Use thrown event histogram when computing IRFs

[kosack]

Description

• when making the IRFs (Aeff and sensitivity), the thrown energy histogram that is stored in the SimTel and DL1 files should be used for weighting, rather than the analytic model of the expected spectrum parameters.
• This will be less error-prone if we combine simulations that used different:
  • input spectra
  • start and ending energy range
  • core offset distribution

In the DL1 files, the simulation/service/shower_distribution table provides this information as a 2D histogram of thrown events per energy and per offset bin. In each row of the table is the information for each run included in the file, so summing the histograms will produce the final distribution.

Examples

THe location of the histograms in the DL1 format:

An example of combining simulations with different spectral parameters: note that N_sim(E_true) is not a pure power-law, but a combination of several

[HealthyPear]

note to self

at the end of the scripts data_training or write_dl2 we reach also the end of the EventSource instance so the source.file_.histograms are filled

for h in source.file_.histograms:
    if h["id"] == 6:
        selected_histogram = h
            break

then add selected_histogram to new table in each HDF5 file
maybe add runheader info so we do not have to read it from performance.yaml