ANÁLISE DA ENTROPIA EM REDES COMPLEXAS

Abstract

In this thesis, we apply network entropy to analyze characteristics of complex networks. It is analyzed the effect of vertices removal on the structure and robustness of directed and undirected networks, as a result of failures and attacks. The variation in the structure and robustness of networks due to this removal is quantified by structural entropies with zero and one order approximation. These entropies are calculated to real networks to verify the robustness, where we have been using biological network as directed and literary network as undirected. The failures are simulated through vertices that are randomly removal, while the attacks are realized on the hubs and according to betweenness centrality. We observe that the network with failures exhibits entropies with values smaller than the original network. When the attack is on the hubs or according to the betweenness centrality we verify that the entropies decrease. However, if the attack is not on the hubs, it is possible to verify values of entropies larger than the original network. Moreover, the measure of the shortest path length showed a correlation with the structural entropy.