Loosely coupled distributed genetic algorithms

Abstract

Iterated, noncooperative N-person games with limited interaction are considered. Each player in the game has defined its local payoff function and a set of strategies. While each player acts to maximize its payoff, we are interested in a global behavior of the team of players measured by the average payoff received by the team. To study behavior of the system we propose a new parallel and distributed genetic algorithm based on evaluation of local fitness functions while the global criterion is optimized. We present results of simulation study which support our ideas.