Title:Phase space maximal entropy random walk: Langevin-like ensemble of physical trajectories

Abstract:As written by statistician George Box "All models are wrong, but some are useful", standard diffusion derivation or Feynman path ensembles use nonphysical nowhere differentiable trajectories of infinite kinetic energy - what seems wrong, bringing question of differences if doing it more right this article is focused on. To consider ensembles of more physical trajectories, we can work in $(x,v)$ phase space like in Langevin equation with velocity controlling spatial steps, here also controlled with spatial potential $V(x)$. There will be discussed and compared 4 approaches to predict stationary probability distributions: using Boltzmann ensemble of points in space (GRW - generic random walk) or in phase space (psGRW), and analogously Boltzmann ensemble of paths in space (MERW - maximal entropy random walk) and in phase space (psMERW). They have qualitatively different predictions, hopefully allowing to decide the most appropriate for various settings by distinguishing them experimentally. Path ensembles have much stronger Anderson-like localization exactly as in quantum ground state, proposed novel in phase space has additionally increased density of velocity as in potential gradient, decreased toward barrier. While MERW is thermodynamically in agreement with quantum mechanics, psMERW suggests slight corrections from consideration of more physical trajectories, which seem natural - reduced velocities toward close barriers, increased down potential gradients.