基于零行列式策略的网络演化博弈动力学研究

项目名称： 基于零行列式策略的网络演化博弈动力学研究

项目编号： No.61473060

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 自动化技术、计算机技术

项目作者： 荣智海

作者单位： 电子科技大学

项目金额： 82万元

中文摘要： 最近迭代囚徒困境博弈研究发现，通过零行列式(ZD)策略可以单方面保证双方期望收益满足线性关系。ZD策略理论为刻画博弈双方作用关系提供了全新的研究视角，将根本改变博弈理论的研究范式。本项目拟系统开展ZD策略的性质、演化稳定性和网络动力学研究。首先，基于两人/多人博弈模型，分析ZD策略可行域；从参数演化角度研究ZD策略中三种重要策略：设定策略、剥削策略和善良策略之间的关系，探索重要ZD策略。然后，针对重要ZD策略，研究其演化稳定性及与种群规模、策略演化时间尺度的关系。在此基础上系统研究网络上的ZD策略：在静态网络上分析各种重要结构对于ZD策略的作用机理，从固定和演化角度研究时间尺度对ZD策略的影响；在动态网络上从移边和空间移动两种模式，分析ZD策略与网络结构的协同演化机制。通过研究与网络系统自适应的ZD策略，揭示群体合作起源与社会网络演化的机理,为多智能体系统优化提供理论和方法上的指导。

中文关键词： 复杂网络；演化博弈；群体动力学；零行列式策略

英文摘要： Recent researches on the iterated Prisoner's Dilemma game have found a new class of strategies, so-called zero-determinant (ZD) strategies, which can enforce a fixed linear relationship between players' expected payoff in a unilateral way. The ZD strategies theory provides a novel viewpoint to depict interaction among players, which fundamentally changes the research paradigm of game theory. This project will systematically study the properties of ZD strategies, analyze their evolutionary stability and networked dynamics. Firstly, based on two-player/multi-player game models, we will analyze the feasible region of ZD strategies, study three kinds of key ZD strategies (pinning strategy, extortion strategy and good strategy), investigate their relationships via parameters evolution, and explore important ZD strategies. Then we will study the evolutionary stability of important ZD strategies and explore their relationships with population size and strategy-evolving time scale. Based on these, we will systematically study ZD strategies on networks. For the fixed network, we will analyze the mechanism of various important structures to ZD strategies, and study the influence of both fixed and evolving time scale to ZD strategies. For the dynamic networks, we will analyze the coevolution between ZD strategies and network structures via two modes, i.e., removing edge mode and migration mode. The investigation about the ZD strategies adapted to networked system will conduce to deep understanding to the origin of population cooperation and the evolution of social networks, and provide theoretical and methodological guidance to the optimization of multi-agent systems.

英文关键词： Complex networks;Evolutionary game;Population dynamics;Zero-determinant strategies