Universality for orthogonal and symplectic Laguerre-type ensembles (Q2473342): Difference between revisions

Ensembles of matrices (matrix probability laws) \(\{M\}\) are considered, with invariant distributions of Laguerre type \[ dP_{n,\beta}(M)=Z_{n,\beta}^{-1}\text{det}(W_{\gamma}(M))e^{-\text{tr}Q(M)}dM.\eqno(1) \] For \(\beta=1,2,4\), the ensembles consist respectively of \(n\times n\) real symmetric matrices, \(n\times n\) Hermitian matrices and \(2n\times 2n\) Hermitian self-dual matrices and they are called orthogonal, unitary and symplectic ensembles. In (1) \(dM\) is the Lebesgue measure on the algebraically independent entries of \(M\), \(\gamma>0\), \(W_{\gamma}(x)=x^{\gamma}\mathbf{1}_{\mathbb{R}_+}(x)\), \(Q\) is a polynomial with positive leading coefficient and \(Z_{n,\beta}\) is a normalization constant. The main results of the paper state that in the cases \(\beta=1\) and \(\beta=2\) the appropriately rescaled local eigenvalue statistics for these ensembles are universal (i.e. independent of \(Q\)) in the limit \(n\rightarrow\infty\) (where in the case \(\beta=1\) only matrices of even dimensions are considered). Corresponding results for \(\beta=2\) have been proved by the fourth author in a previous article. Analogous results have been obtained for Hermite-type ensembles by the first two authors in other papers and some of their methods have been adapted in the present approach. The proof relies on the orthogonal polynomial method for invariant matrix ensembles. The distinct cases of the eigenvalues in the bulk of the spectrum and at the spectral edges, as well as of the lower and upper spectral edges are examined. Two corollaries of the main theorem demonstrate the relevance of this theorem for the understanding of the local eigenvalue statistics in the limit \(n\rightarrow\infty\).