A status report of the QCDSF N f =2+1 Project Yoshifumi Nakamura (NIC/DESY) for the QCDSF collaboration Lattice Regensburg Aug. 3, 2007
Introduction QCDSF Motivation: QCDSF Collaboration has performed simulations of lattice QCD with 2 flavors of dynamical quarks Real world: 6 flavors (3 light, 3 heavy) We want to deal with u, d and s quark dynamically Expectation: Improve: the hadron and quark masses weak matrix element Moments of generalized distributions ・ ・ ・ ・ ・ lattice 2007
Action QCDSF lattice 2007 Gauge: Improved gauge action Tadpole improved Symanzik glue: decrease O(a 2 ) lattice discretization error (N f =3) HPQCD
Action QCDSF Fermion: Clover fermion action with stout links(FLRC) better chiral properties smaller cut-off effects Mild form of UV filtering: lattice 2007 Morningstar & Peardon
Algorithm QCDSF Partition function Even odd preconditioning lattice 2007
Algorithm QCDSF Multi mass preconditioning We modify action lattice 2007 Hasenbusch Horváth, Kennedy & Sint
Algorithm QCDSF Multiple time integration scheme: lattice time scales Sexton & Weingarten Chronological inverter by Minimal Residual Extrapolation: Brower et al., NPB484
Algorithm test QCDSF lattice 2007
Test QCDSF lattice 2007
Partially quenched QCDSF lattice 2007
Partially quenched QCDSF lattice 2007
Vector meson QCDSF lattice 2007
Nucleon QCDSF lattice 2007
QCDSF lattice 2007 Our 2+1 flavors action is more continuum like than our 2 flavors
Conclusion QCDSF QCDSF 2+1 project started We did good choice for action lattice 2007 Plan: TODO: More elaborate tests Implementing new idea Performance: ~20% of 20-25% of
RHMC QCDSF optimal coefficients are found by the Remez algorithm for a given range MC step MD step lattice 2007 Horváth, Kennedy & Sint