first anniversary of theRIKEN/BNL Research Center
I first learned about the RIKEN/BNL Research Center from its distinguishedDirector T.D. Lee and co-founder Nick Sarnios. At first I suspected them of exaggeratingbecause it was difficult for me to understand how a Center so recently established couldhave accomplished so much. But the accomplishments have been tangible: workshops,publications, the formation of a small but vibrant research community, even the actualconstruction of a “superComputer” optimized for quantum chromodynamics calculations,and advances in soflware needed to exploit it.
~~f~~ lnncr the mvderv nf <uch rnnid accnmnlichment ht=mwne elem- tn me Each.-.-O --- --.= ---- J ‘- “---- .-.=..- _--=_a r---------- W-----W -.-s .= . ..w. _w4A
partner had been inspired by the vision of Director Lee about what could be. .. A—---L.. L-.J_2&L _.-L - ---*-- -. AL.. nd--l.L-.. -— xTeL. _—*l T ~l-––-*— -- m.- !J. - -L- .ilUAJIll~llWICU W ILU SUUU d LXIILCI ill LUG DKKKIILIVCIl lNdUUIldl LdU(J17dLOI’y. 1 Ile l(lCZl 01 iirl
intellectual community to take advantage of the discoveries soon to be made at BNL’sReiativktic Heavy Ion Coiiidmis very compelling. The administrations at BNL and atRIKEN moved immediately to take advantage of Professor Lee’s willingness to lead theCenter. T.D. Lee’s visio~ energy, and charismatic leadership drew intellectual supportfrom many quarters, and soon the exciting and moductive group was formed whose=–. . ..accomplishments we are celebrating now.
Few such centers have become so productive in such a short time. AIready in itsfirst year this Center has become an inciispensabie part of the Brookhaven N-ationaiLaboratory mission. I pledge to you my support to keep up this momentum. As thevision of RHIC becomes a reality during 1999, the vision of the RIKEN/BNL Center willgrow into a position of leadership in the new world of physics RHIC will unveil. I wishto extend my deep gratitude to the leaders of RIKEN to make this marvelous assetmmil~hle tn IIC And T Innlc fnrwarrl eauerlv tn anniw=rcnrv mlmher twn !-. —----- ..- —. - —- - -., -.. . . . . . . . - W---J .- —- . “-u-J ..--uv. ... v.
John Marburger
an essential role in the development of quantum mechanics in the 1920’s. At thattime, Bohr had already formulated his quantization rule and Einstein had completedhis Theory of Relativity. A new generation of physicists went through the BohrInstitute, including Heisenberg, Dirac, Pauli, Nishina, and others. While Bohr andEinstein were still active at that time, they did not discover quantum mechanics.Quantum mechanics was created by this new group of young physicists, mostly intheir twenties.
In the 1930s and 1940s, there were new challenges in nuclear forces andquantum electrodynamics. At that time, the heroic generation who created quantummechanics was still active. Yet these new challenges were met and resolved byanother group of younger theorists and experimentalists, Yukawa, Tomonaga atIUKEN, Schwinger, Lamb, Kusch at Columbia and Feynrnan, Dyson at Cornell.Again, almost all were in their twenties and thirnes.
In the 1950s came the strange particles and their puzzling behavior. Once more,the new physics was made by yet another young generation of physicists: GeU-Mann,Narnbu, Yang, myself and others. Then, in the late 1960’s and 70s, there remainedthe tasks of completing the unification of the electromagnetic and the weakinteractions into a single electroweak force and the formulation of the stronginteraction in terms of quantum chromodynarnics. These problems were solved bystill another generation of young physicists, Weinberg, Glashow, Politzer, ‘t Hooftand others.
The progress of physics depends on young scientists facing new challenges. Tocreate the new physics, there has to be the right place at the right time for these youngscientists. This pattern will not change.
II
I
I
I mentioned to Dr. &ima that the time is now ripe for another revolution inphysics, one which will connect the microscopic particle with the macroscopic stateof the physical vacuum. When the Relativistic Heavy Ion Collider is turned on in1999 at BNL, we expect a new state of matter to be created in which chiral symmetrywill be restored and the confinement of quarks and gluons in normal nuclear matterwill no longer hold. The actual discovery of this state is yet to be made. If realized, ittill open the door to a new realm in which, for the first time, the coherent propertiesof the vacuum can be changed through experimental means. These changes willbecome an integral part of the new physics that determines the behavior ofsubnuclear particles. RHIC physics is at the forefront of our science. It intersectswith nuclear, particle and relativistic condensed matter physics. And once again, theforging of this ffontier and the search for the new laws of physics will be done byanother generation of young physicists.
Dr. Arima and I both felt that a new center dedicated to this quest should becreated, and that it would be best located near the RHIC accelerator. We also agreedthat the center should have both theorists and experimentalists so that the theoreticaland experimental discoveries will come hand in hand.
We then discussed how large the theoretical component should be. I mentionedthat, in order for the young scientists to have close interaction with each other, the
First Anniversary
The Tao generates one ~ &m %
oOne generates two, o
Two generates three, 0@ Y 2Three generates
ten thousand things. : Y ;$b- Lao Tse (6’h century BC)
Physics is everywhere at the Center, in the offices, at the corridors, around thecoffee machine and during lunches. These weekly seminars provide the main arteriesof communications indepth discussions. They are all well attended. Besides theRBRC members, active scientists from BNL, Columbi% Stoney Brook are regularparticipants.
The Tuesday spin physics seminm organized by Bob Jaffe and Gerry Buncedesemes a special mention. The participants consists half of theorists and half ofexperimentalists. In the short span of a year, it has unified the scientific communityactive in this field. With the RHIC accelerator to be on-line next year, BNL will bethe world center of spin physics research. The RBRC Tuesday seminars havesucceeded in defining the critical spin experiment which have to be done and infocusing on the important physics information that should be extracted. This is anexcellent example of the close working relationship between the theory andexperimental divisions of the RIKEN BNL Research Center.
The research activities and the intellectual intensity of the Center are furtherenriched by the RBRC Workshops. Each Workshop centers on a specific importantproblem in strong interaction physics. The participants consist of the Center scientistsand other leading experts from all over the world. A useful feature is to have rapidpublication of the proceedings, so that the results of the workshops can be readilydisseminated. Here, at the Center the proceedings are available within a fewweeks after the workshops. This is another unique feature of the Center.
To date, twelve workshops have already taken place at RBRC, of theses, four areon spin physics and eight on RHIC and other QCD physics. These are listed in Table1, together with the upcoming RBRC workshops. The first nine proceedings havealready been distributed world-wide. Within this Japan Fiscal Year, five moreworkshops will take place. This afternoon, as part of the First AnniversaryCelebration, there will be a workshop organized by Bob Mawhinney on the “Physicsof one Teraflop RHCEN - BNL - Columbia QCD Project.”
A test of the quality of any research institution is that of its scientificpublications. The lastest count gives31 scientific papers written by RBRC members.Table 2 gives the list of these papers.
I
1
I
I
I
I
I
The construction of the 0.6 Teraflop RIKEN BNL Research CenterSupercomputer began on February 19th of this year and was completed on August28th. This must be a world record for speed of construction. It has just been awardedthe SC% Golden Bell Finalist for Price Performance. I wishtithank the BrookhavenCCD scientists for their help and suppom
The RBRC 0.6 Teraflop is working in tandem with the Columbia 0.4 Teraflopforming the combined RIKEN-BNL-Columbia one Teraflop QCD project. TheColumbia component was funded through the U.S. Department of Energy. We arefortunate to have the leadership of Norman Christ and Bob Mawhinney for this joint
,
I
1998 isthe year of
the tiger.
New Tenure Track Strong Interaction TheoryRHIC Physics Fellow Program
● Each Fellow spends half time at RBRC and half at the participating institution.● Each participating institution must hold a tenure track position in strong interaction
physics (high energy nuclear theory, RI-UC phenomenology, perturbative and latticeQCD, hadronic spin physics, hadronic spectra and matrix elements).
To date seven institutions have joined this initiative; that means seven new tenure-track positions have been created. In this new program, RBRC supports each Fellowonly for the time that the Fellow is at the Center. Thus, the six unfilled positionsbecome 12 RHIC Physics Fellows. This new initiative aims at the entire physicscommunity globally. The international character and especially the balance of thestrong Japanese component of the present theory group will be maintained andstrengthened.
Ian confident, with this new initiative, RBRC will attract the best young talents,and be ready for the physics that will come out of the RHIC accelerator. What arethe challenges in modem physics? Two of these challenges are listed below:
● Half of the elementary particles are unfree.● Lack of precision QCD calculations.
How should we meet them. This is the Year of the Tiger, so I made two sketches.
Profkssor T. D. Lee presenting bouquets to Fern Simes, Rae Greenberg, Doria Ruegerand Pam Esposito at the First Anniversary Celebration Symposium
Table 1
Proceedings of RIKEN BNL Research Center Workshops
Volume 1
X7 *volume 2
Volume 3
XT1
volume 4
Volume 5
Volume 6
Open Standards for Cascade Models for RHIC (BNL-64912)T&e 23-27.1997 — &&niy~r* ~iklnc ~T@~~~~. . . ..-. —-— . >— --- -.-. —.-,- . - . ---= - &
l?erturllative QCD as a Probe of Haclron Structure (BNMM723)July 14-25, 1997- Organizers: Robert Jaffe and George Sterman
Hadron St)in-Fli~ at RHIC Energies (BNL-64724)Tnlv 7.1-A/&~q ;~, ~~~~ — ~pani7;rc- Wlint T ~;d~r ~nd T T ~p~~~l~~.*.J -. ~-’””~-’ u. --.AAwb W“WU”* ~~1= * ● -e
Non-Equiliiiriurn Many Body Dynamics (BN-I-649 i 2)September 23-25, 1997— Organizers: M-ichaelCreutz and Miklos Gyulassy
Color Superconductivity, Instantons, and Parity (Non?)-Conservation at High-.~~ly~n &@/ @NL-~~ lo~)NTa.,a-La. 11 1007lXUVV1llUGL Ll, 1771 n-nn:qav. nfi;l.lfi. fzxrl.in..xr— Wl&lll14JGL . lV1.lNWb Uy UlClb3y
Quarks and Gluons in the Nucleon (BNL-65234)November 28-29, 1997 — Organizers: Toshiaki Shibata and Koichi Yazaki—
. , ,
Volume 13 Physics of the 1 Teraflop RIKEN-BNL-COLUMBIA QCD Project — FirstAnniversary Celebration (in preparation)October 16, 1998 — Organizer: Robert Mawhinney
Volume 14 Quantum Fields In & Out of Equilibrium (in preparation)October 26-30, 1998 — Organizers: Dan Boyanovsky, Hector De Vega andRob Pisarski
Volume 15 QCD Phase Transitions (in preparation)November 4-7, 1998 — Organizers: Thomas Schaeffer and Edward Shuryak
. .
RBRC/#
10. D. Kharzeev, “Production of Heavy Mesons,” in l%ceedings Quarksand Gluons in the Nucleon, Wake, Japan, eds. T. Shibata and K. Yazaki,RIKEN BNL Research Center, 1997, p. 67.
11. S. E. Vance, Y. Csorgo, and D. Kharzeev,” Observation of PartialUA(l) Restoration from Two-Pion Bose-Einstein Correlation,” [nucl-
th/9802074]; Phys. Rev. Lett. & 2205 (1998).
12. R. D. Mawhinney, “The QCD Hadron Spectrum and the Number ofm amical Quark Flavors,” NUCLPhysics &A-C (Proc. SuppZ),212 (1998).
13. R. D. Mawhinney and C. Jung, “An Investigation of Semiclassical andMonopole Confinement Mechanisms,” in preparation.
14. D. Rischke, “Instabilities and Inhomogeneities in the Early Stage ofUltrarelativistic Heavy-Ion Collisions, “ in Proceedings of Non-EquilibriumMany Body Dynamics, eds. M. Creutz and M. Gyulassy, RIKEN BNLResearch Center, p. 47,1997.
15. J. Kodaira, T. Nasuno, H. Tochimura, K. Tanaka, and Y. Yasui,“Renormalization of the Twist-3 Flavor Singlet Operators in a CovariantGauge,” to appear in the Proceedings of Deep Inelastic Scattering off PolarizedTargets: Theoy Meets Experimentr DESY, Gemmny, 1997.
16. J. Kodaira, T. Nasuno, H. Tochimura, K. Tanaka and Y. Yasui,“Renormalization of Gauge-Invariant Operators for the Structure Function ‘g& Q2),”Prog.~eor. Phys. W_ 315 (1998).
17. D. H. Rischke, “Forming Disoriented Chiral Condensates ThroughFluctuations,” [nucl-th/9806045]; I?hys. Rev. C= 2331 (1998).
18. A. Dumitru, D. H. Rischke, “Collective Dynamics in HighlyRelativistic Heavy-Ion Collisions,” [nucl-th/9806003]; Phys. Rev. C= 354-63 (1999).
19. D. H. Rischke, “Fluid Dynamics for Relativistic Nuclear Collisions,”[nucl-th/9809044], Proc. of the 11 Chris Enge2brecht Summer School inTheoretical Physics: Hadrons in Dense Matter and Hadrosynthesis, Cape Town,South Africa, Feb. 4-13,1998.
RBRC/#
29. C. Bernard, T. DeGrand, C. DeTar, Steven Gottlieb, Urs M. Heller,J.E. Hetrick, N. Ishizuka, C. McNeile, R. Sugar, D. Toussaint, andM. Wingate (MILC Collaboration), “Heavy-Light Decay Constants:Conclusions from the Wilson Action,” [hep-lat/9809109] to appear inProceedingsof Lattice ’98, Boulder, Colorado, 1998.
30. S. Kim and S. Ohta, “Chiral Limit of Light Hadron Mass in QuenchedStaggered QCD,” [hep-lat/9809184]; to appear in Proceedings Lattice ’98,Boulder, Colorado, 1998.
31. D. Kharzeev, R. D. Pisarski, and M. Tytgat, “Possibility ofSpontaneous Parity Violation in Hot QCD,” Phys. Rev. Lett. 81_512 (1998).
32. D. Kharzeev, R. D. Pisarski, and M. Tytgat, “Parity-odd Bubbles inHot QCD,” [hep-ph/9808366]; to appear in l+oc. Continuum Advances inQCD, Minneapolis, 1998.
33. D. Kharzeev, “Workshop on Quarkonium Production in RelativisticNuclear Collisions: Sum.mary,” [hep-ph/9812214]; to appear in l%wc.Quarkonium Production, Seattle, 1998.
34. T. D. Lee, “Generalization of Classical Yang-Mills Fields in Ultra-relativistic Nuclear Collisions,” to appear in Proc. 3rd WorMop on
Continuous Advances in QCD, Minneapolis, 1998.
‘ 35. S. Sasaki and O. Miyamura, “Lattice Study of U(l)A Anomaly: TheRole of QCD-Monopoles,” ~ep-lat/9810039]; Phys. Lett. B= 331-337(1998).
36. N. K. Glendenning and J. Schaffner-Bielich, “Kaon Condensation andDynamical Nucleons in Neutron Stars,” Phys. Rev. Leti. U 4564,1998.
37. A. J. Baltz, Alred Scharff Goldhaber, and Maurice Goldhaber, “TheSolar Neutrino Puzzle: An Oscillation Solution with Maximal NeutrinoMixing,” Phys. Rev. Lett. &l_5730 (1998).
38. A. J. Baltz and L. McLerran, “Two Center Light Cone Calculation ofPair Production Induced by Ultrarelativistic Heavy Ions,” Phys. Rev. Cm1679 (1998).
i\ II
,)
I
RBRC/#
48. J. T. Lenaghan and D. Rischke, The O(N) Model at FiniteTemperature: Renormalization of the Gap Equations in Hartree and Large-N Approximation,” [nucl-th/9901049].
49. S. Sasaki and O. Miyamura, “Topological Aspect of Abelian ProjectedSU(2) Lattice Gauge Theory,” [hep-lat/9811029], Phys. Rev. D.(submitted).
50. J. Schaffner-Bielich and J. Randrup, “DCC Dynamics with the SU(3)Linear Sigma Model,” [nucl-th/9812032].
51. K. Schertler, S. Leupold and J. Schaffner-Bielich, “Neutron Stars andQuark Phases in the NJL Model;’ [UGI-98-41, astro-ph/9901152], Jan. 1999.
52. T. Blum, A. Soni, and M. Wingate, “Calculation of the Strange QuarkMass Using Domain Wall Ferrnions,” [hep-lat/9902016], Phys. Rev. D(submitted).
53. T. Blu.m,“QCD with Domain Wall Quarks,” to appear in ProceedingsYKB ’98, Kyoto, Japan.
54. D. Boer, Investigating the Origins of Transverse Spin Asymmetries atRHIC,” [hep-ph/9902255], Feb. 1999.
—
c!!!!!). . RIKEN BNL Research CenterBldg. 510, Brookhaven National Laboratory UPTON, NY 11973, USA
Workshop on Physics of the 1 TeraflopRIKEN-BNL-Columbia QCD Project
October 16, 1998
. L ----- ---—-
..
J
—-..—— --—-LXJN’lEN’IS
Preface to theSeries......... . . . . .. ....*...... ........ ........ .. . .....*... ...... ....
IntroductionRobert Mawhinney .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..
Workshop Agenda .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The QCDSP ComputerRobert Ikfawhinney . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Grassmann IntegrationonQCDSPMichael Creutz . . . . ..*...... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Open Problems in WeakMatrixElementsAmqjit Soni . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Non-Perturbative Improvement with WiIson FermionsChristopher Dawson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
QCD on Anisotropic LatticesTimothy R. Klassen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..
Strange Quark Mass: First Results with Domain Wall FermionsMatthew Wingate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .*. .
Issues with the Aoki PhaseAkira Ukawa . . . . . . . ...*. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Lattice Study of Correlation Between Instantons and MonopolesShoichi Sasaki . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..
Fermion Zero-Modes in Quenched QCDNorman H. Christ . . . . . . . . . . . . . . . . . . . . . . . . . . . ... . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
UA(l) Effects with Domain Wall FerrnionsG. Siegert . . . . . . . . . . . . . . ● . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..
List of Workshop Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..*...... . . ...0....
Media Coverage/Press Clippings/Award . . . . . . . . . . . . . . . . . . . . . . . ...*. . . . . . . . . ..
Forthcoming RIKEN BNL Research Center Workshops . . . . . . . . . . . . . . . . . . .
i
1
3
5
11
19
25
33
41
49
57
65
73
81
83
107
Preface to the Series
The RIKEN BNL Research Center (RBRC) was established in April 1997 atBrookhaven National Laboratory. It is fimded by the “Rikagaku Kenkysho”(Institute of Physical and Chemical Research) of Japan. The Center is dedicated tothe study of strong interactions, includinghardQCD/spinphysics, lattice QCD andRHIC physics throughnurturingof a new generationof youngphysicists.
During the first year, the Center had only a Theory Group. At present, theTheory Group consists of nine Postdocs and Fellows and has an active VisitingScientist program. In addition,the Center organizesworkshopscentered on specificproblems in stronginteractions.
Now, at the Center’s fust anniversarycelebration,an Experimental Group hasbeen started and constructionof a 0.6 teraflopparallelprocessor which was begun atthe Center on February 19, 1998 was completedon August28, 1998. In addition, anew Tenure Track Strong InteractionTheoryRHIC Physics Fellow Program is underway. To this date eight institutionshave joined this initiative. The Center had hasheld twelve workshopsthusfar.
Each workshop speaker is encouragedto select a few of the most importanttransparencies from his or her presentation, accompaniedby a page of explanation.This material is collected at the end of the workshopby the organizer to form aproceedings, which can thereforebe availablewithina shorttime.
T. D. Lee
October 16, 1998
Introduction
A workshop was held at the R,IKEN-BNL Research Center on the af-ternoon of October 16, i 998, as part of the first anniversary ceremony forthe center. Titled “Workshop on Physics of the 1 Teraflop RIKEN-BNL-Columbia QCD Project”, this meeting brought together the physicists fromRTUF,N.RNT, RNT, .nrl P,nll,mhi. whn S.O ~~~ng ~~e Q~~~~ (~JaQ~Jm4--.- .--T -...4, -., - -.Au .J”. -... ”.- . . ..” w. “
Chromodynamics on Digital Signal Processors) computer at the RIKEN-BNLResearch Center for studies ot Q!LU. in addition, Akira Ukawa, a ieader of
,. .7-
the CP-PACS project at the University of Tsukuba in Japan, attended andgave a talk on the Aoki phase. There were also others in attendance who~xmrn in+cirnc+d in mnma mamarml rw-nnar+inc nf +ha ~CT)CD namn.,+n.,“b. b &LLVb&bo.bu . . . ,..”. b ~bALbL u. FL U~GA WALO WA UUL ~VtiUA VVHA~U ~GL .
The QCDSP computer and lattice QCD had been presented during the.-. —,-.
morning ceremony by Shigemi Ohta ot KhK and the RIKEN-BNL ResearchCenter. This was followed by a tour of the QCDSP machine room and a for-mal unveiling of the computer to the attendees of the anniversary ceremony.-A +h. . . . . Th. ..n:A .a-=In+:Am ,4--.+.,..+:- .r+hn nnncm .-,.. -.+..Ouu IJUG pl GDO. J. lIC 1 Gplu uulllplculuu U1 GULMJOL UG UIU1l vi IJJ.lc ~wlJuL Lulupul)cl
was made possible through many factors: 1) the existence of a complete de-sign and working hardware at Columbia when the RIKEN-BNL center wasbeing set up, 2) strong support for the project from RIKEN and the centerand 3) aggressive involvement of members of the Computing and Communi--n+:n=. n:.,:.:- -+ QRTT TXT;4V. tl..:n na..,nr$,.l ..-.., ..no,-.....fin +1.. n -n-l. n_e ,.CUCLULUUO LJLV LLILULl O,b ULIU. VY LLLL UU.J.= I#u WCLLUL LLCW LC=WULUC, bllC L1lG1ll UCLS U1
the RIKEN-BNL-Columbia, QCD project are looking forward to advances inour understanding of QCD.
Many of the talks in the workshop were devoted to domain wall fermions,a discretization of the continuum description of fermions which preserves the-1,.~.l --.——-~-:-- -r AL- ..--4 :-.. -— ___gluutll >y lllllleblltm U1 IJlle Culllmluulll,
_—_-: -—‘eVwl at finite Mike SpdLiIlg.
nv.:_J. lllb
formulation has been the subject of analytic investigation for some timeand has reached the stage where large-scale simulations in QCD seem verypromising. Pavlos Vranas (until recently at Columbia) had investigated the2-d Schwinger model and Tom Blum and Amarjit Soni had done a first——-.-.-—-—-——.-r Al-- --- 1--- r–– LL- n _L --- l?– r,- T–7 --- ? -z.. –
IIl~iiSUI”HIltXIL U1 Lilt! value lUI” Llle D~ pd,ramecer 101- n — n rmxmg. DOLI1—-—-—. n –.l-
calculations were successful and showed the formulation to be promising.With the computational power available from the QCDSP computers, we
2rp lnnkin u forward ~0 ~~ ~xc~~~~g ~~~.~ f~~ ~IJ~-~1--c~!S~rnAIU~~~~ODSOf Q~~.--- --------o --- -----
The expertise available from RIKEN-BNL, BNL and Columbia physicists
1
RIKEN BNL Research Center
Workshop on Physics of the 1RIKEN-BNL-Columbia QCD
c October 16, 1998
TerafiopProject
Physics Department - Room 2-160Organizer: Robert Mawhinney
AGENDA
2:00 pm
10 min Mawbinney
30 min Creutz:
30min soni:20 min C. Dawson:
20 min Tim Klasserx
20 min Wingate:
4:10-4:30
30 min Ukawa:
20min Sasaki:
30min Christ:
20 min Siegert:
7:00 pm
The QCDSP mmputer
Gmssmann integmtion on QCDSP
Open problems in weak matriz elements
Non-perturbative Renormalization
Anisotmpic lattices
Quark masses
Break
Aoki Phase
Lattice study of wrmlation between
instantons and monopoles
D WF zem modes in quenched theories
U(l)A e@cts with domain wail fermions
Dinner
3
The QCDSP Computer
Robert MawhinneyColumbia University
5
!11
I
I
QCDSP Operating System
Two major components: Q-SHELL and QCDSP ~
kernels.
1. The Q-SHELL runs on the host and is a
derivative of tcsh with built-in commands
to access QCDSP.
2. The QCDSP kernels are loaded when the
machine boots and provide 1/0 from host
to QCDSP as well as user support. Each
node runs its own kernel, so QCDSP is
fully MIMD.
During machine boot, the hardware is checked
as the kernels are loaded. The four-dimensional
configuration of the machine is determined as
well as the topology of the boot network.
7
QCDSP kernels
. No attempt to put UNIX on each node
since it takes too much memory and pro-
vides unnecessary features.
. QCDSP kernels support standard C library
calls. fopeno,
host file system
to host screen.
fcloseo, fprintfo access
from QCDSP. printfo prints
● Kernel calls handle 4d communications
● Kernels allow arbitrary mapping of physics
4d coordinate axes to machine 4d coordi-
nate axes.
. MPI not implemented, but could be.
9
Grassmann Integration on QCDSP
Michael CreutzBrookhaven National Laboratory
11
● Si is linear in az● ~Si = l+S.i
Z = (Ole3i (1 - ‘?2i)(1+ Si)lF)
1 – 7Zi is a projection
● (.22(1-722)=0
● fji(l-ni)=O
● Si(l–n..)=S(l-n~)o replace S’i with full action S
Z ‘= (O\es (1 – T2i)(1+ Si)l.l?)
Repeat for all sites
Z=(Ol~i(l–n~)(l+Si)lF)
Algorithm:
● initialize with full state
● loop over sites:8 apply 1 + Si0. apply 1 – ni● return coefficient of empty state
With a local interaction
● finished sites empty● new sites out of range filled
* nontrivial only in tra-nsverse volumee sparse state space ideal for hash storage
13
. repeat to stability
. works in parallel
On QCDSP
Based on my communication routines
●
s8●
●
●
●
#define DATASIZE element
#include communication.C
allocate(n) get space for n elements
store(i,x) store element x in location i
fetch(i) return element from location i
worksynco resolve all stores/fetches
http: //thy. phy.bnl.gov/~creutz/qcdsp
also included
● add(i,x) like store but adds to existing x● onnode-(i) true if element on current node● utilities (global sum, global and, . ..)
Enhance with new functions
●
●
●
●
●
●
#define hindex Is)
#define hvalue @#include hashcom.Chstore(]s),~)hadd(ls),~)
hfetch(ls))
Comments
On receiving a store or fetch message, processor checks his part of the table, forwards
unresolved requests to next processor.
A processor does/needs not know where an element will wind up.
A processor does/needs not know where an element came from.
Processor knows what he has locally, apply 1 + Si and send out for storage.
No wait before next state8 communication in parallel
● no non-local fetches
Test on 1+1 dimensional model
● Z= ~(d~d~”)~s,+sh+sf
I
I
I
I
15
Open Problems in Weak Matrix Elements
Amarjit SoniBrookhaven National Laboratory
19
Oi.
$Jlmww
21
. . .. .“,. ---... - ., .“. ..”. ,
[‘4*-—... . ... - J/-..+. @+~k - :,A...’ ...._ ,.. ,-..
j@’’\o\v;:kti!Dlva;‘-& ~p:iq’ = “Jw@%J.... .. ,’. ~~‘“”“.:*mf ‘ 1“,.- .,.-.-,. ----.. ~... ‘---~*+T?\,[b-L’[l$;’ii:, .,@wyyv ““”............ . . . .. ,--.
p%wlq; “ -~j’~l%!i!i:, ...—.~-” ,. ..... . -’ -.&’..- .. ..
23
Non-Perturbative Improvement with Wilson Fermions
Christopher DawsonBrookhaven National Laboratory
25
I
Non-Perturbative Improvement with Wilson Fermions
Christopher DawsonH.E. T.
Brookhaven National LaboratoryUpton, New York 11973-5000
The technique of non-perturbative renormalisation using momentum spacepropagators has been highly successful in calculating the renormalisation fac-tors for lattice QCD calculations. An attempt to extend this technique tothe non-perturbative improvement of operators in the on-shell improved the-ory is described. The vital role of B. R.S.T. symmetry in determining theoperators that must be included in the mixing is described, and the resultsof a preliminary study of the flavour non-singlet fermion bilinear operators ispresented. In particular the limitations of the technique, due to large of-shellO(a) terms, are pointed out.
27
.,\
4
la9
e “ Wu9Ym?&.
m &,-&,
b
—Sm -KQ
J’ IL.r .-—
A&d .
SO&$
— (kSQ3 .a. - if! CW!’4AS mad.— - ----- -.—‘- -d-
1– –w h [QJid. . .3
--
1
(M5
0.9
0.85
Z* 0.8
0.75
0.7
0.65
0.6
0.9
0.85
0.s
z~ 0.75
0.7
0.65
..
I 1 1 I I I
o 0.5 1 2.5 3 3.51“~aplM)2
figure 1: Z~ in the chiral limit
0.60 0.5 1 1.5” 2 2.5 3 3.5
I t 1 1 1 1
(ap~~)’ ““
Figure 2 Zv in the chid limit
QCD on Anisotropic Lattices
Timothy R KlassenColumbia University
33
QCD on Anisotropic Lattices
Timothy R. Klassen
Department of Physics, Columbia University, New York,NY 1UW’1. . . ..-
A -:.-4-A-;- T..++;”Ao . . ..+L . +,..%-C.7.-1 10++; (.0 .7-..3, -:-,7. .Wslla. *~Larl +hc, “-.+;-1 ,_.V.f.nlllauulupl~ lab bluco, w LULL a lJG1.lpwzaL aau UIG= .ycl?u+j DLLLUILGL IJLLG D~aU1O,l VAAG
(~ = a/aO > 1), are useful, often crucial, for the study of a variety of phenomenologicallyimportant questions in iattice QCiJ. First, there is the simpie practicai advantageof having (many) more time slices to identify plateaux before the signal gets lost
exponentially fast – in the noise, which is crucial for accurate glueball studies(Morningstar and Peardon), transport coefficients of the QGP, and, perhaps to a lesserextent, P-wave mesons or other particles with bad signal-to-noise properties.
More conceptual problems related to discretization errors arise for latticethermodynamics, and especially heavy quarks.We will concentrate on the latter, wherethe issue is simply that standard, relativistic actions break down when amq >1, and itis currently too costly to study a sequence of lattices with arn~ <<1 for charm or bottomquarks, even in the quenched approximation.
At present two approaches are widely used for quarks with arn~ >1: (i) NRQCDand (ii) the Fermilab or heavy relativistic approach. NRQCD seems to break down forcharmonium, due to large quantum and relativistic corrections to the action (Trottier).The approach we pursue is similar to the Fermilab one, in that we also use a relativisticSyman.zik improvement framework and give up the manifest O(4) symmetry of the actionby going to an anisotropic lattice, where the quark action has only an O(3) symmetry.The same is true for a heavy quark on an isotropic lattice; the Fermilab approach is thespecial case of an isotropic lattice in our approach.
The crucial question in a relativistic Symanzik approach to heavy/anisotropic quarksis the mass dependence of the coefficients. To eliminate all O(a) errors, so thatobservable really scale like U2towards the continuum, requires the coefficients in theaction to be not just coupling but also mass dependent There are three coefficientswhose coupling and mass dependence has to be tuned for a heavy /anisotropic Wilson-type quark action: a bare velocity of light v and two clover coefficients. The former can betuned by demanding that the dispersion relation of the pseudc-scalar, say, be relativistic
.1. - 1- L*. — .-–:.-— l_- o-l—-.:>:--—— r-. —-.:---l I-. -1. —---.-J 12-l J .— ALL--I
for srndi momenta; cne m LLer uwug me mxmumgtx I I.IW WIW uiwqyu UILU U13U Iuevnu u
and the chiral Ward identity (PCAC). On isotropic lattices the coefficients have a strongmass dependence, even classically, which is currentiy ignored in the Fermiia-b approach.
In our exploratory quenched anisotropic simulations of the charmonium spectrum(see below and [1,2]) we only tuned v exactly, and used a Landau link tadpole estimatefor the clover coefficients. We ignored the mass dependence of the clover coefficients:If it were linear this would be legitimate, since it would only introduce O(a2 ) errors.fll=caicallx, linoari+v nf tha Pnd%Pim-I+Q hnldc WOII nn anicntrnnie Iat.tirac-nn t.h~ nllant.llmWHMJ”. ”Q.. J , 1...”-. .“J v. .-” .-” ”-.”----- ..v --- . . “.. .,.. -... ”-.. .-~ .“ .- ..-””-, u.. . ..V q . ..- . -----
level this issue has to be further investigated [2].Note that in the quenched approximation the bare anisotropy in the gauge action
can be determined independently of the quark action, (O = fo({, /3). We did so in [3].
1. T.R. Klassen, hep-lat/9809174.-. ~.?w. Klaccrm ~p. nr~nara+inn7 . ..-”-.., y. “=— -------
3. T.R. Khssen, Nucl. Phys. B533 (1998) 557 [hep-lat/9803010].
●
●
●
●
●
!
Symanzik Improvement on (An) Isotropic Lattices
In the scaling region, the lattice theory can be described by an
effective continuum action of form:
Seff = S()+U~C~2S~a +U2~4C2~S22 + ...
% ‘i
Sni are all operators allowed by (lattice) symmetries, modulo
redundant operators that can be eliminated by field equations.
In isotropic light quark case: So = ~(p + ~)v. For suitablychosen Cli = Clz(g2 ) lattice artifacts of observable scale like a2.
For heavy /anisotropic quarks: So = ~(~0 + VP + m)+.
v is a “bare velocity of light”, to restore space-time exchange symm.
Due to reduced manifest symmetry, O(4) ~ O(3), more terms Sni.
Claim (Fermilab): For suitable (v, Cli) = (v, cli) (g2,son-z,~) errorsare
O(a2), no errors in mass, Clear if aom <1. ‘hue also for aom >>1 ?1
Tim Klassen
oI I I I I I I I I I I I I I I I
p$
------
m=
4tA, ,,11, l,, l,, ,11, ,l,l
Pm
G’
100
I I I I I I I
t\/ii
I
I
I
I
I
:g
II
%
I
II
I
II
iI
IIIEl
0.0
●
.,,;,:
o
0
x
❑
IDl
❑
Nature
Aniso SW,
Aniso SW,
Fermilab
NRQCD
NRQCD
v’,V6,
❑
oLLLLLl—
❑cIlpJ
I
up
u~
,11110,1
a2 (fm2
0.2
Figure 11: (Quenched) Charmonium HFS from vario Is formalisms.
Tim Klassen
Strange Quark Mass: First Results with Domain Wall Fermions
Matthew WingateBrookhaven National Laboratory
41 I
\ .
Strange Quark Mass: First Results with Domain WallFermions
Matthew Wingate, with Tom Blum and Amarjit Soni
In this talk I summarize our exploratory calculation [1, 2] of the strange quarkmass using domain wall fermions. We have computed the one-loop mass renormal-
ization and display the result in the first slide. In addition to the multiplicativerenormalization of the quark mass m ‘AT the domain wall mass M is shifted. In thesecond slide, the shift MC is estimated f~om the size of the tadpole graph, which issignificantly larger than the half-circle graph. However, the shift MC coincides withthe additive renormalization of ordinary 4-d Wilson fermions. The third slide showsthe pion mass squared for different values of M at @= 6.0 and N. = 14. The dottedlines are fits to the ansatz that the renormalization of M is purely additive. Of coursethis is not strictly correct, but for this range of M and at the present level of preci-sion, this assumption fits the data quite well. In slide 4 we plot the same data as afunction of m to demonstrate that the pion mass squared extrapolates linearly to zero
at m = O within the errors. Finally, we determine the strange quark mass by settingthe kaon to its physical value and apply the one-loop renormalization factor. Slide5 shows our resultsresult for this work
in comparison to those using otheris m. (2 GeV) = 82(15) MeV in the
fermion discretizations. Our~ regularization scheme.
References
[1]T.on
[2]T.
Blum, A. Soni, and M. Wingate, parallel talk at the “International SymposiumLattice Field Theory 1998”, Boulder, CO, USA, hep-lat/9809109.
Blum, A. Soni, and M. Wingate, manuscript in preparation.
43
. . .
45
I
I
I
i
I
coO*o
—
I
I I “’1’ ’”1-
CDO* C9
000
o o*\
G
1,, ,,1,,,,I I I I
02O*o
No0
0 0Q30240o“ o* o“
m0.0
mI I ‘1’’”1: ~
U34“
-
00
CD cooo*
oo“
+o0
020 N
060o*
o 0111111111111111’
000
g(m)
47
Issues with the Aoki Phase
Akira UkawaUniversity of Tsukuba
49
Riken-BNL98.10.16
Issues with the Aoki Phase
Akira Ukawa
Center for Computational Physics
University of Tsukuba
51. A brief reminder
52. Existence of the phase -
- De~endence on N -m t
S 3 Finite T phase transition
- r)qpndmre on .jvf -------- ---. .-”
S 4. Continuum limit
f~~vvithX! Gmss+Jeveu model =
S 5. Summary.
51
● explicit check of parity-flavor breakinginside the Aoki phase
symmetry breaking field: L.xt= ~~~Y5~ Y
expect:li%(~++” “rderparame’er
lim m2+= O NG boson of
h-A ‘- broken symmetry
2.5, 1“ I Jr
2.0 1
[i.5
I
1.0
I
0.51
I
....%..
❑ ✍✍✍✍H=O.1
I...~
A“’””” ““05H=O.02
1
1H=O.01
...0.0 t
3.4 3.6 3.8 4.~ 4.2 4.4 4.6 ~.~ 5.0 5.2
I-.......... {1.0
K==.27 . ... ““...
!
..--------....... ~..
-.. ........-----Q--”-“0.. ‘-%.
A!@@lK=O.21 ----.---.-—–’”<;~%%=-–-.--–------.------...---”.--;~;-””--”---””””
T
K=O.26 . ....%....O
.......“:’::;> 1°5.........
E+=:- 1K=O.24
0.0Q.@ 0:01 0.02 0.03 0.04 0.05 0.06
Hl/K
● puzzling result for 4-flavor finite-T transtion:
1st order transition turns into a crossover close to the critical line
0.30
0.25
K
0.20
0.15
p= 4f) ~---’+....2.5 - “m*A (a) mxz
p= 3.0A...&_”A..--A.....4
2.0- fs=2.5 ....V-...v....... .. A
:p=2.0 + ,,,.....7 ,,~...1.5- =-w..-
---U..= ?! .& ; ......’....Q- -“, \ ● ..:j~~.+:
1.0- y y ;,($.....-A.*:!:. : .....,/
0.5-.Q : f{#~ y+
@0.0 , 1 13.0 3.5 4.0 4.5 5.0 5.5 6.0
l/K
I n 1 l\ 1 1 i
- KJNf = 4)
,...
... .;1 1 1 I 10.0 1.0 2.0 3.0 4.0 5.0 6.0
P
55
Lattice Study of Correlation Between Instantons and Monopoles
Shoichi SasakiBrookhaven National Laboratory
57
Lattice Study of Correlationbetween Instantons and Monopoles*
Shoichi SASAKI
RIKEN BNL Research Center, Brookhaven National Laboratory,
Upton, NY 11973, U.S.A.
The several lattice QCD simulations have showed the existence of the non-trivial cor-
relation between instantons and QCD-monopoles in spite of the fact that these topological
objects originate from different homotopy groups. However, nobody knows whether it has
some physical si~ificance, or not, from the topological viewpoint [1].
In 1982, Ezawa and Iwaz~I advocated the hypothesis of abelian dominance, which is
essentially composed of two sentences [2]:
● Only the abelian component of gauge fields is relevant at a long-distance scale.
● The non-abelian effects are mostly inherited by magnetic monopoles.
Actually, lattice Monte Carlo simulation indicates that abelian dominance for some physical
quantities reveals in the maximally abelian (MA) gauge [1].
Here, an unavoidable question arises relating to the non-trivial correlation between in-
stantons and QCD-monopoles. In the abelian dominated system, is it possible for the non-
abelian topological nature to surwive? For such an essential question, Ezawa and IwazaM have
proposed a remarkable conjecture [2]: once abelian dominance is postulated, the topological
feature is preserved by the presence of monopoles.
An analytical study [3] is made to find the relationship between the topological charge
and QCD-monopoles in the lattice formulation base on the hypothesis of abelian dominance.
The topological charge is explicitly represented in terms of the monopole current and the
abelian component of gauge fields in the abelian dominated system [3]. This is consistent
with the previous study, which shows that QCD-monopoles strongly correlate with the
presence of fermionic zero-modes in the MA gauge [4].
[1] For a review article, see M.I. Polikarpov, Yucl. Phys. B (Proc. Suppl.) 53 (1997) 134.
[2]Z.F. Ezawa and A. Iwazaki, Phys. Rev. D25 (1982) 2681; ibid D26 (1982) 631.
[3] S. Sasaki and O. Miyamura, hep-lat/9811029.
[4] S. Sasaki and O. Miyamura, to be published in Phys. Lett. B; hep-lat/9810039.
*This talk is based on some recent works (Ref.2 and Ref.3) with O. Mlyamura.
59
:,\..
,’ :
‘ v-.
2()[1) Ad *O-W-.-=
4 ------ .— --.-,o
61
I—
@i,(s)=+4’’(s)
“** -he ellipsis sfwk
-..
+k &ssian -JIutuatioti .
=16XA;(SJ4M(S)+---”-”
for%e *ta/ divepyme,
%HORO(S)= - 167c4’w2$2(s)
7hiS im itx #a+$
by wzompk in +heabdtw ubmimtedsysiem “,
63
.
Fermion Zero-Modes in Quenched QCD
Norman H. ChristColumbia University
65
I
I
Fermion Zero-Modes in Quenched QCD 1NormanH. Christ
October 16, 1998
As we study QCD more deeply, the properties of the eigenvalue spectrum of theDirac operator become increasingly important. Viewing QCD from the perspective ofthe Feynman path integral, we have known for some time that the spectrum of small
eigenvalues of the Dirac operator in the typical gauge field background appearing inthe path integral has far-reaching physical consequences. The Banks-Cssher formula
demonstrates that the phenomena of chid symmetry breaking can be understood as
directly arising from a non-zero density of Dirac eigenvalues at the point A = O. In
addition, the physical consequences of the axial anomaly are closely related to the zero
modes of the Dirac operator in topologically non-trivial gauge field configurations.In this talk I would Iike to summarize recent work of the Columbia group using our
new QCDSP machines to study the zero-modes of the Dirac operator in quenched QCD.Here, without the usual Dirac determinant representing the effects of fermion loops, gaugeconfigurations with small eigenvalues will be enhanced relative to simulations including
dynamical quarks. Through such a study, we expect to learn something of the distributionof topologically non-trivial gauge configurations in pure QCD and their effect on thefermionic degrees of heedom.
We propose to study the zero modes of a particularly interesting lattice Dirac operator,that defined by the 5-dimensional domain wall fermion formalism. In the limit of large
extent in the ii.fth dimension, this formulation is capable of realizing exact zero modes incontrast with staggered fermions where a “zer~mode shift” smears potential zero modesaway from zero or Wilson fermions were there is no uniquely identified mass value atwhich physical “zero-modes” become zero eigenvectors of the Wilson Dirac operator.As an efficient tool to identify possible zero modes, we examine the quenched chidcondensate, (~~). Dirac zero- (or near-zero-) modes will show up in (~t$) as a dramaticl/m divergence as the fermion mass m approaches zero.
Let us discuss our results. ,First, in Figure 1, we show the behavior of (~~) in thebackground of a single instanton-like configuration with superimposed gauge noise. Thefigure demonstrates a clear l/m signal coming from the expected zero mode. Figures 2and 3 show a similar I/m behavior, now in a zero temperature simulation on 83 x 32 and163 x 32 configurations. It is important to notice that the coefficient of this l/m termdecreases horn 3.3(3) x 10-6 to 0.6(1) x 10-6 as the volume is increased by a factor of 8.This is the behavior that might be expected from zero modes arising born a dilute gasof instantons where the net topological charge per unit space-time volume decreases asI/@, for space-time volume C?<,
Finally in Figure 4, we show a similar comparison of (&b) computed on 163 x 4 and323 x 4 lattices at/3 = 5.71, just above the finite temperature, Nt = 4, phase transition.Here the l/m coefficient remains constant as the volume increases suggesting this l/mterm is a feature of the thermodynamic limit for the quenched transition. This is expectedon theoretical gounds as in argued in hep-lat/98070 18. A somewhat more sophisticatedargument is presented on the next page.
1This ia the joint work of Ping Chen, George Fleming, Adrian Kaehler, Catalin Mahlreanu, RobertMawhinne-y,Gabriclc Sicgwt, ChcxigZhongSui, Pa.vlos Vranas and Yuri Zhwtkc.w
67
A
I 1111111 I I I 111111 I I I [11111 I 1 111111 I I 111111 I
.t \
I
10-’7’
10-27
10-3~
IV=164, 1 <=0.1 ‘
=!l\ -1
i
\
111111 I I 111111 I I 111111 I I 111111 I I 111111 I
10-5 10-4 10-3 10-2 10-1
Figure 1
69
163X32, ~=5.85, quenched, 83 configs
fit to cYW = clmf + CO+ c-lm~-l
0“0” ~~
0.012
0.008
0.004
z
1 0 rnO=l.65, L~=32. — ~ /dof=2.66 . -
---- ~2/dof=3.48
t
10-2
I I I I I0.000 1.‘ 0.00 0.04 0.08 0.12 0.16 0.20 0.24
1’””’ I I f
0.0672(2)m~ + 0.89( I)xIO+ + 0.6(1)x10-m~-7---- 0.0670(2)m~ + 0.92(1 )xIO+
/
Figure 371
$.,
UA(l) Effects with Domain Wall Fermions
Gabriele SiegertColumbia University
73
UA(1) effects with domain wall fermions
G. Siegert12
Domain wall fermions (DWF) are very suited for a search for ~A (1) effectssince they have the full symmetry in the limit LS iniinite. In the other fermionformulations one must always disentangle the UA(1) effects from finite a effects.
The U.4(1) rotates the pseudoscalar pion n = 7V5#@ into the scalar delta6 = i@a~. If we had full U.4(1) symmetry, these two particles would be degen-erate.
For the susceptibility of the pion one finds
/d4zd4y($(z)75T’@ (z)@(v)75T*ti(y))= –(Tr(#T” #---7%~—LA)
=+=.(.+ _D:~+--)).With Du = Au and the mesm spectral density P(A) this is
/
co= –r5.~Nff2 dAp(A) 1~2+m2”
-Co
For the susceptibility of the delta one finds similarity
/&zd4y(zJ(z)T(1+(z)it(g)T~@(y)) = (T7-(T”—DL72T*5H
=c$abNfOJmdAp(A). 1
(A+ m)’
:= &bivf~ /“ 3’ 2m2d&5(A)(A2+ ~2 + (}2 + @)~ )-co
To their difference.
J
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
in the m -+ O limit only zeromodes contribute. Following you find our resultson the lattice.
lsupported by Max-Kedle-Foundation~done in ~o~aboration with p. Chen, N. Christ, G. Fleming, A. Kaehler, T- Kl=.en,
C. Malureanu, R. Mawhinn~ey, C. Sui, P. Vranas, L. Wu, Y. Zhestkov
75
I
I
E“
)
(),LJ
0.3
082
0,’1
(0
-0,1
163X413=5.45Mo= ,9 L$=l 6
mb- m~
.m6- m~Point source—— intercept = 2,771278e-02 +/- 1.000745e-02
■m&- m~Box source“-..’”-----intercept = 1.452831 e-02 +/- 2.435335e-02
_.i-, _-l--
0 0.05 0.1
‘1=mq0.15 0,2
Tue oct 1314:54:41519’98
, ..,,,.,., ...
_——.. —
2
1.5
?1?-!
0.5
0
163X4~=5.40MO=I ,9L~=l 6
/
I /
■ X.-X3— intercept = 3,255795e-01 +/- 4.754008e-02 1/
1 /
7’
I I I
o 0.05 0.1 0.15 0.2m, = m~
,3.*
Tue Ott 1314:14:321998
,,.,.,,..,,.,.,....
WORKSHOP ON PHYSICS OF 1 ~RAFLOP RIKEN-BNL-COLUMBL4QCDPROJECTOCTOBER 16,1998
ORGANIZERS: ROBERT MAWHINNEY
PARTICIPANTS
NAME miation andAddw E-u ADMESS
TomBlum RIKEN BNL Research Center tblum@bnLgovPhysics Department, Bldg. 51OABrookhaven National LaboratoryP.O. Box 5000Upton, NY 11973-5000 USA
Ping Chen Columbia University [email protected] Department538 West 120th StreetNew York, NY 10027 USA
Norman Christ Columbia University [email protected] Department538 West 120th StreetNew York, NY 10027 USA
Mike Cruetz Physics Department, Bldg. 51OA [email protected] National LaboratoryP.O. Box 5000Upton, NY 11973-5000 USA
Chris Dawson Physics Department Bldg. 5 IOA cdawson@bnLgovBrookhaven National LaboratoryP.O. Box 5000Upton, NY 11973-5000 USA
George Plerning Columbia University [email protected] Department538 West 120th StreetNew York, NY 10027 USA
Tlm Klassen Columbia University [email protected] Department538 West 120th StreetNew York, NY 10027 USA
Robert Mawhinney Columbia University [email protected] Department538 West 120th StreetNew York, NY 10027 USA
Shigemi Ohta RIKEN BNL Research Center and KEK [email protected] pPhysics Department Bldg. 510ABrookhaven National LaboratoryP.O. Box 5000Upton, NY 11973-5000 USA
81
Media Coverageof the Unveiling of the RIKEN-BNLSupercomputer
,- J-.,
COLUMBIA UNIVERSITYIN THE CIT1’ OF NEW YORK
PRESIDENT’S ROOM
October 21,1998
.
Dear T. D.:
Iam writingto extend my congratulations to you and your colleagues on the remarkablysuccessfid first year of operations of the RIKEN-Brookhaven Research Center. In my remarks att&e~n~qy~m_~~~ Of t~~ ~~~~~~ 2 vem aun 1 mrrm=scd confidence that the f%nter wi-mlrl mak=J ‘- ~-> - ---------- -- —.=----- —- u. v-w.-. . . ___ ____
valuable contributions to physics. But I had no idea that it would do so in so little time!
Three aspects of the Center deserve special’mention. FirsL it provides a dramaticdemonstration of the possibilities and the benefits of international collaboration in modernscience. Recent ftilures of international projects have received considerable attention and mi-@lead some to question the wisdom of embadcing on new ventures, but you and the leaders ofDTVETNT hmra .h,-i,.m +hm+ aafimax+:am n.- .-+-.... .-A +hm+ ..rha_ :+ A...-. -..—,--- hc._aC+. Q.,..*..S.SlUAUA 1 lILL v w ouu w u UKLL QUU~l CLUUU QCLL1 UUbUL C3UU LLKJL W llG1l 1 L UUG~ G V GL Y UllG UGllG1.llS. UGMJll~
the Center has focused on the nurture and encouragement of a new generation of youngphysicists; more such institutions are needed to assure that the flourishing of science willcontinue into the coming century. Third, the Center stands as a symbol that Brookhaven NationalLaboratory remains at the forefront of basic research, even as it lives through troubled times. Iknow that the new management of the Laboratory, and certainly Columbia’s part of thatmanagement, is committed to preserving and enhancing Brool&aven’s role as one of the world’spreeminent scientific laboratories.
Columbia is proud of the role it has played in the early life of the Center. Yourparticipation, together with that of others in the Columbia physics dep@rnenL has been centito the rapid and productive start of the research program. The collaboration between the Centerand Columbia on the construction of the supercomputer has been accomplished with speed andeffectiveness that was hoped for but could not have been predicted.
The accomplishments of the Center have forced me to revise my already highexpectations even higher still. I wish you and the Center good fortune as it moves into the fhture.And again, congratulations on an impressively successfid first year!
hcereiy.-.
/4lrGeorge upp
Professor T.D. Lee----WY FupinMC 5208
Low Library, Rmm 20Z Mail Code 4309 535 West 116th SWecc New York NY 10027 212-854-2825 Fax 212-8546466
PAGE 2 — BNL UNVEILS ULTIU-FAST SUPERCOMPUTER
The machine is a finalist for the Gordon Bell prize for price-performance at the upcoming SC98High Performance Networking and Computing conference in November.
“All of us at the RIKEN-BNL Research Center are eager to put this supercomputer to the test in ourattempts to solve some of the most pressing questions of modern physics,” said the center’s director,Nobel Prize winner T.D. Lee.
A hique Machine for Physics
The supercomputer stands almost nine feet high and is mounted in six large racks that are water-cooled to keep the machine horn overheating.
There are a total of 12,288 nodes, or processors, in the computer, providing the calculational powerneeded to handle the demands of tracking the movement of millions of virtual subatomic particles.
A specially designed custom computer chip called a node gate array, or NGA, handlescommunications between the nodes and is at the heart of the supercomputer’s design. Each NGA is pairedwith a Texas Instruments 50-megahertz processor and two megabytes of DRAM to form a singleprocessing unit or “node” of the machine. Each node is constructed on a small printed circuit board calleda “daughterboard.” Sixty-four nodes are combined and attached to a larger structure called a“motherboard.” There are 192 motherboards in all.
“Essentially, this computer turns space and time into a four-dimensional lattice, which can be thoughtof as a three-dimensional grid at any moment of time,” said Robert Mawhinney, one of the Columbiaphysicists who led the design team for the FUKEN-BNL machine and its 0.4-teraflop sister machine atColumbia’s physics department. “The computer can be used for many grid-oriented problems and in ourproblem, the grid gives reference points for calculating where particles are at any given moment.”
‘The smaller the boxes in the grid or the lattice,” Mawhinney continued, “the more precise we can bein our calculations. Of course, the smaller and more numerous the boxes, the more computing power isrequired. But with this machine, the calculations will be more precise than ever before.”
Shopping Locally
The supercomputer maybe of world-class stature, but it has local roots. Nearly one-third of thecomponents used to build the machine were purchased from Long IsIand f- in a competitive biddingprocess.
The two largest contracts were awarded to electronic components distributors Marshall Industries ofHauppauge, for $540,000, and Nu Horizons Electronics Corp. of Melville, for $236,000. In a criticalstep, the assembly of all the daughterboards and motherboards was performed by AJC Electronics ofSyosset under an $89,000 contract. The final process of installing and debugging the complete systemwas handled largely by BNL’s Computing & Communications staff.
Other Long Island companies that supplied components include: Hadco Corp. of Uniondale, formore than $115,000 in printed circuit boards; Bell Microproducts of Smithtown, $7,800; AnthemElectronics of Cornmack, $6,000; and Dove Electronics of East Setauket, $500.
“We were glad that much of what we needed was available through local suppliersfl said Lee. “Tobuild such a large computer from scratch is a massive task, but the job was made easier by the excellentservice we received from these companies.”
The RIKEN laborato~, whose name is short for “Rikagaku Kerdqwsho, ” meaning Institute ofPhysical and Chemical Research, is supported by the Japanese Science and Technology Agency. It islocated near To~o.
Brookhaven National Laboratory earn.es out basic and applied research in the physical biomedicaland environmental sciences and in selected energy technologies. Brookhaven is operated by BrookhavenScience Associates, a nonprofit research rnunagement organizatiofi under contract with the U.S.Depa*nt of Energy.
More information on the supercomputer is available on the Web athttp: //www.ccd.bn1.gov/riken_bnl/qcd_project/qcdp.html
87
Brookhaven Review - October 22, 1998
BNL Gets A Supercomputerby Joel ReitmanThe fastest multipurpose non-
commercial supercomputer in theworld was unveiled at BrookhavenNational Laboratory (BNL). Theco-rnputer was desi~ed and builtby scientists and computer spe-cialists from Columbia Universityand BNL.
Called the RIKEN-BNL QCDsupercomputer, the machine is op-timized for advanced research intoa model of matter based on quarksand gluons, the particles that makeup the center of every atom in the
universe.The computer’s speed will allo’w
scientists to predict and analyzethe behavior of subatomic particlesthat will be produced at BNL’snewest atom smasher, the Rela-tivistic Heavy Ion Collider (RHIC),now under construction.
Scientists will use the machineto carry out research in the fore-front of physics, so complicatedthat it can only be done using theworld’s fastest computers.
“This generation of machine andlarger ones to follow will be the
new tools of discovery and innova-tion to solve complex problems inglobal climate, energy, technolo-gies, and basic research,” said Sec-retary of Energy Bill Richardson.
Over $1,000,000 in componentsfor the supercomputer were pur-chased from Long Island firms.
This is what BNL is all about,”said BNL Director JohnMarburger. According toMarburger, 13NL is strengtheningLong Island’s economy throughdirect investment.
The supercomputer maybe ofworld-
class stature, but it has local roots.Nearly one-third of the componentswere purchaseclllvm Long Island firms.l!hetwolargestcontracta were awardedto Marshall Industries of Hauppaugefor $540,000 and Nu Horizons Elec-tronics Corporation of Melville for$236,000.
Other Long Island companies thatsupplied components include AJC Elec-tronics of Syosset, Hadco Corporationof Uniondale, Bell Microproducts ofSmithtown, Anthem ElectronicsofCommack, and Dove Electronicsof East Setaulcet.
I
World’s Fastest Computer Debuts at Brookhaven LabThe fmtest multi-purpose non-commercial
% supercomputer in the world was unveiled
yesterday at Brookhaven National Laboratoryin North Shirley/Upton.
With a top operating speed of600 billirm
calculations per second, or 0.6 teraflops, the
supercomputer is the 12! fastest overall in the
world. At a cost of ‘$1.8 million, it is also one
of the least expensive, according to lab offi-
cials.
Scientists will use the machine to carry outforefront physics research, some so compli-
cated it can only be done using the such an
ultra-fast computer.The computer was designed and built by
scientists and computer specialists from
“Columbia University and BNL, and fundedby the Japanese RIKEN laboratory as part of
its support of the RIKEN-BNL ResearchCenter established in 1997 at BNL.
‘This computer is a tribute to the creativity
and resourcefirlness of the Brookhaven Na-
tional Laboratory, Columbia University and
RIKEN Laboratory scientists who created it,”
said Secretary of Energy Bill Richardson.“This generation of machines and much larger
ones soon to follow will be the new tools ofdiscovery and innovation for science and
society to solve complex problems in global
climate, energy, technologies and basicresearch. ”
Called the RIKEN-BNL QCD supercom-
puter, the machine is optimized for advancedresearch into quantum chromodynarnics, or
QCD, the model of matter basedon the “strong
force” that binds quarks and gluons in the
particles that make up the center of every atom
in the universe.
Among’ other projects, the computer’s
speed will allow scientists to predict and
analyze the behavior of subatomic particles
and phenomena that will be produced at BNL’snewest “atom smasher,” the Relativistic HeavyIon Collider (RHIC), now under construction.
“To have such a strong synergy betweentwo facilities at Brookhaven, RHIC and the
RI KEN BNL center, while at (hc same lime
supporting our local tiigh-technology sec!or,
is truly satisfy ing,” said BNL Director Johi]
Marburger. Over $1,000,000 in componentsfor the supercomputer were purchased from
Long Island firms.”
South Silorc Press - October 2u, 1998
&!!Yf&+EiLLE”S
TVCLIPS
DATETIME
STATIONLOCATION
PROGRAM
.
75 EAST ~R7WFlELD ROAD / LIVINGSTON I NEW JERSEY 07039(973) 992-66m/@oo)S31-11s0
.
October 16, 19985:00-6:00 PMRainbow News 12Long Island, N.Y. IINews 12 Evening Edition
ACCOUNT NUMBER 67/6508NIELSEN AUDIENCE N/A
Scott Feldman, co-anchor:
One of the world’s fastest computers is now on the Island. Engineers atBrookhaven National Lab and Columbia University built this supercomputer thatcan make six hundred billion calculations a second. The 1.8 million dollarcomputer was unveiled today.
lr:~... . . . . . ..-. . . ..AAA r,.. +.s,. ,.,CA.C f~~~ ~M~ .ftili.ta. vbfc /917)7%.701 n
w>LIPS..
~.A.~~
TIMEQTATT~Ne ----- ----
LOCATIONDU~Cl? AM. . . . . . . . .
75 EAST hOR~FIELD ROAD / LIVINGSTON I NEW JERSEY 07039I
(973) 99246cM/(8w) 631-1160
.
fktnher 16 1 CX3R------- .-, . ..” A rrnrmvr NTTMRHI-------- . . . w. ----- ~7!~5~g
6:00-6:30 PM NIELSEN AUDIENCE 721,000WNBC.TV ~hannel l%tjr-..-...”. . “-.
New YorkN(=wc rhanntd A. .- . . - -........”. .
Tan- Uancnn en.nnchnr.---- . . . ...””... ““ -,”..”. .
FI~pA!!v WP all Imnw it’ch~tt~r W~~R ~~’Q hnmemach= hilt take a Innlc nt th~J 9 .. - ‘-- ‘-.” ‘v . . “ “-.-”. “ . . . . . . . . . . ..-. --- . . . ..- . .“”.. -. u.”
world’s fastest homemade computer. This supercomputer was designed and builthv the scientists Of ~~Q&-~~vc~ N~~~Qn~! ~~~ ~ud ~Q]Urn.~~~ lJl~jv~rs@ (v~~~~-, —.- -----------
of the supercomputer). This baby features twelve thousand two hundred andei!zhtv-ei~ht chins ~~’s cq~~!~ of SLX hundred hi 11ion Calcu] ar~~n~ p~~––w––-, ..=. -. ----r_ . -_—_—_. — --------
second, took six months to build at a cost of 1.8 million dollars with localparts. [t wi!! he used to test Lhetheories of quantum physics, such things_————-——
as the Big Bang theory.
Nielsen Audience Source: Nielsen Media Research, July 1998, Persons 2+, NSI 4-week Averages.—
&?7@5LLPS——?SEAST M2RlHflELD ROAD / LIVINGSTON I NEW JERSEY 07029
w
(973) 99246tW/(800) 631-11s0
CLIPS
DATE October 19, 1998TIME 11:00-11 :30PM
STATION WLNY-TV (Ind.) Channel 55LOCATION Riverhead, N.Y. I
PROGRAM News 55 Live at Eleven
ACCOUNT NUMBER 67/6508NIELSEN AUDIENCE 4,000
Jamie Colby, anchor:
A major breakthrough in the world of science. Brookhaven National Laboratoryis now home to a machine that may be able to solve questions that haveperplexed physicists for centuries.
Shari Bregman reporting:
You are looking at the fastest non-commercial supercomputer in the world. Thenew computer, which was unveiled here Friday, will allow scientists at theBrookhaven National Laboratory to conduct physics research faster than everbefore.
Edward McFadden (Advanced Computer Analyst): Nothing that we’ve ever had evenis in the same class as what this computer is capable of providing in terms ofcalculations per second.
Bregman: Six hundred billion calculations per second to be exact. It tookscientists and computer specialists a year and a half to put together theadvanced system, and physicists are looking forward to doing anaIysis thatwasn’t possible with slower computers.
Shigerni Ohta (Physicist): We will investigate how the universe started andthe big bang or how the protons spin, why the protons spin so rapidly.
Bregman: The supercomputer is worth between ten and fifteen million dollars,but it only cost about 1.8 million to build, the reason It’s homemade. Overone-half of the components were bought from local Long Island companies, whichkept the costs way down.
In Upton, Shari Bregman, News 55.
Colby: Construction of the supercomputer was funded by the Japanese RIKENLaboratories as part of its support of the Brookhaven National Lab ResearchCenter.
24 Clips
95
Nielsen Audience Source: Nielsen Media Research, July 1998, Persona 2+, NSI *week AveragesVideo cassettes are available for four weeks from our affiliate: VMS (212)736-2010
Bz?q!&iuE?s—-——~FADIOCLIPS
DATETIME
STATIONLOCATION
PROGRAM
7513STNORlHFlEL0 ROAD I LIVINGSTON I NEW JERSEY 07039
(973] 992- I (800)631-1 160
October 16, 1998 ACCOUNT NUMBER 67/65085:30-6:00 PMWCBS 880AMNew YorkNews
!,
Deborah Rodriguez, ancho~
WCBS News time 5:52. It’s nine feet tall, needs lots of water to keep cooland can answer six hundred billion questions in one second. No it’s notMichael Jordan at a press conference. Sam Liskinzer-Lis-Lissinger says it’sa computer that does things that you can hardly imagine, and it’s just madeits debut on Long Island.
Sam Lissinger reporting:
It’s billed as the fastest non-commercial supercomputer on the planet. It’sname The RIKEN-BNL QCD, but maybe we can just call it Ricky. Anyway, it’snine feet tall, needs water to keep it from overheating,and it can answersixhundredbillion questions in one second. Ricky’s just gone to work at theBrookhaven National Laboratory on Long Island and will be used for physicsresearch among other things, allowing scientists to predict and analyze thebehavior of subatomic particles. Humans can’t even begin to process suchcomplicated calculations. Ricky handles them in the blink of an electroniceye.
Sam Lissinger, CBS News.
RADIOCLIPS
DATE October 16, 1998TIME 2:00-2:19 PM
STATION WINS 1010 AMLOCATION New York City
PROGRAM News
Ralph Howardreporting:
ACCOUNT NUMBER 67/6508
On Long Island, Brookhaven National Lab unveiled it’s new super computer. Thelab calls it the twelfth-fastest computer in the world and the fastest innon-commercial use. We’re told the computer can answer six hundred billionquestions in one second. The outfitters, the energy department, said thecomputer cost just 1.8 million dollars.
97
NEWS MEDIA CONTACT: FOR IMMEDIATE RELEASEJeff Sherwood, 202/586-5806 November 13, 1998
Energy Department Lab Scientists SweepSupercomputhg Awards
Orlando, FL. -- Researchers from four U.S. Department of Energy (DOE) laboratories this weekwere honored by the h.igh-performance scientific computing community as recipients of topawmrdspresented at SC98, the annualhigh-pcx%ormancenetworkingand computing confmence.
An international team of scientists including the department’s Oak Ridge and Lawrence BerkeleyNational Laboratories won the 1998 Gordon Bell prize for best perfommnce of a supercomputingapplication. The team won the award for their modeling of 1,024 atoms of a metallic magnet.
Although the team won for its 657Gigaflops performance level (657 billion calculations persecond), it subsequentlywas able to have the applicationrun at more than one trillioncalculations per second.
Another 1998 Gordon Bell prize recognizes scientists who achieve the best price/perfotmmcelevel on a computer system. The winning team in this category is a collaboration of universitiesand DOE national laboratories led by Columbia University and involving the department’sBrookhaven National Laboratory and Fermi National Accelerator Laboratory. The winningmachine, costing only S1.8 million, is a multi-purpose,non-commercial supercomputer with atop operating speed of 600 bilIion calculations per second. It will carry out forefront physics
research. The machine was built at Brookhaven and fundedby the Japanese RKEN laboratoryas part of its research center at 13rookhaven supporting the Relativistic Heavy Ion Collider.
The two other finalist entries for the Gordon Bell prizes also includedDOE laboratories. Thedepartment’s Sandia National Laboratories, the University of California at Berkeley and IntelCorp. used DOE’s ASCI Red, a supercomputer with more than 4600 dual-processor Pentium Pronodes at Sandi~ to calculate electronic structures. The computer sustained a pefionnance of 605Gigaflops. The finalist for best price/performance was the department’s Los Akunos NationalLaboratory’s Avalon computer costing $150,000 and performing 20 billion operations persecond.
Gordon Bell, who has both designed high-performance computers and administered nationalresearch programs, sponsors the annual prize.
101R-98-172
@PmMMtscyhk OmmCjdMPWar
■ us. Depunlna’ntof Energy ● O#ice uj Public .4ff& ● [email protected] ●
BNL SupercomputerI
wins prizeItmaybe aaignofthe timeathatwe
award prises to computers iustead ofpeople, but recently the supermachine built by Edward McFaddenof Bayport and scientists fromBrookhaven Lab took top honors at acomputing conference heid at Orlsn-do, Florit@ early in November.
Just weeks after unveiling theworld’s fastest multipurpose non-commercial eupercomputsr, the teamthat built the 0.6-teraflop physicsmachine showcased its unique archi-tecture and capabilities at the SC98High Performance Networking and
Da winnah! Supercomputer builtby BNL and Columbia raata with ita
The Long Island Advance - December 3,1998
Computing conference.The team from Brookhaven National
Laboratory and Columbia Universityshowed off a sample crate of mother-boards from the massively parallelmachine at its booth in the researchexhibits hall. They also pointed outthe inexpensive do-it-yourself con-struction that kept toss of the super-machine to around $1.8 million for theentire project. ‘l’hecomputer has a topoperating speed of 600 billion calcula-tions per second, and is designed tocarry out cutting-edge physicsresearch. The machine was built atBrookhaven and funded by the Japan-ese RIREN laboratory m part of itsresearch center at Brookhaven sup-~d~ *e Reiativiatic HeavyIon COL
McFadden, the project manager forthe construction and installation ofthe supercomputer, said, “l’his week Ihave completed my thirfy-6fth year ofservice at BNLj and this has been mymost exciting project and year. It’snot every day you get to build theworld’s fastest non-commercial super-computer for a Nobel Prize winner(Professor T. D. Lee). This projectalso required .CIOSScollaboration withthe chairman of the physics depart-ment at Columbia University and sci-entists &om around the world”
McFadden added that Paul Poles@also of E@P@ was one of the hard-ware technicians who built themachine. Poleski and four other tech-nicians have maintained the otheraupercomputers used at Brookluwenfor the ps9t 20 years.
For their efforts, the scientistsreceived the prestigious Gordon Bellprize for achieving the beatprice@er-formance level on a computer system.Gordon Bell, who has both designedhigh-performance computers andadministered nation research pro-m, sponsors the annual prize.
According to the Department ofEnergy, supercomputers wiU play soincressin gly central role in maintain-ing the safety and reliability of thenation’s nuclear deterrent improvingthe efficiency of combustion systems,improving our understanding of theatmosphere and the oceans, develop-ing advanced materirds, and advsnc-ing many other scientific and engi-neering areas central to the depsrt-ment’s missions.
-Chuck Andsrson
105
