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 1608.07575cs.GTcs.DS
收藏

稳定婚姻问题中的策略博弈

Strategic play in stable marriage problem

Mircea Adrian Digulescu

摘要:

稳定婚姻问题,由盖尔和Shapely [1]解决,包括提供n个“男孩”和n个“女孩”之间的二分匹配-每个人都有一个完全有序的偏好列表,这样就不存在“男孩”和“女孩”在匹配中会喜欢对方而不是他们的伴侣。在本文中,我们分析的情况下,战略发挥的“男孩”在游戏中直接启发了这个问题。我们提供了一个O(n^3)的算法来确定一个匹配,它在盖尔-Shapely意义上不一定是稳定的,但它是联盟稳定的,因为没有参与者有自私的利益离开产生的大联盟加入任何可能可行形成的潜在替代者,也是人最优的。因此,在一个现实的假设下,没有玩家有兴趣“破坏”匹配,即使他理论上可以。对于一些(不是全部)“男孩”来说,结果匹配通常比天真的盖尔-Shapely匹配更好,对于其他人来说也不会更差。这种匹配比顶部交易周期方法产生的匹配更真实(稳定),因此比后者提供了质的改进。此外,我们分析的情况下,玩家被允许进行战略威胁(即愿意牺牲自己的结果,以伤害他人),提供一个相关的例子来说明这种形式的游戏的好处,并最终提供一个指数时间算法,试图确定一个好的威胁决策策略。然后,我们简要地研究了一些其他非传统的可能性,一个球员必须影响他的结果。最常见的变化的博弈模型也描述和分析方面的适用性的方法在本文中。最后,给出了几个可以用本文方法建模和求解的实际问题的例子。

英文摘要:

The stable marriage problem, as addressed by Gale and Shapely [1] consists of providing a bipartite matching between n " boys " and n " girls "-each of whom have a totally ordered preference list over the other set-such that there exists no " boy " and no " girl " that would prefer each other over their partner in the matching. In this paper, we analyze the cases of strategic play by the " boys " in the game directly inspired by this problem. We provide an O(n^3) algorithm for determining a matching which is not necessarily stable in the Gale-Shapely sense, but it is coalition-stable, in that no player has a selfish interest to leave the resulting grand coalition to join any potential alternative one which might feasibly form, and is also man-optimal. Thus, under a realistic assumption set, no player has an interest to " destabilize " the matching, even though he theoretically could. The resulting matching is often better than the naïve Gale-Shapely one for some (not all) of the " boys " , being no worse for the rest. This matching is more realistic (stable) than the one produced by top-trading-cycles method, thus offering a qualitative improvement over the latter. Furthermore, we analyze the situation when players are allowed to make strategic threats (i.e. be willing to sacrifice their own outcome to hurt others), offer a relevant example to illustrate the benefits of this form of play, and ultimately provide an exponential time algorithm which tries to determine a good threat-making strategy. We then briefly examine a few other non-conventional possibilities a player has to affect his outcome. Most common variations to the game model are also described and analyzed with regard to applicability of the methods in this paper. Finally, a few examples of real-life problems which can be modeled and solved with the methods in this paper are presented.

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。

↑