Fig. 3: The molecular dynamics simulation results of GeTe and Li10GeP2S12 using the proposed NVNMD.
From: Accurate and efficient molecular dynamics based on machine learning and non von Neumann architecture
In panel a, the phase transition processes of GeTe is reproduced, which contains initial crystalline phase (I), liquid phase (II), amorphous phase (III), and recrystallization (IV) shown in the insets. In panel b, the mean square displacement (MSD) of Li10GeP2S12 is computed using NVNMD and the diffusion coefficient is extracted from its fitting line. The components of x, y, and z direction show the anisotropic diffusion.
