Distributed Coevolutionary Genetic Algorithms for Multi-Criteria and Multi-Constraint Optimisation

This paper explores the use of coevolutionary genetic algorithms to attack hard optimisation problems. It outlines classes of practical problems which are difficult to tackle with conventional techniques, and indeed with standard ‘single species’ genetic algorithms, but which may be amenable to ‘multi-species’ coevolutionary genetic algorithms. It is argued that such algorithms are most coherent and effective when implemented as distributed genetic algorithms with local selection operating. Examples of the successful use of such techniques are described, with particular emphasis given to new work on a highly generalised version of the job shop scheduling problem.