Numerical Simulation of Quark-Gluon Systems with Massvely Parallel Computer QCDPAX

点击翻译按钮获取中文摘要

In order to perform a large scale computation of lattice QCD with the massively parallel computer QCDPAX, which is constructed through the Grant-in-Aid for Specially Promoted Research in 1989, the environment for the development of QCDPAX-programs is further equipped by purchasing a workstation. On this environment, a high states tics Monte Carlo simulation of the gluon system at finite temperature is performed. First, fixing the lattice size in the temporal direction to be 4, two lattices with the spatial size of 12X12X24 and 24X24X36 are simulated. With the gauge coupling constant approximately at the phase transition point, 910, 000 and 710, 000 iterations are performed respectively. Then, increasing the temporal lattice size up to 6, lattices with the spatial size of 24X24X24 and 36X36X48 are simulated with 500, 000 and 1, 200, 000 interactions respectively. The latter simulation on the 36X36X48X6 lattice improves the previous computations to a large extent both in the size of the lattice and in the number of iterations. With this simulation, we can, for example, conclude that the value of the latent heat associated with the first order deconfining phase transition is about one half of a previous estimation on a small lattice for the case of the temporal lattice size 6. Secondly, we calculated the mass of hadrons on a 24X24X24X54 lattice. With a computation with higher precision than previous simulations, we find that the phenomenological mass formula works well for the quark mass down to about 1/3 of the strange quark mass. Thidly, we studied the confinement of quarks with varying the number of flavors and find that the confinement is lost for the flavor number larger than 7 even with the infinitely large gauge coupling constant.