See also the documentation.
This package provides Python tools for the interaction with the atomic energy network (ænet) machine-learning interatomic potential software package. To learn more about ænet, see the ænet website and GitHub repository.
Common use cases for the tools within the aenet-python package are
- Extraction of structures, energies, and forces from the output of first-principles calculations;
- Interconversion of atomic structure formats, especially conversion to ænet's XSF format;
- Manipulation of atomic structures, e.g., for generating reference data;
- Operations on the featurized reference data produced by ænet's
generate.x; and - Analysis of the outputs generated by ænet's
train.x, i.e., from machine-learning potential training.
Install the Python package as usual, for example with pip
$ pip install --user .
To use the featurization capabilities, the main ænet package needs to be installed separately as described elsewhere. And aenet-python needs to be made aware of the ænet installation path, for example, using its interactive configuration tool
$ aenet config --set-aenet-path [path-to-aenet]
where [path-to-aenet] is the path pointing to the aenet root directory.
aenet-python can be used to interconvert between atomic structure formats.
$ aenet sconv --format
Supported file formats:
Format description read write extensions
--------------------------------------------------------------------
aims FHI-aims geometry.in format yes yes in aims
aimsout FHI-aims output file format yes no out
ase ASE's trajectory (traj) format yes yes traj
atat ATAT's structure format yes yes
cp cp.x trajectory (QE) yes no pos
crystal CRYSTAL's fort.34 format yes yes crystal gui
espresso Quantum Espresso input format yes yes espresso
gap GAP reference structure format yes no gap
lammps LAMMPS atomic structure dump yes yes dump lammps lmp
lammpsdata subset of LAMMPS data format no yes data
pdb Protein Data Base format no yes pdb
prism Prismatic XYZ format yes yes prism
pw pw.x output format (QE) yes no
runner RuNNer input.data format yes yes
tinker Tinker XYZ format yes yes tinker arc
vasp VASP's POSCAR/CONTCAR format yes yes vasp
vasprun VAPS's vasprun.xml archive yes no xml
xdatcar VASP's XDATCAR format yes no
xsf XCrysDen Structure Format yes yes xsf
xyz XYZ Cartesian coordinates yes yes xyz
For more details, see the documentation.
If you make use of this package, please cite the ænet reference publication
[1] N. Artrith and A. Urban, Comput. Mater. Sci. 114 (2016) 135-150, http://dx.doi.org/10.1016/j.commatsci.2015.11.047
If you are using the featurization tools, please also cite
[2] N. Artrith, A. Urban, and G. Ceder, Phys. Rev. B 96, 2017, 014112, https://doi.org/10.1103/PhysRevB.96.014112
[3] V. Gharakhanyan, M. S. Aalto, A. Alsoulah, N. Artrith, A. Urban, ICLR 2023, https://openreview.net/forum?id=4Hl8bjobpl9