Power spectrum of the circular unitary ensemble (Q2111749)

The sequence of \(N\) ordered energy levels of a given quantum system can naturally be interpreted as a discrete time random process, the times being the indices of the ordered energy levels. A usual key assumption is then the independence of the mean values on their index (``stationarity''). Without making such a restrictive assumption, the authors first develop a nonperturbative theory to compute the Fourier transform of the covariance matrix (``power spectrum''). Applying it to the circular unitary ensemble (CUE) of random matrices, they express this power spectrum with the tau function of the most general Painlevé function, the sixth one, thus defining a universal probability law. The additional interest of this paper is the representation of the above result by three different, equivalent methods: a differential equation (the sixth Painlevé), a Fredholm determinant, a Toeplitz determinant. The paper, which avoids unnecessary esoteric terminology, is clearly written and very pleasant to read.