DiBoson Analysis

1
DiBoson Analysis
Thomas Barber University of Cambridge
2
Overview

• Why is the ZZ?nnll channel useful to study?
• Summary of analysis algorithm used.
• Summary of Datasets used.
• Details of private MC_at_NLO signal sample.
• Comparison of csc signal and private sample.
• Signal and background comparisons.
• Cut comparison of full sim and fast sim data.

3
Why is the ZZ?nnll useful?

• Comparison to the 4-lepton channel ZZ?llll
• BR(Z?invisible) 20.00 0.06
• BR(Z?ee) 3.363 0.004
• BR(Z?mm) 3.366 0.007
• So the overall diboson branching ratio
• BR(ZZ?llll) 0.453
• Compared to the missing energy channel
• BR(ZZ?llnn) 2.81
• This gives a ratio of branching fractions
• BR(ZZ?llnn) / BR(ZZ?llll) 6.2
• Around 6 times as many events to be produced
  compared to the 4-lepton channel.
• For now concentrate on the electron channel.

4
Analysis Algorithms

• Custom written algorithms based on
  AnalysisSkeleton.
• Electron ID requies that isEM 0
• Overlap removal done by EventView.
• Algorithms run over EventView containers.
• Initial cut on jets Pt gt15 GeV
• Initial electron cut h lt 2.5

EVToolLooper.EventViewTools
"EVElectronInserter", "EVPhotonInserter,
"EVMuonInserter", "EVTauJetInserter",
"EVParticleJetInserter"
5
Dataset Details
6
Dataset Details
Calculated from MC_at_NLO (see next slide)
Changed from web values
http//jarguin.home.cern.ch/jarguin/dc3requests_sm
.html
Events in Cambridge
7
Notes on Data (1)Private MC_at_NLO Sample

• Small sample of 2682 ZZ?eenn events produced.
• Underlying diboson event done by MC_at_NLO
• Using the CTEQ6M pdf (Les Houches Accord library
  10000)
• Decay done by Jimmy, with Tauola Photos.
• Sim, Digi Reco done in 12.0.4
• Detector description 'ATLAS-CSC-01-02-00
• Material distortions in Inner Detector and
  Calorimeters
• Cross section for ZZ production from MC_at_NLO is
• s 14.74 0.08 pb
• Using BR calculate the ZZ?eenn cross section
• s 198 fb
• Use this sample to compare with LO Pythia sample.

8
MC_at_NLO Events in Atlantis
9
Notes on Data (2)csc Signal Sample

• Csc signal sample 5981 ZZ?llnn.
• csc11.005981.Pythiazzllnunu.recon.AOD.v11004213
• Uses pdf set CTEQ6L
• 35500 events generated using Pythia in 11.0.4
  (leading order)
• Unfiltered cross section for ZZ production from
  Pythia is
• s 11.14 pb
• This is lower than the MC_at_NLO cross section
  (14.74 pb) as expected.
• Setting the decay mode to lepton and neutrino
  mode only gives
• s 330.9 fb
• Filter applied at production level, requiring
• 2 leptons in the event with
• Lepton Pt gt 4.5 GeV
• Lepton h lt 2.7
• nb doesnt appear to have Pt miss gt 50 GeV as
  described in http//jarguin.home.cern.ch/jarguin/d
  c3requests_sm.html (see later)
• Yields a filter efficiency of 66.7 reducing the
  cross section to
• s 221 fb
• Events filtered to leave electron channel
  (ZZ?eenn) only. (17702 events)

Wider problem How do you find cross sections for
csc events?
10
Pythia vs MC_at_NLO

• Comparison on csc Pythia sample and private
  MC_at_NLO sample.
• Plot parameter distributions with the cut values
  of atlfast study by S.Hassani (ATL-COM-PHYS-2002-0
  12)
• Cuts are not applied in the following plots.
• Plots are normalised to respective cross sections
  calculated above.

11
Electron Distribution
Electron Pt gt 20 GeV
Electron h gt 2.5
Pythia red vs MC_at_NLO brown
12
ee Composite Particle

• Pythia seems to give more electrons than MC_at_NLO
  but bear in mind the pythia generator cuts
• 2 leptons in the event with
• Lepton Pt gt 4.5 GeV
• Lepton h lt 2.7

Likely that more Pythia Electrons will
survive the reconstruction.
13
Pt miss distribtuion
nb no generator pt cut on Pythia sample
Pt miss gt 50 GeV
14
Jet Distribution
VETO
VETO

• Atlast applies jet veto, excluding events with
  jets in the region
• Jet Pt gt 30 GeV
• Jet h lt 3

15
Z Pt Distribution
Z Pt gt 150 GeV
16
Other cuts?
cos(q) lt 0.99?

• Cuts not in the atlfast study
• Match missing energy and Z pt
• (MET-Zpt)/Zpt
• Require missing pt and Zt to be back-to-back
• cos(fMET-fz)
• Assumes no initial ZZ Pt.
• Neither of these are actually cut on.

17
After Cuts
18
Atlfast Signal Comparison
(S Hassani, ATL-COM-PHYS-2002-012) For ZZ?llnn
includes muon channel.
19
Pythia vs ttbar

• Comparison on csc Pythia sample and ttbar
  background.
• DQ2 dataset csc11.005200.T1_McAtNlo_Jimmy.recotri
  g.AOD.v11000505
• Cross section for ttbar from MC_at_NLO output is
• s 773.6 pb
• ttbar filter requires at least one leptonic decay
  with Pt gt 1GeV (no all-hadronic decays), with
  efficiency of 52.0
• Cross section for ttbar is therefore
• s 402.3 pb
• Compare to website value (http//jarguin.home.cer
  n.ch/jarguin/dc3requests_sm.html)
• s 461 pb
• Same order of magnitude, but not an exact match.
  Use 461 pb in the following plots.
• Also give a reminder of the atlfast cuts values,
  but again, cuts are not implemented in plots.

20
Electron Z Distributions
Pythia red vs ttbar pink
Electron h gt 2.5
Electron Pt gt 20 GeV
mll 91.2 GeV lt 10 GeV
21
ttbar Distributions
Pythia red vs ttbar pink
Veto events containing jets with Pt gt 30GeV Jet
h lt 3
Z Pt gt 150 GeV
Pt miss gt 50 GeV
22
Atlfast ttbar Comparison
(S Hassani, ATL-COM-PHYS-2002-012)
Low Statistics!
23
Pythia vs Zjets

• Comparison on csc Pythia sample and Zjets
  background.
• DQ2 datasets for all Z decay channels separately
• csc11.005185.PythiaZee_pt100.recon.AOD.v11004103
• csc11.005186.PythiaZmumu_pt100_fixed.recon.AOD.v11
  004205_tid002215
• csc11.005187.PythiaZtautau_pt100.recon.AOD.v110041
  03
• csc11.005183.PythiaZnunuJet.recon.AOD.v11004206
• Filters are applied at generation level
• Z?ee channel
• Electron Pt gt 10 GeV
• Electron h lt 2.7
• Z Pt gt 100 GeV
• Z?mm channel
• Muon Pt gt 5 GeV
• Muon h lt 2.8
• Z Pt gt 100 GeV
• Z?tt channel
• Tau Pt gt 5 GeV
• Tau h lt 2.8
• TauTau ? ll, with ll mass gt 60 GeV

Cross sections from web
s(Z?eejets) 1.2 pb
s(Z?mmjets) 1.2 pb
s(Z?ttjets) 0.12 pb
s(Z?nnjets) 723 pb
24
Electron Z Distributions
Pythia vsZ?eejets, Z?mmjets, Z?ttjets,
Z?nnjets
Electron h gt 2.5
Electron Pt gt 20 GeV
Should there be a peak here?
mll 91.2 GeV lt 10 GeV
25
ZJets Distributions
Pythia vsZ?eejets, Z?mmjets, Z?ttjets,
Z?nnjets
Veto events containing jets with Pt gt 30GeV Jet
h lt 3
Pt miss gt 50 GeV
Z Pt gt 150 GeV
26
Atlfast ZJets Comparison
(S Hassani, ATL-COM-PHYS-2002-012)
27
Atlfast ZJets Comparison (2)
Jet Veto and Z Pt cut reduces any remaining
background from jets
28
Things To Do

• Investigate other sources of background.
• Migrate from MyZeeOnAOD ? EventView Code.
• Start looking at the muon channel.
• Tune cut parameters and values.
• Calculate/understand cross sections properly.
• Advanced analysis techniques(likelihood, ANN,
  boosted decision trees etc)

29
Tauola Bug

• Aside discovered bug in Tauola (when trying to
  generate ZZ?ttmm events)
• Tauola tries to back-navigate from the
  tau?Z?intermediate process to calculate
  polarisation.
• Intermediate process has no pdgID number.
• Tauola doesnt like this very much.
• Tauola gives a cryptic error message and then
  crashes.
• Fixed by adding additional pdgID switch
  statements into Tauola.