Universality of Random-Matrix Results for Non-Gaussian Ensembles

We study random-matrix ensembles with a non-Gaussian probability distribution $P\left(H\right)\sim \exp [-N\mathrm{tr}V(H\left)\right]$, where $N$ is the dimension of the matrix $H$ and $V\left(H\right)$ is independent of $N$. Using Efetov's supersymmetry formalism, we show that in the limit $N\to \infty$ both energy level correlation functions and correlation functions of $S$-matrix elements are independent of $P\left(H\right)$ and hence universal on the scale of the local mean level spacing. This statement applies to each of the three generic ensembles (unitary, orthogonal, and symplectic). Universality is also found for correlation functions depending on some external parameter. Our results generalize previous work by Brezin and Zee [Nucl. Phys. B402, 613 (1993)].