Status of the Hadronic Top Search

1
Status of the Hadronic Top Search
P. Azzi, A. Castro, G. Cortiana,T. Dorigo, A.
Gresele, J. Konigsberg, G. Lungu, A.
Sukhanov

• The all hadronic channel
• Samples data and MonteCarlo
• Kinematical Selection
• Tag Rate
• Background and Systematics
• Btagging Efficiency
• Cross section
• Conclusions

2
Dataset and kinematical selection optimization

• Dedicated trigger N(jet)gt4 with Etgt15 and
  ?Etgt125 GeV
• ?Lint 165pb-1 with all relevant subdetectors on
  and ok
• RAW and CORRECTED (L7) JET energy information
  used
• Signal MC Herwig Detector Simulation TrigSim
• We apply some pre-requisites for a minimal
  clean-up of the sample (see CDF Note 6808)
• Optimization of S/?B for jet multiplicity gt6 and
  for the following quantities (see CDF Note 6808)
  with these results
• ?Et ? 320 GeV
• ?Et/?s (centrality) ? 0.77
• (Aplanarity 0.0037 x ?3Et) ? 0.85

3
Summary table of kinematical selection using at
least 6 jets
Cut ttbar evt (165 pb-1) MJ evt Eff.() (inclusive ttbar) S/B
Pre-requisites 631 1134030 54.7 1/2000
6 ? Njet ? 8 244 83219 22.0 1/340
(Apla0.0037x?3Et)?0.85 97 4221 8.7 1/43
Centrality gt 0.77 72.0 1930 6.5 1/26
SumEt gt 320 GeV 69.0 1237 6.3 1/23
4
Systematics on the kin. selection
Source what ??/? ()
Fragmentation HERWIG vs PYTHIA 1
ISR modeling PYTHIA vs PY.noISR 13
FSR modeling PYTHIA vs PY.noFSR 5
Energy scale Change by ? 1? 26
Total All contributions 30
?incl (6.3 ? 0.04 (stat) ? 1.9 (syst))
5
Secondary Vertices (btags) .

• We use SECVTX (Summer 2003) in a method 1 line
  approach
• define a tag rate and a parametrization which can
  provide a bgr estimate
• compare positive OBServed tags to EXPected tags
  from the tag rate parametrization
• We do so before the application of any
  kinematical selection and derive a systematic
  uncertainty on the bgr evaluation
• Finally we apply a tight kinematical selection
  and look for an excess of tags w.r.t. the bgr as
  expected from top

6
Tag Rate vs Et, Eta, Ntrk and Apla
Eta 3 bins
Et 8 bins


Ntrk 11 bins
Apla 8 bins
7
Summary table for the parametrization (8?11?3?3)
4 jets 5 jets 6 jets ? 7 jets
Events 532634 249573 66807 15599
Taggable jets 1274233 742349 238532 67039
OBS ( tags) 38421 21313 6680 1708
EXP ( tags) 38421?191 21150?116 6524?46 1733.5?18
Nobs-Nexp/Nexp() 0 ? 0.8 0.3 ? 0.9 2.3 ? 1.4 -1.4 ? 2.6
8
(OBS EXP) / EXP vs Jet Multiplicity

• If we plot the ratio
• (OBS EXP) / EXP
• as a function of the jet
• multiplicity ( for 4, 5, 6 or ? 7)
• Tags EXP is consistent with OBS

9
Systematics uncertainties on the bgr estimate

• Different control samples
• pick events with the SMALLEST possible presence
  of ttbar events
• highly populated
• Three cases
• 2 with reverse cuts
• 1 check stability

10
Systematics on Njet

• If we compare OBS and EXP for Njet ? 6,
• (Apla0.0037x?3Et) ? 0.85, Centrality ? 0.77 and
  ?Et ? 320 GeV we
• see that
• Nobs 3542
• Nexp 3550 ? 47
• (Nobs Nexp)/Nexp (0.2 ? 0.1)
• ? We consider a systematic uncertainty on Njet
  0.2

11
Systematics on ?Et,
We consider all events with 5 jets and
(Apla0.0037x?3Et) ? 0.85, Centrality ? 0.77.
Systs. on ?Et 1.4
1.6
ALLOW SLOPE
0.6
CONVOLUTION
12
on (Apla0.0037x?3Et) and on Centr.
Syst. (Apla0.0037x?3Et) 4.7
Syst. Centr. 0.3
1.2
1.4
0.8
0.6
13
Systematics on Inst.Luminosity,

• We consider as a
• control sample
• all events with 4 jets.
• Syst. Inst. Lum. ltlt 1

14
on Run and on jet-?
Syst. Run ltlt 1
Syst. jet-? ltlt 1
15
Total Systematic uncertainty

• We now combine all systematics (sum in
  quadrature)
• Njet 0.2
• SumEt 1.4
• Centrality 0.3
• (Apla0.0037x?3Et) 4.7
• Inst. Luminosity ltlt 1.0
• Run ltlt 1.0
• Jet-? ltlt 1.0
• Total systematic uncertainty 5

16
Background estimate after KIN SEL
4 jets 5 jets 6 jets ? 7 jets
Events 60 420 773 883
EXP () 7.9?0.4?0.4 62.9?1.4?3.1 126 ? 2 ? 6 152 ? 2 ? 7.6
OBS () 11 70 170 156
Nobs- Nexp 3.1 ? 0.6 7.1 ? 3.4 44 ? 6 4 ? 8
17
some preliminary results

• For Njet ? 6 jets
• (SIGNAL REGION) we see
• Nobs 326 tags
• Nexp(bgr) 278.0 ? 2.8 (stat)
• ? 13.9(syst) tags
• ? NobsNexp 48.0 ? 14.0 tags

18
Btagging Efficiency

• We can follow two methods
• factorization method where
• ?overall,evbtag ? evt btag (1 - ?evt
  btag) ? ?evt mistag
• with ? evt btag F2b ? ?btag ? SF ? (2- ?btag ?
  SF) F1b ? ?btag ? SF
• and SF 0.86 ? 0.07.
• We have done a cross-check with the single lepton
  analysis
• (following CDF Note 6598)
• counting method where we degrade the tagged jets
  with the SF
• If we compare the two methods they give
  consistent results.

19
Btagging efficiency per event and per jet
Njet ?evtbtag()
4j 53.1 ? 3.5
5j 53.6 ? 3.5
6j 54.7 ? 3.5
gt6j 54.9 ? 3.5
gt6jkin 59.3 ? 3.7
Njet ?jetbtag ()
4j 63.5 ? 5.1
5j 64.4 ? 5.2
6j 66.6 ? 5.5
gt6j 66.6 ? 5.5
gt6jkin 73.7 ? 6.0
Eff. b-evt (59.3 ? 3.7)
Eff. b-jet (73.7 ? 6.0)
If NO matching with b-jet Eff. jet (83.7 ?
8.2)
20
Efficiencies plot
21
Cross Section

• The presence of tt events in the pretag sample
  leads to an
• overestimate of the background. We account for
  it with an
• iterative procedure and then we rescale the
  background
• Nexp Nexp ? ((N Ntt) / N)pretag 266.1
• and the corrected excess would be
• Nobs- Nexp 326 266.1 60

22
Final summary table
4 jets 5 jets 6 jets ? 7 jets
Events 60 420 773 883
Backg. 7.9?0.4?0.4 62.9?1.4?3.1 126 ? 2 ? 6 152 ? 2 ? 7.6
Backg. corrected 266.1 ? 16.7 266.1 ? 16.7
Top MC (6.7 pb) 0.3 ? 0.1 6.5 ? 2.0 51.3 ? 15.9 51.3 ? 15.9
Tags Expected 8.2 69.4 317.4
OBS () 11 70 170 156
23

• We build the following likelihood function

with the following input values
Lumin L 165-10 pb-1
Obs tags n 326
Exp tags b 278-14
Exp tags corr b 266.1-16.7
Kin eff. ek 6.3-1.9
Tag eff. eb 83.7-8.2
bb(N-Ntt)/N Npretag events Nttpretag tt
events
24

• The maximization of the likelihood gives, as
  central value

• The cross section (iterative) amounts to

25
Conclusions

• First full pass with Run I method top cross
• section.
• To do next
• brush up the systematic especially jet energy
  scale and state of the cut PSR, FSR and PDF
• Seek preblessing next March

26
Kinematic cuts optimization Apla vs ?3Et
We reject the bottom left corner. By cutting on
Aplanarity K x ?3Et. We pick up the best value
for k and look for the maximum of S/?B
mj
Aplanarity
tt
Projection
SumEt3
Optimization
27
Kinematic cuts optimization Centrality
After the cut (Aplanarity 0.0037 x ?3Et) ? 0.85
we find the best value to cut on the Centrality
S/B
mj
S/?B
28
Kinematic cuts optimization ?Et
After the cut (Aplanarity 0.0037 x ?3Et) ? 0.85
and Centrality ? 0.77 we find the best value to
cut on ?Et
S/B
tt
mj
S/?B
29
Parametrization using matrix of Jet50

• As a cross-check , we apply the matrix of Jet50
  on our data
• sample even if it is not much appropriate
  because
• it comes from a sample with 2 jets and low SumEt
  and we use a data with at least 6 jets at higher
  SumEt
• the statistic of the Jet50 sample is very small
  in our signal region
• and this is reflected in the bigger
  Nobs-Nexp/Nexp

4 jets 5 jets 6 jets ? 7 jets
Nobs-Nexp/Nexp() 4.0 ? 0.7 4.7 ? 0.9 7.0 ? 1.4 4.6 ? 1.5
30
Negative Tags before and after kin. sel.
4 jets 5 jets 6 jets ? 7 jets
OBS (- tags) 9774 5593 1745 515
EXP (- tags) 9774?98 5232?58 1593?23 420?9
OBSEXP/EXP() - 7 9 22
EXP(JET50) (-tags) 10508?107 5715?72 1748?26 462?9
4 jets 5 jets 6 jets ? 7 jets
OBS (- tags) 5 15 42 52
EXP (- tags) 3 ? 0.4 18.1 ? 0.6 34.4 ? 1.0 39 ? 1.0
EXP(JET50) (-tags) 3.4?0.2 21.4?0.8 39.2?1.0 44.1?1.3
31
Cross Check with matrix from Jet50
32
Systematics on ?Et, Centr and (Apla0.0037x?3Et)

1. for events passing the kin. sel., we drop the 6th
   jet and reconstruct new ?Et, Centr and
   (Apla0.0037x?3Et) distributions (6-to-5,,
   distributions)
2. in the corresponding control sample we fit the
   distributions of the Nobs/Nexp ratio with a first
   degree polynomial
3. convolute the polynomial function with the
   corresponding 6-to-5,, normalized distribution
4. the integral of the convolution gives the total
   systematic uncertainty