Density Profiles in Open Superdiffusive Systems

Abstract

We numerically solve a discretized model of Levy random walks on a finite one-dimensional domain in the presence of sources and with a reflection coefficient $r$. At the domain boundaries, the steady-state density profile is non-analytic. The meniscus exponent $\mu$, introduced to characterize this singular behavior, uniquely identifies the whole profile. Numerical data suggest that $\mu =\alpha/2 + r(\alpha/2-1)$, where $\alpha$ is the Levy exponent of the step-length distribution. As an application, we show that this model reproduces the temperature profiles obtained for a chain of oscillators displaying anomalous heat conduction. Remarkably, the case of free-boundary conditions in the chain correspond to a Levy walk with negative reflection coefficient.