Metropolis Importance Sampling for Rugged Dynamical Variables

Preprint
Publisher Website
Google Scholar
  • 17 September 2002
Abstract
A funnel transformation, which acts recursively from higher towards lower temperatures, is introduced. It alters the a-priori probabilities of a canonical Metropolis simulation, such that they zoom in on the global energy minimum if a funnel exists indeed. As a first, simple approximation to the full transformation, a factorized version, called rugged Metropolis one (RM1), is tested. Although RM1 is only a crude approximation of the general transformation, the gain in computational performance is already substantial and, due to its simplicity, RM1 is well suited to replace the conventional Metropolis updating for these kind of systems.

This publication has 0 references indexed in Scilit: