Parallel single front genetic algorithm: performance analysis in a cluster system

In this paper a performance analysis in a cluster system of the parallel single front genetic algorithm (PSFGA) is carried out. The PSFGA is a parallel evolutionary optimizer for multiobjective problems that use a structured population in the form of a set of islands. The SFGA, an elitist evolutionary algorithm with a clearing procedure that uses a grid in the objective space for diversity maintaining purposes, is performed on each subpopulation (island) associated to a different area in the search space. Experimental results show that PSFGA outperforms SFGA and SPEA (strength Pareto evolutionary algorithm) in the cases studied.