1: \begin{abstract}

2:   Consider a square random matrix with independent and identically distributed

3:   entries of mean zero and unit variance. We show that as the dimension tends

4:   to infinity, the spectral radius is equivalent to the square root of the

5:   dimension in probability. This result can also be seen as the convergence of

6:   the support in the circular law theorem under optimal moment conditions. In

7:   the proof we establish the convergence in law of the reciprocal

8:   characteristic polynomial to a random analytic function outside the unit

9:   disc, related to a hyperbolic Gaussian analytic function. The proof is short

10:   and differs from the usual approaches for the spectral radius. It relies on

11:   a tightness argument and a joint central limit phenomenon for traces of

12:   fixed powers.

13: \end{abstract}

14: