add full5x5 shower variables to reco::Photon #5906
Conversation
…moved from pat::Photon for this reason)
|
A new Pull Request was created by @bendavid (Josh Bendavid) for CMSSW_7_2_X. add full5x5 shower variables to reco::Photon It involves the following packages: DataFormats/EgammaCandidates @cmsbuild, @nclopezo, @StoyanStoynev, @slava77 can you please review it and eventually sign? Thanks. |
|
One disclaimer: In order to keep the code change minimal, I've reused the ShowerShape struct (as was done for the electron) Since the HoverE related variables are actually not depending on full5x5 or not, I left these extra data members as zero. This can be further cleaned up by splitting the pure EM shower shapes into a separate struct (but then we should do the same for the GsfElectron) and I would suggest to leave that for 73x if we care. |
|
the code looks ok |
|
+1 for #5906 7aadc58 tested in CMSSW_7_2_0 (test area sign720a) Here's one example comparing e5x5 default (blue) with the full5x5 e5x5 just as a proof of existence Event size of the gedPhoton branch is up by 10% on 100 events (should be less, I guess on more events); the non-ged photon branch size is up by 1% (the cost of zeros). @cmsbuild note that I've updated the validate.C script to make plots in my https://github.com/slava77/cms-reco-tools ... how do we get jenkins in sync, we may need to discuss this in the core meeting On a 72X related subject, I see a random difference in my tests of this PR in a csv variable in wflow 4.37: running valgrind now, to be followed up in slava77#53 |
|
This pull request is fully signed and it will be integrated in one of the next CMSSW_7_2_X IBs unless changes (tests are also fine). @nclopezo can you please take care of it? |
add full5x5 shower variables to reco::Photon
Correcting an inconsistency which entered in 71x and propagated to 72x.
For history purposes:
In CMSSW_7_0_X the full5x5 shower variables were added to pat::Electron and pat::Photon so they would be easily accessible in the MiniAOD (the need for these variables arose after the reco part of the release was closed and they were added during the miniaod development stage)
In 71X the full5x5 variables were added to the reco::GsfElectron, but due to some oversight/miscommunication never to the reco::Photon. The changes adding the full5x5 variables to the pat::Electron AND photon classes were nevertheless intentionally kept out of 71x, resulting in these variables being nowhere accessible for photons in 72x MiniAOD. (Well they could be recomputed from the rechits, but this is a nasty regression compared to 70x from an end user point of view.)
@lgray @arizzi @gpetruc