Simulating the Island Model of Evolutionary Algorithms

The bmigrate utility simulates the island model of evolutionary algorithms, allowing for real-time visualisation, analysis, and control. At its simplest, bmigrate orchestrates simultaneous simulations consisting of repeated simulation runs. Each run is a sequence of generations over a finite set of islands.

In each generation, (1) each island's individuals reproduce with a Poisson process with a mean of the individual's utility function. The offspring then (2) migrate to an island (which may be the birth island). On each island, (3) a single native individual and (4) single immigrant individual are selected, then the native individual is (5) replaced by the immigrant. The remaining immigrants are discarded, keeping islands' populations constant.
Single Generation of Process

The system is open source, though we kindly ask that researchers using it cite our work with the given BibTeX citation. It has an extensive manual describing the simulator's theoretical underpinnings, implementation, installation, configuration, and operation. For patches, questions, comments, and so on, please contact , k-Consulting, who originally wrote bmigrate for , Stockholm School of Economics, with financial support from the Knut and Alice Wallenberg Research Foundation. What follows is a list of recent releases.