Orthogonal and symplectic matrix models: universality and other properties (Q647374)

The author considers the ensembles of random matrices whose joint eigenvalue distribution is \(p_{n, \beta}(\lambda_1, \dots, \lambda_n)= Q^{-1}_{n, \beta} [V]e^{\beta H(\lambda_1, \dots, \lambda_n)/2},\) where the function \(H\) is called Hamiltonian in analogy with statistical mechanics and \(Q_{n, \beta}[V]\) is the normalizing constant containing the function \(V\) (called the potential) being a real valued Hölder function. This distribution corresponds for the cases \(\beta = 1, 2, 4\) to real symmetric, Hermitian and symplectic matrix models respectively. The bounds for these matrix models with multi-interval supports of the equilibrium measures are found for the rate of convergence of linear eigenvalue statistics and for the variance of linear eigenvalue statistics. The logarithms of partition functions \(Q_{n, \beta}[V]\) up to the order \(O(1)\) are presented. Finally, the universality of local eigenvalue statistics in the bulk is proved.