Multimodal optimization using crowding-based differential evolution

Multimodal optimization is an important area of active research within the evolutionary computation community. The ability of algorithms to discover and maintain multiple optima is of great importance - in particular when several global optima exist or when other high-quality solutions might be of interest. The differential evolution algorithm (DE) is extended with a crowding scheme making it capable of tracking and maintaining multiple optima. The introduced CrowdingDE algorithm is compared with a DE using the well-known sharing scheme that penalizes similar candidate solutions. In conclusion, the introduced CrowdingDE outperformed the sharing-based DE algorithm on fourteen commonly used benchmark problems.