Elsevier Science Home
Computer Physics Communications Program Library
Full text online from Science Direct
Programs in Physics & Physical Chemistry
CPC Home

[Licence| Download | New Version Template] aabt_v1_0.gz(19 Kbytes)
Manuscript Title: Efficient multitasking of the SU(3) lattice gauge theory algorithm on the CRAY X-MP.
Authors: D.W. Kuba, K.J.M. Moriarty
Program title: SU3
Catalogue identifier: AABT_v1_0
Distribution format: gz
Journal reference: Comput. Phys. Commun. 36(1985)351
Programming language: Fortran.
Computer: CRAY X-MP-4 (MODEL 48).
Operating system: COS (VERSION X.15).
RAM: 2700K words
Word size: 64
Keywords: Particle physics, Elementary, Quantum chromodynamics, Lattice gauge theory, Su(3) gauge theory, Monte carlo methods, Vector processors, Multitasking.
Classification: 11.5.

Nature of problem:
The Monte Carlo lattice gauge theory algorithm with the metropolis et al. Updating procedure is vectorized and multitasked on the four processor CRAY X-MP and results in a code with a link-update-time, in 64-bit arithmetic and 10 hits-per-link, of 11.0 musec on a 16**4 lattice, the fastest link-update-time so far achieved. The program calculates the Wilson loops of size up to L/2 x L/2 for an L**4 lattice for SU(3) gauge theory. The formulation of a gauge theory on a lattice was proposed by Wilson to regulate the divergences of quantum field theory.

Solution method:
SU(3) gauage theory is simulated on a hypercubical lattice by Monte Carlo methods on a multitasking vector processor, the CRAY X-MP-4.

For a lattice of L sites in each of the space-time directions we need storage for 4L**4 SU(3) matrices, or 4L**4 *18 real variables. This is the only restriction on the program.

Running time:
The test run on a 16**4 lattice for SU(3) with 40 Monte Carlo iterations calculating all Wilson loops up to size 8*8 took 156.9 s for execution on the CRAY X-MP-4.