2016-05-272024-05-14https://scholars.lib.ntu.edu.tw/handle/123456789/656414摘要：這個計畫研究一個在環形結構網路（cyclic network）中的重複集體行為。在這個重複集體行為中，參賽者只能知道自己和鄰居對參與此集體行為的態度。而在重複的過程裡，參賽者只觀察到自己和鄰居的行動。給定網路結構是固定的、有限的、連接的、非指向性的、已知的、和非環形的，Chen 2015指出：如果參賽者足夠重視未來的效用，那麼，對於所有滿足強連結（strong connectedness）的型態分布，對於所有的網路結構，都可以找到一組均衡使得參賽者在有限期數內達成事後柏拉圖效率配置。在這裡，強連結的定義是說，對任兩個想參加此集體行為的參賽者，必定有一條路徑可以連結他們，而且此路徑上的每個節點皆是想參加此集體行為的參賽者。換句話說，如果網路是非環形的，且想參加此集體行為的參賽者之間必定有一條路徑可供他們以行動溝通，那麼他們必定有一組策略帶領學習到關於此集體行為的知識。 但是，如果網路是環形的，就可能產生搭便車問題（free rider problem）。因為可能存在搭便車問題，環形網路的解決方法比起非環形網路更為困難。這個計畫的意圖是想找到處理環形網路的方法，希望對所有環形網路而言，都可以找到一組均衡，使得想參加此集體行為的參賽者最後都可以學習到關於此集體行為的知識。 <br> Abstract: This project studies a repeated collective action within cyclic networks, which aims to extend the result in Chen 2015. In this repeated action, Players only know their neighbors’ inclination to participate as well as monitor their neighbors’ past actions. Given that the networks are fixed, finite, connected, commonly known, undirected, and without circles, Chen 2015 shows that, for all priors with full support on the strong connectedness states, there is a (weak) sequential equilibrium path in which the ex-post efficient outcome repeats after a finite time T when discount factor is sufficiently high. Here, we say that a state has strong connectedness if and only if for every two players who incline to participate, there is a path consisting of players who have the same types to connect them. In other words, in tree network, Chen 2015 suggests that players can learn the relevant information if the routes for communication are not interrupted and if players are sufficiently patient. However, if the networks are with cycles, the free rider problem will occur. Due to the free rider problem, the proof to show the existence of such (weak) sequential equilibrium is much harder than the acyclic case. This project is meant to be a solution in finding such equilibria in the cyclic networks.網路協調問題集體行為重複賽局不完全資訊溝通networkcoordinationcollective actionrepeated gameincomplete informationcommunication