Materials Project Trajectory Dataset of r²SCAN (MPr²SCAN)

This JSON file represents the MPr2SCAN dataset, containing 34,927 material IDs and 238,247 structures. The dataset was used to fine-tune the pretrained CHGNet on the r2SCAN data. The structures and labels were parsed from all r2SCAN static and relaxation trajectories available in the Materials Project Database in March 2024, using a selection method to avoid duplicate structures.The format of the JSON file looks like this:MPr2SCAN-'mp-id-0'-'frame-id-0'-'structure': dictionary of pymatgen.core.Structure-'uncorrected_total_energy': [eV] raw r2scan energy from VASP output-'energy_per_atom': [eV/atom] r2scan energy per atom, this is the energy label used to fine-tune CHGNet-'force': [eV/A] force on the atoms-'stress': [kBar] stress on the cell-'magmom': [muB] magmom on the atoms-'bandgap': [eV] bandgap-'mp_id': corresponding materials ID in Materials Project<br>-'frame-id-1'...<br><br>-'mp-id-1'...