Parallel kinetic Monte Carlo simulation of Coulomb glasses

Abstract

We develop a parallel rejection algorithm to tackle the problem of lowacceptance in Monte Carlo methods, and apply it to the simulation of thehopping conduction in Coulomb glasses using Graphics Processing Units, forwhich we also parallelize the update of local energies. In two dimensions, ourparallel code achieves speedups of up to two orders of magnitude in computingtime over an equivalent serial code. We find numerical evidence of a scalingrelation for the relaxation of the conductivity at different temperatures.