changzhiai/dft-databases
Database DFT trajectories grouped by exchange-correlation functional. All files are ASE .traj format. Each file is a single source database and you can directly use it to train your own model. Potentials trained on these databases with the IANN trainer are released at changzhiai/iann-foundation-models; see the documentation. PBE File Name File Size Image Size Functional Details Year Source MPtrj.traj 3.50 GB 1,579,091 PBE MPtrj dataset 2022 Materials… See the full description on the dataset page: https://huggingface.co/datasets/changzhiai/dft-databases.
Database
DFT trajectories grouped by exchange-correlation functional. All files are ASE .traj format. Each file is a single source database and you can directly use it to train your own model.
Potentials trained on these databases with the IANN trainer are released at changzhiai/iann-foundation-models; see the documentation.
PBE
RPBE
r2SCAN
Usage
from ase.io import Trajectory
traj = Trajectory("PBE/MPtrj.traj")
print(len(traj))
atoms = traj[0]
print(atoms.get_potential_energy(), atoms.get_forces())Note
Outlier structures were removed from each database (less than 1%).