arXiv++ Combinatorics

We introduce a new technique to prove bounds for the spectral radius of a random matrix, based on using Jensen's formula to establish the zerofreeness of the associated characteristic polynomial in a region of the complex plane. Our techniques are entirely non-asymptotic, and we instantiate it in three settings: (i) The spectral radius of non-asymptotic Girko matrices -- these are asymmetric matrices $\mathbf{M} \in \mathbb{C}^{n \times n}$ whose entries are independent and satisfy $\mathbb{E} \mathbf{M}_{ij} = 0$ and $\mathbb{E} |\mathbf{M}_{ij}^2| \le \frac{1}{n}$. (ii) The spectral radius of non-asymptotic Wigner matrices -- these are symmetric matrices $\mathbf{M} \in \mathbb{C}^{n \times n}$ whose entries above the diagonal are independent and satisfy $\mathbb{E} \mathbf{M}_{ij} = 0$, $\mathbb{E} |\mathbf{M}_{ij}^2| \le \frac{1}{n}$, and $\mathbb{E} |\mathbf{M}_{ij}^4| \le \frac{1}{n}$. (iii) The second eigenvalue of the adjacency matrix of a random $d$-regular graph on $n$ vertices, as drawn from the configuration model. In all three settings, we obtain constant-probability eigenvalue bounds that are tight up to a constant. Applied to specific random matrix ensembles, we recover classic bounds for Wigner matrices, as well as results of Bordenave--Chafaï--García-Zelada, Bordenave--Lelarge--Massoulié, and Friedman, up to constants.

Given an arbitrary subgraph $H=H_n$ and $p=p_n \in (0,1)$, the planted subgraph model is defined as follows. A statistician observes the union a random copy $H^*$ of $H$, together with random noise in the form of an instance of an Erdos-Renyi graph $G(n,p)$. Their goal is to recover the planted $H^*$ from the observed graph. Our focus in this work is to understand the minimum mean squared error (MMSE) for sufficiently large $n$. A recent paper [MNSSZ23] characterizes the graphs for which the limiting MMSE curve undergoes a sharp phase transition from $0$ to $1$ as $p$ increases, a behavior known as the all-or-nothing phenomenon, up to a mild density assumption on $H$. In this paper, we provide a formula for the limiting MMSE curve for any graph $H=H_n$, up to the same mild density assumption. This curve is expressed in terms of a variational formula over pairs of subgraphs of $H$, and is inspired by the celebrated subgraph expectation thresholds from the probabilistic combinatorics literature [KK07]. Furthermore, we give a polynomial-time description of the optimizers of this variational problem. This allows one to efficiently approximately compute the MMSE curve for any dense graph $H$ when $n$ is large enough. The proof relies on a novel graph decomposition of $H$ as well as a new minimax theorem which may be of independent interest. Our results generalize to the setting of minimax rates of recovering arbitrary monotone boolean properties planted in random noise, where the statistician observes the union of a planted minimal element $A \subseteq [N]$ of a monotone property and a random $Ber(p)^{\otimes N}$ vector. In this setting, we provide a variational formula inspired by the so-called "fractional" expectation threshold [Tal10], again describing the MMSE curve (in this case up to a multiplicative constant) for large enough $n$.

Motivated by the community detection problem in Bayesian inference, as well as the recent explosion of interest in spin glasses from statistical physics, we study the classical Glauber dynamics for sampling from Ising models with sparse random interactions. It is now well-known that when the interaction matrix has spectral diameter less than $1$, Glauber dynamics mixes in $O(n\log n)$ steps. Unfortunately, such criteria fail dramatically for interactions supported on arguably the most well-studied sparse random graph: the Erdős--Rényi random graph $G(n,d/n)$, due to the presence of almost linearly many outlier eigenvalues of unbounded magnitude. We prove that for the \emph{Viana--Bray spin glass}, where the interactions are supported on $G(n,d/n)$ and randomly assigned $\pmβ$, Glauber dynamics mixes in $n^{1+o(1)}$ time with high probability as long as $β\le O(1/\sqrt{d})$, independent of $n$. We further extend our results to random graphs drawn according to the $2$-community stochastic block model, as well as when the interactions are given by a "centered" version of the adjacency matrix. The latter setting is particularly relevant for the inference problem in community detection. Indeed, we use this to show that Glauber dynamics succeeds at recovering communities in the stochastic block model in a companion paper [LMR+24]. The primary technical ingredient in our proof is showing that with high probability, a sparse random graph can be decomposed into two parts -- a \emph{bulk} which behaves like a graph with bounded maximum degree and a well-behaved spectrum, and a \emph{near-forest} with favorable pseudorandom properties. We then use this decomposition to design a localization procedure that interpolates to simpler Ising models supported only on the near-forest, and then execute a pathwise analysis to establish a modified log-Sobolev inequality.