Graph spectra and the detectability of community structure in networks

We study networks that display community structure – groups of nodes within which connections are unusually dense. Using methods from random matrix theory, we calculate the spectra of such networks in the limit of large size, and hence demonstrate the presence of a phase transition in matrix methods for community detection, such as the popular modularity maximization method. The transition separates a regime in which such methods successfully detect the community structure from one in which the structure is present but is not detected. By comparing these results with recent analyses of maximum-likelihood methods we are able to show that spectral modularity maximization is an optimal detection method in the sense that no other method will succeed in the regime where the modularity method fails.

💡 Research Summary

The paper investigates the fundamental limits of community detection in networks that exhibit a modular structure, where groups of nodes are more densely connected internally than externally. Using the stochastic block model (SBM) with two equally sized communities, the authors decompose the adjacency matrix A into its expectation ⟨A⟩ and a zero‑mean random fluctuation X. The expectation matrix carries the low‑rank signal that encodes the community partition, while X behaves like a Wigner random matrix whose eigenvalue distribution follows the semicircle law.

A key insight is that the second eigenvalue λ₂ of ⟨A⟩ must separate from the bulk of X’s spectrum in order for any linear or nonlinear algorithm to recover the planted communities. The bulk edge is given by λ_c = 2σ, where σ² is the variance of the entries of X. The authors derive explicit expressions λ₂ = (p_in – p_out)N/2 and σ² =