A Study on Parallel Distributed Genetic Algorithms. Discussion on a Randomized Migration Island Model of Distributed Genetic Algorithms.
Mitsunori Miki, Tomoyuki Hiroyasu, Yasunori Nakamura · TRANSACTIONS OF THE JAPAN SOCIETY OF MECHANICAL ENGINEERS Series A · 2000
This paper discusses about the characteristics of distributed genetic algorithms (DGAs). Among the several types of models for DGAs, this paper focused on a rendomized migration island model. In this model, the island of the migration is decided as every migration opportunity at random. When there are a lot of islands, this model is very useful. The efficiency of this model is discussed through the numerical examples. In this study, Distributed Genetic Algorithm with Distributed Environment (DEGA) is also introduced. Usually, the parameters in GAs are the same in each island but they are different in DEGAs. This approach makes designers free from setting appropriate GA parameters. Applying this algorithm to solve numerical examples, it is also clarified that there are several advantages in this approach.