training and test data for the machine learning interatomic potential construction of six solid electrolytes

This repository contains the training and test datasets used to train the machine-learning interatomic potentials (deep-potential models) reported in "Disentangling Cation–Polyanion Coupling in Solid Electrolytes: Which Anion Motion Dominates Cation Transport?" The dataset covers six systems: Na2B10H10, LiBH4, Na3OBH4, Li3PS4, Na11Sn2PS12, and Na4P2S6. For each system, separate training and test sets are provided. For Na3OBH4, the test set additionally includes configurations containing sodium vacancies. The data are stored in the DeepMD-kit format. Each configuration set contains: - coord.npy : atomic coordinates - force.npy : DFT forces - energy.npy : DFT total energies - box.npy : simulation cell vectors - type.raw : atom type indices - type_map.raw : mapping from type indices to chemical species All reference energies and forces were computed with VASP using the PBE exchange-correlation functional and the projector augmented-wave (PAW) method.Plane-wave kinetic-energy cutoffs: 650 eV for Na2B10H10 and LiBH4; 600 eV for Na3OBH4; 500 eV for Li3PS4 and Na4P2S6; and 400 eV for Na11Sn2PS12.Gamma-centered k-point meshes: 2x2x2 for Na2B10H10, LiBH4, and Li3PS4; 3x3x3 for Na3OBH4 and Na4P2S6; and 2x2x1 for Na11Sn2PS12.