Pre-trained ALIGNN-FF¶
ALIGNN-FF is a universal graph neural network force-field. Several pretrained versions
are shipped with the package under
alignn/ff/ and on
Figshare.
CLI¶
Common tasks:
# Single-point energy on the unrelaxed structure
run_alignn_ff.py --file_path POSCAR --task="unrelaxed_energy"
# Structure optimization
run_alignn_ff.py --file_path POSCAR --task="optimize"
# Energy vs. volume curve
run_alignn_ff.py --file_path POSCAR --task="ev_curve"
Additional tasks supported by the CLI include phonon calculations, interface
gamma-surface scans, and simple MD — run -h to see the full set.
Available checkpoints¶
Model folders bundled with the package:
v10.30.2024_dft_3d_307k— trained on 307k JARVIS-DFT configurations (Oct 2024)v12.2.2024_dft_3d_307k— updated December 2024 releasev2024.12.12_dft_3d_multi_prop— multi-property (energy + atomwise targets)alignnff_wt01— weighted loss variantalex_band_gap,jv_mbj_bandgap_alignn— bandgap-oriented models
Additional models are listed in
all_models_alignn_atomwise.json.
Default path helper¶
from alignn.ff.ff import default_path
print(default_path()) # directory containing the default checkpoint
Using it programmatically¶
See the ASE calculator page for a full worked example covering relaxation and an EV curve.