Study On Several Dynamic Processes On Complex Networks

In recent years, academic research on complex networks is in the ascendant. Almost all of the complex systems made of interacting particles can be characterized by complex networks. Complex network exists in both human society and nature. Some infrastructure that are closely related with us, such as power systems, Internet, telephone communication networks, transportation networks, etc., can be described by complex networks. Besides these man-made systems, human society itself is a huge complex networks. From the different perspectives of studies, human social networks can be seen as friend networks, sexual networks, epidemic networks, and scientific networks. In the biological systems, networks still exist, such as cell networks, protein-protein interaction networks, neural networks. Therefore, the study of complex networks is of great significance for human life and work, and for the understanding of nature and evolution of human society.The study of complex systems by complex networks adopts overall point of view. By studying the microscopic features and interactions among individuals, we can predict the overall macroscopic behavior of complex systems. Hence, network method provides us with a very useful tool to treat various social challenges. Now, the network research has penetrated into various fields of science, for example, sociology, ecology, natural subjects. Current research includes the network-based personalized recommendation, cooperative behavior and evolutionary game on complex networks, cascading failures on complex networks, and so on. With the current hotspots, the author of this thesis has done some relevant research follows.For the evolutionary game on complex networks, we have studied an evolutionary prisoner's dilemma game on a scale-free network by a modified Fermi updating rule, where each player is assigned an inertia index that controls its learning activity. An interesting finding is that the cooperation level can be significantly improved when the individual inertia is introduced. More importantly, a parameterβis also introduced to control the diversity level of inertia among the individuals. It is found that there exists an optimal value ofβleading to the highest cooperation level. The observed results are explained by the feedback mechanism and the characteristic of the Fermi function. Our analysis also reveals that the players with moderate degrees play a critical role in the evolution of the whole system. Besides this, we have also studed the effects of heritability on the evolution of spatial public goods games. In our model, the fitness of players is determined by the payoffs from the current interactions and their history. Based on extensive simulations, we find that the density of cooperators is enhanced by increasing the heritability of players over a wide range of the multiplication factor. We attribute the enhancement of cooperation to the inherited fitness that stabilizes the fitness of players, and thus prevents the expansion of defectors effectively.For the naming game on complex networks, we have investigated the naming game on small-world geographical networks. The small-world geographical networks are constructed by randomly adding links to two-dimensional regular lattices, and it is found that the convergence time is a nonmonotonic function of the geographical distance of randomly added shortcuts. Apart from this, we have proposed a negotiation strategy to address the effect of geography on the dynamics of naming games over small-world networks. Communication and negotiation frequencies between two agents are determined by their geographical distance in terms of a parameter characterizing the correlation between interaction strength and the distance. A finding is that there exists an optimal parameter value leading to fastest convergence to global consensus on naming.For the network-based recommendation, we present a recommendation algorithm based on the resource-allocation progresses on bipartite networks. In this model, each node is assigned an attraction that is proportional to the power of its degree, where the exponentβis an adjustable parameter that controls the configuration of attractions. In the resource-allocation process, each transmitter distributes its each neighbor a fragment of resource that is proportional to the attraction of the neighbor. Based on a benchmark database, we find that decreasing the attractions that the nodes with higher degrees are assigned can further improve the algorithmic accuracy. More significantly, numerical results show that the optimal configuration of attractions subject to accuracy can also generate more diverse and less popular recommendations. In addition, we propose a novel method to compute the similarity between congeneric nodes in bipartite networks. Different from the standard cosine similarity, we take into account the influence of a node's degree. Substituting this new definition of similarity for the standard cosine similarity, we propose a modified collaborative filtering (MCF). Based on a benchmark database, we demonstrate the great improvement of algorithmic accuracy for both user-based MCF and object-based MCF.