Mismatch of photon information between "Photons" and "ReconstructedParticles" collections

Dear all,

There seems to be a mismatch in reconstructed photons between the “Photons” and “ReconstructedParticles” in FCC samples in EDM4HEP format. This issue has been raised in an old thread Photons at Delphes FCC-ee but there was no clear conclusion.

If we select photons from the ReconstructedParticles using the Photon index, i.e.

.Define("photons",   "ReconstructedParticle::get(Photon0, ReconstructedParticles)")

all photons in this collect have energy > 2 GeV. This is expected as in the IDEA DELPHES the photon efficiency is set to 0 if they have energy < 2 GeV.

If the photons are selected from ReconstructedParticles not with the Photon index, but with ReconstructedParticles.type==22, there is no cut-off of photon energy and the distribution extends to 0.

(Please note that these two plots are made from different sample files so the number of entries do not mean anything. But this behavior (distribution) is the same for different samples. The ones we checked are pythia Zee, Zmumu at 91 GeV, Whizard ttbar sample at 365 GeV, MadGraph Heavy Neutral Lepton samples at 91 GeV.)

I have two specific questions related to this.

  1. Is the presence of extra low energy photons in the ReconstructedParticles expected and reasonable? In comparison, the 2 GeV cut-off is also applied to electrons and muons in IDEA DELPHES, and those low energy electrons or muons are indeed cleaned from the ReconstructedParticles collection.
  2. This is more related to FCCAnalysis. Nominally when we do jet clustering and missing energy calculation, we use the full ReconstructedParticles collection, i.e. including those low energy photons. Would this lead to an over-optimistic jet cluster or missing energy resolution?

Would be great if people could share more information/insight on this.

Best,
Xunwu

Hi Xunwu,

ReconstructedParticles are all PFCandidates, whereas Photons are intended to be well-identified photons with some (undefined) quality criterion. The cut at 2 GeV is arbitrary at this stage.

Of course including everything down to few MeV leads to optimistic resolutions and should be studied by applying minimum energy cuts on that collection. I did make a simplistic version of such study here (slide 11):

As you see this impacts mostly H->gg that has a large soft pi0 component.

I let the ECAL expert comment more, but from what I understand it is not unreasonable to expect reconstructed photons down to ~ 100 MeV which would be allowed by very low noise term. Of course to be proven also considering the material in front.

Dear Michele,

Thanks a lot for the material! This is exactly what I was wondering.
Yes, I will take note of this. In our studies, for the things sensitive to jet and missing energy resolution, we will try a few thresholds and get a sense of the impact.

Best,
Xunwu