A search of ZZ signal in 3 lepton channel

1
A search of ZZ signal in 3 lepton channel

• OUTLINE
• Reminder of the Analysis status in release 13
  (14 TeV)
• Validation in release 14 (10 TeV)
• Conclusion and future plans

• Azeddine Kasmi
• Bob Kehoe
• Southern Methodist University

Thanks to H. Ma, M. Aharrouche
2
Motivations

• 3 leptons channel has higher acceptance than 4
  leptons reconstructed.
• The egamma sliding window algorithm does not
  cover the forward regions
• In this talk, the emphasis is on a cluster based
  approach to get the full acceptance
• However, background is higher

Medium electron N3e 2 N4e
Number of medium electrons/event
3
The fourth lepton
e1
e2
Z
e2
Z
e4

• We tried many algorithms.
• Tau (54) , Egamma (60), EMTopo-Cluster (75)
• Jets C4 (efficiency 93)
• CaloCalTopoCluster_P4 (92).
• Focus on Jet C4 and CaloCalTopoCluster
• Define a variable M_ZBest.
• check the two combinations M13 and M23.
  The closest of them to the M_ZPDG is the
  M_ZBest
• The Z second is from the reconstructed lepton
  and the jet candidate or the caloTopocluster
  candidate
• Use a 2D cut on the two Z peaks

e3
b
b
Z
e1
ZZ Zbb Zb
arbitrary units
ZZ Zbb Zb
4
Particle Identification
Jet candidate with a truth match in ZZ

• Jet C4 Algorithm
• The EM fraction The fraction of energy left by
  the jet candidate in the EM calorimeter
• CaloCalTopoCluster_P4
• We built a likelihood method using four moments
  of the Topoclusters
• Longitudinal moment, Lateral, Isolation, and
  Energy fraction of the most energetic cell
• Pdf's for signal are from Z?ee sample
• Pdf's for background are from ttbar
• Likelihood in different eta regions showed an
  efficiency form 85 to 95 and a fake rate from
  20 to 7
• Focus is only on a cluster without regard to a
  track

No match In ZZ
Zbb
S B
5
Final Selection cuts while using the jet
algorithm and the calo Topocluster

• Pt of the 3 leptons PT3e gt 10 GeV and PT2m1e gt
  10 GeV
• Medium electrons, STACO for muons
• Impact parameter significance
• do/so lt 5.5 for electrons
• do/so lt 5 for electrons
• do/so lt 3 for muons
• Isolation etcone20/et lt 0.14
• Missing ET ETmiss lt 25 GeV ( to deal with WZ)
• SV2 lt 0 and 3DSV1comb (w_cmb gt -2) (w_cmb lt
  3.5)) .
• EMF_B gt 0.8 and EMF_EC gt 0.85
• Take the first leading jet candidate
• The likelihood cut on TopoCluster candidate gt 0.5
• MZ_Best between 75 GeV to 100 GeV
• MZ_second between 85 GeV to 110 GeV
• MZ_Best between 80 GeV to 100 GeV
• MZ_second between 80 GeV to 100 GeV

6
Results (1fb-1) for the Jet Algorithm and
TopoCluster
Signal 5 events Background 1 event
In the case of 4l channel in release
12 Signal13 events Background0.2 event
Signal 5 events Background 1 event
7
LHC will run _at_ 10 TeV

• Data Sample (10k events) reconstructed in release
  14.4.0
• mc08.105931.McAtNlo0331_JIMMY_ZZ_llll.simul.HITS.e
  367_s462
• The cross section is less by 40 in 10 TeV than
  it is at 14 TeV 40.61 fb versus 67 fb
• The PT spectrum of the unfound electron seems to
  be lower in 10 TeV sample
• In the following slides
• Will validate the 10 TeV sample of ZZ?4l
• Compare with Aharrouches forward electron

10 TeV 14 TeV
PT spectra of the unfound electron at CM 10 TeV
and 14 TeV
8
Performance of the CaloCalTopocluster and the C4
jet algorithm

• Jet C4 algorithm the truth unfound Z electron
  and the matched jet candidate have a similar PT
  spectra
• The matching is also good in the
  CaloCalTopoCluster.
• The low PT is most likely due to the splitting
  in the crack regions

Jet C4
CaloCalTopoCluster PT gt 7 GeV
9
Forward Electron and CaloCalTopoclusters in h gt
2.5
The matching is quite well except that the
current implementation of the Forward electrons
is done with a PT cut of 15 GeV
ETA
Fwrd electron
MZBest
CaloCalCluster
PT cut of 15 GeV Mohamed has changed to 5 GeV
MZsecond
10
Summary table of performances for ZZ Sample
11
Conclusions and future plans

• The ZZ?3lX Analysis contribution to the 4l
  channel is about 13 in term of the significance
• The same techniques developed for the 3lX
  searches in samples of 14 TeV can be used for 10
  TeV samples relaying on cluster based approaches
• For the first 200 pb-1 assuming the same
  selection efficiencies, 0.55 signal event and
  0.1 background event
• Run on the 10 TeV background samples or at least
  rescale the 14 TeV background sample to have a
  better estimate for S/B
• Need to Optimize the cuts