Review this package in light of gammapy recent developments - Githubissues

cta-observatory / iact-estimator

Estimate the observability properties of a source with an IACT system

https://iact-estimator.readthedocs.io/latest/?badge=latest

2 stars 0 forks source link

Review this package in light of gammapy recent developments #3

Open HealthyPear opened 8 months ago

HealthyPear commented 8 months ago

Since the original script was developed, gammapy evolved quite a bit.

At some point we should review the entire package to see if we can substitute some features with gammapy and/or integrate some more advanced features.

HealthyPear commented 2 months ago

This would basically entail to not use the current scripts, but the gammapy API top use the IRF of the instrument which the user wants to use for the proposal and run an estimation of the flux / sky map

[ ] 1D spectral analysis ( e.g. https://docs.gammapy.org/1.2/tutorials/analysis-1d/spectrum_simulation.html) (#33)
[ ] 3D full analysis (e.g. https://docs.gammapy.org/1.2/tutorials/analysis-3d/simulate_3d.html + spectra for whole extended source and parts of it)

HealthyPear commented 1 week ago

OK I am splitting this issue in smaller ones after the first tests:

I have a working version from MAGIC point-like DL3 observations (gammapy dataset) for the estimation of 1D spectrum but no events
same from CTA example irfs from gammapy
I tested gammapy.data.ObservationsEventsSampler and with some fixes it is possible to simulate observations based on full-enclosure IRFs and possible perform the full 3D estimation analysis

HealthyPear commented 1 week ago

The configuration system might be very low-level at the beginning, but I am trying to use the same configuration schema as provided by gammapy's high-level interface.

Unfortunately this issue upstream prevents us to use it in one go - that would be awesome.

HealthyPear commented 1 week ago

ObservationsEventsSampler is not contemplated directly in the HLI of gammapy, but if we write to file the fake observation then it should work out of the box!

HealthyPear commented 1 week ago

gammapy devs explained me that ObservationsEventsSampler is not really needed unless for data challenges or time-variable data

in fact simulating a 3D map dataset already allows to get events sampled in the fov from just faking the dataset counts

the only limitation that seems to be remaining is doing a plot like the one in this tutorial section when we do not have simulated events, but just counts from an on/off spectrum dataset

HealthyPear commented 1 week ago

An additional check that I want to add is for the target source altazimutal trajectory against the same coordinates valid fro the IRFs input by the user