The Mean First Passage Time Of Random Walk In Large Scale Generic Complex Networks

Shi Pei-Teng,Meng Jun-Yong,Wang Jun,Huang Xiao-Han,Deng Su,Wu Ya-Hui,Zhang Jiang
2015-01-01
Abstract:Random walk' has been heavily investigated in the past years for its widely application in science and engineering. The Mean First Time (MFPT) is an important indicator in depicting the information diffusion process and underlying topology behind the complex networks, attracting more and more scholars' attention. However, many scholars neglect the direct flow and dissipation for each vertex in the previous random walk model and the computation of the fundamental matrix approach for MFPT is quite expensive when the network size is large. In this paper, we deduce another explicit expression of MFPT according to the Markov Process, which is the same with the conventional representation but own high computational efficiency. Based on the new expression of MFPT, we provide a sub-matrix module algorithm to solve the low memory problem in the large-scale MFPT calculation. Experiments on the clickstream network show the algorithm's feasibility. This paper provides a framework to conduct MFPT analysis in large-scale generic complex networks and may be meaningful to many research fields like the spread of public opinions, epidemics diffusion, community detection and so on. Keywords: random walk model, mean first passage time, computation efficiency, sub-matrix module algorithm, large-scale complex network
What problem does this paper attempt to address?