Performance tuning of evolutionary algorithms using particle sub swarms

Particle swarm optimization (PSO) technique proved its ability to deal with very complicated optimization and search problems. This paper proposes a new particle swarm variant which deals with sub-populations. This algorithm is applied for solving the well known class of mathematical problems: geometrical place problems (also known as locus problems). Finding the geometrical place can be sometimes a hard task and in almost all situations the geometrical place consists in more than one single point. The performance of the sub-swarm based PSO method is compared with evolutionary algorithms. The main advantage of the PSO technique is its speed of convergence. Also, we propose a hybrid algorithm by combining PSO and EA. This combination is able to detect the geometrical place very fast for difficult problems for which EA´s need more time and PSO technique even with sub-populations could not find the geometrical place.