Abstract
Genetic algorithm (GA) is a stochastic metaheuristic process consisting onthe evolution of a population of candidate solutions for a given optimizationproblem. By extension, multipopulation genetic algorithm (MPGA) aims forefficiency by evolving many populations, or islands, in parallel and performingmigrations between them periodically. The connectivity between islandsconstrains the directions of migration and characterizes MPGA as a dynamicprocess over a network. As such, predicting the evolution of the quality of thesolutions is a difficult challenge, implying in the waste of computer resourcesand energy when the parameters are inadequate. By using models derived fromstatistical mechanics, this work aims to estimate equations for the study ofdynamics in relation to the connectivity in MPGA. To illustrate the importanceof understanding MPGA, we show its application as an efficient alternative tothe thermalization phase of Metropolis-Hastings algorithm applied to the Isingmodel.