Frontiers in Contemporary Physics III

1
Frontiers in Contemporary Physics III May 23-28
Vanderbilt University, Tennessee
Julien Donini University of Padova and INFN On
behalf of the CDF collaboration
2
Search for Single Top
Top has been discovered at Tevatron in tt
pairs Single top production predicted by theory
Wtb coupling

• Wt-associated
• Tevlt 0.1 pb
• negligible at Tevatron

s-channel ?Tev 0.88 0.11 pb
t-channel ?Tev 1.98 0.25 pb
(for Mtop 175 GeV)
Z. Sullivan hep-ph/0408049, T. Tait hep-ph/9909352

• Observation of single top
• cross section is about 40 of ttbar
• more difficult to observe due to large
  background
• - important Wjets background

3
Single Top Physics
Why look for single top ?

• Observation of single top allows direct access to
  Vtb CKM matrix element
• cross section Vtb2
• study top-polarization and EWK top interaction
• Probe b-quark PDF (t-channel)
• Test non-SM phenomena
• heavy W boson
• anomalous Wtb couplings
• FCNC couplings like t ? Z/? c
• 4th generation
• Potentially useful for Higgs searches
• single top has same final state as HiggsW
  (associated) production

4
Run I Results

• Tevatron Run I results at ?s 1.8 TeV
• Theoretical cross-sections were 30 smaller
• ?t 1.40 pb
• ?s 0.76 pb
• Single top has not been observed both CDF and DØ
  set 95 limits
• t-channel 22 pb (DØ) 13 pb (CDF)
• s-channel 17 pb (DØ) 18 pb (CDF)
• combined 14 pb (CDF)
• Run II
• higher beam energy (1.96 TeV)
• higher rate
• more luminosity CDF/DØ have more than 800 pb-1
  on tape
• better detector acceptance

5
Signal Topology

• General strategy
• Separated search of signal of s and t-channels
  single top production in 1-tag and 2-tag samples
• Combined search of st signal
• Event selection we look for W2 jets channel
• one high pT isolated lepton (from W) Et gt 20
  GeV, ? lt 1
• veto Z, dilepton, convertion events
• missing Et (? from W) MET gt 20 GeV
• exactly two jets w/ Et gt 15 GeV, ? lt 2.8
• ask for at least 1 b-tag
• Note s and t-channel both produce 2 b-jets but
  bbar from t-channel usually lost into beam pipe
• Additional cuts
• cut on mt mass window 140 lt Ml?b lt 210 GeV
• for 1-tag sample, leading jet Et gt 30 GeV

6
Invariant Mass Ml?b Calculation

• b choosing
• in 1-tag events, choose b-tag jet as b from top
• in 2-tags evt, choose tight jet with larger Q.?
  as the b from top
• Recipe is 97 successful for t-channel
• 53 (75) successful for
  s-channel 1-tag sample (2-tags)
• Choosing correct neutrino solution
• Et(?) MET
• pz (?) is obtained from the constraint Ml? MW
  
• Two solutions pick the one with lower pz (?)
• i.e most central solution.
• Successful matching rate 74 t-channel, 71
  s-channel

7
b-tagging at CDF

• Three methods are currently used and well tested
  in high pt physics
• Soft lepton tagging
• ? Use semi-leptonic decay of b-quarks
• b ? l?c (BR 20), b ? c ? l?s (BR 20).
• ? Leptons with softer pt spectrum than W/Z, less
  isolated.
• Secondary vertex identification
• ? Iterative fit on tracks with significant
  impact parameter
• ? Efficiency is 45-50 for central top b-jets
• ? Mistag rates are kept typically at 4-5
• Jet-probability
• ? Probability P that tracks associated with a
  jet come from primary vertex
• ? Heavy flavors P tends to 0.

8
Signal and Background Modeling

• Signal modeling
• Pythia (or Herwig) do not reproduce correct NLO
  pt distribution for the b jets in the t-channel
• Use MadEvent generator
• generate bq ? tq and gq ? tbq separately
  and merge them to reproduce the b pt spectrum
  from NLO calculations (ZTOP)

• Background modeling
• ttbar and non-top background
• Top pairs Pythia, ?(tt) 6.7 pb.
• Non-top
• W heavy flavor (62) Alpgen
• mistags (25) and non-W (10) estimated from
  data
• diboson (WW, WZ, ZZ) production (3) Pythia

9
Signal and Background Modeling
Expected number of signal and background events
compared with observation for L 162 pb-1
Published CDF results Phys. Rev. D 71, 012005
(2005)

• Combined and 1-tag signal mostly dominated by
  t-channel (65-70)
• 2-tags signal only s-channel
• Total background is dominated by non-top events
  (89)
• Observations are in good agreement with
  predictions

10
Separate Search for Single Top
Maximum likelihood technique to extract the
signal content from the data Perform search to
separate s-channel and t-channel events use
variables that help discriminate the two channels.
d
z direction

• For t-channel use kinematical boost
• In proton-antiproton collisions
• top production light quark jet goes in p
  direction
• anti-top light quark jet goes in p-bar
  direction
• Correlation between
• pseudorapidity of untagged jet ?
• lepton charge Q
• Q. ? distribution asymmetric for t-channel
• s-channel
• Count double b-tags

u
t
W
g
b
11
Q.? Distributions for Separate Search
Data and stacked MC templates weighted by the
expected number of events in the 1-tag sample.
12
Likelihood Function
Joint likelihood function for Q.? distribution in
the 1-tag sample and number of events in 2-tag
samples
k bin index j process index
Poisson
t or s-channel
bg gauss. constraints
Syst. Uncert.

• 4 processes t-channel (j1), s-channel (j2),
  ttpair and non top (j3, 4)
• ?k mean number of events in bin k of Q.?
  distribution
• ?d mean number of events in the 2-tags sample
• nk, nd observed number in data
• the background is allowed to float but is
  constrained to SM predictions
• 7 sources of systematic uncertainties

13
Systematic Uncertainties
Systematic acceptance uncertainties for
t-channel, s-channel and combined signal

• b-tagging (7)
• luminosity (6)
• top mass (4)
• JES (4)

Main systematics
14
Expected Sensitivity for Separate Search

• Perform N MC experiments to estimate expected
  sensitivity
• For each exp. we integrate out all nuisance
  parameters (all variables except ?sig) from
  Lsig, and calculate the upper limit at 95 C.L.
• Calculate the median of N individual upper
  limits

t-channel
s-channel
15
Posterior probability densities for data
t-channel
s-channel
The maxima of the probability densities give the
most probables values for the cross
sections Since all these results are compatible
with zero, we set upper-limits (at 95 CL)
?t-ch 0.02.4-0.0 pb ?s-ch 4.6 3.8 pb
?t-ch lt 10.1 pb ?s-ch lt 13.6 pb
16
Combined Search for Single Top

• Measurement of the combined t-channel plus
  s-channel signal in the data
• use kinematic variable that discriminates st
  channels from the background
• total transverse energy HT distribution (scalar
  sum of MET, Et of lepton and jets)
• same HT distribution for s and t-channels
• different distributions for ttbar and non-top
  background

Make a likelihood function similar to that in
separate search Use smoothed HT distributions
17
HT Distributions for Combined Search
Data Distributions versus SM Expectation for
combined sample
18
Limits for Combined Search
Expected sensitivity
Probability density from data
?st (MPV) 7.7 5.1-4.9 pb ?st (95 CL) lt
17.8 pb
?st (a-priori) lt 13.6 pb at 95 CL
19
Summary
95 CL limits and most probable x-sections values
(pb), with 162 10 pb-1

• No significant evidence for single top quark
  production
• set first limits on single top cross section
• compared with Run I improvement of upper limits
  of 28 (t-channel) and 20 (s-channel)
• Technical improvements of the analysis
• improved MC modeling for single top events
• full Bayesian treatment of systematic
  uncertainties

20
Multivariate Analysis

• New analysis improvements
• Resolution of Ml?b
• Factors contributing to poor resolution
• Jet energy, b-jet choosing, neutrino choosing,
  lepton energy.
• use a kinematic fitter to find the neutrino
  solution most consistent with measured values of
  the b-jet and MET

• Multivariate likelihoods
• use a combination of variables to optimize
  signal and background separation
• variables include Q?, dijet invariant mass, cos
  ?lepton other jet, HT, ET
• likelihood function computed from distributions
  of each variable
• compute sensitivities (i.e how much luminosity
  for discovery ?)

21
Conclusion

• CDF accumulated more events than in entire Run I
• Single top search at the Tevatron is more
  challenging than anticipated
• CDF prepares advanced analysis techniques
• multivariate (neural-network) analysis
• CDF expects 3? signal evidence with 1.5 fb-1
• Many improvement are needed to observe single top
  in the next few years but it is feasible in Run
  II !

22
Backup slides
23
(No Transcript)
24
(No Transcript)