有限理性下的进化博弈与合作机制  

Evolutionary game and cooperation mechanisms based on bounded rationality

作　　者：王先甲[1] 刘伟兵[2] 

机构地区：[1]武汉大学经济与管理学院,武汉430072 [2]武汉大学政治与公共管理学院,武汉430072

出　　处：《上海理工大学学报》2011年第6期679-686,508,共8页Journal of University of Shanghai For Science and Technology

基　　金：国家自然科学基金资助项目(71071119)

摘　　要：在介绍传统博弈论基本假设的基础上,指出完全理性是传统博弈论均衡选择的最基本假设和完全理性在动态博弈中的局限,提出有限理性与动态学习是动态博弈中多个智能体选择的基础;探讨了不断改善有限理性的进化学习方法和智能体选择机制;解释了进化博弈与进化稳定策略的本质;指出了传统博弈论与进化博弈论的区别;提出了有限理性下进化博弈与合作机制研究的问题,给出了有限理性下进化博弈与合作机制的研究结果.The paper indicates that complete rationality is the basic assumption of traditional game theory and leads to limitations in dynamic game.It was pointed out that bounded rationality and replicator dynamics are the foundation of multi-person selection,and the evolutionary learning methods for improving bounded rationality and selection mechanism in dynamic game were explored.In addition,the essence of evolutionary game and evolutionary stable equilibrium was revealed and the difference between traditional game theory and evolutionary game theory was described.The problems to be studied about evolutionary game theory and cooperation mechanisms were proposed and some concerned research results under the assumption of bounded rationality were provided.

关 键 词：进化博弈 有限理性 强化学习 复制动态 进化稳定策略 

分 类 号：O225[理学—运筹学与控制论]