Kramers equation algorithm with Kogut-Susskind fermions on lattice

We compare the performance of the Kramers equation Monte Carlo (KMC) algorithm with that of the Hybrid Monte Carlo (HMC) algorithm for numerical simulations with dynamical Kogut-Susskind fermions. Using the lattice Gross-Neveu model in 2 space-time dimensions, we calculate the integrated autocorrelation time of different observables at a number of couplings in the scaling region on 162 and 322 lattices while varying the parameters of the algorithms for optimal performance. In our investigation the performance of KMC is always significantly below than that of HMC for the observables used. We also stress the importance of having a large number of configurations for the accurate estimation of the integrated autocorrelation time.