摘要
本文提出一种基于Mealy自动机的多人重复博弈演化模型。该模型用Mealy自动机表示博弈参与人的战略行为,并构造一个基于遗传算法的自动机演化程序。博弈初始时,博弈参与人提交一个Mealy自动机。博弈参与人的有限自动机在当前状态下与竞争对手的自动机进行博弈,并且依据其获取的支付转换到下一状态,然后在新的状态下开始新一周期的博弈。在进行一段时间博弈后,则利用自动机演化程序对自动机进行演化,演化的标准是依据其在这一段时间博弈中的获得的平均支付及战略成本。通过仿真可以验证,在参与人理性程度较低且存在信息干扰的情况下,博弈参与人为了降低在博弈过程中的信息失真率,战略成本需要以较大的幅度下降。
A multi-person repeated game evolution model base on mealy automata was proposed. In this model, the player＇s strategic behavior was expressed by Mealy automaton, and then the evolutionary program of automata based on genetic algorithm was constructed. At the beginning of game, the player submits a Mealy automaton. Mealy automaton was updated by evolution program of automaton after game for a period of time. Automaton evolution is according to mean payoff and strategic cost achieved in stage game. The strategic cost can be measured by the number of automaton＇s states. At last, the process of model and simulation based on Mealy automaton was illustrated though an example of iterated prisoner＇s dilemma game. In order to reduce distortion rate of information at the same time when the information is imperfect, the simulation results validate that it is necessary to cut down strategic cost significantly in the process of game.
出处
《管理工程学报》
CSSCI
北大核心
2010年第1期158-162,共5页
Journal of Industrial Engineering and Engineering Management
基金
国家自然科学基金资助项目(90510016)
关键词
有限理性
Mealy自动机
重复博弈
战略演化
仿真
bounded rationality
mealy automata
repeated game
strategic evolution
simulation
