Research data for: "Data-efficient machine-learning interatomic potential for studying radiation effects in germanium"

Attahced are the training dataset and potential files for the publication "Data-efficient machine-learning interatomic potential for studying radiation effects in germanium".

The description of the uploaded files is as follows:

Ge-TGAP_dataset.tar.gz: This dataset supplies training data for a Gaussian Approximation Potential for germanium, developed specifically for radiation damage studies. It encompasses 451 structures from dimers, multiple bulk crystal phases, liquid configurations at various temperatures, a diverse range of defect structures, and other relevant configurations. All structures are stored in extended XYZ format, with each configuration annotated by total energy, atomic forces, and virial stresses calculated via DFT at the PBE level using VASP. Additional details on dataset generation and DFT calculations are provided in the published paper.

Ge-TGAP_LAMMPS.tar: The potential files for LAMMPS.

The unique identifier of the potential is: "GAP_2025_3_8_120_22_43_25_989".