Explanation of Stagnation at Points that are not Local Optima in Particle Swarm Optimization by Potential Analysis
Alexander Raß, Manuel Schmitt, Rolf Wanka · 2015
This paper investigates the frequently observed phenomenon of stagnation which appears on particle swarm optimization (PSO). We introduce a measure of significance of single dimensions and provide experimental and theoretical evidence that the classical PSO, even with swarm parameters known (from the literature) to be good, almost surely does not converge to a local optimum (stagnation) if too few particles are used. Stagnation is an undesirable property of PSO.