Invertibility of Sparse non-Hermitian matrices

Abstract

We consider a class of sparse random matrices of the form $A_n =(\xi_{i,j}\delta_{i,j})_{i,j=1}^n$, where $\{\xi_{i,j}\}$ are i.i.d.~centered random variables, and $\{\delta_{i,j}\}$ are i.i.d.~Bernoulli random variables taking value $1$ with probability $p_n$, and prove a quantitative estimate on the smallest singular value for $p_n = \Omega(\frac{\log n}{n})$, under a suitable assumption on the spectral norm of the matrices. This establishes the invertibility of a large class of sparse matrices. For $p_n =\Omega( n^{-\alpha})$ with some $\alpha \in (0,1)$, we deduce that the condition number of $A_n$ is of order $n$ with probability tending to one under the optimal moment assumption on $\{\xi_{i,j}\}$. This in particular, extends a conjecture of von Neumann about the condition number to sparse random matrices with heavy-tailed entries. In the case that the random variables $\{\xi_{i,j}\}$ are i.i.d.~sub-Gaussian, we further show that a sparse random matrix is singular with probability at most $\exp(-c n p_n)$ whenever $p_n$ is above the critical threshold $p_n = \Omega(\frac{ \log n}{n})$. The results also extend to the case when $\{\xi_{i,j}\}$ have a non-zero mean. We further find quantitative estimates on the smallest singular value of the adjacency matrix of a directed Erd\H{o}s-R\'{e}yni graph whenever its edge connectivity probability is above the critical threshold $\Omega(\frac{\log n}{n})$.