Signal Definition

Introduction

For theorists, it is more precise and easier to calculate inclusive photon cross section, i.e. photon cross section without any isolation requirements.

For experimentalists, however, one has to apply isolation cuts when doing analysis, in order to suppress background from hadron decays. If the underlying event and fragmentation are well modeled in the simulation, we could use simulation to obtain the efficiency of applying isolation cuts at the reconstruction-level, and report an inclusive photon cross section.

However, we know underlying events and fragmentation are not modeled correctly in current version of CMSSW. In order to minimize systematic uncertainty on the efficiency, the best way is to report isolated photon cross section. Then, the efficiency becomes:

           eff =        recoIso < xxxx GeV + other cuts
                       -------------------------------------------

                          genIso <  yyyy GeV  
 

If xxxx and yyyy are equivalent, then, systematic uncertainties from UE and fragmentation can be minimized.


How do we determine yyyy? How do we determine how isolated the photons have to be when we report this x-section?

Answer 1: Find the best isolation cuts at the reco-level that gives the best signal and background separation, then, do a mapping from the reco-level to gen-level.

Answer 2: Determine a priori where to make a cut at the generator-level, by scanning the signal efficiency of isolation at the generator-level.

A priori definition by scanning cuts on generator-level isolation

  • Introduction
    • Generate Photon_Pt15 samples, plot generator-level isolation distribution for photon pt > 20 GeV, |η| < 2.5
    • Integrate the isolation distribution to find the point that gives 90% and 95% efficiencies
    • Use generator-level isolation for cone Δ R = 0.4
    • Conclusion: we could use 5 GeV as a first value for getting 95% signal efficiency
    • ATLAS tune 90% efficiency:
      atlas_geniso_10k.root_0.90.gif
    • ATLAS tune 95% efficiency:
      atlas_geniso_10k.root_0.95.gif
    • D6T tune 90% efficiency:
      d6t_geniso_10k.root_0.90.gif
    • D6T tune 95% efficiency:
      d6t_geniso_10k.root_0.95.gif

Comparison of generator-level and reconstruction-level isolation

  • Comparison of experimental generator-level isolation and reco-level isolation: studies by Chia-Ming, using similar reconstruction algorithms at the generator-level
  • Studies from Eiko (reco-level isolation includes ecal and hcal only)
    • Use generator-level isolation for cone Δ R = 0.4
    • Reco-level isolation = ecal + hcal isolation with Δ R = 0.4 (ecalRecHitSumEtConeDR04 + hcalTowerSumEtConeDR04)
    • Within MPA, the matching is done by looking for the highest gen-photon that has ΔR < 0.15 from the reco-photon
    • Basic requirements are: photon 15 < pt < 30 GeV, |η| < 2.5, hadronicOverEm < 0.15
    • D6T tune Photon_Pt15 MPA samples.
      The upper figure studies profile of reco isolation vs gen isolation, where the error bars are spreads.
      The lower figure studies the spread perpendicular to the red line fit (rotate to the coordinate of principle axis)
      bestGenCalIsoDR04.gif
  • Studies from Eiko (reco-level isolation includes ecal and hcal only)
    • Use generator-level isolation for cone Δ R = 0.4
    • Reco-level isolation = ecal + hcal isolation with Δ R = 0.4 (ecalRecHitSumEtConeDR04 + hcalTowerSumEtConeDR04)
    • Within MPA, the matching is done by looking for the highest gen-photon that has ΔR < 0.15 from the reco-photon
    • Basic requirements are: photon 15 < pt < 30 GeV, |η| < 2.5, hadronicOverEm < 0.15
    • D6T tune Photon_Pt15 MPA samples.
      The upper figure studies profile of reco isolation vs gen isolation, where the error bars are spreads.
      The lower figure studies the spread perpendicular to the red line fit (rotate to the coordinate of principle axis)
      bestGenCalIsoDR04_trackIso.gif

-- ShinShanYu - 05-May-2010

Topic attachments
I Attachment History Action Size Date Who Comment
GIFgif atlas_EffGenIso_10.root_0.90.gif r1 manage 14.3 K 2010-05-06 - 14:00 ShinShanYu  
GIFgif atlas_EffGenIso_10.root_0.95.gif r1 manage 14.4 K 2010-05-06 - 14:00 ShinShanYu  
GIFgif atlas_geniso_10k.root_0.90.gif r1 manage 14.5 K 2010-05-05 - 14:51 ShinShanYu  
GIFgif atlas_geniso_10k.root_0.95.gif r1 manage 14.5 K 2010-05-05 - 14:52 ShinShanYu  
GIFgif bestGenCalIsoDR04.gif r2 r1 manage 11.0 K 2010-05-11 - 20:20 ShinShanYu  
GIFgif bestGenCalIsoDR04_trackIso.gif r1 manage 9.7 K 2010-05-11 - 20:21 ShinShanYu  
GIFgif d6t_geniso_10k.root_0.90.gif r1 manage 14.1 K 2010-05-05 - 14:52 ShinShanYu  
GIFgif d6t_geniso_10k.root_0.95.gif r1 manage 14.2 K 2010-05-05 - 14:52 ShinShanYu  
Edit | Attach | Watch | Print version | History: r4 < r3 < r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r4 - 2010-05-11 - ShinShanYu
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Main All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback