Parallel genetic algorithms: a useful survey

Abstract

In this article, we encompass an analysis of the recent advances in parallel genetic algorithms (PGAs). We have selected these algorithms because of the deep interest in many research fields for techniques that can face complex applications where running times and other computational resources are greedily consumed by present solvers, and PGAs act then as efficient procedures that fully use modern computational platforms at the same time that allow the resolution of cutting edge open problems. We have faced this survey on PGAs with the aim of helping newcomers or busy researchers who want to have a wide vision on the field. Then, we discuss the most well-known models and their implementations from a recent (last six years) and useful point of view: we discuss on highly cited articles, keywords, the venues where they can be found, a very comprehensive (and new) taxonomy covering different research domains involved in PGAs, and a set of recent applications. We also introduce a new vision on open challenges, and try to give hints that guide practitioners and specialized researchers. Our conclusion is that there are many advantages to using these techniques, and lots of potential interactions to other evolutionary algorithms, as well as we contribute to creating a body of knowledge in PGAs by summarizing them in a structured way, so that the reader can find this article useful for practical research, graduate teaching, and as a pedagogical guide to this exciting domain.