| Abstract: |
| In this talk we discuss the implementation of numerical schemes for the approximation of stationary Mean Field Games (MFG) with a variational structure. By casting the MFG as a PDE-constrained optimization problem, we apply a proximal method for its solution combined with a finite difference approximation of the state equation, leading to a fully discrete, first-order iterative scheme. The proposed scheme has several interesting features, such as robustness with respect to different viscosity values and modelling of congestion effects. Moreover, it can be easily implemented for time-dependent MFG. I this talk we will discuss computational aspects behind each of these features. |
|