Coevolutionary Game-Theoretic Multi-Agent Systems

Multi-agent systems based on N-person games with limited interaction are considered. We are interested in the global behavior of the team of players, measured by the average payoff received by a player. To evolve a global behavior in the system, we propose a coevolutionary algorithm, where only local fitness functions are evaluated while the global criterion is optimised. The multi-agent system is applied to develop a distributed algorithm of dynamic mapping tasks in parallel computers.