Effects of hierarchical migration in a parallel distributedparameter-free GA

摘要

Effects of hierarchical migration methods in a paralleldistributed Parameter-free GA (PfGA) are described. The PfGA is acompact and robust algorithm that extracts a local population from thewhole population and evolves it by adaptively changing the size of thesubpopulation. We propose hierarchically parallel distributedarchitectures with migration methods for the PfGA implementing them in aparallel machine. In evaluating many function optimization problems asrecent benchmark tests, we verified that the proposed parallel PfGAarchitectures with the migration methods effectively decrease the numberof evaluations to converge with the success rates held or improved byincreasing the number of subpopulations