Investigating in scalability of opposition-based differential evolution
Shahryar Rahnamayan, G. Gary Wang · 2008
Differential Evolution (DE) is an effective, robust, and simple global optimization algorithm. Opposition-based differential evolution (ODE) has been proposed based on DE; it employs opposition-based population initialization and generation jumping to accelerate convergence speed. ODE shows promising results in terms of convergence rate, robustness, and solution accuracy. This paper investigates its performance on large scale problems. A recently proposed seven-function benchmark test suite for the CEC-2008 special session and competition on large scale global optimization has been utilized for the current investigation. Results interestingly confirm that ODE outperforms its parent algorithm (DE) on all high dimensional (500D) benchmark functions (F1-F7). By these supporting results, ODE is recommended by authors as an appropriate candidate for cooperative coevolutionary algorithms (CCA) to tackle with large scale problems. All required details about the testing platform, comparison methodology, and also achieved results are provided.