-
Notifications
You must be signed in to change notification settings - Fork 4.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fixing the calculation of MET Uncertainties by using PFMuons, PFElectrons, PFTaus, and PFPhotons #26302
Fixing the calculation of MET Uncertainties by using PFMuons, PFElectrons, PFTaus, and PFPhotons #26302
Conversation
…and PFPhotons only
The code-checks are being triggered in jenkins. |
+code-checks Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-26302/8996
|
A new Pull Request was created by @eioannou (Emilios Ioannou) for master. It involves the following packages: PhysicsTools/PatUtils @cmsbuild, @perrotta, @santocch, @slava77 can you please review it and eventually sign? Thanks. cms-bot commands are listed here |
please test |
The tests are being triggered in jenkins. |
Comparison job queued. |
Comparison is ready Comparison Summary:
|
@Sam-Harper , FYI |
Here you re-define pfElectrons, pfPhotons, pfMuons, pfTaus, and then you use them to calculate the MET uncertainties. Isn't it quite error prone doing so? What happen if any of those objects gets re-defined for the MET calculation sometime in the future? Still you will end up with having decoupled the calculation of the MET and the one of its its error, as it was the origin of the issue pointed out in #25751. The original python configurations are quite convoluted and I can hardly reconnect all pieces... However, I would say that the real fix would be to reuse the very same objects that entered the calculation of the MET for the calculation of the MET uncertainties, which means import'ing for the errors the same configuration which are used (elsewhere) for the MET. Is there any technical hindrance in doing so? |
If I change the name of the collections (pfElectrons, pfPhotons, pfMuons and pfTaus) which will be used to the met uncertainties calculation, will it be okay? |
Sorry, my point was kind of the opposite: why cannot you import the very same collections which were used (elsewhere, I guess) to compute the MET? As such, you wouldn't need to modify them here by hand if some of those collections will get re-defined for the MET calculation (e.g. different thresholds, etc.). |
I guess the issue here is that PF candidates and packed candidates have slightly different interfaces for the cuts, so one needs to somewhere define a selection for packed candidates: |
If I am correct, the MET is computed in this config file cmssw/RecoMET/METProducers/python/PFMET_cfi.py Lines 1 to 11 in e2a0510
So, the collection which is used to compute the MET is the "particleFlow". However, if I try to compute the MET uncertainties by using only the particleFlow, I guess, I have to modify the whole code which calculates the met uncertainties in the MET tool. So, an idea was, before the calculation of the met uncertainties, to create new collections of PF Objects only. The current code uses all objects (Muons, Electrons, Taus, Photons) to compute the met uncertainties which it is wrong. So, my idea was to create collections which include only PF objects before the calculation of met uncertainties. |
@eioannou this PR modifies in a significant way the errors on MET. |
I will prepare a presentation and will include the link here when it is ready to be presented at MET meeting. |
@eioannou |
ping @eioannou |
Hi @perrotta, I am trying to prepare the presentation about this PR, by performing some validation tests, and I am going to present the results in the next JETMET meeting. Sorry for the delay. |
type bugfix |
@eioannou it would be nice to include this bugfix in 10_6. I don't think there will be any JetMET meeting by tomorrow, the deadline set for the the latest open pre of 10_6: if you have any validation ready, and if you are confident enough that it does correspond to what the MET uncertainties must be, please post it here. |
I will try to post the plots here today. |
I have performed validation tests for MET uncertainties of Type1 MET. The results are shown below. The green lines correspond to MET Uncertainties which are calculated by using PF particles only and the red lines stand for the MET uncertainties which are calculated by using all particles (as it is done until now). The plots show that the changes are minor between of two calculations, however, the correct method to compute the MET uncertainties is to use only PF particles in the calculations. Validation tests were performed by using a DY sample of 50K events (DYJetsToLL_M-50_TuneCP5_13TeV-amcatnloFXFX-pythia8). |
+1
|
+1 |
merge |
+1 |
This pull request is fully signed and it will be integrated in one of the next master IBs (tests are also fine). This pull request will be automatically merged. |
This PR is submitted in order to fix a bug in the calculation of MET Uncertainties. MET Uncertainties must be calculated by using PF Electrons, PF Photons, PF Taus, and PF Muons only.
The additional snippet of the code finds PF objects and stores them into the collections and then, those collections are used for the calculations of MET uncertainties
This PR is related to the issue #25751