A CLT for the characteristic polynomial of random Jacobi matrices, and the G\(\beta\)E (Q6045830)

The authors establish a central limit theorem for the real part of the logarithm of the characteristic polynomial of random Jacobi matrices. Let \(J_{n,\mathbf{a},\mathbf{b}}\) denote a random \(n\times n\) tridiagonal Jacobi matrix constructed using two independent sequences of independent random variables with variances defined in terms of a parameter \(v\). Letting \(D_n(z,J_{n,\mathbf{a},\mathbf{b}})=\det(zI_n-J_{n,\mathbf{a},\mathbf{b}}/\sqrt{n})\) denote the characteristic polynomial of \(J_{n,\mathbf{a},\mathbf{b}}/\sqrt{n}\), and for a fixed \(z\in(-2,2)\setminus\{0\}\), the authors show that \[ \frac{\log|D_n(z,J_{n,\mathbf{a},\mathbf{b}})|-n(z^2/4-1/2)}{\sqrt{v\log n/2}} \] converges in distribution to a standard Gaussian as \(n\to\infty\). Although this result does not include the case \(z=0\) (i.e., the law of the determinant of \(J_{n,\mathbf{a},\mathbf{b}}\)), the authors note that their proof requires only minor changes to be adapted to that case. This central limit theorem does cover the case of the G\(\beta\)E ensemble for \(\beta>0\), which is the distribution on configurations of \(n\) particles given by the density \[ p(\lambda_1,\ldots,\lambda_n)=C_{n,\beta}\exp\left\{-\frac{\beta}{4}\sum_{i=1}^n\lambda_i^2\right\}\prod_{1\leq i<j\leq n}\left|\lambda_i-\lambda_j\right|^\beta\,, \] which generalizes the law of the eigenvalues of orthogonal, Hermitian and symplectic random matrices with Gaussian entries.