Full text
| |
| Author(s): |
Alexandre Hannud Abdo
Total Authors: 1
|
| Document type: | Doctoral Thesis |
| Press: | São Paulo. |
| Institution: | Universidade de São Paulo (USP). Instituto de Física (IF/SBI) |
| Defense date: | 2009-09-18 |
| Examining board members: |
Alessandro Paulo Sérvio de Moura;
Nestor Felipe Caticha Alfonso;
Tania Tome Martins de Castro;
Osame Kinouchi Filho;
Renato Vicente
|
| Advisor: | Alessandro Paulo Sérvio de Moura |
| Abstract | |
This thesis presents a study of models of network contagion and network growth, relating topological properties of the network with dynamical properties of the evolution. Of network growth models, an extension to the concept of clustering coefficient was applied to better determine the validity of some models, demonstrating the practical use of this extension and revealing the non-trivial role of the cycle topology between real-world networks and models. Of the network contagion models, an extension of Dodds-Watts generalized model of contagion was used to approach the practical problem of planning seeding strategies for large-scale interventions, where one wishes to infect some population through diffusion from a smaller initially infected group. It is shown that the extension introduced, which parametrizes the ratio between the time invested by the influential and the influenced, determines the direction and intensity of the relationship between the degree of the seeds and the success of strategies. This second problem can also be interpreted as a study of the stability with respect to initial conditions, where the relevant variable, the infected set, takes values on the subsets of the graphs vertices. (AU) | |