A parallel genetic algorithm with distributed environment scheme

Abstract

Introduces an alternative approach to relieving the task of choosing optimal mutation and crossover rates by using a parallel and distributed GA with distributed environments. It is shown that the best mutation and crossover rates depend on the population sizes and the problems, and those are different between a single and multiple populations. The proposed distributed environment GA uses various combination of the parameters as the fixed values in the subpopulations. The excellent performance of the new scheme is experimentally recognized for a standard test function. It is concluded that the distributed environment GA is the fastest way to gain a good solution under the given population size and uncertainty of the appropriate crossover and mutation rates.