We briefly describe some of our recent results for the mass spectrum and matrix elements using $O(a)$ improved fermions for quenched QCD. Where possible a comparison is made between improved and Wilson fermions.
The improvement of fermionic operators for Ginsparg-Wilson fermions is investigated. We present explicit formulae for improved Greens functions, which apply both on-shell and off-shell.
We discuss the improvement of bilinear fermionic operators for Ginsparg-Wilson fermions. We present explicit formulae for improved Greens functions, which apply both on-shell and off-shell.
We present new results from our ongoing study of flavor singlet pseudoscalar mesons in QCD. Our approach is based on (a) performing truncated eigenmode expansions for the hairpin diagram and (b) incorporating the ground state contribution for the connected meson propagator. First, we explain how the computations can be substantially improved by even-odd preconditioning. We extend previous results on early mass plateauing in the eta channel of two-flavor full QCD with degenerate sea and valence quarks to the partially quenched situation. We find that early mass plateau formation persists in the partially quenched situation.
In view of the recent excitement about a tension between determinations of f_Ds from experiment and from simulations of lattice QCD with dynamical quarks, we try to clear up the picture of lattice determinations in the continuum limit of the quenched approximation. For O(a) improved Wilson quarks we see linear scaling in the squared lattice spacing a^2 only for a<~0.08fm. For coarser lattices we observe significant contaminations from higher order cutoff effects. As an aside we also study the scaling of the charm quark mass and the ratio of the vector to the pseudo-scalar decay constant and the spin-splitting.
We present an exact dynamical QCD simulation algorithm for the $O(a)$-improved Wilson fermion with odd number of flavors. Our algorithm is an extension of the non-Hermitian polynomials HMC algorithm proposed by Takaishi and de Forcrand previously. In our algorithm, the systematic errors caused by the polynomial approximation of the inverse of Dirac operator is removed by a noisy-Metropolis test. For one flavor quark it is achieved by taking the square root of the correction matrix explicitly. We test our algorithm for the case of $N_f=1+1$ on a moderately large lattice size ($16^3times48$). The $N_f=2+1$ case is also investigated.