Feature #753
Updated by Mayer Michael over 11 years ago
For model analysis in HESS the 68% containment radius of the PSF is computed by integrating over the actual shape. In fact, this shape differs significantly from a single Gaussian. One way to parameterise the PSF is a composition of three Gaussians with different widths, similar to _GCTAPsf2D_. The other way, which also has been done before, is to parameterise the PSF with a Student's t-distribution. This function has only one parameter which can be calculated using the 68% containment radius.
Since the R68 of the PSF is the only parameter which is stored in model lookup tables, it would be nice to have a proper parameterisation of the PSF with only one parameter. My suggestion would be to include a class _GCTAPsfStudentsT_ which inherits from _GCTAPsfVector_ and just overwrites the functions like _operator()_, _mc()_ and _delta_max()_.
This class would also help to better compare results from gammalib with HESS-Analyses because the PSF would be modelled properly. Attached is a residual map (data-model)/model (data-model)/data of the Crab Nebula using the a single Gaussian PSF. You can clearly see a ring-like excess originating from a bad parametrisation of the PSF.
!res_map.png!
Since the R68 of the PSF is the only parameter which is stored in model lookup tables, it would be nice to have a proper parameterisation of the PSF with only one parameter. My suggestion would be to include a class _GCTAPsfStudentsT_ which inherits from _GCTAPsfVector_ and just overwrites the functions like _operator()_, _mc()_ and _delta_max()_.
This class would also help to better compare results from gammalib with HESS-Analyses because the PSF would be modelled properly. Attached is a residual map (data-model)/model (data-model)/data of the Crab Nebula using the a single Gaussian PSF. You can clearly see a ring-like excess originating from a bad parametrisation of the PSF.
!res_map.png!