复杂网络的结构与动力学分析-凝聚态物理专业论文.docx

ii ii Abstract It is recognized that complex network is novel and efficient tool for describing and studying the structures of various complex systems. Nowadays, the structure and dynamics of complex network, as an issue filling with physical ideas and principles, has become one of the hot research topics in statistical physics. These studies are of great significance for understanding both structures of complex systems and various dynamical behaviors on them, and then for developing the effective ways to control these behaviors. In this dissertation, our main works relating with three aspects: networked structures, dynamics on networks and interactions between structures and dynamics, are introduced and summarized as follows. Firstly, in the issue of networked structures, we focus our attentions on the recently rising branch: link prediction. Consider that previous algorithms of the link prediction can only predict the likelihood of the existence of a link, while cannot estimate the directions of links. Many real networks, however, have directed links. We extend twelve local- information-based similarity indices to the corresponding directed versions, and build up the basic framework of directed link prediction. In addition, based on the statistical analysis of network motifs, we construct a generalized common neighbour index, and proposed a simple combination of the preferential attachment index and the generalized common neighbour index. By testing the fourteen prediction algorithms designed by these indices on real-world directed networks and analyzing the prediction accuracies, some valuable conclusions are obtained. Particularly, due to the strongly improved prediction accuracy and low computational complexity, the generalized common neighbour index and the combining index may be indeed useful for link mining in various realistic networked systems. Secondly, dynamical behaviors or controls about four typical dynamical processes taking placing on complex networ