Split up production configuration tools for parameters derived from ‘trigger’ and ‘reconstruction’ distributions

[GernotMaier]

Suggest to split up production configuration tools for parameters to avoid mixing up two very different questions:

1. parameters like energy range, view cone or core scatter radius are derived from triggered event distribution and should make use of the sim_telarray generated histograms (read with simtools.simtel.simtel_io_histogram(s)). Requires as input simulations generated with very wide intervals on energy range, view cone, core scatter area. Metric here is the ‘fraction of events lost’ by the choice of the intervals (e.g., one would not bother a fraction of 10^-5 events lost due to a view cone radius cut)
2. parameters like the total number of events to be simulated. Depends on 1 and uses reconstructed events to determine e.g. the expected statistical uncertainty on the effective area after gamma/hadron separation cuts (queries DL2 files; although one could use the sim_telarray histograms plus a simply function for the signal efficiency). Does not require simulations of increased ranges as discussed in 1). Might / should use as much any functionality provided by ctapipe / IRF production pipeline (especially if we consider other parameters than the uncertainty on the effective area)?

Especially @tobiaskleiner , please comment.

[tobiaskleiner]

Do you suggest to add another module to read a lookup table generated for (1.) or to completely split the tool in two seperate modules? Currently 1. is not ready and would require additional simulations as you said (also in dependence to az, el, nsb, configuration).
See also #1219, where there is a link to a lookup table that was once used, but is most likely too simplistic.

2. The dependence is only partly ( event scaling is per energy bin, so there is not really a dependence on energy range). For the computation of the signal efficiency we would need the full input anyway and also with configuration parameter dependence. Or do you mean to calculate this outside of simtools and provide a lookup table? Yes, could be worth it to look into ctapipe for some further uncertainty calculations, also if we want to use 2d spatial bins.

[GernotMaier]

I am not discussing lookup tables above, this is a small thing to add after we have the values. We first need the algorithms and code to derive both statistics and ranges.

[GernotMaier]

Attach the white board discussions we had today - just to make we sure we can find them again.