Niching in Evolutionary and Swarm-Based Optimization for Multimodal Problems: A Survey

Abstract

Niching is an essential mechanism in multimodal optimization (MMO), extending evolutionary algorithms (EAs) and particle swarm optimization (PSO) to trace multiple global and local optima and preserve them throughout the search process, rather than converging on a single solution. This paper surveys niching techniques across PSO, differential evolution (DE), genetic algorithms (GA), evolution strategies (ES), and gravitational search algorithms (GSA), organizing them by population structure (single- versus multi-population) and by their reliance on explicit niching parameters. We examine parameter-free ring-topology PSO technique, describing its neighborhood-based update rule and explaining how niches emerge without an explicit niche-radius parameter. As reported consistently across the surveyed literature, this parameter-free approach performs comparably to, and on several benchmark functions more consistently than, parameterdependent alternatives such as SPSO and FER-PSO. The survey further reviews hybrid and archive-guided niching methods that combine PSO with differential evolution, evolution strategies, and gravitational search, and closes with a synthesis showing that the same parameter-free principle central to ring-topology PSO extends naturally across these related families.