Methods of statistical physics and complex mathematical problems

Abstract

Discusses some nontraditional applications of the methods of statistical physics. Such methods provide useful approaches to various complex mathematical problems. The authors discusses in detail the simulated annealing optimisation method based on the Monte Carlo Metropolis method which proved to be efficient for solving combinatorial optimisation problems. The author investigates the parallel implementation of the annealing method and argues that one thus arrives at a new class of computer architectures-the statistically coupled processor schemes-which can handle nonlinear problems of great complexity.