Edge universality of sparse Erdős-Rényi digraphs (2304.04723v4)

Abstract: Let $\mathcal A$ be the adjacency matrix of the Erd\H{o}s-R\'{e}nyi directed graph $\mathscr G(N,p)$. We denote the eigenvalues of $\mathcal A$ by $\lambda_1^{\cal A},...,\lambda^{\cal A}N$, and $|\lambda_1^{\cal A}|=\max_i|\lambda_i^{\cal A}|$. For $N^{-1+o(1)}\leq p\leq 1/2$, we show that [ \max{i=2,3,...,N} \bigg|\frac{\lambda_i^{\mathcal A}}{\sqrt{Np(1-p)}}\bigg| =1+O(N^{-1/2+o(1)}) ] with very high probability. In addition, we prove that near the unit circle, the local eigenvalue statistics of ${\mathcal A}/\sqrt{Np(1-p)}$ coincide with those of the real Ginibre ensemble. As a by-product, we also show that all non-trivial eigenvectors of $\mathcal A$ are completely delocalized. For Hermitian random matrices, it is known that the edge statistics are sensitive to the sparsity: in the very sparse regime, one needs to remove many noise random variables (which affect both the mean and the fluctuation) to recover the Tracy-Widom distribution. Our results imply that, compared to their analogues in the Hermitian case, the edge statistics of non-Hermitian sparse random matrices are more robust. The edge of non-Hermitian matrices possesses the cusp singularity, which was believed to be a technical difficulty of the model. Our first novelty is the observation that the cusp singularity is in fact an advantage instead of an obstacle, and when used properly, it can make the computation easier for non-Hermitian matrices. The second novelty is an use of integration by parts formula inside the Girko's Hermitization, which completely avoids the study of the Green function at larger spectral scales. The third novelty is the self-similarity of the self-consistent equations of certain Green functions, which eliminates the effect of large expectations of the matrix entries.