Catalog ultralegacy corrections to use
See SUS workshop 31.01.24 presentation.
Tau 2018 UL
-
Genuine tau, e faking tau, and muon faking tau energy scales: "tau_energy_scale"
-
DeepTau vs. electron: DeepTau2017v2p1VSe
-
DeepTau vs. muon: DeepTau2017v2p1VSmu
-
DeepTau vs. jet: DeepTau2017v2p1VSjet
-
Trigger efficiencies for mutau: tau_trigger
, typemutau
(also needs to knowDeepTauVSjet
working point used). The Embed UL TWiki says the trigger values are provided by the Tau POG (and the files they link in the little table give the same values as the Tau POG file) so I only load and use the Tau POG file. -
Trigger efficiencies for etau: tau_trigger
, typeetau
. Not implementing UL until I know how to do the single electron trigger efficiency (issue #102). -
Emb Energy scale: not available, work in progress. Currently using MC values for now. -
Emb DeepTau vs. jet, in tau_embed.json
Muon 2018 UL
The necessary information is split between the MUO repository and the central NanoAOD JSON repository so you need both repositories. From the MUO repository we only need Efficiency_muon_generalTracks_Run2018_UL_trackerMuon_avg.json
, which is simple enough to hard-code, so I do that.
-
Muon energy scale in MC and Embed: After reading this thread, I think the averaged SFs across pT bins are here with the avg in the file name, so I think in principle "avg" files. So, this one for 2018 UL. I stared at it some more and the pre-UL uncertainties that we use are actually more conservative than this recommendation (we bin in only three eta bins, but it's conservative enough to accommodate the four eta bins). So I'm just going to keep the pre-UL setup. -
ID efficiencies: we use mediumID, so we want NUM_MediumID_DEN_genTracks
.sf
is nominal. While the 'systup' and 'systdown' are up/down variations with total stat+syst uncertainties, in HIG-22-007 the uncertainty for this is a yield uncertainty. -
Iso efficiencies: we used mediumID for the ID. In tne mutau
channel, we hadMuon_pfRelIso04_all < 0.15
, which is theTight
working point defined here. In theemu
channel the signal region also uses theTight
working point. So we want to useNUM_TightRelIso_DEN_MediumID
, and 'sf' is nominal. While 'systup' and 'systdown' are up/down variations with total stat+syst uncertainties, in HIG-22-007 the uncertainty for this is a yield uncertainty. -
Single muon trigger efficiencies, MC and Embed, for mutau channel): there is only one available in the Muon POG file: NUM_IsoMu24_DEN_CutBasedIdTight_and_PFIsoTight
. This is a bit awkward because we use the medium ID and tight PF isolation. In the Muon-only repo, they have a mediumID and PFIsoMedium trigger efficiency, but neither of these are exactly what we use. -
Electron+muon cross-trigger efficiencies: TBD (issue #102).
Not used/considered for this analysis as far as I know:
- Muon reconstruction scale factors: near 1.0 and not recommended for analyses which aren't precision analyses.
- Rochester energy and momentum scale corrections.
- The muon energy scales that are also binned by pT: files without avg in the filename are binned in eta and pT which seems like overkill.
- Muon tracking efficiencies are essentially 1.0 and no corrections are recommended
- Muon energy scale in Embedded: negligible according to 2018 pre-UL TWiki: "The uncertainty and correction on the muon energy scale is negligible in comparison."
Electron 2018 UL
The necessary files are split between the EGM repository and the central NanoAOD JSON repository. I cloned both and copied by hand the files needed from the EGM repository to where I have checked out the central NanoAOD JSON repository.
-
Electron energy scales: nominal values are applied already in NanoAOD v9 and up. The uncertainties need to be taken from these .json files which are in the EGM repository. I was scouring this forum thread which dates to before the .json were introduced so it's not entirely relevant, but was good for mulling over and this forum thread which only pertains to photons, but gives me confidence that NanoAOD v9 has the central energy scale already applied. -
Electron MVA ID scale factor: MC and Embed: There is wp80iso, wp80noiso, wp90iso, wp90noiso
inUL-Electron-ID-SF
in/eos/user/s/skkwan/jsonpog-integration/POG/EGM/2018_UL/electron.json
. We useElectron_mvaFall17V2noIso_WP90
, sowp90noiso
. The uncertainty for this is a yield uncertainty in HIG-22-007 in the datacards here for MC and here for Embed -
Electron energy scale in Embed: not available yet. I will just apply the MC values for now. -
Single electron trigger efficiency: MC: we may have to derive our own (issue #102). As a temporary stopgap, either apply a value of 1.0, or see if we can use values that Cécile derived in 2023 on this TWiki to at least get the implementation code in there. -
Electron trigger efficiency: Embedded: ? -
Electron+muon cross-trigger efficiency: see issue #102 .
Not used/considered for this analysis as far as I know:
- Reco efficiencies.
- "Isolation" efficiency - I think this is all wrapped up in the MVA ID scale factor.