A Multigrid Algorithm for Sampling Imaginary-Time Paths in Quantum Monte Carlo Simulations

摘要

We describe a novel simulation method that eliminates the slowing-downproblem in the Monte Carlo simulations of imaginary-time path integrals nearthe continuum limit. This method combines a stochastic blocking procedure withthe multigrid method to rapidly accelerate the sampling of paths in a quantumMonte Carlo simulation, making its dynamics more ergodic. The effectiveness andefficiency of this method are demonstrated for several one-dimensional quantumsystems and compared to other standard and accelerated methods.