maml.apps.gbe package

Implementation of GB energy prediction paper Weike Ye, Hui Zheng, Chi Chen, Shyue Ping Ong “A Universal Machine Learning Model for Elemental Grain Boundary Energies” arXiv preprint arXiv: 2201.11991 (2022).

maml.apps.gbe.describer module

Module implements the describers for GB entry.

class maml.apps.gbe.describer.GBBond(gb: GrainBoundary, loc_algo: str = ‘crystalnn’, bond_mat: np.ndarray | None = None, **kwargs)

Bases: MSONable

This class describes the GB Bonding environment using provided local environment algorithm. Available algorithms: GBBond.NNDict.keys(), default CrystalNN.

NNDict(_ = {‘brunnernn_real’: <class ‘pymatgen.analysis.local_env.BrunnerNN_real’>, ‘brunnernn_reciprocal’: <class ‘pymatgen.analysis.local_env.BrunnerNN_reciprocal’>, ‘brunnernn_relative’: <class ‘pymatgen.analysis.local_env.BrunnerNN_relative’>, ‘covalentbondnn’: <class ‘pymatgen.analysis.local_env.CovalentBondNN’>, ‘critic2nn’: <class ‘pymatgen.analysis.local_env.Critic2NN’>, ‘crystalnn’: <class ‘pymatgen.analysis.local_env.CrystalNN’>, ‘cutoffdictnn’: <class ‘pymatgen.analysis.local_env.CutOffDictNN’>, ‘econnn’: <class ‘pymatgen.analysis.local_env.EconNN’>, ‘jmolnn’: <class ‘pymatgen.analysis.local_env.JmolNN’>, ‘minimumdistancenn’: <class ‘pymatgen.analysis.local_env.MinimumDistanceNN’>, ‘minimumokeeffenn’: <class ‘pymatgen.analysis.local_env.MinimumOKeeffeNN’>, ‘minimumvirenn’: <class ‘pymatgen.analysis.local_env.MinimumVIRENN’>, ‘nearneighbors’: <class ‘pymatgen.analysis.local_env.NearNeighbors’>, ‘openbabelnn’: <class ‘pymatgen.analysis.local_env.OpenBabelNN’>, ‘voronoinn’: <class ‘pymatgen.analysis.local_env.VoronoiNN’>_ )

_get_bond_mat(gb: GrainBoundary)

  • Parameters gb (GrainBoundary) – the grain boundary structure object.

  • Returns bond matrix matrix of bond lengths

    1. bm[i][j] = bond length between atom i&j if i & j is bonded (determined by the loc_algo)

    2. bm[i][j] = bm[j][i]

    3. If not bonded, the bm[i][j] = np.nan

as_dict()

Dict representation of the GBond class.

  • Returns str, “bond_mat”: bond matrix}

  • Return type dict of {“loc_algo”

property bond_matrix(: ndarra )

The (padded) bond matrix.

classmethod from_dict(d)

  • Parameters d (dict) – Dict representation.

  • Returns GBBond

get_breakbond_ratio(gb: GrainBoundary, b0: float, return_count: bool = False)

Get the breakbond ratio, i.e the ratio of shorter bonds vs longer bonds compared to the bulk bond length (b0) The algo to find the bonds can vary Meme: if use get_neighbors, require a hard set cutoff, which adds to

arbitrariness.

  • Parameters

    • gb (GrainBoundary) – a GrainBoundary object

    • b0 (float) – cutoff to determine short vs. long bonds, default the bulk bond length

    • return_count (bool) – whether to return count of

  • Returns shorter_bond / longer_bonds if return_count: shorter_bond: # of short bonds longer_bond: # of long bonds

  • Return type ratio (float)

get_mean_bl_chg(b0: float)

Function to calculate the mean bond length difference between GB and the bulk :param b0: the bond length in bulk. :type b0: float

Returns: the mean_bl_chg

property max_bl(: floa )

The maximum bond length.

property min_bl(: floa )

The minimum bond length.

class maml.apps.gbe.describer.GBDescriber(structural_features: list | None = None, elemental_features: list | None = None, **kwargs)

Bases: BaseDescriber

The describers that describes the grain boundary db entry with selected structural and elemental features.

abc_impl( = <_abc.abc_data object )

sklearn_auto_wrap_output_keys( = {‘transform’_ )

generate_bulk_ref(gb_df: DataFrame, inc_target: bool = True)

Generate the bulk reference for given gb entry :param gb_df: data for gb :type gb_df: pd.DataFrame :param inc_target: whether to include target. :type inc_target: bool

  • Returns data for bulk

  • Return type bulk_df (pd.DataFrame)

transform_one(db_entry: dict, inc_target: bool = True, inc_bulk_ref: bool = True, mp_api: str | None = None)

Describe gb with selected structural and elemental features :param db_entry: entry from surfacedb.All_GB_properties_parallel_copy :type db_entry: dict :param inc_target: whether to include target in the df, default: False :type inc_target: bool :param inc_bulk_ref: whether to generate bulk reference

bulk reference: i.e. the entry of the origin bulk of the GB,

the rotation angle (theta) = 0, gb_energy = 0
  • Parameters mp_api (str) – MP api key.

Returns: pd.DataFrame of processed data, columns are the feature labels

The task_id is included and serve as the unique id of the data

maml.apps.gbe.describer.convert_hcp_direction(rotation_axis: list, lat_type: str)

four index notion to three index notion for hcp and rhombohedral axis :param rotation_axis: four index notion of axis :type rotation_axis: list :param lat_type: the :type lat_type: str

  • Returns rotation axis in three index notion.

maml.apps.gbe.describer.convert_hcp_plane(plane: list)

four index notion to three index notion for hcp and rhombohedral plane :param plane: four index notion. :type plane: list

  • Returns three index notion of plane

maml.apps.gbe.describer.get_elemental_feature(db_entry: dict, loc_algo: str = ‘crystalnn’, features: list | None = None, mp_api: str | None = None)

Function to get the elemental features :param db_entry: db entry :type db_entry: Dict :param loc_algo: algorithm to determine local env.

Options: see GBBond.NNDict.keys()

  • Parameters

    • features (List) – list of feature names e_coh: cohesive energy G: G_vrh shear modulus a0: bulk lattice parameter a ar: atomic radius b0: the bond length of the metal bulk mean_delta_bl: the mean bond length difference

between GB and the bulk

bl: the mean bond length in GB

* **mp_api** (*str*) – api key to MP.
  • Returns pd.DataFrame of elemental features

maml.apps.gbe.describer.get_structural_feature(db_entry: dict, features: list | None = None)

The structural features: d_gb: interplanal distance of the gb_plane d_rot: interplanal distance of the gb_plane

w/ smallest integer index ) normal to rotation axis

theta: rotation angle (sin and cos).

  • Parameters

    • db_entry (Dict) – db entry

    • features (List) – list of features.

  • Returns pd.DataFrame of structural features

maml.apps.gbe.presetfeatures module

Module defines feature objects for GB energy model.

class maml.apps.gbe.presetfeatures.my_quant(str_name: str, latex_name: str | None = None, latex_unit: str | None = None)

Bases: object

An object to describe GB quantities.

property latex()

Returns: latex rep of the quant.

property name()

Returns: string rep of the quant.

property unit()

Returns: latex rep of the quant unit.

maml.apps.gbe.utils module

Module implements helper functions to retrieve data for GB energy prediction paper.

maml.apps.gbe.utils.load_b0_dict()

Helper function to retrieve the b0 data. b0 is the bulk bond length.

  • Returns b0}

  • Return type returns the dict of {element

maml.apps.gbe.utils.load_data(filename: str | None = None)

Helper function to load the data Default is to load the 361 data :param filename: the filename of the data. :type filename: str

  • Returns data list

maml.apps.gbe.utils.load_mean_delta_bl_dict(loc_algo: str = ‘crystalnn’)

Helper function to load the mean_delta_bl data :param loc_algo: name of the algorithm. :type loc_algo: str

  • Returns mean_delta_bl(float)}

  • Return type {task_id (int)

maml.apps.gbe.utils.update_b0_dict(data_source: list, mp_api: str | None)

Helper function to update the b0 dictionary Requires api key to MP :param data_source: list of GB data :type data_source: list :param mp_api: API key to MP :type mp_api: str

  • Returns {el: b0}.

  • Return type b0_dict (dict)

© Copyright 2022, Materials Virtual Lab