1: \begin{abstract}

2: %     We consider adjacency and Laplacian matrices of random Erd\H{o}s-R\'{e}nyi $r$-uniform hypergraphs on $n$ vertices, in the setting where $r$ can vary with $n$ such that $r / n \to c \in [0, 1)$. Adjacency matrices of hypergraphs are contractions of the adjacency tensor along the vector $\bone = (1, \ldots, 1)^{\top}$. The entries of such matrices exhibit long range correlations. We show that the expected empirical spectral distribution of appropriately centered and scaled hypegraph adjecency matrices converge weakly to the semi-circle law with variance $(1 - c)^2$. In the course of proving this, we establish a universality result under a Pastur-type condition, which turns out to be $d_{\avg}/r^7 \to \infty$. \SSM{Mention $d_{\avg}$.} We also establish exponential concentration inequalities for the empirical spectral measure in the regime $r/n \to 0$ and as a consequence, deduce in-probability (resp. almost sure) convergence of the ESD to the semi-circle law in the regime $r/ n \to 0$ (resp. $r \sqrt{\log n} / n \to 0$).

3: 

4: %     In contrast with the Erd\H{o}s-R\'{e}nyi random graph ($r = 2$), two eigenvalues stick out of the bulk of the spectrum of the normalised adjacency matrix. In a Gaussian version of the model we show that an appropriately scaled largest eigenvalue converges in probability to $\frac{c}{2} + \sqrt{c - \frac{3c^2}{4}}$. When $c = 0$, we need a rescaling and the rescaled largest eigenvalue converges in probability to $1$. We prove similar results for the smallest eigenvalue. We also give estimates on the rates of convergence.

5: 

6: %     We establish analogous results for the bulk and edge eigenvalues of the associated Laplacian matrices.

7: % \end{abstract}

8: