English

Low-Degree Hardness of Detection for Correlated Erd\H{o}s-R\'enyi Graphs

Data Structures and Algorithms 2025-11-11 v1 Probability Statistics Theory Statistics Theory

Abstract

Given two Erd\H{o}s-R\'enyi graphs with nn vertices whose edges are correlated through a latent vertex correspondence, we study complexity lower bounds for the associated correlation detection problem for the class of low-degree polynomial algorithms. We provide evidence that any degree-O(ρ1)O(\rho^{-1}) polynomial algorithm fails for detection, where ρ\rho is the edge correlation. Furthermore, in the sparse regime where the edge density q=n1+o(1)q=n^{-1+o(1)}, we provide evidence that any degree-dd polynomial algorithm fails for detection, as long as logd=o(lognlognqlogn)\log d=o\big( \frac{\log n}{\log nq} \wedge \sqrt{\log n} \big) and the correlation ρ<α\rho<\sqrt{\alpha} where α0.338\alpha\approx 0.338 is the Otter's constant. Our result suggests that several state-of-the-art algorithms on correlation detection and exact matching recovery may be essentially the best possible.

Keywords

Cite

@article{arxiv.2311.15931,
  title  = {Low-Degree Hardness of Detection for Correlated Erd\H{o}s-R\'enyi Graphs},
  author = {Jian Ding and Hang Du and Zhangsong Li},
  journal= {arXiv preprint arXiv:2311.15931},
  year   = {2025}
}

Comments

40 pages