Passive Oppositional Differential Evolution for Global Optimization

Abstract

This paper presents an algorithm named PODE which is a significant improvement of a previous algorithm with the same name by the first author of this paper. Its key idea is to introduce a set of opposite vectors, not included in the population, but used in the mutation step to gain diversity. Several techniques are also introduced mainly to enhance global exploration capability, which include use of quasi-opposition, introduction of global iterations and local iterations, and the unique design of mutation and crossover operators which are different between global and local iterations. The obtained algorithm is free of parameters. Numerical results on an extensive set of functions show that this algorithm has excellent robustness and has better overall performance than classic DE, ODE, and old PODE.