Infinite random matrices and ergodic measures (Q5953695)

This paper is concerned with measures of infinite-dimensional unitary measures and associated spectral measures corresponding to the Dyson ensemble as the distribution of eigenvalues for finite-dimensional unitary matrices. The problem is transformed from unitary to Hermitian matrices by the Cayley transform. A 2-parameter family of measures is studied where the first parameter corresponds to the mean of the eigenvalues in the finite case, and one of the main theorems proves that the spectral measure is concentrated on the subset where the second parameter is zero. The other main theorem identifies the sine process, known to be a limit of the radial parts of the spectral measures in the finite-dimensional part, as the point process naturally obtained from a spectral measure in the infinite-dimensional case.