Aspectos dinâmicos e estruturais em modelos de redes para sistemas complexos

Abstract

In this thesis we study systems where some form of disorder or non-homogeneity has a significant role at the complexity of the structural building or of the dynamics regulation of the system. First, we study the dynamics of Boolean networks, where the rules to update the state of the nodes are randomly chosen and control the global behavior of the system. At the critical threshold, and near to it, we propose that the transition to the critical regime can be characterized by the divergence of the relaxation time Tr. Based on simple scaling arguments, we show that the cumulative probability distribution of Tr decays as a power-law , with exponent iqual -1, for the annealed model at the critical region. Then, we study a novel method for network decomposition, which we apply to scale-free networks, that have the broad degree distribution as a fundamental feature. This method consists in a simultaneous and iterative remotion of groups of nodes with degree K until there are no more nodes with this degree in the network. Thus, we define new variables that characterize the process of decomposition and we obtain a set of well define exponents and parameters. From the behavior of these variables we can see, through some mathematical manipulations, that our method is self-consistent, serving as a useful tool for the study of the structural features of the network. At last, we study the backbones of the percolation cluster, where we use a network model with layers arranged in a disorderly way to represent some kind of anisotropy resistance to the percolation. Our numerical results indicate a break at the universality class on the fractal dimension and on the mass distribution of the backbones.