inclination_index

probnumeval.multivariate.inclination_index(approximate_solution, reference_solution)[source]

Compute the inclination index (II).

The II is a version of the NCI that additionally indicates whether an estimate is

  • Underconfident if \(\text{II} < 0\) holds. The estimated error is way larger than the actual error.

  • Overconfident if \(\text{II} > 0\) holds. The estimated error is way smaller than the actual error.

Parameters

  • approximate_solution (Union[Normal, _RandomVariableList]) – Approximate solution as returned by a (Gaussian) probabilistic numerical method.

  • reference_solution (ndarray) – Reference solution. This is an array, because it must be a deterministic point-estimate.

Returns

Return type

Inclination index.

See also

anees()

An alternative calibration measure.

non_credibility_index()

Non-credibility index.