Abstract
Various types of genetic algorithms have been proposed in literature. One of the active topics is on the distribution of genetic algolithms. The distributed genetic algolithms generally fall into two categories: Coarse-grained genetic algorithms and fine-grained genetic algorithms. In this paper, we propose a hybrid method of the two distributed genetic algolithms. In computer simulations on function optimization problems, we examine the performance of the proposed method. Specifically, we examine the effect of the position of elites, the method of selection and the topology of subpopulations.
