Phase Transitions in Planted k-Factor Recovery

Abstract

This paper studies the problem of inferring a $k$-factor, specifically a spanning $k$-regular graph, planted within an Erdos-Renyi random graph $G(n,\hat{I}»/n)$. We show that as the average degree $\hat{I}»$ surpasses the critical threshold of $1/k$, the inference problem undergoes a transition from almost exact recovery to partial recovery. Moreover, as $\hat{I}»$ tends to infinity, the accuracy of recovery diminishes to zero. In addition, we characterize the recovery accuracy of a linear-time iterative pruning algorithm and show that it achieves almost exact recovery when $\hat{I}»<1/k$. A key component of our analysis is a two-step cycle construction: we first build trees through local neighborhood exploration and then connect them by sprinkling using reserved edges. Interestingly, for proving impossibility of almost exact recovery, we construct $\hat{I}˜(n)$ many small trees of size $\hat{I}˜(1)$, whereas for establishing the algorithmic lower bound, a single large tree of size $\hat{I}˜(\sqrt{n\log n})$ suffices.

36 pages, 5 figures. Extended abstract appeared at the 2025 Conference on Learning Theory, Lyon, France; accepted for publication in the Annals of Applied Probability