1: \begin{abstract}

2: We prove that the empirical spectral distribution of a $(d_L, d_R)$-biregular, 

3: bipartite random graph, under certain conditions, converges

4: to a symmetrization of the Mar\v{c}enko-Pastur distribution of random

5: matrix theory. This convergence is not only global (on fixed-length

6: intervals) but also local (on intervals of increasingly smaller length).  Our method parallels the one used previously by Dumitriu and Pal (2012).  

7:     %If $d=n^{o(1)}$ and $d\to\infty$, the 

8:    % normalized eigenvalues of a random $d$-regular graph converge to the

9:    % semicircle law  \cite{DuP}, just as with Wigner matrices.

10:    % We consider a random bipartite graph in which all vertices of one

11:    % vertex class have degree $d_L$, and all vertices of the other vertex

12:    % class have degree $d_R$.  By analogy with the case of

13:     %random $d$-regular graphs, the eigenvalues of these graphs should

14:    % be related to the Mar\v{c}enko-Pastur distribution.

15:    % We confirm this suspicion, proving that under similar conditions, 

16:     %the squares of the normalized eigenvalues of

17:    % a semiregular bipartite graph converge to the 

18:    % Mar\v{c}enko-Pastur distribution.

19:    % We also prove a short scale convergence law for the eigenvalues.

20:   \end{abstract}

21: