Axelrod竞赛.pptVIP

* * * * * * * * * 张醒洲，大连 * 张醒洲，大连 * 非完全信息动态博弈 Axelrod’s 竞赛 * 张醒洲，大连 * 概 要 Axelrod’s 竞赛 一个理性的参与人和从一组人中抽取的一个人 两阶段一般囚徒困境 三阶段一般囚徒困境 * 张醒洲，大连 * Axelrod’s竞赛: 第一轮 Robert Axelrod is a political scientist at the Univ. of Michigan. In the late 1970s, he invited some economists, psychologists, mathematicians, and sociologists familiar with the repeated Prisoner’s Dilemma to submit strategies (written in computer code) for a finitely repeated Prisoner’s Dilemma with payoffs of (3,3) for (C,C),(5,0) for (F,C), (0,5) for (C,F) and (1,1) for (F,F). He received 14 entries, which he pitted against each other and against a strategy that randomly chooses C and F each with probability 1/2, in 200-fold repetitions of the game. He paired each strategy against each of the others five times. Strategies could involve random choices, so a pair of strategies could generate different outcomes when paired repeatedly. The strategy with the highest payoff was tit-for-tat, submitted by Anatol Rapoport, a member of the Department of Psychology of the Univ. of Tornoto. Axelrod’s竞赛: 第一轮 20世纪70年代后期，密歇根大学的政治科学家Robert Axelrod邀请 了一些熟悉重复囚徒困境博弈的经济学家、心理学家、数学家和社会学 家进行一个竞赛，参与竞赛的人以计算机程序的形式提交自己的策略。 他收到了14个程序，自动运行这些程序，与其他程序进行博弈,包括一个博弈，其中参与人分别以1/2的概率选择策略 C (合作)和 F(坦白)。博弈重复进行200次，博弈的胜利者是所有博弈中的支付累计值最高的策略。 最后获胜的策略是针锋相对策略，该策略由多伦多大学的心理学家 Anatol Rapoport提交。 * 张醒洲，大连 * * 张醒洲，大连 * 针锋相对策略 针锋相对(Tit-for-Tat) 策略 它使重复博弈以相互合作开始并在其后模仿对方的行动 C: C D: D ( ?, D ) ( ?, C ) * 张醒洲，大连 * Axelrod’s 竞赛: 第二轮 Axelrod subsequently ran a second tournament. He invited the participants in the first tournament to compete again and also recruited entrants by advertising in journals read by microcomputer users (a relatively small crowed in the early 1980s). Contestants were informed of the results of the first round. Sixty-two strategies were submitted. The contest was run slightly different this time: the length of each game was determined probabilistically. Again tit-for-tat (again submitted by