Parts | Days | Selection | Search | Updates | Downloads | Help

DY: Fachverband Dynamik und Statistische Physik

DY 67: Networks: From Topology to Dynamics (joint session DY/ BP/SOE)

DY 67.11: Talk

Friday, March 20, 2015, 12:15–12:30, BH-N 128

Efficient sampling of networks with high clustering — •Rico Fischer¹, Jorge Leitao¹, Tiago Peixoto², and Eduardo Altmann¹ — ¹Max-Planck-Institut für Physik komplexer Systeme — ²University of Bremen

The problem in network generation is to obtain networks which satisfy specified properties but that are otherwise random. Traditional Markov Chain Monte Carlo methods (like Metropolis-Hastings) can be used in this problem but often fail in important cases, e.g., they do not correctly sample random networks with high clustering coefficients due to a rough >> landscape, which typically leads to abrupt phase transitions, metastable states and hysteresis. In this talk we show how an efficient sampling of high-clustering networks is obtained using multicanonical Monte-Carlo methods. We characterize the efficiency of this method, we use it to investigate the phase transition methods, and we explore different applications.