On the second-order correlation of characteristic polynomials of Hermite \(\beta \) ensembles (Q990915): Difference between revisions

A variant of the Gaussian unitary ensemble, the Hermite \(\beta\) ensemble, is studied. The Hermite \(\beta\) ensemble is a family of probability measures of \(n\) unordered real-valued points \(\lambda_1, \lambda_2, \dots, \lambda_n\) on \({\mathbb R}^n\) whose probability density is given by \[ p_{n, \beta}(x_1, \dots,x_n) = \frac{1}{Z_{n, \beta}} \prod_{1 \leq j < k \leq n} |x_j - x_k|^{\beta} \prod_{j = 1}^n e^{-x_j^2/2}, \] where \(x_1, \dots,x_n \in {\mathbb R}\) and \(Z_{n, \beta}\) is the appropriate normalizer. When \(\beta = \) 1, 2 and 4, it reduces to the classical Gaussian ensembles. Applying the discovery by \textit{I. Dimitriu} and \textit{A. Edelman} [J. Math. Phys.~43, No.~11, 5830--5847 (2002; Zbl 1060.82020)] of the so-called matrix models, for all values \(\beta\) the second-order correlation function of the characteristic polynomials is calculated. Then, the asymptotic behavior of the second-order correlation of the characteristic polynomials both in the bulk \((0 < \beta < 4)\) of the spectrum and at the edge \((\beta > 0)\) is obtained. When proving these results the basic ideas of \textit{F. Götze} and \textit{H. Kösters} [Commun. Math. Phys. 285, No.~3, 1183--1205 (2009; Zbl 1193.15035)] on general Hermitian and real Wigner matrices under the fourth-order moment condition were followed.