Kramers-like escape driven by fractional Gaussian noise.

We investigate the escape from a potential well of a test particle driven by fractional Gaussian noise with Hurst exponent 0<H<1. From a numerical analysis we demonstrate the exponential distribution of escape times from the well and analyze in detail the dependence of the mean escape time on the Hurst exponent H and the particle diffusivity D. We observe different behavior for the subdiffusive (antipersistent) and superdiffusive (persistent) domains. In particular, we find that the escape becomes increasingly faster for decreasing values of H , consistent with previous findings on the first passage behavior. Approximate analytical calculations are shown to support the numerically observed dependencies.