arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2201.12612cs.GTmath.OC
收藏

关于非合作完全信息半马尔可夫博弈

On Non-Cooperative Perfect Information Semi-Markov Games

K. G. Bakshi, S. Sinha

AI总结:

本文证明N人非合作完全信息半马尔可夫博弈在极限比率平均收益下存在纯半平稳纳什均衡,通过归约为SMDP简化证明,并用Mondal算法求解。

摘要:

我们证明,在极限比率平均收益准则下,N人非合作半马尔可夫博弈存在纯半平稳纳什均衡。在较早的一篇论文中,已处理了零和两人情形。证明方法是将这类完全信息博弈归约为一个相关联的半马尔可夫决策过程(SMDP),然后利用SMDP理论中的存在性结果。利用这一归约过程,可以得到以下结论的简单证明:(a)零和两人完全信息随机(马尔可夫)博弈在贴现和未贴现收益准则下都具有值,并且双方玩家都具有纯平稳最优策略。(b)类似的结论也适用于N人非合作完全信息随机博弈。所有这类博弈都可以使用归约后的SMDP(对于随机博弈情形为MDP)的任何高效算法来求解。在本文中,我们实现了Mondal算法,以在极限比率平均收益准则下求解SMDP。

英文摘要:

We show that an N-person non-cooperative semi-Markov game under limiting ratio average pay-off has a pure semi-stationary Nash equilibrium. In an earlier paper, the zero-sum two person case has been dealt with. The proof follows by reducing such perfect information games to an associated semi-Markov decision process (SMDP) and then using existence results from the theory of SMDP. Exploiting this reduction procedure, one gets simple proofs of the following: (a) zero-sum two person perfect information stochastic (Markov) games have a value and pure stationary optimal strategies for both the players under discounted as well as undiscounted pay-off criteria. (b) Similar conclusions hold for N-person non-cooperative perfect information stochastic games as well. All such games can be solved using any efficient algorithm for the reduced SMDP (MDP for the case of Stochastic games). In this paper we have implemented Mondal's algorithm to solve an SMDP under limiting ratio average pay-off criterion.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑