Unitary $n$-correlations with restricted support in random matrix theory

Abstract: We consider the $n$-correlation of eigenvalues of random unitary matrices in the alternative form that is not the tidy determinant common in random matrix theory, but rather the expression derived from averages of ratios of characteristic polynomials in a method that can be mimicked in number theoretical calculations of the correlations of zeros of $L$-functions. This alternative form for eigenvalues of matrices from $U(N)$ was proposed by Conrey and Snaith and derived by them when the test function has support in (-2,2), derived by Chandee and Lee for support (-4,4) and here we calculate the expression when the support is (-6,6).