Emergence of communities in an Erdos-Renyi network

Abstract

We examine the emergence of community structure from an Erdos-Renyi network utilizing “preferential reattachment”, where each agent discon- nects from a nearest neighbor most different from itself, and connects to a next nearest neighbor to which it is most similar. The differences between two agents correspond to the absolute deviation between the ranks assigned to them initially. The ranks across the population are Gaussian- distributed with standard deviation σ. We find that for low-diversity populations, preferential reattachment causes the network to break into separate components, while for high-diversity populations, the network remains intact but a significant community structure is still attained. Our research provides insights on how agent diversity can affect the formation of community structures in a network where links are purely random.