NUMERICAL STUDY OF LATTICE QCD WITH THREE DYNAMICAL QUARKS

点击翻译按钮获取中文摘要

This work aimed to develop methods for lattice QCD simulations including dynamical strange quark in addition to dynamical up and down quarks. The main results we have obtained are as follows.1. Algorithm developmentTreating odd number of dynamical quarks requires an algorithmic development. We have shown that the polynomial HMC, in which the inverse of a single Dirac operator needed for the strange quark is approximated by a polynomial, provides a viable method for this purpose. We have also extended the algorithm to the Kogut-Susskind action, and has formulated an exact PHMC algorithm for it; such an exact algorithm had not been known before.2. phase structure of three-flavor lattice QCDWe have applied the algorithm to study the phase structure for three-flavor lattice QCD at finite lattice spacings. It was found that the combination of the plaquette and clover quark action develops a line of first-order phase transition which is a lattice artifact. While the transition terminates bef … More ore reaching the continuum limit, the lattice spacing at the termination point turned out to be about 4GeV. Since we need to start simulations closer to the continuum limit to avoid unphysical effects of such phase transitions, the plaquette-clover combination appears impractical for carrying out actual simulations. Switching the gauge action to an RG-improved action, it was found that the phase transition disappears, presumably because gauge field fluctuation becomes more continuum-like for improved actions.3. Determination of the clover coefficient for RG-improved gauge actionTo carry out a continuum extrapolation using lattice spacing 1/a=1-2GeV, an O (a) improved action is needed. Since the clover coefficient for the RG-improved gauge action is not known, we have decided to apply the Schrodinger functional method for this purpose. It turned out that O(a/L) errors need to be controlled precisely for this purpose, since bare gauge coupling is large. We treated this problem through simulations with a fixed L(in physical units), and determined the clover coefficient as a function of the bare gauge coupling.4. Physics runsAs a first step toward realistic three flavor QCD simulations, we have started runs at l/a=2GeV using 16^3x32 lattices. We plan to carry out the light hadron spectrum run, followed by determinations of light quark masses and the QCD coupling constant. Further work on heavy quark physics shall utilize gauge configurations generated in those studies. Less