一种行为识别声誉更新机制下的演化博弈特征  

Some characteristics of population evolutionary games under a rule of reputation updating with action discrimination

作　　者：董飞[1] 周永辉[1] 

机构地区：[1]贵州师范大学数学与计算机科学学院,贵州贵阳550001

出　　处：《贵州师范大学学报（自然科学版）》2013年第2期69-74,共6页Journal of Guizhou Normal University：Natural Sciences

基　　金：国家自然科学基金(11161011);贵州省自然科学基金[2010]C2147)资助

摘　　要：一次性囚徒困境博弈中,合作是不理性的行为。群体中要实现或维持合作需要有效的机制,间接互惠就是依赖于声誉可以实现合作的一种机制。首先针对不同行为施予者,给出一种间接声誉评价规则,然后基于一种声誉更新方程,讨论了群体策略的最佳回应区域、纳什平衡及最佳回应动力学演化特征,指出了存在对应的混合策略内部区域,经过长期演化,识别者或背叛者占据整个种群。Cooperating in prisoner＇s dilemma game is irrational and some supporting mechanism is needed to achieve or maintain cooperation. Indirect reciprocity based on reputation is one such mechanism. In this paper we first propose an evaluating rule of indirect reputation for all plays with different actions, that is, giving a good reputation to the only one who cooperate twice before playing a next game. Then we investigate some characteristics of the evolutionary games, including the origins of best reply mixed strategies to pure strategies, Nash equilibrium points and some best reply dynamics. It shows that there exists corresponding inner range of mixed strategies such that discriminators or defectors may dominate the whole population.

关 键 词：演化 背叛 合作 识别 最佳回应 

分 类 号：O178[理学—数学]