Elements of simulated annealing in Pareto front search

Abstract

Determination of the Pareto front of location problem solutions represents one of the very complex and computational time demanding tasks, when solved by exact means of mathematical programming. This paper is motivated by possible application of the metaheuristic simulated annealing to the process of obtaining a close approximation of the Pareto front by a set of non-dominated solutions of the p-location problem. Contrary to the other approaches, the suggested method is based on minimization of non-dominated solution set area, which directly describes quality of the approximation. Elements of the simulated annealing method are used for random breaking some limits imposed on local characteristics of the improving process. The presented results of the numerical experiments give an insight to relations among the simulated annealing parameters and optimization process efficiency.