Local population model

also diffusion model, neighborhood model, or fine grained model. For each individual a neighborhood is defined (not to be mixed up with search space neighborhood). Only individuals sharing the same neighborhood can mate and produce offspring. The selection pool is composed of all individuals of this neigborhood. The use of the local population model leads to a flowing subdivision of the population. The individuals are isolated by distance. The flow (diffusion) of information inside the population is determined by the topology/structure of the neighborhood. Local population models are able to maintain the population's diversity for a longer time period (compared to classical EAs, see global population model) by the effect of niching. Thus, it can prevent the EA from premature convergence.