Electron / photon scale and resolution corrections for ICHEP 2016

[emanueledimarco]

Implemented the corrections from https://twiki.cern.ch/twiki/bin/viewauth/CMS/EGMSmearer To use them, you need to:

• cd EgammaAnalysis/ElectronTools/data
• git clone -b ICHEP2016_approval_4fb https://github.com/ECALELFS/ScalesSmearings.git ; cd -
• scram b

To configure the corrections

• electrons:

  lepAna.doElectronScaleCorrections = { 'data' : 'EgammaAnalysis/ElectronTools/data/ScalesSmearings/80X_Golden22June_approval',
  'GBRForest': ('$CMSSW_BASE/src/CMGTools/RootTools/data/egamma_epComb_GBRForest_76X.root', 'gedelectronp4combination'+era), 'isSync': sync }

• photons:

  photonAna.doPhotonScaleCorrections = { 'data' : 'EgammaAnalysis/ElectronTools/data/ScalesSmearings/80X_Golden22June_approval', 'isSync': sync }

[schoef]

@emanueledimarco

thanks a lot. era should be '25ns', sync should be False. I sanity-tested the PR, change in pt for photon, electron is within expectation.

[emanueledimarco]

@schoef yes, sorry, I omitted the values for those parameters. They are like you say. I also tested it and it runs, but I am now running on full DoubleEG and SinglePhoton to make a more extensive validation (also because there were rare crashes)

[schoef]

@emanueledimarco OK, please let me know because I'm about to launch a full scale tupelization Thanks for the heads-up

[emanueledimarco]

And this should also fix the other problem reported by Philipp, i.e. that if the category is not found, the corrector tried to access an incorrect memory location, corrupting in irreproducible way the subsequent modules (electron ID modules, for example)

[schoef]

@emanueledimarco I processed almost all available mAODv2 with this branch (including f38d7ff ) and have so far not seen a failure.

[emanueledimarco]

I have also validated the scale and resolution corrections on the golden JSON with 7.65 fb-1 and they work as expected.