GTES: A SIMULATION METHOD BY GAME AND LEARNING FOR ANALYSIS SYSTEMS OF ACTORS

This paper proposes an approach towards modeling an actor system, especially suited to describe a company's organization, based on game theory [11] and learning-based (evolutionary) local optimization. This method relies on the combination of three techniques: sampling for simulation (Monte-Carlo), game theory as far as the search for equilibrium is concerned and heuristic local search methods, such as genetic algorithms. This combination is not original as such, although it is rarely used with the full combined expressive power of this array of techniques. Our contribution with this paper is twofold. On the one hand we propose a model which is a natural framework for the collaboration between these three techniques. On the other hand, we use genetic algorithms to extend the search of Nash equilibrium, obtained as fixed-points of an iterative transformation. This remains a simulation tool, not intended to solve problems but to validate a given model and to study its properties.