English

Exact Matching of Random Graphs with Constant Correlation

Statistics Theory 2022-07-08 v2 Data Structures and Algorithms Probability Machine Learning Statistics Theory

Abstract

This paper deals with the problem of graph matching or network alignment for Erd\H{o}s--R\'enyi graphs, which can be viewed as a noisy average-case version of the graph isomorphism problem. Let GG and GG' be G(n,p)G(n, p) Erd\H{o}s--R\'enyi graphs marginally, identified with their adjacency matrices. Assume that GG and GG' are correlated such that E[GijGij]=p(1α)\mathbb{E}[G_{ij} G'_{ij}] = p(1-\alpha). For a permutation π\pi representing a latent matching between the vertices of GG and GG', denote by GπG^\pi the graph obtained from permuting the vertices of GG by π\pi. Observing GπG^\pi and GG', we aim to recover the matching π\pi. In this work, we show that for every ε(0,1]\varepsilon \in (0,1], there is n0>0n_0>0 depending on ε\varepsilon and absolute constants α0,R>0\alpha_0, R > 0 with the following property. Let nn0n \ge n_0, (1+ε)lognnpn1Rloglogn(1+\varepsilon) \log n \le np \le n^{\frac{1}{R \log \log n}}, and 0<α<min(α0,ε/4)0 < \alpha < \min(\alpha_0,\varepsilon/4). There is a polynomial-time algorithm FF such that P{F(Gπ,G)=π}=1o(1)\mathbb{P}\{F(G^\pi,G')=\pi\}=1-o(1). This is the first polynomial-time algorithm that recovers the exact matching between vertices of correlated Erd\H{o}s--R\'enyi graphs with constant correlation with high probability. The algorithm is based on comparison of partition trees associated with the graph vertices.

Keywords

Cite

@article{arxiv.2110.05000,
  title  = {Exact Matching of Random Graphs with Constant Correlation},
  author = {Cheng Mao and Mark Rudelson and Konstantin Tikhomirov},
  journal= {arXiv preprint arXiv:2110.05000},
  year   = {2022}
}

Comments

55 pages, 1 figure

R2 v1 2026-06-24T06:46:51.870Z