A New Technique for Sampling Multi-modal Distributions

Abstract

In this paper we demonstrate that multi-modal probability distribution functions (PDFs) may be efficiently sampled using an algorithm originally developed for numerical integration by Monte Carlo methods. This algorithm can be used to generate an input PDF which can be used as an independence sampler in a Metropolis–Hastings chain to sample otherwise troublesome distributions. Some examples in one, two, and five dimensions are worked out. We also comment on the possible application of our results to event generation in high-energy physics simulations.