Performance of Differential Evolution and Particle Swarm Methods on Some Relatively Harder Multimodal Benchmark Functions
The IUP Journal of Computational Mathematics, Vol. III, No. 1, pp. 7-18, March 2010
Posted: 28 Mar 2010
Date Written: March 24, 2010
Abstract
Particle swarm optimization and Differential Evolution (DE) optimization methods of global optimization are two of the very versatile methods of minimization/maximization of multimodal non-convex continuous nonlinear functions that find their application in many fields of engineering, natural sciences and social sciences. In most cases they perform better than their rival methods, such as the genetic algorithms, simulated annealing, tabu search and clustering algorithms. However, these methods are not a panacea for all problems of non-convex optimization. The present paper tests their performance on certain harder multimodal functions and finds that in a majority of cases they falter. There is a need, therefore, to improve these methods.
Keywords: Global optimization, Non-convex functions, Particle swarm, Differential evolution, Hard benchmark functions
Suggested Citation: Suggested Citation