Comprehensive Comparison of Genetic Algorithm and Pattern Search Method

Abstract

At present, the comparisons of derivative-free optimization methods are mostly about a certain problem, there are few comprehensive comparison. This paper comprehensively compared genetic algorithm (GA)and the pattern search method (PS) in terms of the solution quality and function evaluations. The results show that the performance of these two methods strongly depends on the problem type and number of decision variables. For largesized complex problem, GA is more robust, especially the objective function is multiobjective, multimodal, discontinuous, complex constraints. However, it required more function evaluations. The pattern search method is better than GA for small-sized problems, having comparable solution quality, meanwhile requiring fewer function evaluations.