arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2003.11307math.OC
收藏

不确定多智能体博弈中纳什均衡的后验概率可行性保证

A posteriori probabilistic feasibility guarantees for Nash equilibria in uncertain multi-agent games

University of Oxford(牛津大学)

George Pantazis, Filiberto Fele, Kostas Margellos

AI总结:

该研究基于场景方法的后验进展,提出无分布方法为多智能体博弈纳什均衡提供概率可行性鲁棒保证,规避高计算量的不可约支持子样本求解,并通过两类仿真验证了理论结果。

摘要:

本文提出了一种无分布方法,用于为多智能体博弈的纳什均衡(NE)提供鲁棒性保证。借助所谓场景方法(scenario approach,Campi等人,2018)的最新后验研究进展,我们为具有多面体约束的可行性问题提供了概率保证。该结果随后被应用于多智能体博弈场景,能够为给定博弈的任意纳什均衡的约束违反情况提供鲁棒性证书。我们的保证可与任何返回均衡解的纳什均衡求解算法配合使用。最后,通过利用问题的结构,我们无需采用计算成本极高的算法来寻找不可约支持子样本——这是场景方法的核心概念。我们的理论结果辅以仿真研究,针对两类不同问题(二维可行性问题与电动汽车(EV)充电控制问题)的解的鲁棒性展开了研究。

英文摘要:

In this paper a distribution-free methodology is presented for providing robustness guarantees for Nash equilibria (NE) of multi-agent games. Leveraging recent a posteriori developments of the so called scenario approach (Campi et al., 2018), we provide probabilistic guarantees for feasibility problems with polytopic constraints. This result is then used in the context of multi-agent games, allowing to provide robustness certificates for constraint violation of any NE of a given game. Our guarantees can be used alongside any NE seeking algorithm that returns some equilibrium solution. Finally, by exploiting the structure of our problem, we circumvent the need of employing computationally prohibitive algorithms to find an irreducible support subsample, a concept at the core of the scenario approach. Our theoretical results are accompanied by simulation studies that investigate the robustness of the solutions of two different problems, namely, a 2-dimensional feasibility problem and an electric vehicle (EV) charging control problem.

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。

↑