Algoritmo ultra rápido para calcular propriedades críticas no problema de percolação em redes bidimensionais

Abstract

Through studies on percolation, can determine if a percolates dimensional network, covering only part of the borders of the agglomerates by checking if there are two border sites connected on opposite sides of the net, i.e. without the need to fill all the sites form agglomerates. The object of this thesis is an algorithm for this purpose. Given the speed that this algorithm will, covering only part of the borders of the clusters, we saw that it would be possible to study never reached sizes of networks (more than one trillion websites), with less complexity than 1 and a low computational cost compared to the algorithms already developed on the theme percolation. We pass, therefore, to want to study the behavior of percolation threshold and the fractal dimension of the border into networks of different sizes and with a lot of simulations, which results allowed comparisons and confirm the predictions made by laws scales known in the literature.