LHCEWWG-MB + VBScan: VBS WZ Analysis discussion

40/2-A01 (CERN)



Show room on map

EWK LHC + VBScan WZ meeting(minutes)

link to the agenda: https://indico.cern.ch/event/749140/




  • VBSCan eager to welcome more people than already involved and more activities, acting as a driving force for VBS studies in the LHC EW WG as well.




  • ptj > 40GeVin the definition of N_jet, one of the inputs to the MVA
  • interference is included in the signal and used as uncertainty
  • MathieuPellen: do you observe significant discrepancies between generators for the signal?
    • Emmanuel Sauvan: Sherpa sample is used as well
    • Joany Manjarres: the difference between Sherpa and Madgraph is smaller here(12-20%) wrt SSWW. The difference is still to be understood, though
  • MP:why didn’t you quote a result for WZjj putting together QCD+EW?
    • for the publication, planning to publish QCD+EW XS
  • MP: why didn’t you define a loose selection region as well, where to measure the cross-section?
    • ES: the loose selection of CMS corresponds to the one of ATLAS
    • MP: a simpler phase space would be useful, easier for theory computations, for the use of the results from theory
    • ES: Rivet routines will be included as well
  • Senka Duric: are you planning to release difference among generatorrs for the paper?
    • ES: discussing, probably yes
  • Kenneth Long: sl. 8, what are the inputs of the the Asimov for the bkg?
    • ES: the bkg is postfit, therefore the expected signifcance of 3.3 already accounts for the fact that the bkg is small wrt predictions, and is only due to the excess in the signal refion
    • Q: what would be the a-priori significance then? 
    • ES: about 2.1 sigma
    • KL: to properly compare the analysis, one should run the MC thorough the same Rivet routines, to have an apple-to-apple comparison: CMS expected is purely MC
    • ES: agrees
  • SD how was the optimisation done? Completely blindly? No, the background ctrl regions were unblinded while the signal region was blind
  • Kristin Lohwasser(KrL):still about the point of running through the same Rivet routines. Discussed in the LHC EW WG, it seems ATLAS and CMS could do it, it could be a good case to do it for the first time
    • KL: for CMS the numbers are already produced w/ Rivet routines(theones containing the selections only), public from LHE workshop studies. Cuts can be changed and routines rerun if needed
    • KrL: who could do it for ATLAS? 
    • KrL: Could this be done for other analyses as well?
    • SD: Rivet routines will become available when the paper is out
    • KL: the CMS routine is already public, the ATLAS should still be shared
    • KrL: proposes to take CMS routine(whichis the outcome of a LesHouces workshop) and run it on ATLAS MC, and produce a comparison plot
  • Pietro Govoni: do you use the Stewart-Tackmann recipe for the theory scale uncertainty on Njet?
    • Louis Portales, ES: the scale variation by itself is used only, this is a point to be investigated though
  • Nick Q: How was the undertainty through BDT propagated? Samplying only Up/Down or more values? A: Only Up/Down were propagated.




  • ES: about the XS meas: there’s a ptmiss cut at 30 GeV, while the fiducial region does not have the cut implemented
    • KL: prefer to extrapolate in only one direction, while setting pt of neutrino to 30 GeV migrations can happen in two directions, therefore the cut is left out
    • ES: does this introduce theory acceptance uncertainties?
    • KL: yes, it does and it’s accounted for, probably the impact of this particular cut is at the order of few percent
  • ES: what’s the model uncertainty for the extrapolation to the loose fiducial region?
    • KL included in the extrapolation, at the order of 5%




  • SD: also the inclusive WZ analysis is out. Did you do any consistency checks in the overlapping region of the two analyses?
    • ES: yes. The two analyses are completely coupled, ad the VBS analysis is a subset of the inclusive analysis, where jet-related variables are measured(andcomparison to generators are performed).
    • PG: do you see the same scale factor for QCD WZ?
    • ES: in the 2jet bin there’s a similar behaviour(evenif there the selection is on jets w/ pt > 25 GeV) of the Sherpa simulation. Better agreement w/ Sherpa 2.1
  • SD: was the difference in modeling due to different generators included, as a source of uncertainty in the VBS analysis?
    • ES: no, the comparison between the two Sherpa(2.1and 2.2) is not used as a source of systematic uncertaintybut it is within the already used uncertainties
    • JM: Behaviour observed seems consistent also for other, similar analyses in ATLAS
  • KL: what are the plans for the combination?
    • SD: in VBSCan we’re getting ready for the combination, generating privately signal models and fitting by using public information. In the LHC EW WG it is done internally, i.e. including private information of the collaboration. All relevant uncertainties are needed of course
    • ES: the differential unfolded distributions are published with the aim of easing the combination
  • Action items:
    • KrL: writing an email to Kenneth to ask for the Rivet routine for WZ
    • LHC EW WG leaders: contact ATLAS and CMS to find reference people for SSWW routines
      • Compile list of routines/final states available(speakingalso to Jonas), 
      • circulate within ATLAS group for MC-Multiboson studies / circulate with CMS people to gauge interest from their side
There are minutes attached to this event. Show them.
    • 4:00 PM 4:20 PM
      VBScan: Possible Combination based on public information 20m
      Speakers: Pietro Govoni (Universita & INFN, Milano-Bicocca (IT)), Senka Duric (Kansas State University (US)), Senka Duric (Kansas State University (US))
    • 4:20 PM 4:40 PM
      VBS WZ Analysis: ATLAS 20m
      Speaker: Louis Portales (LAPP (FR) - Université Savoie Mont-Blanc)
    • 4:40 PM 5:00 PM
      VBS WZ Analysis: CMS 20m
      Speaker: Kenneth Long (University of Wisconsin Madison (US))