复杂网络中基于记忆长度的合作行为

A memory-based cooperation behavior research in complex network

网络结构和群体总数对合作行为有重要影响.为进一步研究个体的学习行为和记忆长度如何影响群体的合作率,构建了三种不同类型的网络,采用基于代理计算实验方法,将网络的节点看作有限理性的个体,并构造了个体的适应函数和个体学习和策略的演化方式,研究系统的合作行为.计算实验结果表明偏好选择学习策略提高了无标度网络的合作水平,降低规则网络的合作水平.此时,记忆长度显著改变群体的合作水平,而个体采用随机选择策略时,记忆长度对群体的合作率产生很小的影响.另外,群体总数越高在无标度网络中合作水平越高,而小世界网络中合作水平越低,规则网络中则呈现不规则变化.从而证明偏好选择学习策略选择有一定的适应性,记忆长度和群体总数针对不同的学习策略和网络结构会带来截然相反的结果.

Different networks and group sizes have important influence on cooperation behavior. To further study how the individual learning behavior and memory size influence the cooperators, three heterogonous networks are established. On the basis of agent-based computational experiments, the vertexes of network are regarded as the bounded-rationality individual, and the fitness function and evolution of individual strategies and learning behavior are also constructed. The results show that preferential learning rule improves the cooperation level in BA scale-free network while reduces the regular networks. And in this environment, memory size changes the cooperation level in all the networks greatly, while has little impact when individual adopt random selection learning rule. Besides, the increase of the group size makes the cooperation level in BA scale-free network higher while it decreases the proportion of cooperators in small-world network and makes the irregular movement in the regular network. The results prove that preferential selection rule has its adaptability, memory length and group size can make inverse results in different network when different learning rules are used.