Search Space Division in GAs using Phenotypic Squares Estimates

In this article, we study a new type of forking GA (fGA), the phenotypic forking GA (p-fGA). The fGA divides the whole search space into subspaces depending on the convergence status of the population and the solutions obtained so far; and is intended to deal with multimodal problems which are difficult to solve using conventional GA. We use a multi-population scheme, which includes one parent population that explores one subspace, and one or more child population(s) exploiting the other subspace. The p-fGA divides the search space using phenotypic properties only, and defines a search subspace (to be exploited by a child population) by a neighborhood hypercube around the current best individual in the phenotypic feature space. Empirical results on complex function optimization problems show that the p-fGA performs fairly well compared to a conventional GA. Two other variants of the p-fGA, the moving window p-fGA (to accelerate the speed of convergence in the child populations) and the variable resolution p-fGA (to solve multimodal problems with high precision) are also studied in this article.