Y.: A New Model of Distributed Genetic Algorithm for Cluster Systems

Abstract In this study, a new model of distributed genetic algorithm (DGA) for cluster systems is pro-posed. That is called Dual Individual Distributed Genetic Algorithm: DuDGA. DGA is a very good parallel model of genetic algorithms, because the necessary network traÆc is not so heavy. On the other hand, DGA needs additional parameters to simple GA such as migration rate and migration intervals. In the DuDGA, each island of DGA has only two individuals. Because of this constitu-tion, Necessary parameters decrease markedly. At the same time, the searching ability improves and the network traÆc is also decrease. These char-acteristics are suitable for the PC cluster systems. Through the typical test functions, these character-istics are conrmed and discussed