Heuristic pattern search and its hybridization with simulated annealing for nonlinear global optimization

Abstract

In this article, we present a new approach of hybrid simulated annealing method for minimizing multimodel functions called the simulated annealing heuristic pattern search (SAHPS) method. Two subsidiary methods are proposed to achieve the final form of the global search method, SAHPS. First, we introduce the approximate descent direction (ADD) method, which is a derivative-free procedure with high ability of producing a descent direction. Then, the ADD method is combined with a pattern search method with direction pruning to construct the heuristic pattern search (HPS) method. The last method is hybridized with simulated annealing (SA) to obtain the SAHPS method. The experimental results through well-known test functions are shown to demonstrate the efficiency of the proposed method SAHPS.