Evolutionary Game Dynamics And Game Mechanisms Based On Complex Networks

Evolutionary game is a new research field, which was developed from the classical game theory in economics. Due to its strict logic and mathematical framework, it has recently drawn great attention from many disciplines, such as mathematics, physics, and evolutionary biology. And there are many excellent research achievements till now. Compared with the study of classical game, we will pay more attention to evolution characteristics, dynamics process and co-development with other directions.At the same time, another face that can not be neglected is the existence of various complex networks, such as neural network, protein network, internet, communication network, traffic network, and so on. Hence, every one is an aggregation of networks or lies on a node of different networks. Naturally, based on these realistic networks and individual interaction scale, researchers introduce players of game theory into net-works and find that evolution puzzles could be effectively resolved. With physicists subsequently introducing nonlinear dynamics and other physics approaches into spatial game, the study of evolutionary game fully reaches a new peak. On the other hand, evolution biologists and experimental economists also do many evolutionary game ex-periments, which further perfect the study of evolutionary game.Combined with these considerations, the thesis mainly explores the study of evo-lutionary game from three aspects. Firstly, we study characteristics of evolutionary dy-namics in well-mixed populations. In previous works, players in game were regarded as perfectly rational individuals, which is obviously is inconsistent with realistic sit-uations. We begin with approximate best response dynamics, introduce the function expression of bounded rationality, and examine the evolution process. We find that compared with the results of classical replicator dynamics, the rock-scissor-paper cycle also occurs, but characteristic with evolution process (e.g. evolution way and equi-librium point) is different. Simultaneously, approximate best response dynamics can avoid the limitation possessed by replicator dynamics, and make the simulation results approach to experimental results. In addition, we investigate the mechanisms of spatial game on the complex net-works. Compared with classical game, every player only interacts with limited neigh-bors, and then updates its strategy according to relative dynamics. We introduce fitness mechanism and heterogeneity mechanism into spatial game, and find that cooperation can be improved greatly. Importantly, these facilitation mechanisms are robust for dif-ferent game models and networks, and they can be explained via exploring effective interaction structure and evolutionary characteristics during evolution process.Lastly, we study the coevolution of game which becomes a mushrooming avenue. The so-called coevolution does not only reflect the strategy updating, but also includes the evolvement of network topology or updating rule over time. We investigate the coevolution of game through considering aspiration-induced reconnection and incor-porating individual age in the evaluation of fitness. Compared with spatial effect alone, cooperation could be greatly improved.Since evolutionary game research belongs to the research of interdiscipline, it not only attracts interest of physicists, mathematicians and evolutionary biologists, but im-portantly it is related with many realistic questions in society, which also attracts great attention from sociologists and engineers. We believe that with the obtained prominent achievements, it will bring new energy to these fields.