Robust Particle Swarm Optimizer based on Chemomimicry

Particle swarm optimizers (PSO) were first introduced by Kennedy and Eberhart as stochastic algorithms which seek optimal solutions to functions through the use of swarm intelligence [1]. The main theme of PSO is that many particles are allowed to explore a function space. As each particle relocates it inputs its coordinates into the objective function for evaluation. Particles are assigned directions and magnitudes for motion based on distances to the best global functional evaluation outcome (g), and/or their individual best locations (p). Traditionally the positions (X) of particles and their velocities (V) are updated as follows, Xi = Xi + Vi Vi = Vi + c1 · runif(0, 1) · (pi − xi) + c2 · runif(0, 1) · (g − xi)