ParameterSensitivityEstimator

class gammapy.estimators.ParameterSensitivityEstimator

Bases: object

Estimate the sensitivity to a given parameter.

Computes the TS distribution in the non-null hypothesis using the log likelihood of the Asimov dataset (i.e. a dataset with counts = npred) and the non-central chi2 distribution. Once the TS distribution under the testing hypothesis is known, one can compute the required parameter value to have 50% of measurements above a given significance threshold.

Parameters:

parameterParameter

Parameter to test

null_valuefloat or Parameter

Value of the parameter for the null hypothesis.

n_sigmafloat, optional

Number of required significance level. Default is 5.

n_free_parametersint, optional

Number of free parameters. Default is None, which utilises len(parameters).

rtolfloat, optional

Relative precision of the estimate. Used as a stopping criterion. Default is 0.01.

max_niterint, optional

Maximal number of iterations used by the root finding algorithm. Default is 100.

References

• Cowan et al. (2011), “Asymptotic formulae for likelihood-based tests of new physics”

Attributes Summary

tag

Methods Summary

parameter_matching_significance(datasets)	Parameter value matching the target significance.
run(datasets)	Run the parameter sensitivity estimator.

Attributes Documentation

tag = 'ParameterSensitivityEstimator'

Methods Documentation

parameter_matching_significance(datasets)

Parameter value matching the target significance.

run(datasets)

Run the parameter sensitivity estimator.

Parameters:

datasetsDatasets

The datasets used to estimate the parameter sensitivity.

Returns:

resultfloat

Parameter sensitivity given as the difference between the value matching the target significance and the null value.

__init__(parameter, null_value, n_sigma=5, n_free_parameters=None, rtol=0.01, max_niter=100)

classmethod __new__(*args, **kwargs)