Systematic uncertainties

[HealthyPear]

Context

This issue will focus on the management of systematic uncertainties.
If you are instead interested in implementing statistical uncertainties, please refer to issue #87

Current status

Within the CTA-ASWG-IRF working group, there is already a sub-task group that works on this aspect (link to IRF WG main page).

I think @FrancaCassol manifested interest in this and she is already using the IRF bracketing code present under the organization account.

I think a point to clarify is if such code (or any similar one developed from it) should be part of pyirf in form of a dedicated module (if its authors agree of course) or if this task should be performed using directly the science tools since it requires only output from pyirf.

[FrancaCassol]

Hi @HealthyPear,
indeed, I confirm my interest for this task. I am at present adding to the bracketing code the possibility to independently shift the energy bias and the energy resolution (at present the shift of the pdf, which depends on the ratio E_reco/E_true, shifts them together). I will push soon this code in the original repository and then try to move it pyirf

[FrancaCassol]

Hi,

I developed some scaling functions. The idea is to give as input the irf to scale in gammapy format (e.g. EffectiveAreaTable2D, EnergyDispersion2D) and to get as return the scaled irf. I propose to add a module systematics.py containing these functions. Is this compliant with the present pyirf code?

[HealthyPear]

Hi!
Yes, it seems a good place to start!