Evaluation of a New Resampling Scheme for f Monte Carlo Methods

Abstract:

A new class of methods for reducing the number of particles in delta-f Monte Carlo simulations is presented. The reduction of particles is necessary when there is a continuous growth of the number of particles during the simulation. The method is based on resampling the particles distribution in local partitions of the phase-space. The resampling is accomplished by replacing particles with fewer numbers of new particles in each partition while ensuring that the moments of distribution are conserved. It’s demonstrated that the method well preserves the distribution function.