Ab initio validation on the connection between atomistic and hydrodynamic description to unravel the ion dynamics of warm dense matter

Abstract

Ion dynamics exhibits inherent multiscale characteristics because it contains both atomistic and hydrodynamic behaviors. Although atomic-scale ab initio molecular dynamics is the subject of intense research on warm dense matter, the macroscopic relaxation process contained in the zero-frequency mode of the ionic dynamic structure factor (DSF) cannot be demonstrated due to the limitation of simulation sizes. Here, we fill this gap via the machine-learning deep potential method. To capture the ion dynamics near the hydrodynamic limit with ab initio accuracy, an accurate and efficient electron-temperature-dependent interatomic potential was constructed. We quantitatively verify the consistency of thermal diffusivities obtained from hydrodynamics and the fluctuation-dissipation theorem and further provide a microscopic perspective of energy transport to understand the zero-frequency mode of DSF. As implemented in two temperature states, a competitive mechanism is found to account for the damping of the zero-frequency mode.