On near optimal trajectories for a game associated with the \(\infty\)-Laplacian (Q662827)

A conjecture of Peres, Schramm, Sheffield and Wilson is considered wherein a discrete time tug-of-war game parametrized by a small parameter \(\varepsilon\) is conjectured to lead in the \(\varepsilon \downarrow 0\) limit to a zero sum stochastic differential game whose value function satisfies a given Dirichlet problem involving the \(\infty\)-Laplacian. Under suitable technical conditions that include regularity of the solution to this Dirichlet problem, it is established that the conjectured continuous time game dynamics indeed arises as the limiting optimal dynamics of a stochastic differential game interpreted in the sense of Elliott and Kalton, when both players play \(\delta\)-optimally in the approximating games. The limit law for a sequence of near saddle points is also identified.