Model extensions
Genetic and evolutionary algorithms

When the objective function exhibits several equivalent optima, a classical evolutionary algorithm typically converges to only one of them, in a random manner. However, in many applications, it is important to identify all such optima, rather than merely locating a single optimal solution. Extensions of classical EA, inspired by the natural phenomenon of ecological niche formation, therefore rely on the implicit or explicit management of subpopulations (or niches). Resource-sharing schemes are of particular interest: when individuals within the same subpopulation are required to share limited resources, the growth of that subpopulation is constrained, and, in case of overpopulation, individuals are encouraged to explore and colonise new regions of the...