Bayesian Learning in Mean Field Games

Résumé

We consider a mean-field game model in which the cost functions depend on a fixed parameter referred to as the “state,” which remains unknown to the players. Players acquire information about the state through private signals received during the course of the game. We derive a mean-field system that characterizes the equilibrium payoff of the game and demonstrate the existence of a solution under standard regularity assumptions. Furthermore, we establish the uniqueness of the solution when the cost function satisfies the monotonicity assumption proposed by Lasry and Lions at each state.

Mots-clés

Référence

Eran Shmaya et Bruno Ziliotto, « Bayesian Learning in Mean Field Games », SIAM Journal on Control and Optimization, vol. 63, n° 3, 2025.