PROCEEDINGS OF THE RIKEN BNL RESEARCH CENTER WORKSHOP ON LARGE SCALE COMPUTATIONS IN NUCLEAR PHYSICS USING THE QCDOC, SEPTEMBER 26 - 28, 2002.

Abstract

The massively parallel computer QCDOC (QCD On a Chip) of the RIKEN BNL Research Center (RI3RC) will provide ten-teraflop peak performance for lattice gauge calculations. Lattice groups from both Columbia University and RBRC, along with assistance from IBM, jointly handled the design of the QCDOC. RIKEN has provided $5 million in funding to complete the machine in 2003. Some fraction of this computer (perhaps as much as 10%) might be made available for large-scale computations in areas of theoretical nuclear physics other than lattice gauge theory. The purpose of this workshop was to investigate the feasibility and possibility of using a supercomputer such as the QCDOC for lattice, general nuclear theory, and other calculations. The lattice applications to nuclear physics that can be investigated with the QCDOC are varied: for example, the light hadron spectrum, finite temperature QCD, and kaon ({Delta}I = 1/2 and CP violation), and nucleon (the structure of the proton) matrix elements, to name a few. There are also other topics in theoretical nuclear physics that are currently limited by computer resources. Among these are ab initio calculations of nuclear structure for light nuclei (e.g. up to {approx}A = 8 nuclei), nuclear shell model calculations, nuclear hydrodynamics, heavy ion cascade and other transport calculations for RHIC, and nuclear astrophysics topics such as exploding supernovae. The physics topics were quite varied, ranging from simulations of stellar collapse by Douglas Swesty to detailed shell model calculations by David Dean, Takaharu Otsuka, and Noritaka Shimizu. Going outside traditional nuclear physics, James Davenport discussed molecular dynamics simulations and Shailesh Chandrasekharan presented a class of algorithms for simulating a wide variety of femionic problems. Four speakers addressed various aspects of theory and computational modeling for relativistic heavy ion reactions at RHIC. Scott Pratt and Steffen Bass gave general overviews of how qualitatively different types of physical processes evolve temporally in heavy ion reactions. Denes Molnar concentrated on the application of hydrodynamics, and Alex Krasnitz on a classical Yang-Mills field theory for the initial phase. We were pleasantly surprised by the excellence of the talks and the substantial interest from all parties. The diversity of the audience forced the speakers to give their talks at an understandable level, which was highly appreciated. One particular bonus of the discussions could be the application of highly developed three-dimensional astrophysics hydrodynamics codes to heavy ion reactions.