Fit

class gammapy.modeling.Fit(datasets, store_trace=False)

Bases: object

Fit class.

The fit class provides a uniform interface to multiple fitting backends. Currently available: “minuit”, “sherpa” and “scipy”

Parameters

datasetsDatasets

Datasets

Methods Summary

confidence(parameter[, backend, sigma, …])	Estimate confidence interval.
covariance([backend])	Estimate the covariance matrix.
minos_contour(x, y[, numpoints, sigma])	Compute MINOS contour.
optimize([backend])	Run the optimization.
run([backend, optimize_opts, covariance_opts])	Run all fitting steps.
stat_profile(parameter[, values, bounds, …])	Compute fit statistic profile.
stat_surface(x, y, x_values, y_values[, …])	Compute fit statistic surface.

Methods Documentation

confidence(parameter, backend='minuit', sigma=1, reoptimize=True, **kwargs)

Estimate confidence interval.

Extra kwargs are passed to the backend. E.g. iminuit.Minuit.minos supports a maxcall option.

For the scipy backend kwargs are forwarded to brentq. If the confidence estimation fails, the bracketing interval can be adapted by modifying the the upper bound of the interval (b) value.

Parameters

backendstr

Which backend to use (see gammapy.modeling.registry)

parameterParameter

Parameter of interest

sigmafloat

Number of standard deviations for the confidence level

reoptimizebool

Re-optimize other parameters, when computing the confidence region.

**kwargsdict

Keyword argument passed ot the confidence estimation method.

Returns

resultdict

Dictionary with keys “errp”, ‘errn”, “success” and “nfev”.

covariance(backend='minuit', **kwargs)

Estimate the covariance matrix.

Assumes that the model parameters are already optimised.

Parameters

backendstr

Which backend to use (see gammapy.modeling.registry)

Returns

resultCovarianceResult

Results

minos_contour(x, y, numpoints=10, sigma=1.0)

Compute MINOS contour.

Calls iminuit.Minuit.mncontour.

This is a contouring algorithm for a 2D function which is not simply the fit statistic function. That 2D function is given at each point (par_1, par_2) by re-optimising all other free parameters, and taking the fit statistic at that point.

Very compute-intensive and slow.

Parameters

x, yParameter

Parameters of interest

numpointsint

Number of contour points

sigmafloat

Number of standard deviations for the confidence level

Returns

resultdict

Dictionary containing the parameter values defining the contour, with the boolean flag “success” and the info objects from mncontour.

optimize(backend='minuit', **kwargs)

Run the optimization.

Parameters

backendstr

Which backend to use (see gammapy.modeling.registry)

**kwargsdict

Keyword arguments passed to the optimizer. For the "minuit" backend see https://iminuit.readthedocs.io/en/latest/api.html#iminuit.Minuit for a detailed description of the available options. If there is an entry ‘migrad_opts’, those options will be passed to iminuit.Minuit.migrad().

For the "sherpa" backend you can from the options method = {"simplex",  "levmar", "moncar", "gridsearch"} Those methods are described and compared in detail on http://cxc.cfa.harvard.edu/sherpa/methods/index.html. The available options of the optimization methods are described on the following pages in detail:

• http://cxc.cfa.harvard.edu/sherpa/ahelp/neldermead.html

• http://cxc.cfa.harvard.edu/sherpa/ahelp/montecarlo.html

• http://cxc.cfa.harvard.edu/sherpa/ahelp/gridsearch.html

• http://cxc.cfa.harvard.edu/sherpa/ahelp/levmar.html

For the "scipy" backend the available options are desribed in detail here: https://docs.scipy.org/doc/scipy/reference/generated/scipy.optimize.minimize.html

Returns

fit_resultFitResult

Results

run(backend='minuit', optimize_opts=None, covariance_opts=None)

Run all fitting steps.

Parameters

backendstr

Backend used for fitting, default : minuit

optimize_optsdict

Options passed to Fit.optimize.

covariance_optsdict

Options passed to Fit.covariance.

Returns

fit_resultFitResult

Results

stat_profile(parameter, values=None, bounds=2, nvalues=11, reoptimize=False, optimize_opts=None)

Compute fit statistic profile.

The method used is to vary one parameter, keeping all others fixed. So this is taking a “slice” or “scan” of the fit statistic.

See also: Fit.minos_profile.

Parameters

parameterParameter

Parameter of interest

valuesQuantity (optional)

Parameter values to evaluate the fit statistic for.

boundsint or tuple of float

When an int is passed the bounds are computed from bounds * sigma from the best fit value of the parameter, where sigma corresponds to the one sigma error on the parameter. If a tuple of floats is given those are taken as the min and max values and nvalues are linearly spaced between those.

nvaluesint

Number of parameter grid points to use.

reoptimizebool

Re-optimize other parameters, when computing the fit statistic profile.

Returns

resultsdict

Dictionary with keys “values”, “stat” and “fit_results”. The latter contains an empty list, if reoptimize is set to False

stat_surface(x, y, x_values, y_values, reoptimize=False, **optimize_opts)

Compute fit statistic surface.

The method used is to vary two parameters, keeping all others fixed. So this is taking a “slice” or “scan” of the fit statistic.

Caveat: This method can be very computationally intensive and slow

See also: Fit.minos_contour

Parameters

x, yParameter

Parameters of interest

x_values, y_valueslist or numpy.ndarray

Parameter values to evaluate the fit statistic for.

reoptimizebool

Re-optimize other parameters, when computing the fit statistic profile.

**optimize_optsdict

Keyword arguments passed to the optimizer. See Fit.optimize for further details.

Returns

resultsdict

Dictionary with keys “x_values”, “y_values”, “stat” and “fit_results”. The latter contains an empty list, if reoptimize is set to False