The Research Of Forming Complex Network Based On Coordination Game

In society, friendly individuals tend to have the same hobbies. So that they are more likelyto take part in many activities together and get benefits. But due to the restriction of variousfactors, the individuals maybe can’t find the suitable group, and then these individuals wouldchange their strategies by adjusting their appearances to find like-minded friends withthemselves. So as continually join and dynamically game among individuals, some networkstructure would be changed. At the meantime, the change of network structure also makes adifference on the process of coordination game. Therefore, how to influence each other betweenthem and some key factors’ influence have become a new research subject.Complex networks are widely existed in nature and human society. The development ofcomplex network theory provides a well application framework for the performance of theevolutionary game. Nodes of the network present the individuals in society. The game betweenindividuals is affected by the network topology structure and the rules of the game. And edges ofthe network present relative relationship among individuals. The nodes with large degrees aremore likely to be selected to cooperation, instead that the nodes with small degrees are morelikely to be disconnected.The evolutionary game on the complex network focuses on the cooperation amongindividuals. By the establishment of some game rules to study that how to interact between theevolutionary game and network structure, to understand the formation of the complex networkcommunity laid a solid foundation.We propose a new model and algorithm to form a community-like network by imitating theprocess of agents who have common interests getting together. The interaction among agents canbe depicted by coordination game. Accompanying with network evolution, agents’ positionskeep changing and network scale grows gradually by adding new agents. In the process of theinteraction among individuals, by the measures of game learning and adjusting the networkconnection to maximize personal profits, so as to promote the cooperation among the individuals.At the same time, the structure of the network model in the process of individual game takes acertain degree of change. The cluster of game strategies among individuals, and the emergenceof the network community structure both show agents’ strategies and network structure changecontinuously and form the phenomenon of co-evolution. The results of theoretical analysis andsimulation experiments show that the network structure has strayed from the initial network onthe degree distribution and other features. Through constant evolution and game, both thecooperation rate and average payoff among the individuals are larger and occur the phenomenonof the individual strategies cluster that is communization. Firstly, the degree distribution of the network with constant nodes doesn’t follow Poisson distribution, and tend to the network of BAscale-free network model. There are some nodes with larger degrees in the network. In addition,the average degrees of the network nodes also present the phenomenon of rising rapidly, fallingfast and rising stable. Accompanied by the occurrence of evolution, the most nodes’ neighbors ofthe network have changed. And then the network model with the changed number of nodes issimilar to the real network in respect of both degree distribution and several characteristics. Thehigh cooperation rate among agents demonstrates the community-like gather. Secondly, theinfluence on network construction and degree of community from key parameters is explored.The parameter of adjusting network link can promote the cooperation among agents. The rate ofcooperation between the individuals is the biggest when the parameters of adding edges anddecrease edges in a middle value. They are too big or too small that are not conducive tocooperation between the agents. The parameter of game learning can accelerate communityindependence and make closer connections within the community. The agents within samecommunity tend to have higher the rate of cooperation, but it is not conducive to communicationamong the community. Finally, GN community split algorithm is applied to divide thewell-formed network and the results indicate the number of community keeps in accordance withthe scale of strategy space in coordination game.