- history_matching(obs, predictions, threshold=3.0, discrepancy=0.0, rank=1)[source]#
Perform history matching to compute implausibility and identify NROY and RO points. This implementation performs history matching as a single run, completing the process in one execution without iterative refinement or staged waves.
The implausibility is calculated as the absolute difference between the observed and predicted values, normalized by the square root of the sum of the variances of the observed and predicted values. The implausibility is then compared to a threshold to classify the points as NROY or RO. The discrepancy value(s) can be provided as a scalar or an array to account for model discrepancy. The rank parameter is used to select the number of observations to consider for implausibility calculation. The default value is 1, which corresponds to the most recent observation. :param obs: Observations as (mean, variance). :type obs: tuple :param predictions: Predicted (mean, variance). :type predictions: tuple :param threshold: Implausibility threshold for NROY classification. :type threshold: float :param discrepancy: Discrepancy value(s). :type discrepancy: float or ndarray :param rank: Rank for implausibility calculation. :type rank: int
- Returns:
Contains implausibility (I), NROY indices, and RO indices.
- Return type:
