C-GAP-20U

The complete dataset used for training the C-GAP-20U interatomic potential for carbon. Suitably converged labels were obtained with revised DFT settings, see CAM.840-6.

>>> from load_atoms import load_dataset
>>> load_dataset("C-GAP-20U")
C-GAP-20U:
    structures: 6,088
    atoms: 400,275
    species:
        C: 100.00%
    properties:
        per atom: (forces)
        per structure: (config_type, energy, free_energy, stress, virial)

License

This dataset is licensed under the GPLv3 license.

Citation

If you use this dataset in your work, please cite the following:

@article{Rowe-20-07,
    title = {An Accurate and Transferable Machine Learning Potential for Carbon},
    author = {Rowe, Patrick and Deringer, Volker L. and Gasparotto, Piero and Cs{\'a}nyi, G{\'a}bor and Michaelides, Angelos},
    year = {2020},
    journal = {The Journal of Chemical Physics},
    volume = {153},
    number = {3},
    pages = {034702},
    doi = {10.1063/5.0005084},
}

Properties

Per-atom:

Property	Units	Type	Description
forces	eV/Å	ndarray(N, 3)	force vectors (DFT)

Per-structure:

Property	Units	Type	Description
energy	eV	float64	total structure energy (DFT)
free_energy	eV	float64	total structure free energy (DFT)
virial	eV	ndarray(3, 3)	virial stress tensor (DFT)
stress	eV Å\({}^{-3}\)	ndarray(3, 3)	stress tensor (DFT) (- virial / cell.volume)
config_type		str	category of structure

Miscellaneous information

C-GAP-20U is imported as an InMemoryAtomsDataset:

Importer script for C-GAP-20U

from load_atoms.database.backend import BASE_GITHUB_URL, SingleFileImporter
from load_atoms.database.internet import FileDownload


class Importer(SingleFileImporter):
    @classmethod
    def file_to_download(cls) -> FileDownload:
        return FileDownload(
            url=f"{BASE_GITHUB_URL}/C-GAP-20U/C-GAP-20U.xyz",
            expected_hash="da0462802df1",
        )

DatabaseEntry for C-GAP-20U

name: C-GAP-20U
year: 2020
description: |
    The complete dataset used for training the
    `C-GAP-20U <https://pubs.aip.org/aip/jcp/article/153/3/034702/1062660/An-accurate-and-transferable-machine-learning>`_
    interatomic potential for carbon.
    Suitably converged labels were obtained with revised DFT settings, see `CAM.840-6 <https://doi.org/10.17863/CAM.84096>`_.
category: Potential Fitting
minimum_load_atoms_version: 0.2
citation: |
    @article{Rowe-20-07,
        title = {An Accurate and Transferable Machine Learning Potential for Carbon},
        author = {Rowe, Patrick and Deringer, Volker L. and Gasparotto, Piero and Cs{\'a}nyi, G{\'a}bor and Michaelides, Angelos},
        year = {2020},
        journal = {The Journal of Chemical Physics},
        volume = {153},
        number = {3},
        pages = {034702},
        doi = {10.1063/5.0005084},
    }
license: GPLv3
per_atom_properties:
    forces:
        desc: force vectors (DFT)
        units: eV/Å
per_structure_properties:
    energy:
        desc: total structure energy (DFT)
        units: eV
    free_energy:
        desc: total structure free energy (DFT)
        units: eV
    virial:
        desc: virial stress tensor (DFT)
        units: eV
    stress:
        desc: |
            | stress tensor (DFT)
            | (:code:`- virial / cell.volume`)
        units: eV Å\ :math:`{}^{-3}`
    config_type:
        desc: category of structure


# TODO: remove after Dec 2024
# backwards compatability: unused as of 0.3.0
files:
     - name: C-GAP-20U.xyz
       hash: da0462802df1