Mattione Update 04302014

From GlueXWiki
Jump to: navigation, search

n3pi Cut Study (Shown at Physics Meeting)

  • Run 9003 (no EM background), ~5 MegaEvents, searching for γp→π+π+π-(n)
  • Reconstructed Signal Combos (DCutAction_ThrownTopology, DCutAction_AllTruePID) (% is Signal Remaining)
    • All is 34361 combos (events) of 59562 thrown n3pi in ~5 MegaEvent sample: 57.7% Acceptance
    • +/- 5σ cut on Thrown <---> Detected matching for DCutAction_AllTruePID.
  • PROBLEM: Way too many events lost due to PID cut.
Mattione Update 04212014 n3piCutStudy Signal.png


π+ PID

  • 1D Pulls of skim tracks (tThrown - tReconstructed)/σt (Reconstructed)
    • With DCutAction_ThrownTopology, DCutAction_AllTruePID @ +/- 5 σ matching.
    • TBT is time-based tracking.
    • For track time, order of preference is: BCAL, TOF, FCAL, TBT. Histograms only filled when that detector is selected.
      • Large # of FCAL matches: perhaps track/TOF matching too restrictive?
      • Large # of TBT tracks: no detector match
    • Uncertainties used for pull (and PID) calculations are hard-coded.
    • Histograms from DHistogramAction_PID
BCAL
TOF
TBT
FCAL


  • 2D Pulls of π+'s (tThrown - tReconstructed)/σt (Reconstructed) with no cuts except +/- 5 σ thrown matching.
    • Yellow line at pull = 0.0 just to guide the eye.
    • Low-θ tracks being matched to ST & BCAL hits indicates problems with hit-matching criteria.
    • For track time, order of preference is: BCAL, TOF, FCAL, TBT. Histograms only filled when that detector is selected.
    • Histograms from DHistogramAction_GenReconTrackComparison
BCAL Pull vs p
BCAL Pull vs θ


TBT Pull vs p
TBT Pull vs θ


ST Pull vs p
ST Pull vs θ


TOF Pull vs p
FCAL Pull vs p


Alternate n3pi Cut Study

  • Run 9003 (no EM background), ~5 MegaEvents, searching for γp→π+π+π-(n)
  • If no PID cut & very loose mass: 86.3% of signal, skim size = 46.6% (through grey: -3.0 < MM2 < 5.0, Tracking FOM +/- 5σ)
  • If no PID cut & tighter mass: 83.8% of signal, skim size = 37.6% (through red: -2.0 < MM2 < 4.0, Tracking FOM +/- 5σ)
  • Reconstructed Signal Combos (DCutAction_ThrownTopology, DCutAction_AllTruePID) (% is Signal Remaining)
    • Originally had 34361 combos (events) of 59562 thrown n3pi in ~5 MegaEvent sample: 57.7% Acceptance
    • +/- 5σ cut on Thrown <---> Detected matching for DCutAction_AllTruePID.
Mattione Update 04302014 n3piCutStudy Signal.png


  • All Reconstructed Combos (% is Skim Size)
Mattione Update 04302014 n3piCutStudy All.png


Very Alternate n3pi Cut Study (No BCAL)

  • Run 9003 (no EM background), ~5 MegaEvents, searching for γp→π+π+π-(n)
  • If PID cut (no BCAL): 70.0% of signal, skim size = 21.7% (through blue: -2.0 < MM2 < 4.0, Tracking FOM +/- 5σ)
    • 15% signal loss due to PID cut without BCAL, compared to 35% signal loss with BCAL
    • Worse than above study (no PID cut: 83.8% of signal, skim size = 37.6%)
  • Reconstructed Signal Combos (DCutAction_ThrownTopology, DCutAction_AllTruePID) (% is Signal Remaining)
    • Originally had 34361 combos (events) of 59562 thrown n3pi in ~5 MegaEvent sample: 57.7% Acceptance
    • +/- 5σ cut on Thrown <---> Detected matching for DCutAction_AllTruePID.
  • BCAL time is ignored: for large-angle tracks, uses CDC (time-based tracking) time instead.
Mattione Update 04302014 n3piCutStudy SignalNoBCAL.png


  • All Reconstructed Combos (% is Skim Size)
Mattione Update 04302014 n3piCutStudy AllNoBCAL.png


Extremely Alternate n3pi Cut Study (No BCAL, KinFit)

  • Run 9003 (no EM background), ~5 MegaEvents, searching for γp→π+π+π-(n)
  • Reconstructed Signal Combos (DCutAction_ThrownTopology, DCutAction_AllTruePID) (% is Signal Remaining)
    • Originally had 34361 combos (events) of 59562 thrown n3pi in ~5 MegaEvent sample: 57.7% Acceptance
    • +/- 5σ cut on Thrown <---> Detected matching for DCutAction_AllTruePID.
  • BCAL time is ignored: for large-angle tracks, uses CDC (time-based tracking) time instead.
Mattione Update 04302014 n3piCutStudy SignalNoBCALKinFit.png


  • All Reconstructed Combos (% is Skim Size)
Mattione Update 04302014 n3piCutStudy AllNoBCALKinFit.png


  • Problem: No/oddly-shaped sidebands remain for sideband-subtraction analyses.