“Coarse correlated equilibria in mean field games”
Martedì 5 Dicembre 2023, ore 17:00 - Aula 2AB40 - Federico Cannerozzi (Università degli Studi di Milano)
Abstract
In the context of continuous-time stochastic mean field games (MFGs), we introduce a generalization of MFG solution, called coarse correlated solution, which can be seen as the mean field game analogue of a coarse correlated equilibrium. The latter is a generalization of Nash equilibrium for stochastic games, which allows for correlation between the strategies of non-cooperative players.
Our notion of solution can be justified by showing that approximate N-player correlated equilibria can be constructed starting from a correlated solution to the mean field game, and existence can be proved by means of a minimax theorem.
If time allows, an application to an abatement game between greenhouse gas emitters will be presented, in which coarse correlated solutions lead both to greater abated quantities and higher payoffs than the usual MFG solution.
The talk is based upon joint works with L. Campi (University of Milan “La Statale”), M. Fischer (University of Padua) and F. Cartellier (ENSAE Paris).
