基于Agent的演化博弈下的合作行为研究

Cooperation under Agent-based Evolutionary Games

文章在演化博弈的视角下,利用重复囚徒困境博弈(IPD)对合作的演化进行分析。在平均场假设下,合作策略无法在竞争中生存。而在基于Agent的演化博弈中,合作策略有可能获得成功。其原因在于通过限制合作行为的范围,可以使合作者聚集成簇状,从而防止合作的红利被背叛者占有。通过理论推导与计算机仿真,文章证明了利他策略获得成功的条件是c/b/1/m,并对此结果做出了解释。

This article used IPD to analyze the evolution of eeoperafive behavior in the field of evolutionary game theory. Under the mean-field assumption, cooperator cannot survive in competition with defectors. However, in agent- based evolutionary game theory, cooperators can thrive. The reason for that is by restricting the scope of cooperation, cooperators can survive by living in clusters, thereby preventing the benefit of cooperation from being exploited by defectors. This article not only compares the difference between spatial evolu- tionary theory and classical theory, but also analyzes the conditions needed for the emergence and maintenance of cooperation using PD game simulation on a regular network.