ExtrapolationSelectionParameters

class ExtrapolationSelectionParameters(extrapolation_grade_algorithm=<class 'NL.ComputerScienceUtilities.NLFlag._NLFlag.MaxvolStandard'>, selection_basis_size=None, descriptor_cutoff=0.2, descriptor_basis_size=100, committee_size=6, write_atomic_error_estimates=False)

Additional parameters for candidate selection and extrapolation grade calculations.

Parameters:
  • extrapolation_grade_algorithm (MaxvolStandard | MaxvolLinearized | MaxForce | QueryByCommitteeForces | QueryByCommitteeEnergy) – Select which extrapolation grade algorithm should be used. The MaxForce algorithm is only available with MD active learning. For multi-element systems QueryByCommitteeForces is the recommended method.
    Default: MaxvolStandard

  • selection_basis_size (int) – The basis size used for candidate selection and extrapolation grade calculation. Only used if MaxvolStandard is used for the extrapolation grade.
    Default: Same size as MTP basis size specified in fitting parameters.

  • descriptor_cutoff (float) – The cutoff value for the distances in normalized descriptor space, above which two configurations are considered different. Only used by query-by-committee and max. force algorithms.
    Default: 0.2

  • descriptor_basis_size (int) – The MTP basis size that should be used for the configurational descriptor. Only used by query-by-committee and max. force algorithms.
    Default: 100

  • committee_size (int) – The number of committee members that should be used when the query-by-committee method is selected. Only used by query-by-committee algorithms.
    Default: 6

  • write_atomic_error_estimates (bool) – Write the per-atom error estimates as measurements to the trajectory whenever the extrapolation grade is calculated. This is only available with query-by-committee methods. Only used by query-by-committee algorithms.
    Default: False

committeeSize()
Returns:

The number of committee members when the query-by-committee method is used for the extrapolation grade.

Return type:

int

descriptorBasisSize()
Returns:

The MTP basis size that is used in the descriptor.

Return type:

int

descriptorCutoff()
Returns:

The descriptor cutoff value.

Return type:

float

extrapolationGradeAlgorithm()
Returns:

Which extrapolation grade algorithm should be used.

Return type:

MaxvolStandard | MaxvolLinearized | MaxForce | QueryByCommitteeForces | QueryByCommitteeEnergy

selectionBasisSize()
Returns:

The basis size used for candidate selection and extrapolation grade calculation.

Return type:

int | None

uniqueString()

Return a unique string representing the state of the object.

writeAtomicErrorEstimates()
Returns:

True if the per-atom error estimates should be written to the trajectory.

Return type:

bool