1: \begin{abstract}

2:   We investigate the spectrum of the infinitesimal generator of the continuous

3:   time random walk on a randomly weighted oriented graph. This is the

4:   non-Hermitian random $n\times n$ matrix $L$ defined by $L_{jk}=X_{jk}$ if

5:   $k\neq j$ and $L_{jj}=-\sum_{k\neq j}L_{jk}$, where $(X_{jk})_{j\neq k}$ are

6:   i.i.d.\ random weights. Under mild assumptions on the law of the weights, we

7:   establish convergence as $n\to\infty$ of the empirical spectral distribution

8:   of $L$ after centering and rescaling. In particular, our assumptions include

9:   sparse random graphs such as the oriented Erd\H{o}s-R\'enyi graph where each

10:   edge is present independently with probability $p(n)\to0$ as long as

11:   $np(n)\gg (\log(n))^6$. The limiting distribution is characterized as an

12:   additive Gaussian deformation of the standard circular law. In free

13:   probability terms, this coincides with the Brown measure of the free sum of

14:   the circular element and a normal operator with Gaussian spectral measure.

15:   The density of the limiting distribution is analyzed using a subordination

16:   formula. Furthermore, we study the convergence of the invariant measure of

17:   $L$ to the uniform distribution and establish estimates on the extremal

18:   eigenvalues of $L$.

19: \end{abstract}

20: