The ergodic Mean Field Game system for a type of state constraint condition

Abstract

This paper investigates the well-posedness of a type of state constraint ergodic Mean Field Game system in a bounded domain in which the Hamilton-Jacobi-Bellman equation is paired with an infinite Dirichlet boundary condition. In this setting, the infinite boundary condition prevents the underlying stochastic trajectories from reaching the boundary, resulting in a state constraint solution. We consider a class of superlinear power-like Hamiltonians and establish existence and uniqueness results under appropriate assumptions on the coupling. We treat both non-local and local couplings, and establish the uniqueness of the system by exploiting the decay of the density near the boundary.