Competition between quenched disorder and long-range connections: A numerical study of diffusion

Abstract

The problem of random walk is considered in one dimension in the simultaneous presence of a quenched random force field and long-range connections the probability of which decays with the distance algebraically as p_l ~ \beta l^{-s}. The dynamics are studied mainly by a numerical strong disorder renormalization group method. According to the results, for s>2 the long-range connections are irrelevant and the mean-square displacement increases as <x^2(t)> ~ (ln t)^{2/\psi} with the barrier exponent \psi=1/2, which is known in one-dimensional random environments. For s<2, instead, the quenched disorder is found to be irrelevant and the dynamical exponent is z=1 like in a homogeneous environment. At the critical point, s=2, the interplay between quenched disorder and long-range connections results in activated scaling, however, with a non-trivial barrier exponent \psi(\beta), which decays continuously with \beta, but is independent of the form of the quenched disorder. Upper and lower bounds on \psi(\beta) are established and numerical estimates are given for various values of \beta. Beside random walks, accurate numerical estimates of the graph dimension and the resistance exponent are given for various values of \beta at s=2.