Dual of DE: a new scheme of differential evolution algorithm

Differential Evolution DE is a class of powerful evolutionary algorithms for global numerical optimisation. Using the mechanisms of complementarities that exist in nature, opposition-based learning has been successfully used to generate opposite number to enhance the search ability of DE. In this paper with the opposite concept for the scheme of DE, a completely new scheme of DE is proposed. Based on the formal analysis of traditional DE and the concept of duality borrowed from mathematics, Dual of DE is derived. Then the Dual of DE has been tested on 18 commonly used benchmark problems and the results obtained were compared to those of the traditional DE. Experimental results show that the Dual of DE can serve as a good complementary scheme for the traditional DE in most of the considered problems. Thus it can be an alternative in solving real-world optimisation problems.