Scaling Up Game Theory: Achievable Set Methods for Efficiently Solving Stochastic Games of Complete and Incomplete Information

Under certain restricted MARL addresses the RL problem in these environments using circumstances these results can be directly applied, stochastic (Markov) games (Littman 1994) as the formal while without scalable algorithms game theoretic results are model. The field of game theory has long addressed the question Once we have efficient algorithms the last two of how to model and predict the behavior of multiple self interested shortcomings can be overcome by modeling bounded rationality agents acting in the same environment. Game theory and the need for exploration as unknown variables in is a mature discipline containing some of the best answers a known game of incomplete information. MARL has taken much, as it should, from game theory. My solution revolves around using achievable set functions as the basis for computation and communication.