The Impact of Network Degree Correlation on Parrondo's Paradox

摘要

A multi-agent Parrondo's model based on complex networks is studied to analyze the influence of network degree correlation on Parrondo's paradox. The model includes a zero-sum Game A, representing cooperation and competition behaviors between agents, and a two-branch Game B, representing the capital of a node and all of its neighbors. Then the parameter space of Parrondo's paradox pertaining to cooperation and competition behavioral patterns, and the gradual change of the parameter space from an assortative random network to a disassortative random network is analyzed. The simulation results suggest that the size of the region of the parameter space that elicits Parrondo's paradox is negatively correlated with the degree correlation of the network. For two distinct sets of probability parameters, the microcosmic reasons underlying the occurrence and non-occurrence of the paradox under the disassortative and assortative random network are elaborated, respectively. Common interaction mechanisms of the behavioral patterns, the asymmetric structure of Game B, and network topology are also revealed.