A viscous ergodic problem with unbounded and measurable ingredients. Part 2: Mean-Field Games

We address the problem of existence and (non-)uniqueness of solutions $(c,u(\cdot ),\mu )$ to ergodic mean-field games in the whole space ${\mathbb{R}}^m$ with unbounded and merely measurable data, and for non-separable Hamiltonian. The payoff functional satisfies a new monotonicity condition, different from the usual one due to Lasry and Lions. The method we use is also different from classical approaches. It relies on duality theory and optimization in abstract Banach spaces together with maximal dissipativity of diffusion operators.