FluxPointEstimator

class gammapy.spectrum.FluxPointEstimator(obs, groups, model)

Bases: object

Flux point estimator.

Computes flux points for a given spectrum observation dataset (a 1-dim on/off observation), energy binning and spectral model.

A spectral model shape is assumed, all parameters except the norm are fixed A 1D amplitude fit is done, to find the amplitude so that npred equals excess.

This is done for each group independently.

The method is used for example in this FERMI-LAT catalog paper https://ui.adsabs.harvard.edu/#abs/2015ApJS..218…23A or the HESS GPS paper https://ui.adsabs.harvard.edu/#abs/2018A%26A…612A…1H

Parameters:

obs : SpectrumObservation or SpectrumObservationList

Spectrum observation(s)

groups : SpectrumEnergyGroups

Energy groups (usually output of SpectrumEnergyGroupMaker)

model : SpectralModel

Global model (usually output of SpectrumFit)

Attributes Summary

fit	Instance of SpectrumFit
obs	Observations participating in the fit

Methods Summary

compute_flux_point(energy_group)
compute_flux_point_sqrt_ts(fit, stat_best_fit)	Compute sqrt(TS) for flux point.
compute_flux_point_ul(fit, stat_best_fit[, …])	Compute upper limits for flux point values.
compute_points()
fit_point(model, energy_group, energy_ref[, …])

Attributes Documentation

fit

Instance of SpectrumFit

obs

Observations participating in the fit

Methods Documentation

compute_flux_point(energy_group)

compute_flux_point_sqrt_ts(fit, stat_best_fit)

Compute sqrt(TS) for flux point.

Parameters:

fit : SpectrumFit

Instance of spectrum fit.

stat_best_fit : float

TS value for best fit result.

Returns:

sqrt_ts : float

Sqrt(TS) for flux point.

compute_flux_point_ul(fit, stat_best_fit, delta_ts=4, negative=False)

Compute upper limits for flux point values.

Parameters:

fit : SpectrumFit

Instance of spectrum fit.

stat_best_fit : float

TS value for best fit result.

delta_ts : float (4)

Difference in log-likelihood for given confidence interval. See Example below.

negative : bool

Compute limit in negative direction.

Returns:

dnde_ul : Quantity

Flux point upper limit.

Examples

To compute ~95% confidence upper limits (or 2 sigma) you can use:

from scipy.stats import chi2, norm

sigma = 2
cl = 1 - 2 * norm.sf(sigma) # using two sided p-value
delta_ts = chi2.isf(1 - cl, df=1)

compute_points()

fit_point(model, energy_group, energy_ref, sqrt_ts_threshold=1)