Level Correlations for Metal-Insulator Transition

Abstract

We study the level-statistics of a disordered system undergoing the Anderson type metal-insulator transition. The disordered Hamiltonian is a sparse random matrix in the site representation and the statistics is obtained by taking an ensemble of such matrices. It is shown that the transition of levels due to change of various parameters e.g. disorder, system size, hopping rate can be mapped to the perturbation driven evolution of the eigenvalues of an ensemble subjected to Wigner-Dyson type perturbation with an initial state given by a Poisson ensemble; the mapping is then used to obtain desired level-correlations.