1. Monica D’Onofrio University of Liverpool
Boson+ jet production as a background for searches in ATLAS
Monica D’Onofrio
University of Liverpool
Standard LHC 2011 Workshop
Durham, April 13th 2011

2. Monica D'Onofrio, SM@LHC 2011
Introduction
Associated production of W/Z boson and jets extensively studied:
 As a direct measurement
test of perturbative QCD at high Q2
Sensitive to underlying events and hadronization modeling
NLO prediction available
1 mb
1 nb
1 pb
As a background for rare SM processes and for most of the searches for physics beyond SM. Among others:
Searches for supersymmetry
Searches for SM and MSSM Higgs boson(s)
13/4/2011
Monica D'Onofrio, 2011

3. Monica D'Onofrio, SM@LHC 2011
Outline
The ATLAS detector and data-taking in 2010
SM measurements in a glance
W/Z+jets as backgrounds focusing on specific searches for new physics:
Searches for SUSY in final state events with large ETMiss, high pT jets with and without leptons
Data-driven or partially data-driven techniques
Searches for SM Higgs boson: HWW* as example
A few words on top as a background for new physics:
Searches for SUSY in final state events with large ETMiss and b-jets
Summary and conclusions
13/4/2011
Monica D'Onofrio, 2011

4. Monica D'Onofrio, SM@LHC 2011
The ATLAS detector
Spectrometer coverage up to |h|<2.7
Trigger and measurement for m with momentum resolution < 10% up to Em ~ 1 TeV
Coverage up to |h|=4.9
EM calorimeter, e/g trigger, ID, measurement
Resolution: s/E ~ 10%/√E
HAD calorimeter (jets, MET)
Tiles(central), Cu/W-Lar (fwd)
E-resolution: s/E ~ 50%/√E
Fwd cal: s/E ~ 90%/√E
Length: ~45 m
Radius: ~12 m
Weight: ~7000 tons
3-level trigger
rate to tape ~200 Hz
Inner Detector (|h|<2.5)
Tracks and Vertex reconstructions
s/pT ~ 3.8 x 10-4 pT (GeV)
2 T magnetic field
13/4/2011
Monica D'Onofrio, 2011

5. ~ 35 pb-1 of data used in the analyses presented here
The 2010 ATLAS pp data
~ 35 pb-1 of data used in the analyses presented here
Profiting at best from the excellent LHC performance:
Maximum values of 6 pb-1 luminosity per day
Instantaneous luminosity values up to 2 × 1032 cm-2 s-1
Detector efficiency above 90%
13/4/2011
Monica D'Onofrio, 2011

6. W/Z+jets cross section
See talk from Ellie Dobson
W(en,mn)+jets cross section
Z(ee,mm)+jets cross section
 arXiv:
(Accepted by Phys Lett. B )
Comparison with NLO predictions (MCFM)
and several Monte Carlo simulation tools
ME(LO)+Parton Shower
now updated with
full 2010 data-set
13/4/2011
Monica D'Onofrio, 2011

7. W/Z+jets as background
Most of the final states investigated in searches for new physics have:
High pT jets (light or heavy flavor jets)
Missing transverse energy (MET)
Possibly high pT leptons
Same signature as W/Z+jets production, in particular:
Z  nn + jets (MET+jets, irreducible background)
W  ln + jets (MET+jets(+leptons) where l = e,m,t)
Different approaches have been followed, depending on the analysis, with some common features:
Whenever possible, use of data-driven techniques
to estimate absolute normalization and/or shapes
defining ‘control’ samples and making closure tests
Note: with 35 pb-1 in several cases use Monte Carlo tools to model observable shapes
Data-driven methods sometimes affected by large statistical uncertainties g q c0 ~
Jets
Missing Transverse Energy
f.i.: SUSY q g
Z/g* n
13/4/2011
Monica D'Onofrio, 2011

8. Monte Carlo samples used
Analyses generally employ MC samples generated with:
Alpgen associated with HERWIG (not ++) and JIMMY for W+jets and Z+jets (including Wbb, Zbb)
MLM matching scheme to combine samples with different final state parton multiplicities (up to 5 for inclusive, up to 3 for Wbb/Zbb)
PYTHIA used for low DY region and Ztt (t-decays with TAUOLA)
SHERPA samples used for cross check in some cases
Large “production” on-going for 2011 analyses
V+jets predictions normalized to NNLO cross sections (FEWZ)
CTEQ6L1 for ALPGEN and SHERPA samples
MRST2007lomod (LO modified) for PYTHIA samples (for low mass DY)
Example from HWW* analysis
13/4/2011
Monica D'Onofrio, 2011

9. Monica D'Onofrio, SM@LHC 2011
Searches for SUSY
Monica D'Onofrio, 2011
13/4/2011

10. Object identifications
Common tools and requirements for ‘good events’ are used
Primary vertex
At least 1 good vertex with Ntracks>4
Electrons
pT > 20 GeV, |h|<2.47
reject events if electron candidates are in transition region (1.37<|h|<1.52)
Jets
anti-kT, R=0.4
pT > 20 or 30 GeV, |h| up to 2.8
Reject events compatible with noise or cosmics
Muons
pT > 20 GeV, |h|<2.4
combined/extrapolated info from ID and Muon spectrometer
Sum pT of tracks <1.8 GeV in DR<0.2
B-Jets
Exploit Secondary vertex reconstruction algorithm
Missing ET
Calculated from objects and clusters
Remove overlapping objects
If DR(jet,e)<0.2, remove jet
If 0.2<DR(jet,e)<0.4, veto electron, if DR(jet,m)<0.4, veto muon
13/4/2011
Monica D'Onofrio, 2011

11. Search in no-lepton final states
arXiv: (Sub. PLB)
Search in no-lepton final states
Select events with jets, missing ET and no lepton (e/m veto)
Signal regions definition on the basis of jet multiplicity
(n ≥2 jets or n≥3 jets), jet pT and ETMiss thresholds and:
Scalar sum of objects pT
Effective mass (meff)
Stransverse mass (mT2)
 Phys.Lett.B463:99-103,1999
 J.Phys.G29: ,2003
4 signal regions
Due to trigger requirements
QCD-multijet rejection
Enhance sensitivity to SUSY
13/4/2011
Monica D'Onofrio, 2011

12. Results for SUSY in jets+MET
[u]=uncorrelated uncertainties (MC statistics, acceptance, jet energy resolutions..) [j]=Jet Energy Scale (6%-10% as function of jet pT), [L]=luminosity (11%)
W/Z+jets bkg dominates
Good agreement between data and SM predictions
13/4/2011
Monica D'Onofrio, 2011

13. W+jets (and top background)
Non-QCD bkg dominated by
Wtn, W(missed)e/m
Top pair production (t  t+jets)
Central value derived from MC:
W+jets: ALPGEN normalized to NNLO
Top: (+HERWIG and JIMMY),
CTEQ6.6 PDF
Cross checks on data:
Control regions with leptons removed from W data
In-situ checks for t background, derived from Wmn events
2 ‘replacement’ methods:
smeared resolution function: hadronic t decay products considered as a single additional t-jet  t-jet smearing function calculated from W → τν MC
full simulation (embedding)
Finally use MC  theoretical/modeling uncertainties comparable to statistical uncertainty from data-driven estimates
derived from selected W --> mu nu events. The muon is removed from the event and replaced by a Monte Carlo tau lepton decay, which is either smeared using resolution functions to emulate the detector response (smeared tau) or processed using the full detector simulation (simulated tau). The yellow band shows the combined uncertainty from the jet energy scale, background subtraction and acceptance corrections. 
13/4/2011
Monica D'Onofrio, 2011

14. Embedding method for tau: an example
Method used for data-driven estimates in searches
for charged Higgs in t-hadronic final states:
t+jets and t+leptons
SM background estimated with data-driven techniques:
Fake t: e,m or jets misidentified as t jets  rate from data
e,m: matrix method; Jets: in g+jet samples
QCD-multijets: in control samples with loose-no-tight t candidates
Real t (relevant for t+jets): from top and W+jets with embedding method
ATLAS-CONF
13/4/2011
Monica D'Onofrio, 2011

15. Estimate of true-t background
Control samples with single and top pair production and W+jets events with muons:
Replace muon with simulated t lepton
Re-reconstruct new hybrid events
Use these events instead of simulation:
 Advantage: whole event is taken from data including pile-up, HF jets etc.
m+jets sample:
one isolated m, pT>20 GeV
at least 3 jets, pT>20 GeV
at least 1 b-tagged jet
m(jj) in [MW± 20 GeV]
MET>30 GeV, S ET > 200 GeV
Method is statistically limited at the moment:
 Use loose selection with respect to baseline
13/4/2011
Monica D'Onofrio, 2011

16. Z + jets in SUSY jets+MET
Z+jets background is dominated by the irreducible Znn + jets
Central value derived from MC:
Z+jets: ALPGEN normalized to NNLO
Control regions with leptons removed from Z data
Z+jets control sample
MET recalculated for each event artificially removing the leptons from Z-decay. Corrections for m vs n coverage done with MC
derived from selected W --> mu nu events. The muon is removed from the event and replaced by a Monte Carlo tau lepton decay, which is either smeared using resolution functions to emulate the detector response (smeared tau) or processed using the full detector simulation (simulated tau). The yellow band shows the combined uncertainty from the jet energy scale, background subtraction and acceptance corrections. 
Additional cross checks also performed in g+jets samples:
 generally poor statistics at medium/high MET, to be further investigated
13/4/2011
Monica D'Onofrio, 2011

17. Search in 1-lepton final states
Phys. Rev. Lett. 106, (2011)
Search in 1-lepton final states
Require exactly 1 lepton (e or m, pT>20 GeV) + ≥3 jets [pT> 60,30,30 GeV]
Privilege signatures from gluino/squark cascade decays with intermediate steps
Isolated lepton suppresses QCD multijet background and facilitates triggering
Use mT as additional discriminating variable, Missing ET and jets and leptons pT
13/4/2011
Monica D'Onofrio, 2011

18. Signal region and results
mT>100 GeV  to suppress W+jets and top pair production
MET/meff > 0.25  to suppress QCD background
meff>500 GeV  to enhance sensitivity to SUSY particles
+ Df(jet, ETMiss) > 0.2
for QCD rejection
After all cuts:
One event observed in each channel
Main background: top ~ 70%, rest = W+jets
Estimated with partially data-driven methods
(similar for the electron channel)
13/4/2011
Monica D'Onofrio, 2011

19. SM background estimation (I)
Exploit use of control regions:
Based on ETMiss VS MT
Define samples enriched in a given process
Constrain MC predictions to data in that region (rely on MC shapes)
Extrapolate to other regions (with MC). Ex.:
Ntt SR (pred) =(NdataCR –NBkgMCCR)× (Ntt SR,MC /Ntt CR,MC )
Systematic uncertainties on extrapolation factors
W-enriched sample(require < 1 b-tagged jet)
top-enriched sample
(require ≥ 1 b-tagged jet)
13/4/2011
Monica D'Onofrio, 2011

20. SM background estimation (II)
Exploit use of control regions:
Based on ETMiss VS MT
Define samples enriched in a given process
Constrain MC predictions to data in that region (rely on MC shapes)
Extrapolate to other regions (with MC). Ex.:
Ntt SR (pred) =(NdataCR –NBkgMCCR)× (Ntt SR,MC /Ntt CR,MC )
Systematic uncertainties on extrapolation factors
Additional control regions at low MT or low Missing ET used to validate the assumption on MC shape.
Pull:
Used to estimate QCD in other CR
Main uncertainties:
Theory/modeling: 50% W+jets (uncertainty on meff NLO shape), 25% top (comparison between generators)
B-tagging: ~ [10-25]%
13/4/2011
Monica D'Onofrio, 2011

21. Likelihood method (1-lepton)
Fill all useful information into a likelihood => minimize to estimate bkgs
One poisson for signal region and for each control region
 simultaneous fit of all regions (signal and control)
Systematic uncertainties treated as nuisance parameters
Fitted predictions in agreement with observation
13/4/2011
Monica D'Onofrio, 2011

22. Monica D'Onofrio, SM@LHC 2011
0 and 1-lepton results
Limit in mSUGRA
0-lepton
If msquark=mgluino
exclude < 775 GeV
1-lepton
If msquark=mgluino
exclude < 700 GeV
13/4/2011
Monica D'Onofrio, 2011

23. Summary of V+jets estimate methods
Baseline for 0-lepton
MC based approach: MC based estimate where both the shape and the rate in the signal region (SR) are estimated from MC
Mixed data and MC based via overall corrections : estimate where the MC rate is constrained by data in a control sample (CS), but the MC is used to extrapolate from the control sample to the signal region.
Pros: remove uncertainties on Lumi and total s, factorize part of detector and theoretical uncertainties (if Control Sample CS ~ similar topology)
Cons: central value possibly affected by large statistical fluctuation in CS. Theoretical uncertainties might be quite large.
Event based correction on data: A quasi data-driven approach, where both the number of events and the shape are taken from a data CS. In case correction factors must be applied to account for the acceptance and ID efficiency of the events in the CS  taken from data when possible, otherwise from Monte Carlo.
Baseline for 1-lepton (simultaneous fits in several CS)
Additional cross checks
13/4/2011
Monica D'Onofrio, 2011

24. Searches for SM Higgs in WW*
Monica D'Onofrio, 2011
13/4/2011

25. SM Higgs →W W*→lν lν (l = e, μ)
ATLAS-CONF
SM Higgs →W W*→lν lν (l = e, μ)
Strong sensitivity in 120 < m(HSM) < 200 GeV
Cut-based analysis
combine H + 0 jet, H + 1 jet and H + 2 jet
Dominant backgrounds: DiBoson, tt,V+jets
(H+0/1j)
(H+2j)
0-jet final states
1-jet final states
2-jet final states
2 leptons, m(ll)<50(65) GeV
pT(ll) > 30 GeV
0.75×mH < mT < mH
2 leptons and exactly 1 jet
same as H+0j selection
pT(jet)>25 GeV, |h|<2.5
B-jet veto (anti-tag)
|PTTOT | < 30 GeV (l1,l2,j,MET)
Ztt suppression
2 leptons and 2 jets
hj1* hj2 < 0, Dhjj>3.8, mjj>500 GeV
m(ll)<80 GeV, pT(ll) > 30 GeV
0.75×mH < mT < mH
B-jet veto and Ztt suppression
|PTTOT | < 30 GeV (l1,l2,j1,j2MET)
Exclusion limits as function of MH has been presented in a mass range of 120 – 200 GeV. At 160 GeV a SM-like Higgs boson with a production rate of 1.2 times the SM value is excluded at 95% Confidence level using PCL.
Signal region
13/4/2011
Monica D'Onofrio, 2011

26. Good agreement data/MC in shape for kinematic distributions
W+jets background
Define control sample enriched in W+jets:
One lepton must satisfy ID and isolation
cuts of the analysis
Require second lepton to satisfy loose
set of cuts (fakeable)
 W+jets expectations in signal region (SR):
ds/dX (SR) = ds/dX (CR) × fl
fl = fake factor = Prob(loose  ID )
Good agreement data/MC in shape for kinematic distributions
13/4/2011
Monica D'Onofrio, 2011

27. Fake factor
Control sample defined in multi-jets events with at least a fakeable lepton.
Real lepton contamination (from W,Z) removed
~ 50% uncertainties, mostly dominated by sample and trigger selection dependence
Luminosity after prescale
13/4/2011
Monica D'Onofrio, 2011

28. W+jets estimates in HWW*
H+0j
H+1j
H+2j: negligible
13/4/2011
Monica D'Onofrio, 2011

29. Z+jets (and low DY) background
Scaling the yield in MC by a ETMiss mis-modeling factor using control regions
Same method used independently on jet multiplicity
In H+2j selection, Z-->ll+jets expected to be dominant before m(ll) selection  checks on pre-selection events for several kinematic distributions:
ET and h jets (1,2,3)
m(jj), Dh(jj), Df(jj)
Good agreement in shape and in absolute normalization (within 10%)
Exclusion limits as function of MH has been presented in a mass range of 120 – 200 GeV. At 160 GeV a SM-like Higgs boson with a production rate of 1.2 times the SM value is excluded at 95% Confidence level using PCL.
13/4/2011
Monica D'Onofrio, 2011

30. Results for SM Higgs →W W*
Upper limit on σxBR(H→WW*)
mH=120 GeV : 54 pb
mH=160 GeV : 11 pb
mH=200 GeV : 71 pb
Exclusion limits as function of MH has been presented in a mass range of 120 – 200 GeV. At 160 GeV a SM-like Higgs boson with a production rate of 1.2 times the SM value is excluded at 95% Confidence level using PCL.
13/4/2011
Monica D'Onofrio, 2011

31. A few words on top as background
Monica D'Onofrio, 2011
13/4/2011

32. Searches in ETMiss+b-jets
arXiv:
Submitted to PLB
Third generation squarks might be lighter than 1st, 2nd generation  possibly high cross sections:
direct pair or gluino-mediated production
Final state enriched in b-jets  search in events with jets (≥1 b-jet) +ETMiss ( + 0/ ≥ 1) leptons b/ ~ /t ~
Event selection
0-lepton
1-lepton
Signal regions:
0-lepton: Meff > 600 GeV 1-lepton: Meff > 500 GeV
13/4/2011
Monica D'Onofrio, 2011

33. SM Background Top pair production dominant for both selections
0-lepton:
QCD: Partially data-driven
Top/Boson+jets: MC estimate
1-lepton:
Data-driven techniques for all bkg
QCD: “matrix-method” as in previous analyses
Top/Boson+jets: exploit low correlation between MT and Meff
Revert Dfmin<0.4
Take Meff shape from Dfmin <0.4
Meff (GeV)
MT (GeV) A B C
D (Signal Region)
100 40
500
Dfmin <0.4
From MC: take fraction of QCD events passing
Dfmin >0.4
40 <mT<100 Gev
“Closure-test” with MC  good agreement with data
13/4/2011
Monica D'Onofrio, 2011

34. Systematic uncertainties
0-lepton analysis: theoretical uncertainties larger than JES at high meff values
Tails of relevant distributions
for SUSY event selection affected by
large theoretical uncertainties.
Meff: 600 GeV  1 TeV
Meff: 600 GeV  1 TeV
March 8, 2011
Monica D'Onofrio, CERN Seminar

35. Summary and conclusion
W/Z+jets background is one of the major background for searches for new physics
Several approaches followed, depending on the analysis variables (jet multiplicity, explicit lepton requirement, with/without b-tagging)
Common features:
define control regions orthogonal to signal samples
use MC tools to estimate shapes, data-driven techniques for normalization
Only a few examples shown here
Larger use of data-driven techniques with more data!
in most cases, use of MC samples unavoidable (acceptance corrections, control sample-to-signal region corrections)  Reducing theoretical uncertainties might be the key-issue for kinematic regimes interesting for searches
13/4/2011
Monica D'Onofrio, 2011

36. Monica D'Onofrio, SM@LHC 2011
Back-up slides
Monica D'Onofrio, 2011
13/4/2011

37. Top mass and pair cross section
Measured in e/m+jets (with b-tag) and dilepton channels (combined)
Measured in e/m+jets channel
Dominant uncertainty due to JES
Employ the ratio of reconstructed top to W mass (R32)
m(t) = ± 4.0 ± 4.9 GeV
μ+jets channel
35 pb-1
ATLAS-CONF
ATLAS-CONF
ATLAS-CONF
ATLAS-CONF
ATLAS-CONF
ATLAS-CONF
σ(tt) = 180 ± 9± 15 ± 6 pb [ 10% total uncertainty ]
13/4/2011
Monica D'Onofrio, 2011

38. Electron Performance Results
Figure right bottom: Control distributions for the forward Z to electrons analysis, Z rapidity distribution. The background is determined from a fit to the invariant mass distribution. The simulation is normalised to the data.
forward-central Zs
electrons above the tracker acceptance
ATLAS-CONF
13/4/2011
Monica D'Onofrio, 2011

39. ID and Muon Combined Performance
Low pT efficiency from J/ψ→μμ decays
High pT efficiency from Z→μμ decays
Very good description of alignment in MC
ATLAS-CONF
ATLAS-CONF
Efficiency understood down to very low pT
Present understanding of ID alignment
Figure bottom left:
Di-muon invariant mass distribution for oppositely charged muon pairs with transverse momentum above 20 GeV and the above track isolation.
The momentum vector is jointly reconstructed in the muon spectrometer and the inner tracking detector.
The measured di-muon mass is shown for the full pp data of 2010 in first-pass reconstruction using preliminary calibration and alignment constants and after the reprocessing of the data with improved calibration and alignment constants. The distribution is compared to MC prediction using Pythia-generated Z->µµ events and reconstructed with the same software version as the original data processing.
pull functions for TRT are smaller than 1, so no additional smearing needs to be added
Smear MC hit uncertainties
ATLAS-COM-CONF
Improved momentum scale and resolution
muon scale uncertainty is < 1%
dimuon mass resolution 1.8% barrel and 3% end-cap
13/4/2011
Monica D'Onofrio, 2011

40. Monica D'Onofrio, SM@LHC 2011
Jet Energy Scale
ATLAS-CONF
In-situ calibrations
barrel 3%
Improved by factor of 2 with respect to previous version
Evaluated up to 3.5 TeV in energy and |η|<4.5
MPF: Missing Et Projection Fraction
Figure 1:
Fractional jet energy scale systematic uncertainty as a function of pjetT for jets in the pseudorapidity region 0.3 ≤ |h| < 0.8 in the calorimeter barrel. The total uncertainty is shown as the solid light blue area. The individual sources are also shown, with uncertainties from the fitting procedure if applicable.
Figure 2:
Figure 12: Jet energy scale uncertainty as a function of pjetT in the central barrel. This plot shows the data to Monte Carlo simulation ratios for several in-situ techniques that test the jet energy scale exploiting photon jet balance (direct balance or using the missing transverse momentum projection technique), the balance of a leading jet with a recoil system of two or more jets at lower transverse momentum (multijets) or using the momentum measurement of tracks in jets.
13/4/2011
Monica D'Onofrio, 2011

41. Jet Energy and Etmiss Resolutions
PbPb data only sample reaching this high in ∑ET
14.5 TeV
Exmiss and Eymiss resolution as a function of the total transverse energy (ΣET) for Pb+Pb and pp collision events. For Pb+Pb, due to low statistics, RMS values are plotted instead of sigma from the fit. The line represents independent fits to the resolution obtained in Pb+Pb data and in pp data at 7 TeV ( ATLAS-CONF ). Exmiss , Eymiss and ΣET are computed from cells with energy > 2σ(noise) calibrated with Global Cell Weighting (GCW). A logarithmic scale is used on the X axis
Advanced calibrations → improve resolution by 10-30%
Monte Carlo agrees with data within 10%
13/4/2011
Monica D'Onofrio, 2011

42. Monica D'Onofrio, SM@LHC 2011
QCD for 0-lepton
QCD-multijet background due to mis-reconstructed jets and neutrinos from HF leptonic decays
ETMiss expected to be aligned to one of the jets
Use partially data-driven estimate:
Rescale MC samples (PYTHIA and ALPGEN)
in control region  Df(jet, ETMiss) < 0.4
Cross-checked with
fully data-driven
techniques (Jet smearing)
 Use control region
based on reversed
ETMiss/meff for rescaling
After rejection:
QCD ~5% of TOT Bkg
13/4/2011
Monica D'Onofrio, 2011

43. Interpretation of the results
Use profile likelihood ratio:
include correlations of uncertainties where appropriate
 Estimate upper limits at 95% C.L. on N signal events and effective cross sections independently of new physics models (background-only hypothesis)
Exclude non-SM: N events 43.9(A), 11.9(B), 37.6(C), 3.5(D)
s of 1.3(A), 0.35(B), 1.1(C), (D) pb
Best sensitivity
Region D
(3j, meff > 1 TeV)
Translate results in limits on MSUGRA/CMSSM
(m0,m1/2)-plane
parameters at GUT scale
1. Unified gaugino(scalar) mass m1/2(m0)
3. Ratio of H1, H2 vevs tanβ
4. Trilinear coupling A0
5. Higgs mass term sgn()
Best sensitivity
Region C
(3j, meff > 500 GeV)
If msquark=mgluino
exclude < 775 GeV
Theoretical uncertainties on SUSY NLO cross sections included in limit calculation
13/4/2011
Monica D'Onofrio, 2011

44. Monica D'Onofrio, SM@LHC 2011
2-leptons analysis
Search for dilepton (e,m) pairs from neutralino/chargino decays
Two search strategies, requiring opposite-sign (OS) and same-sign (SS) dileptons events
Opposite-Sign
Same-Sign
Event selection
exactly two leptons
M(ll) > 5 GeV
Signal regions
OS: ETMiss > 150 GeV
SS: ETMiss > 100 GeV
Main SM Background
OS: top pair (estimate in CR)
SS: misidentified leptons (fakes)  data-driven as in 1-lepton analysis
Loose optimization on the sensitivity based on the study of MSSM models with squark and gluino mass at ~ 400 GeV
13/4/2011
Monica D'Onofrio, 2011

45. Monica D'Onofrio, SM@LHC 2011
Tovey, JHEP 0804 (2008) 034
Polesello, Tovey, JHEP 1003 (2010) 030
Top-tagger: mCT
In the decay of a two pair-produced heavy states which decay via daci
mCT distributions have endpoints defined by m(d), m(a) and the vector sum of transverse momenta of the visible particles upstream of the system for which the contransverse mass is calculated (pb)
For the top-pair system mCT (ll), mCT (jj ), mCT (j l, jl) can be constructed
Efficiency mCT tagger = 85%
control region for ttbar:
mCT-tagged events
60<ETMiss<80 GeV
13/4/2011
Monica D'Onofrio, 2011

46. Monica D'Onofrio, SM@LHC 2011
Top background for OS
Dileptonic top decays tt  l+nb l-nb
“Top tagging” algorithm based on contransverse mass (mCT)
Control Sample
ETMiss [60,80] GeV, ≥2 jets with pT>20 GeV
Calculate mCT from 4-vectors of leptons and jets  must be consistent with tt bounds
m(jet,l1) and m(jet,l2) consistent with top decays
Data CR: 15 top-tagged events
MC CR: 21.3±3.8 (18.8 from ttbar)
Estimation in Signal Region 
13/4/2011
Monica D'Onofrio, 2011

47. Other backgrounds for 2-lepton
Electron charge-flip:
Relevant for Same-Sign dilepton final states
Background from dilepton top events:
Hard bremsstrahlumng process
Charge mis-identified rate taken from Zee MC samples as a function of |h|
Validated in Zee data
Z+jets:
Zem from MC (low statistics in data)
Semi-data driven estimation for Zee,mm
Control region:
81<m(ll)<101 GeV
ETMiss < 20 GeV
Corrected for predicted number of W and top in control region
Cosmics:
2 methods considered 
matrix method based on impact parameter
Trigger Lifetime
Both consistent to zero
Define an upper bound: Ncos < 1.32 at 68% CL, Ncos < 3.45 at 95% CL
13/4/2011
Monica D'Onofrio, 2011

48. Monica D'Onofrio, SM@LHC 2011
Results (I)
Agreement between data and SM expectations within uncertainties:
Use sum of ee,mm,em channel for SS, combination of the three channels for OS
95% C.L. upper limits on effective cross section s ∙ A∙BR from new physics:
SS: s<0.07 pb
ee: 0.09 pb, mm: 0.21 pb, em: 0.22 pb
Interpretation in mSUGRA
13/4/2011
Monica D'Onofrio, 2011

49. Results (b-jets)
Good agreement between data and SM predictions within systematic uncertainties in both channels
0-lepton analysis
Interpret the results as 95%C.L. upper limits on N signal events independently of new physics models:
N(0-lepton) > 10.5
N(1-lepton)>4.7
Effective cross sections:
s (0-lepton) > 0.32 pb
s (1-lepton) > 0.13 pb
1-lepton analysis
13/4/2011
Monica D'Onofrio, 2011

50. 0-lepton bkg details (b-jets)
Breakdown of non-QCD SM-background contributions for 0-lepton analysis at each stage of the selection (per pb-1)
b-tagging
13/4/2011
Monica D'Onofrio, 2011

51. 1 lepton bkg details (b-jets)
QCD estimation from a Matrix method relying on 2 data sets differing only in the lepton ID criteria: tight (standard) and loose (relaxed):
Breakdown of non-QCD SM-background contributions for 1-lepton analysis at each stage of the selection
ereal=measured in Z-samples
efake=measured QCD-enriched sample
13/4/2011
Monica D'Onofrio, 2011