Speaker: Sylvain Gibaud, Institut de Mathématiques de Toulouse
I will present an evolutionnary game model based on a multi-agent system. In it the evolution will not be managed by a generation system whereby after play a new population is drawn but instead by the fact that agents who perform poorly will die.
Our model represents a spatialized prisoner's dilemma where the players move on a torus and play against other players only if they are at the same location. As it is usually done in evolutionnary games, the action of the players will be fixed. A player will be either a cooperative player (who always cooperates) or a defecting player (who always defects).
This system has been explored using simulations (or approximations). The presentation will show a probabilistic framework in which we can prove theoretical results.