Simulated Tempering: A New Monte Carlo Scheme

Abstract

We propose a new global optimization method (Simulated Tempering) for simulating effectively a system with a rough free energy landscape (i.e. many coexisting states) at finite non-zero temperature. This method is related to simulated annealing, but here the temperature becomes a dynamic variable, and the system is always kept at equilibrium. We analyze the method on the Random Field Ising Model, and we find a dramatic improvement over conventional Metropolis and cluster methods. We analyze and discuss the conditions under which the method has optimal performances. ROM2F-92-06, SCCS 241, hep-lat/9205018 1 Simulated annealing is an efficient heuristic method which is used to find the absolute minimum of functions with many local minima: it has been introduced independently in the framework of the Monte Carlo approach for discrete variables in ref. [1], and in the framework of stochastic differential