学术文献库

Mean field games have traditionally been defined~[1,2] as a model of large scale interaction of players where each player has a private type that is independent across the players. In this paper, we introduce a new model of mean field teams and games with \emph{correlated types} where there are a large population of homogeneous players sequentially making strategic decisions and each player is affected by other players through an aggregate population state. Each player has a private type that only she observes and types of any $N$ players are correlated through a kernel $Q$. All players commonly observe a correlated mean-field population state which represents the empirical distribution of any $N$ players' correlated joint types. We define the Mean-Field Team optimal Strategies (MFTO) as strategies of the players that maximize total expected joint reward of the players. We also define Mean-Field Equilibrium (MFE) in such games as solution of coupled Bellman dynamic programming backward equation and Fokker Planck forward equation of the correlated mean field state, where a player's strategy in an MFE depends on both, her private type and current correlated mean field population state. We present sufficient conditions for the existence of such an equilibria. We also present a backward recursive methodology equivalent of master's equation to compute all MFTO and MFEs of the team and game respectively. Each step in this methodology consists of solving an optimization problem for the team problem and a fixed-point equation for the game. We provide sufficient conditions that guarantee existence of this fixed-point equation for the game for each time $t$.