Update on eμ/e/μ LFV resonance search SUSY LFV group 06-05-2014
Jan 29, 2016
Update on eμ/e/μ LFV resonance search
SUSY LFV group 06-05-2014
Updates since EB3
• fixed a bug on jet selection• Correct 2 inconsistency on object selection among different
channels (mu isolation and tau overlap removal)
• Apply suggestions/comments from EB3 and FAR note structure and txt Wjet & QCD further validation & Zμμ bkgd check in μ channel systematics & correlation Bayesian Vs. Frequentist method for limit extraction signal theoretical production cross-section calculation
All numbers and plots are updated
updatesNote edition add a part for trigger efficiency (Sec. 2.1)
add a subsection for signal event selection after object selection(Sec. 3.5) add subsections for signal production cross-section (Sec. 5.2 & 5.3)
Egamma trigger Muon trigger: barrel Muon trigger: endcap
Trigger efficiency study are from Trigger performance group:[58] Tricoli, A. et al., Single Electron Trigger Performance Plots. ATL-COM-DAQ-2012-146, 20 Jun 2012.[59] Muon trigger performances in 2012 (approved plots for CHEP 2013).
updates QCD further validation
validate QCD shape with SS Vs OS (also ++ Vs. --)
Reasonable agreement
updates Wjet further charge validation
Construct a Wjet control region invert the cut on the angle phi between the leptons: dphi> 2.7 1 <dphi< 2.7 Missing Et > 30 GeV (enhance W). Remove opposite sign requirement Njet< 2 (reduce t tbar).‐ Remove Pt cut (increases statistics). all other cuts are kept the same as signal event selection
Reasonable agreement
updates Zμμ bkgd study in μ channel
It seems most of the Zμμ bkgd events in μ channel are kept since a mu is mis-identified as a tau
Good Z peak truth level object match also tell us the tau is from a mu fake
updates Systematics for signal search• correlations between different backgrounds
• Systematics for signal search systematics numbers from table 3 &7&9 in supporting
note are mean values for Mll>200GeV, which are dominated by low mass region
it’s difficult to calculate systematics in each mass window for high mass region considering the statistics
A correction factor from MC
Constant number from data CR WW/Ttbar/Singletop/Zll
Similar as: y = a0 – b*x x and y are anti-correlated
updates Systematics for signal search• Construct a control region where signal events are rejected, comparing data Vs.
MC and take the difference as systematics invert the cut on the angle phi between the leptons: dphi> 2.7 1 <dphi< 2.7 all other cuts are kept the same as signal event selection
Put a 20% systematic uncertainty on total background in signal region
updates Statistical methods test for limit extraction: Bayesian Vs. Frequentist Bayesian method: the using one for this analysis Frequentist method: a standard package also used by Zeμ group http://root.cern.ch/root/html/tutorials/roostats/StandardHypoTestInvDemo.C.html
The central limit value is almost the same , but frequentist method gives smaller uncertainty bands
updates signal theoretical production cross-section calculation
Z prime
k×=LONLO
K-factor and systematics are from Z’ group
Reference:
updates signal theoretical production cross-section calculation
stau neutrino calculated with a NLO generator package
Reference:
updates signal theoretical production cross-section calculation
stau neutrino calculated with a NLO generator package
Systematics:
PDF factorization/ renormalization scale s
https://twiki.cern.ch/twiki/bin/viewauth/AtlasProtected/SUSYSignalUncertainties
Reference: SUSY group common definition
Need to be discussed ?
• How to address the 20% systematics for signal region ?
• Systematics bands from Frequentist method ?• mu mis-identified as tau ?• Z’ k-factor: directly use Z’ group’s results?
Back up