compute_bandstructure_benchmark_metrics

compute_bandstructure_benchmark_metrics#

autoplex.benchmark.phonons.utils.compute_bandstructure_benchmark_metrics(ml_model, structure, mp_id, ml_phonon_bs, dft_phonon_bs, ml_imag_modes, dft_imag_modes, atomwise_regularization_parameter, soap_dict, suffix, displacement=0.01)[source]#

Compute phonon band-structure benchmark metrics and generate associated plots.

Parameters:

  • ml_model (str) – ML model to be used. Default is GAP.

  • structure (.Structure) – A structure object.

  • mp_id (str) – Materials Project ID.

  • ml_phonon_bs (PhononBandStructureSymmLine.) – ML generated pymatgen phonon band-structure object.

  • dft_phonon_bs (PhononBandStructureSymmLine.) – DFT generated pymatgen phonon band-structure object.

  • ml_imag_modes (bool) – Whether the ML-based phonon band structure shows imaginary modes.

  • dft_imag_modes (bool) – Whether the DFT-based phonon band structure shows imaginary modes.

  • displacement (float) – Displacement distance for phonons

  • atomwise_regularization_parameter (float) – Regularization value for the atom-wise force components.

  • suffix (str) – GAP potential file suffix.

  • soap_dict (dict) – Dictionary containing SOAP parameters.

Returns:

Overall root mean squared error between DFT and ML phonon band-structure.

Return type:

dict including