predict_gap_decomp#

mbgdml.predictors.predict_gap_decomp(Z, R, entity_ids, entity_combs, model, **kwargs)[source]#

Predict all \(n\)-body energies and forces of a single structure.

Parameters:

  • Z (numpy.ndarray, ndim: 1) – Atomic numbers of all atoms in r (in the same order).

  • r (numpy.ndarray, ndim: 2) – Cartesian coordinates of a single structure to predict.

  • entity_ids (numpy.ndarray, ndim: 1) – 1D array specifying which atoms belong to which entities.

  • entity_combs (iterable) – Entity ID combinations (e.g., (53,), (0, 2), (32, 55, 293), etc.) to predict using this model. These are used to slice r with entity_ids.

  • model (mbgdml.models.gdmlModel) – GDML model containing all information need to make predictions.

Returns:

  • :obj:`numpy.ndarray`, ndim (1) – Energies of all possible \(n\)-body structures. Some elements can be numpy.nan if they are beyond the criteria cutoff.

  • :obj:`numpy.ndarray`, ndim (3) – Atomic forces of all possible \(n\)-body structure. Some elements can be numpy.nan if they are beyond the criteria cutoff.

  • :obj:`numpy.ndarray`, ndim (2) – All possible \(n\)-body combinations of r (i.e., entity ID combinations).