A Cellular Genetic Algorithm with Disturbances: Optimisation Using Dynamic Spatial Interactions

This paper describes a novel evolutionary algorithm inspired by the nature of spatial interactions in ecological systems. The Cellular Genetic Algorithm with Disturbances (CGAD) can be seen as a hybrid between a fine-grained and a coarse-grained parallel genetic algorithm. The introduction of a ldquodisturbance-colonisationrdquo cycle provides a mechanism for maintaining flexible subpopulation sizes and self-adaptive controls on migration. Experiments conducted, using a range of stationary and non-stationary optimisation problems, show how changes in the structure of the environment can lead to changes in selective pressure, population diversity and subsequently solution quality. The significance of the disturbance events lies in the new ldquoecologicalrdquo patterns that arise during the recovery phase.