Study Of Dynamics In Game Systems On Complex Networks

The intention of this paper is to study three important complex systems, including minority game, animal conflicts and Chinese words system. Based on the complex network theory, it is focused on system model, evolution dynamics and potential applications. Minority game and animal conflicts systems are chosen to research strategies' dynamics on the complex networks. Chinese words network is selected for its potential applications.Firstly, we study the effects of the existence of another type of agents, called spies, in the minority game (MG).Unlike the normal agents in the MG, the spies do not carry any strategy. Instead, they decide their action by scouting some normal agents and take the minority action of the spied group. For a few spies and when there is useful information in the normal agents' actions, the spies can avoid the crowd effect of the normal agents and win more readily. When information becomes less useful and when more spies are present, the spies' crowd effect hurts the success rate of the spies themselves, and the normal agents could have a higher success rate than the spies. More spies actually assist more normal agents to win, as the spies also provide more winning quotas. This leads to a nonmonotonic behavior in the total success rate of the population as a function of the fraction of spies. This model has been focused by Nature China when it was published in PRE.And then, the spy model is studied on the complex networks according to different picking methods used by the spies, including random and preference strategies. When the spies scout normal agents randomly, they got more information than preference strategy, so the pay-off is higher than the other strategy's and the system standard deviation decreases significantly. But more information will make spies' action similar and the crowd effect will cut down spies' success rate. This also leads to a nonmonotonic behavior in the total success rate of the spy as a function of the information they got. In addition, because the nomal agents can not judge whether they are scouted or not by their payoff fluctuation, a new parameter is designed to monitor spy's quantity in the system.Secondly, we simulated animal conflicts on different networks, where five strategies that the animals may take are considered. The result of the evolution of the five strategies on networks shows that whether one strategy dominates or two strategies coexist on the network is determined by the structure of the network. But no matter what structure the network is, the total-war strategy is constrained and never becomes a final winning strategy when it contests with the other four limited-war strategies. This may be the reason that the animals choose the limited-war strategies to fight against other animals of the same species. On the other hand, a new multi-body conflict model is given to simulate more complex animal combats.Finally, we defined two kinds of Chinese words network (CWN) in this paper. The nodes in the network are composed by the Chinese characters, phrases or classical idioms from authority dictionaries. Studying the network characters and giving a new evolution model to simulate the CWN, we found that the phrase construction follows random and preferential choosing method and we found 55 Chinese characters used by ten thousand distinct phrases. The Chinese words network exhibiting Smallworld character will make it easy to search information as fast as English language.