On $r$-to-$p$ norms of random matrices with nonnegative entries: Asymptotic normality and $\ell_\infty$-bounds for the maximizer

Souvik Dhara,Debankur Mukherjee,Kavita Ramanan

2024-03-16

Abstract:For an $n\times n$ matrix $A_n$, the $r\to p$ operator norm is defined as $$\|A_n\|_{r\to p}:= \sup_{\mathbf{x}\in\mathbb{R}^n:\|\mathbf{x} \|_r\leq 1 } \|A_n\mathbf{x} \|_p\quad\text{for}\quad r,p\geq 1.$$ For different choices of $r$ and $p$, this norm corresponds to key quantities that arise in diverse applications including matrix condition number estimation, clustering of data, and construction of oblivious routing schemes in transportation networks. This article considers $r\to p$ norms of symmetric random matrices with nonnegative entries, including adjacency matrices of Erdős-Rényi random graphs, matrices with positive sub-Gaussian entries, and certain sparse matrices. For $1<p\leq r<\infty$, the asymptotic normality, as $n\to\infty$, of the appropriately centered and scaled norm $\|A_n\|_{r\to p}$ is established. When $p \geq 2$, this is shown to imply asymptotic normality of the solution to the $\ell_p$ quadratic maximization problem, also known as the $\ell_p$ Grothendieck problem. Furthermore, a sharp $\ell_\infty$-approximation bound for the unique maximizing vector in the definition of $\|A_n\|_{r\to p}$ is obtained, and may be viewed as an $\ell_\infty$-stability result of the maximizer under random perturbations of the matrix with mean entries. This result is in fact shown to hold for a broad class of deterministic sequences of matrices having certain asymptotic expansion properties. The results obtained can be viewed as a generalization of the seminal results of Füredi and Komlós (1981) on asymptotic normality of the largest singular value of a class of symmetric random matrices. In the general case with $1<p\leq r< \infty$, spectral methods are no longer applicable, and so a new approach is developed involving a refined convergence analysis of a nonlinear power method and a perturbation bound on the maximizing vector, which may be of independent interest.

Combinatorics

What problem does this paper attempt to address?

The main problems that this paper attempts to solve are the asymptotic normality of the random matrix \(r \to p\) norm and the \(\ell_\infty\)-approximation bound of the maximizing vector. Specifically:

1. **Asymptotic normality**:
- For an \(n\times n\) symmetric non - negative random matrix \(A_n\), its \(r \to p\) norm is defined as:
\[
\|A_n\|_{r \to p}:=\sup_{x\in\mathbb{R}^n:\|x\|_r\leq1}\|A_nx\|_p\quad\text{for }r, p\geq1
\]
- The paper proves that when \(1 < p\leq r<\infty\), the appropriately centered and scaled \(r \to p\) norm \(\|A_n\|_{r \to p}\) is asymptotically normally distributed. The specific result is:
\[
\frac{1}{\sigma_n}(\|\bar{A}_n\|_{r \to p}-\alpha_n(p, r))\xrightarrow{d}Z\sim\text{Normal}(0, 2)
\]
where \(\bar{A}_n := n^{-(1/p - 1/r)}A_n\), and
\[
\alpha_n(p, r):=(n - 1)\mu_n+\frac{1}{2}\left(\frac{1}{p - 1}+\frac{1}{r - 1}\right)\frac{\sigma_n^2}{\mu_n}
\]

2. **\(\ell_\infty\)-approximation bound of the maximizing vector**:
- The paper also provides the \(\ell_\infty\)-approximation bound of the maximizing vector \(v_n\). Specifically, for \(1 < p\leq r<\infty\), the maximizing vector \(v_n\) satisfies:
\[
\|v_n - n^{-1/r}\mathbf{1}\|_\infty\leq C\cdot n^{-1/r}\sqrt{\frac{\log n}{n}}\frac{\sigma_n^2}{\mu_n}
\]
where \(\mathbf{1}\) is the all - ones vector and \(C\) is a constant.

These results not only extend the classical results of Füredi and Komlós on the asymptotic normality of the largest eigenvalue, but also provide a new method for dealing with the general \(r \to p\) norm problems, especially when \(p\neq r\), the traditional spectral methods are no longer applicable. In addition, these results are of great significance in statistics and machine learning, especially when dealing with sparse matrices and random perturbations.