Influence of the Migration Policy in Parallel DistributedGAs with Structured and Panmictic Populations

  • Authors:
  • Enrique Alba;José M. Troya

  • Affiliations:
  • Dpto. de Lenguajes y Ciencias de la Computación, Univ. de Málaga, Campus de Teatinos (3.2.12), 29071-MÁLAGA, España. eat@lcc.uma.es;Dpto. de Lenguajes y Ciencias de la Computación, Univ. de Málaga, Campus de Teatinos (3.2.12), 29071-MÁLAGA, España

  • Venue:
  • Applied Intelligence
  • Year:
  • 2000

Quantified Score

Hi-index 0.00

Visualization

Abstract

Parallel genetic algorithms (PGAs) have beentraditionally used to overcome the intense use of CPU and memory thatserial GAs show in complex problems. Non-parallel GAs can beclassified into two classes: panmictic and structured-populationalgorithms. The difference lies in whether any individual in thepopulation can mate with any other one or not. In this work, they areboth considered as two reproductive loop types executed in theislands of a parallel distributed GA. Our aim is to extend theexisting studies from more conventional sequential islands to otherkinds of evolution. A key issue in such a coarse grain PGA is themigration policy, since it governs the exchange of individuals amongthe islands. This paper investigates the influence of migrationfrequency and migrant selection in a ring of islands running eithersteady-state, generational, or cellular GAs. A diversity analysis isalso offered from an entropy point of view. The study uses differentproblem types, namely easy, deceptive, multimodal, NP-Complete, andepistatic search landscapes in order to provide a wide spectrum ofproblem difficulties to support the results. Large isolation valuesand random selection of the migrants are demonstrated as providing alarger probability of success and a smaller number of visited points.Also, interesting observations on the relative performance of thedifferent models are offered, as well as we point out theconsiderable benefits that can accrue from asynchronous migration.