Heuristic Search Value Iteration for Zero-Sum Stochastic Games

In sequential decision making, heuristic search algorithms allow exploiting both the initial situation and an admissible to efficiently for optimal solution, often planning purposes. Such exist problems with uncertain dynamics, partial observability, multiple criteria, or collaborating agents. this article, we look at two-player zero-sum stochastic games (zsSGs) a discounted criterion, in view propose solution tailored fully observable case, while solutions have been proposed particular, though still more general, partially cases. This setting induces reasoning on lower upper bound of value function, which leads us proposing zsSG-HSVI, algorithm based iteration (HSVI), thus relies generating trajectories. We demonstrate that, each player acting optimistically, employing simple initializations, HSVI's convergence finite time ∈-optimal is preserved. An empirical study resulting approach conducted benchmark various sizes.