Learning and Evolution in Games

游戏中的学习与进化

• 批准号: 0426199
• 负责人: Drew Fudenberg
• 金额: --
• 依托单位: Harvard University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2004
• 资助国家: 美国
• 起止时间: 2004-11-01 至 2008-10-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0426199&HistoricalAwards=false
• 关键词: Learning; Evolution; Games

This award will fund work in game theory. The major goal is to develop alternatives to the widely-used concept of Nash equilibrium. This concept and its refinements are the basis for most applications of game theory, yet it is still not clear what predictions to make in games with multiple Nash equilibriua, nor just what equilibrium concept is appropriate in games whose strategies correspond to contingent plans, as opposed to an uncontingent choice of an action. The starting point for this project is that equilibria in games arise as the long-run outcome of a non-equilibrium dynamic process of learning, imitation, or evolution. The project has two components. In the first, the PI will model rational learning by patient, long-lived individuals and will analyze what happens when these individuals interact with a succession of anonymous opponents. The second part will fous on the role of evolution or imitation as a dynamic process. Here individuals will imitate the behavior of successful players (or animals genetically programmed to play a successful strategy will have higher reproductive success). The major goal is to determine what will happen in a large but finite population of these individuals. This project will combine recent advances in economics with work in mathematical biology. The results will be of interest to scholars in many fields that use game theory, including political science, sociology, computer science, and psychology.This award is made as part of the FY 2004 Human and Social Dynamics(HSD) portfolio, following a competition that involved participationfrom all NSF Directorates and Offices. HSD awards are administeredthrough specific programs that may change during the award period;management of all HSD awards is coordinated on an NSF-wide basis.

该奖项将资助博弈论方面的工作。主要目标是发展替代广泛使用的纳什均衡概念。这一概念及其改进是大多数博弈论应用的基础，但我们仍然不清楚在具有多重纳什均衡的游戏中可以做出什么样的预测，也不清楚在策略对应于偶然计划（而不是非偶然行动选择）的游戏中，什么样的均衡概念是合适的。这个项目的出发点是，游戏中的平衡是学习、模仿或进化等非平衡动态过程的长期结果。该项目有两个组成部分。首先，PI将为耐心、长寿的个体的理性学习建模，并分析当这些个体与一连串匿名对手互动时会发生什么。第二部分将关注进化或模仿作为一个动态过程的作用。在这里，个体会模仿成功玩家的行为（或者动物的基因决定了一个成功的策略会有更高的繁殖成功率）。主要目标是确定在数量庞大但数量有限的这些个体中会发生什么。这个项目将把经济学的最新进展与数学生物学的工作结合起来。研究结果将引起许多使用博弈论领域的学者的兴趣，包括政治学、社会学、计算机科学和心理学。该奖项是2004财年人类与社会动力学（HSD）项目的一部分，此前所有NSF理事会和办公室都参与了竞争。HSD奖是通过特定的程序来管理的，在奖励期间可能会发生变化；所有HSD奖的管理都是在nsf范围内协调的。