Continuous time finite state mean field games (Q373001)

In this paper a continuous-time dynamic game is considered where a large number of players can be in any of \(d\) states. The players can switch from state to state and their decisions depend on certain optimality criteria. It is assumed that the game is symmetric with respect to permutation of the players. Players only know their own position and the distribution of players in each of the \(d\) states. Each player can control the transition rate from one state to another and so on. Moreover, the information available to the reference player (fix player) is the distribution of players given by a probability vector \(\theta\in S^{d},\) where \(S^d\) is the probability simplex \[ \begin{cases} \sum_{i=1}^{d}\theta^{i}=1,\\ \theta^{i}\geq 0. \end{cases} \] In this work a limiting mean field model is derived and its main properties are characterized. It is shown that the mean field limit is a system of ordinary differential equations. For a game between \(N+1\) players the Nash equilibrium problem is formulated and the existence of Nash equilibrium solutions and the convergence as \(N\rightarrow\infty\) of the mean field model are established. An estimation of the rate of convergence is obtained. Finally, some examples for potential mean field games are considered.