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
[10_2_X] Fall17 V2 Photon ID and PhotonIDValueMapProducer speedup #25372
[10_2_X] Fall17 V2 Photon ID and PhotonIDValueMapProducer speedup #25372
Conversation
A new Pull Request was created by @guitargeek (Jonas Rembser) for CMSSW_10_2_X. It involves the following packages: PhysicsTools/PatAlgos @perrotta, @monttj, @cmsbuild, @fgolf, @slava77, @peruzzim can you please review it and eventually sign? Thanks. cms-bot commands are listed here |
@guitargeek I leave to reconstruction and release managers the discussion on how to do the backport technically. From the NanoAOD production perspective, the part that is on top of #25347 should be completely transparent. Thanks |
Hi Marco, yep it will be discussed today in the RECO meeting. But can you please explain again what you mean with "the part that is on top of #25347 should be completely transparent"? Thanks! |
Hi @slava77 and @perrotta, so did we agree in the end on the minimal backport solution actually? For 10_2_X, that would mean #25347 and this PR which I'd kindly ask to get tested. About the minimal backport of V2 IDs to 9_4_X: I was not entirely correct with my assessment this afternoon that it would only require new config files. In reality, the variables are slightly different, as the V2 ID takes the absolute value of track-cluster matching variables [2] while V1 does not [1]. That means an additional I suppose I'll just quickly set up the minimal backport to 9_4_X and make a PR, so it can be further discussed there. Jonas [1] https://github.com/cms-sw/cmssw/blob/master/RecoEgamma/ElectronIdentification/data/ElectronMVAEstimatorRun2Fall17V1Variables.txt#L36 |
Ok @perrotta, @slava77, @Sam-Harper and @michelif, again about the 9_4_X backport. I have the minimal backport set up now [1]. In the end, it was not so trivial as expected, since we also need a new cut class for the cut ID and for the electron MVA, I need also to produce the BDT output before the logistic transformation now. Please tell me which option you prefer and I will validate it and make the backport PR:
[1] https://github.com/guitargeek/cmssw/tree/EgammaID_9_4_12_minimal |
@cmsbuild please test with cms-sw/cmsdist#4518 |
The tests are being triggered in jenkins. |
@guitargeek about the backport to 94X, I think we agreed on the "minimal support", and I would concentrate on it by now. The "full backport" can be arranged for a later stage, if needed. About currently open pull request for 10_2_X: since we agreed about backporting both V2 IDs and the code speed up, this is the pull request which contains them. At this point, #25313 is superseeded, and #25347 is only a subsample of this one, and therefore redundant. I think you can close both them, now. |
Comparison job queued. |
Pull request #25372 was updated. @perrotta, @andrius-k, @kmaeshima, @schneiml, @cmsbuild, @jfernan2, @fgolf, @slava77, @peruzzim can you please check and sign again. |
Okay @peruzzim , I tested some workflows with MiniAOD locally and they pass. It should be safe now to test with #25444 cms-sw/cmsdist#4518. Sorry for the initial problems with this. |
please test with #25444 cms-sw/cmsdist#4518 |
The tests are being triggered in jenkins. |
+1 The following merge commits were also included on top of IB + this PR after doing git cms-merge-topic: |
Comparison job queued. |
Comparison is ready Comparison Summary:
|
+xpog for the change to |
+1 |
+1
|
+1 |
merge |
Primarily, this is the backport of #25293 plus changes to make Photon MVA V2 work before #24131 and adapting to new weights files names.
Additionally, since the NanoAOD developers would really appreciate the PhotonIDValueMapProducer speedup from #25092 to be backported to 10_2_X. Here, I suggest to simply synchronize the PhotonIDValueMapProducer (and required new header files) with master to get the speedup.
Validation plots for all photon ID value maps:
https://rembserj.web.cern.ch/rembserj/plots/CMSSW_PRs/25372/
Validation workflow:
Depends on cms-sw/cmsdist#4518.
[1] https://github.com/guitargeek/cmssw/blob/f76b6a2c9425d5ad5df91df5ef04272d0ab40e39/RecoEgamma/EgammaTools/test/testPhotonIDs_cfg.py
[2] https://github.com/guitargeek/cmssw/blob/f76b6a2c9425d5ad5df91df5ef04272d0ab40e39/RecoEgamma/EgammaTools/test/photon_validation_10_2_6.py