Stochastic Equilibria under Imprecise Deviations in Terminal-Reward Concurrent Games
نویسندگان
چکیده
We study the existence of mixed-strategy equilibria in concurrent games played on graphs. While existence is guaranteed with safety objectives for each player, Nash equilibria need not exist when players are given arbitrary terminal-reward objectives, and their existence is undecidable with qualitative reachability objectives (and only three players). However, these results rely on the fact that the players can enforce infinite plays while trying to improve their payoffs. In this paper, we introduce a relaxed notion of equilibria, where deviations are imprecise. We prove that contrary to Nash equilibria, such (stationary) equilibria always exist, and we develop a PSPACE algorithm to compute one.
منابع مشابه
Mixed Nash Equilibria in Concurrent Terminal-Reward Games
We study mixed-strategy Nash equilibria in multiplayer deterministic concurrent games played on graphs, with terminal-reward payoffs (that is, absorbing states with a value for each player). We show undecidability of the existence of a constrained Nash equilibrium (the constraint requiring that one player should have maximal payoff), with only three players and 0/1-rewards (i.e., reachability o...
متن کاملMixed Nash Equilibria in Concurrent Games
We study mixed-strategy Nash equilibria in multiplayer deterministic concurrent games played on graphs, with terminal-reward payoffs (that is, absorbing states with a value for each player). We show undecidability of the existence of a constrained Nash equilibrium (the constraint requiring that one player should have maximal payoff), with only three players and 0/1-rewards (i.e., reachability o...
متن کاملTheory and Methodology Average-discounted equilibria in stochastic games
In stochastic games with ®nite state and action spaces, we examine existence of equilibria where player 1 uses the limiting average reward and player 2 a discounted reward for the evaluations of the respective payo sequences. By the nature of these rewards, the far future determines player 1's reward, while player 2 is rather interested in the near future. This gives rise to a natural cooperat...
متن کاملPolicy Invariance under Reward Transformations for General-Sum Stochastic Games
We extend the potential-based shapingmethod fromMarkov decision processes to multi-player general-sum stochastic games. We prove that the Nash equilibria in a stochastic game remains unchanged after potential-based shaping is applied to the environment. The property of policy invariance provides a possible way of speeding convergence when learning to play a stochastic game.
متن کاملConvergent Learning Algorithms for Unknown Reward Games
In this paper, we address the problem of convergence to Nash equilibria in games with rewards that are initially unknown and must be estimated over time from noisy observations. These games arise in many real-world applications, whenever rewards for actions cannot be prespecified and must be learned online, but standard results in game theory do not consider such settings. For this problem, we ...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2016