Knödlseder JürgenUpdated over 11 years ago by
The model has been validation by generating pull distributions using cspull. The validation has been done using ctools-00-06-00 and GammaLib-00-07-00.
An exponentially cut-off power law following the HESS Crab spectrum was used to generate mock data. Below the pull distributions for all fitted parameters (prefactor, spectral index, cut off energy and 2 background parameters) are shown for 10000 Monte Carlo samples for an exposure time of 18000 seconds. The deadtime correction factor was assumed to 0.95. The file kb_E_50h_v3 was used for the instrumental response function, data have been simulated for 0.1-100 TeV for an ROI of 5 deg.
 |
 |
 |
 |
 |
Apparently, there is some asymmetry in the uncertainty of the cut off energy. I guess that this comes from the fact that the cut off energy enters under the exponential in the spectral model, making the uncertainty inherently asymmetric.
Note: we may explore a model where instead of fitting the cut off energy, one fits the exponential of the cut off energy as parameter.