Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
We derive an exact, all-orders resummation of the hydrodynamic gradient expansion for monoenergetic particle elastic scattering in three dimensions with an arbitrary rotationally invariant kernel. This resummation yields an infinite continued fraction for the frequency- and wave-number-dependent diffusivity, where each term depends on higher angular harmonics of the collision operator. The resulting analytic structure of the diffusivity reveals how microscopic scattering details manifest in macroscopic transport, with deviations from Fick's law decaying exponentially at a rate dictated by the inverse diffusivity's complex-wavenumber singularities.
Exponentially small deviations from Fick's law encode the full microscopic scattering kernel, revealing a direct link between complex-wavenumber singularities and transport properties.
We consider monoenergetic particles propagating in three spatial dimensions and undergoing elastic scattering off an external medium. For an arbitrary rotationally invariant scattering kernel, we derive an exact expression for the frequency- and wave-number-dependent diffusivity, thereby resumming the hydrodynamic gradient expansion to all orders. The result is an infinite continued fraction, with each successive level determined by the relaxation rate of a higher angular harmonic of the collision operator. Thus, while ordinary diffusion probes only the first angular harmonic, the resummed constitutive relation retains information about the full scattering kernel. We show how this microscopic information manifests itself in the analytic structure of the diffusivity and in the response to a localized external source. In particular, deviations from Fick's law decay exponentially with distance, at a rate determined by the nearest complex-wavenumber singularity of the inverse diffusivity. Thus, exponentially small departures from hydrodynamics retain detailed information about the underlying scattering process.