Derivative moments for characteristic polynomials from the CUE (Q1758201): Difference between revisions

The \(N\)-dimensional circular unitary ensemble (CUE) of random matrix theory is studied. Its probability space consists of the set \(\text{U}(N)\) of \(N \times N\) unitary matrices with normalized Haar measure \(\mu^{\text{Haar}}\). The characteristic polynomial of a matrix \(U \in \text{U}(N)\) is given by \(Z_U(\theta) := \prod_{n =1}^{N} (1- \text{e}^{i(\theta_n- \theta)})\), where \(\text{e}^{i \theta_1}, \dots,\text{e}^{i \theta_N}\) are the eigenvalues of \(U\). Then the distribution \[ V_U(\theta) := \exp(iN\frac{\theta + \pi}{2} - i \sum_{n=1}^{N}\frac{\theta_n}{2})Z_U(\theta) \] with real values for \(\theta \in [0,2 \pi]\) is defined. Joint moments of the distribution of \(V_U\) are defined for \(h > -1/2 \) and \(k > h - 1/2\) as \[ \tilde{F}_N(h, k) := \int_{\text{U}(N)}|V_U(0)|^{2k -2h}|V_U'(0)|^{2h}\text{d}\mu^{\text{Haar}}. \] In earlier works, the moments of \(V_U\) have been studied for integer values \(h\). In this work, the study of \(\tilde{F}_N(h, k)\) is extended to the cases for half-integer values of \(h\). The calculations of moments are also carried out for the asymptotic case when the size \(N\) goes to infinity.