Corrected jets in CMSSW

1
Corrected jets in CMSSW

• Can run a FW module to correct jets
• Consistent with 1_2_0 release
• For Midpoint Cone R0.5 and 0.7
• For Iterative Cone R0.5
• Instructions https//hypernews.cern.ch/HyperNews/
  CMS/get/JetMET/83.html
• Makes separate collections
• FW-based only
• FWLite version in the works (function)

2
Editorial Comments

• Making separate collections is overkill
• p4(CorrectedJet) C(PT, ?, ?, em/had, lumi,)
  p4(OriginalJet)
• Why not store one number per jet, or just provide
  a function?
• Function be easily made totally transparent to
  FWLite.
• This is one of the most obvious use-cases for
  FWLite!
• Particularly because there are so many jet
  collections
• Iterative vs MidPoint Cone, ?R0.5 vs 0.7, Kt
  algorithm
• As well as so many possible levels/type of
  correction
• Correct to particles in cone or to original
  parton?
• Apply or dont apply underlying event correction
  (pileup!)?
• b-jets vs light quark jets?
• Only calorimeter based vs trackingcalorimeter
  based (eFlow)
• And my experience is that these can change quite
  rapidly
• Every byte in AOD is precious.why waste them?
• ltend of rantgt

3
Tried to use corrections

• Recipe seems to work
• cvs co -r JetResponseV8 JetMETCorrections/MCJet
• cvs co -r V00-00-07b RecoJets/JetAnalyzers
• Add this incantation to your cmsRun cfg file
• include "JetMETCorrections/MCJet/data/JetCorrectio
  ns120mcone5.cff
• Insert module (sequence?) JetCorrectionsMcone5 in
  your path
• And out you get a new jet collection
  corJetMcone5
• Provided you had the right collection to start
  with, of course

4
Next try to use it with our BBox

• I made it work, but it was clunky
• This is how I would like the jet BBox to
  function
• // get vectors of uncorrected jets and corrected
  jets with no Pt cut
• cms1Cuts jetCut
• jetCut.ptmin 0.
• stdvectorltconst recoCaloJetgt jets
  jets.getJets(mcone05, jetCut)
• stdvectorltconst recoCaloJetgt c_jets
  jets.getJets(correctedmcone05, jetCut)
• Then I have access to both corrected and
  uncorrected jets with a minimum of fuss
• The mumbo-jumbo is all hidden

5

• This, however, is not how it works
• Current jet BBox allows me to access
  one-and-only-one jet collection at a time
• It is possible to choose which one in a cfi file
• In order to access 2nd collection I had to go in
  and modify by hand
• myModule/interface/myModuleFW.h
• myModule/src/myModuleFW.cc
• myModule/data/myModule.cfi
• Jets/src/Jets.cc
• Jets/interface/Jets.h
• ..and how to do it wasnt immediately obvious
• ..and this was for FW only
• ? The Jets BBox is not black enough

6
Lesson

• The Jet BBox should allow me to easily access
  several jet collections in a single job
• The names of these jet collections will probably
  be a moving target
• ? Must maintain flexibility
• This may apply to other object types too
• Did not commit mods to access 2 jet
  collections because Dmytro tells me that things
  will change significantly
• But I did convey to him what I think we need..

7
One random event

• ---- Event 2  ------------ Uncorrected Jets
  Pt 10.3196 Eta -2.816 Phi -0.8323
  Pt 5.40623 Eta -3.48884 Phi 0.0532513
  Pt 4.20876 Eta -0.0143419 Phi 2.61109 Pt
  3.95908 Eta 4.08236 Phi -2.07385 Pt
  3.24885 Eta 1.51637 Phi -2.80799 Pt
  2.73552 Eta 2.37856 Phi -0.179766 Pt
  1.19576 Eta -0.4785 Phi -0.392699 Pt
  1.02987 Eta 3.2265 Phi -0.436332 Pt
  1.00978 Eta 1.0875 Phi -1.61443
• Corrected Jets Pt 19.8062 Eta -2.816
  Phi -0.8323 Pt 11.5128 Eta -3.48884 Phi
  0.0532513 Pt 14.1202 Eta -0.0143419 Phi
  2.61109 Pt 9.93978 Eta 4.08236 Phi
  -2.07385 Pt 12.0436 Eta 1.51637 Phi
  -2.80799 Pt 8.20876 Eta 2.37856 Phi
  -0.179766 Pt 4.56752 Eta -0.4785 Phi
  -0.392699 Pt 2.64958 Eta 3.2265 Phi
  -0.436332 Pt 3.71985 Eta 1.0875 Phi
  -1.61443
• Note corrected jets appear to no longer be Pt
  ordered

8
One more (general) comment

• Usage of BBoxes is now tied to TableMaker
• Configured in TabeMaker cfg/cfi files
• TableMaker is an example of a hi-level analysis
  class
• There will be many more hi-level analysis classes
  for different studies as we proceed
• All of these classes will need BBoxes
• Right now a bit painful to instantiate new
  analysis
• Need to either
• Decouple BBoxes from TableMaker (hi-level anal.
  class)
• OR
• Make it a easier to start a new analysis that
  uses BBoxes, eg, a hi-level-analysis base-class?