Numéro
J. Phys. I France
Volume 4, Numéro 5, May 1994
Page(s) 655 - 673
DOI https://doi.org/10.1051/jp1:1994168
DOI: 10.1051/jp1:1994168
J. Phys. I France 4 (1994) 655-673

Statistical properties of one-point Green functions in disordered systems and critical behavior near the Anderson transition

Alexander D. Mirlin1 and Yan V. Fyodorov2

1  Institut für Theorie der Kondensierten Materie, Universitât Karlsruhe, 76128 Karlsruhe, Germany
2  Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot 76100, Israel


(Received 1 Octoher 1993, accepted 9 Fehruary 1994)

Abstract
We investigate the statistics of local Green functions $G(E, x, x) = \langle x \vert(E - \hat{H})^{-1}\vert x\rangle$, in particular of the local density of states $\rho \propto Im G(E, x, x)$, with the Hamiltonian $\hat{H}$ describing the motion of a quantum particle in a d-dimensional disordered system. Corresponding distributions are related to a function which plays the role of an order parameter for the Anderson metal-insulator transition. When the system can be described by a nonlinear $\sigma$-model, the distribution is shown to possess a specific "inversion" symmetry. We present an analysis of the critical behavior near the mobility edge that follows from the abovementioned relations. We explain the origin of the non-power-like critical behavior obtained earlier for effectively infinite-dimensional models. For any finite dimension $d < \infty$ the critical behavior is demonstraied to be of the conventional power-law type wilh $d = \infty$ playing the rote of an upper critical dimension.



© Les Editions de Physique 1994

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.