Diffusion in the mean for an ergodic Schrödinger equation perturbed by a fluctuating potential (Q496191)

Proving diffusive propagation of the quantum wave function in a weakly disordered background, over arbitrary long time scales in dimension \(d\geq 3\), remains a difficult open problem in mathematical physics. In view of many rigorous results in dimensions \(1\) and \(2\), one knows that it is the recurrence that can be made responsible for, e.g., the Anderson localization. It is often stated that the diffusion occurs and persists more readily for models in which the recurrence is eliminated or significantly reduced. This idea was behind prior work of the author in which diffusive propagation has been established for solutions of a tight binding random Schrödinger equation with a random potential subject to a stochastic evolution (cf. [\textit{Y. Kang} and \textit{J. Schenker}, J. Stat. Phys. 134, No. 5--6, 1005--1022 (2009; Zbl 1232.82010)] and [\textit{C. Musselman} and \textit{J. Schenker}, ``Diffusive scaling for all moments of the Markov Anderson model'', Markov Process Related Fields 21, No. 3 (2015)]). The aim of the present paper is to consider the more general situation, in which the environment is composed of two parts: a large static one that on its own would lead to Anderson localization and a small dynamic part that evolves according to the recipe given in previous publications. Diffusive scaling of position moments and a central limit theorem are obtained for the mean position of a quantum particle hopping on a cubic lattice and subject to the just mentioned random potential. The proof of the existence of a finite positive diffusion coefficient is a consequence of Theorem 1.1 (central limit theorem for single time position marginals) which in turn follows from more general Theorems 2.1 (central limit theorem) and 2.2 (on stationary random potentials) under a number of preliminary assumptions. The proofs employ the augmented space analysis of Markov semigroups.