Research On Robustness During Network Evolution

Many important complex system, such as Protein-Protein Interaction Net-work, Gene Regulatory Network, Internet, WWW, Power grid and Tra?c Net-work, exist in real world. These networks'normal performance are essential tothe stabilization of nature and human society. Studying on network robustness,which refers to the ability to maintain network function or properties under per-turbation, has important inspiration to the design of artificial complex systems.The key work of this thesis consists of several parts listed in the following:Firstly, we study the e?ects of network topology on the network robust-ness from an dynamic points of view. We use Boolean Network to model theRandom Graph and Scale-free Network, and explore robustness to two types ofperturbation on these two categories of networks during network evolution toperform some pre-established function.We demonstrated that scale-free networksperform inferior to homogenous random graph on mutational robustness, whichincrease the evolvability and facilitates the evolutionary search to perform thetarget function; while they are superior on robustness of attractors to randomgraph, which demonstrates the higher tolerance to state inversion.In the second place, we introduced switching target function during evolu-tion, and investigate the network robustness of the two categories of networks.We found that the switching way of evolution improves the mutational robust-ness of the two types of networks, and Scale-free Network increase more; but therobustness of attractors of the two networks decreases. We concluded that theswitching way of evolution facilitates the networks to form some specific networkmotives with robust topology, which increase the mutational robustness of thesetwo networks, and balance the robustness and evolvability.