Global Search and Local Ascent for Large Cournot Games

Abstract

Equilibria detection in large games is a fundamental problem in computational game theory. A memetic algorithm, Global Search and Local Ascent (GSLA), is proposed. GSLA's performance is evaluated by means of numerical experiments within the framework of a Cournot game involving up to 100 players and by comparison with an evolutionary multiobjective optimization algorithm adapted for Nash equilibria detection.