| Abstract 摘要 | Interdependent networks usually undergo catastrophic cascading failures triggered by the failures of a small fraction of nodes. The infrastructure system networks such as Internet and power grid systems are interdependent on each other and often have rich connectivity links within a region but less links among different regions, which can be regarded as communities. For instance, the products exchange network of cities between America and Asia, about 90% of the links are intra-continent links. Natural disasters such as earthquakes, tsunamis or typhoons usually destroy a local region and consequently influence the whole region. How the damage of a local community influences the whole interdependent system is a crucial question. In this talk, I will introduce our new results about the cascading failures of such a system through percolation theory. Our main findings are that certain interdependent network embeds an extreme risk solely due to its community structure from which reconnecting a little bit fraction of links may lead to an abrupt change of the robustness of the whole system. Moreover, we find that increasing the community structure salience dose not necessarily decreases the systematic risk of the interdependent networks, which are essentially different from that of single networks. |
