Population Size Reduction in Particle Swarm Optimization Using Product Graphs

Purpose of this paper is to introduce a population size reduction in particle swarm optimization algorithm, where the reduction is performed by selecting two particles (also donor particles) randomly and replacing these by a new particle with elements determined from a set of pair values obtained by the Cartesian product of both donor particles for each particular element randomly. Average values of each pair values from the donor particles are calculated for corresponding elements of the new particle. The proposed PSOGP was applied on a benchmark function suite consisted of four well-known functions and compared with the original PSO algorithm. The results are very promising and show the potential of the proposed idea.