0. on behalf of the ATLAS and CMS collaborations
Moriond QCD
Early physics with top quarks at LHC
P.Ferrari
CERN
on behalf of the ATLAS and CMS collaborations
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

1. LHC is a tt factory + Total production cross section
(90%) +
(10%)
tt production cross section at LHC:
~833 pb
tt production cross-section
at Tevatron:
6.7 pb
2 tt events per second !
> 8 millions tt events expected per year
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

2. Top Physics day one
In 2008 ECM = 14 TeV few fb-1  already negligible statistical err
1) Top properties and basic SM physics at s = 14 TeV :
Estimate of σtop ~ 20% accuracy
Start to tune Monte Carlo
Measure top mass  feedback on detector performance
2) Understand/calibrate detector and trigger: tt  bl bjj
Light jet energy scale selecting a pure sample of W jj in tt events (< 1%)
b-tag efficiency (~ 5%)
Missing energy calibration
3) Prepare for new physics:
Resonances, MSSM higgses, SUSY, FCNC
Measure differential cross sections (ds/dpT,ds/dMtt) sensitive to new physics
(provides also an accurate test of SM predictions)
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

3. Light jet energy calibration
Template histograms with different E scales a and relative E resolutions b:
W  qq in ~106 PYTHIA tt events
Simple tt  lnb jjb selection with tt events :
1(e/m) pT>20 GeV, ETmiss>20 GeV, = 4 jets pT>40 GeV (2 b-tagged),
150 GeV< mjjb< 200 GeV  W purity ~83%
Fit each template histogram to mjj in the « data », find best c2
a = ± 0.004, b = 1.47 ± 0.05
Statistics limited. Unknown syst limit< 0.5% from combinatorial backg. and templates shape
JES as a function of energy ( n energy bins and nxn matrix template)
a = 1
« Data »
Best fit
Ej/Eq
±2%
@ 1.3 fb-1
b-jets
Light jets
Pamela Ferrari
Rencontres de Moriond 2007 QCD session Eq

4. Calibrating b-tagging
Using semi-leptonic (and fully leptonic) tt events
Optimize the jet pairing efficiency via mass
constraints in kinematic fits and likelihoods.
Only one jet is tagged as b-jet (on Whad side)
TOP CANDIDATE
W CANDIDATE
CMS NOTE
Isolate jet samples with a highly enriched b–jet content, on which the b–jet identification algorithms can be calibrated.
Main systematics :ISR/FSR
For 1 fb-1 (10fb-1) relative accuracy on the b–jet identification efficiency is ~ 6% (4%) in barrel region and about 10% (5%) in the endcaps.
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

5. Day one: can we see the top?
We will have a non perfect detector:
Let’s apply a simple selection
No b-tag
relaxing cut on 4th jet: pT>20 GeV:
doubles signal significance!
No b-tag
|mjj-mW| < 20 GeV
Hadronic top=3 jets
maximising pT top
W =2 jets maximising pT W in jjj rest frame
4 jets pT> 40 GeV
600 pb-1
Isolated lepton pT> 20 GeV
ETmiss > 20 GeV
100pb-1
W+jets
Combinatorial
Siginficance (s)
only with 100 pb-1 (few days)
100 pb-1
Luminosity (pb-1)
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

6. Refining the selections: semi-leptonic case
More refined selection studied with the aim of applying it to
x-section, mass, polarization studies..
Example: CMS NOTE 2006/064
1 isolated lepton pT>20 GeV
≥4 jets ET>30 GeV |h|<2.4
2 b-tagged jets
Coverging kin. fit to mW
stat
esel ~ 6.3%
S/B~26.7
total w/o lumi
total w lumi
@5 fb-1 stt(m)=0.6% (stat)± 9.2% (syst)±5.0%(lumi)
Exploiting new topological variables from D0?
Sphericity S and Aplanarity A
Centrality C
HT =
Df(lep,n)
KTmin=min D(h,f) between 2 jets
1fb-1
Not very useful
to separate
from W+jets
after selection
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

7. Summary of cross-section
The cross-section has also been extracted from in the di-leptonic and fully hadronic channels here examples from:
CMS NOTES /
Dstt/stt
syst (%)
Dstt/stt stat (%)
Dstt/stt lumi (%)
Main
bkg
Eff
(%)
S/B
10fb-1
Semi-Leptonic
9.7
0.4 3
Btag PDF PileUp tt
W+j
6.3
26.7
Dilepton 11
0.9
PDF
Btag
JES
ttll with
(Wtnt, tl) 5
5.5
1fb-1
hadronic 20
JES PileUp
QCD
1.6
1/9
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

8. Top mass measurement in lepton + jets channel
1) minimization of c2  reconstruct mW hadronic & jet E rescaling (a1,a2)
keep W’s if |mW GeV| < 2GmW
chose b-jet that maximises top pT
W purity 56.5%, top purity 45%, e=1.1%
2) Kinematic fit:
10fb-1
Fullsim
CMS NOTE
mtop
Kinematic fit to reconstruct entire tt final state:
c2 based on kinematic constraints (El,j & directions vary within resolution) c2 minimisation, event by event
Mtop fitted in slices of c2
Estrapolation from linear fit: mtop = mtop(2 = 0)
Gaussian/Full Scan Ideogram estimator for mt:
Event by event likelihood method convoluting the event resolution function with expected theoretical template. mtop obtained from maximum likelihood method
PYTHIA
Fullsim
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

9. Top mass measurement (semileptonic channel)
c) Selection of high pT top quarks pT(top) > 200 GeV/c:
t and t tend to be back-to-back  used as constraint to reduce bkg
3 jets in 1 hemisphere tend to overlap: collect E in a cone around candidate top
less sensitive to jet calibration. Mass scale recalibration based on hadronic W,
independent systematic errors  gain in combination
Source of uncertainty
Had. top Mtop (GeV/c2)
Kinematic fit Mtop (GeV/c2)
High PT sample
Mtop (GeV/c2)
Light jet energy scale (1 %)
0.2
b-jet energy scale (1 %)
0.7
b-quark fragmentation
0.1
0.3
ISR
FSR 1.
0.5
Combinatorial background
Mass rescaling
0.9
UE estimate (± 10 %)
1.3
Total
1.6
Statistical
0.05
Comparing the 3 methods
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

10. Di-lepton channel and Hadronic channels
1fb-1
PYTHIA
Dilepton channel: clean channel but need to reconstruct
2 n’s. Reconstruction via 0C fit assuming mW and 2 equal
masses for top mt1=mt2 (6 eq. ,6 unknowns)
The different n solutions are weighted using the
SM prediction for the n and n E spectra
The neutrino solution with the highest weight is chosen mtop
S/B = 12 -
Hadronic channel: full kinematic reconstruction of both
sides but huge QCD multijet background:
6-8 jets, ET>30 GeV
Centrality>0.68,aplanarity>0.024
ETtot- ET of 2 leading j>148 GeV
2 b-tagged jets
Best jet pairing obtained from
likelihood based mainly on
angular distrubution of jets.
CMS NOTE
(stat)
mt (GeV/c2)
(syst)
@1fb-1 dilepton
~1.5
~4.2
@1fb-1 hadronic
~0.6
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

11. Resonances in Mtt pp  X  tt with tt decaying semi-leptonically
Technicolor, Strong EW symmetry breaking models, Z’, SUSY:
usual semi-leptonic events preselection
use W and top mass constraint:
neutrino pZ from mW constraint, solution
giving best top mass is retained
|mjj-mW|  20 GeV
b-jet associated with hadronic top is the
one maximising pTtop
|mbjj-mT|  40 GeV
5fb-1
3xZ’ signal
Since pT of top from resonance decay is
larger than in direct production
Add lower cut on top pT 370,390,500 GeV/c for mZ’ =1,1.5,2 GeV/c2 to increase purity (s/B~ )
@5 fb-1
MZ’= 1 TeV
MZ’= 1.5 TeV
MZ’= 2 TeV CL
~2.75s
~2.96s
~3.3s
x-sec (pb) ~4 ~3
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

12. Flavour Changing Neutral Currents
CMS NOTE 2006/093
No FCNC at tree level in SM:
u (c,t)
5σ sensitivity SM
2HDM
MSSM u
Z/γ
Look for FCNC in top decays:
SN-ATLAS
tqg
(+1l+missing)
tqg
(+1l+1g+missing)
tqZ
(2jets+3l+missing) t
@ 10 fb-1 2 orders of magnitude better than Tevatron/LEP/HERA
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

13. Top spin correlations A=
t and t are produced unpolarized, but spins are correlated
anomalous coupling (technicolor), tH+b, spin 0/2 heavy resonance H/KK gravitons  tt, would move A away from SM expectation
s(tLtL) + s(tRtR) - s(tLtR) - s(tRtL) A=
s(tLtL) + s(tRtR) + s(tLtR) + s(tRtL)
Fit to double differential distribution
Eur.Phys.J.C44S
Semilep. + dilep. (10 fb-1)
Fitting to distribution of
angles bewteen top spin analyser in top rest frame versus angle of t spin
analyser in antitop rest frame
Syst. dominated by b-JES, top mass and FSR
A=0.41 0.014(stat) 0.023(syst)
CMS NOTE 2006/111
Semilep. (10 fb-1)
Fitting to distribution of
lepton angle vs b-quark angle in the tt rest frame
lepton angle vs lower energy quark angle from the W-decay in the tt rest frame
Abt,lt=0.375 ±0.014(stat) (syst)
Aqt,lt=0.346 ± 0.021(stat) (syst)
+0.055
-0.096
+0.026
-0.055

14. Conclusions
LHC startup will require a long period of development and understanding
LHC is a top factory, but before performing precision measurements, a huge effort is needed in order to
Understand the detectors and control systematics
Complete study using full simulations and NLO generators
Early top signal will help
We could get top signal with ~ 100 pb-1
s(tt) to ~13% and Mtop to 1% with 1fb-1
In addition our aim is, as soon as we get a large statistics (few fb-1), to be ready for early discovery of new physics!
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

15. BACK-UP TRANSPARENCIES
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

16. Calibrating the missing energy
We can calibrate the missing Energy since in semileptonic tt events the momentum of the neutrino is constrained from kinematics : MW
known amount of missing energy per event
Calibration of missing energy vital for all
(R-parity conserving) SUSY and most exotics! t
Example from SUSY analysis
Events
Miscalibrated detector or escaping ‘new’ particle
Perfect detector
Range: 50 < pT < 200 GeV
Missing ET (GeV)
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

17. Effect of cut on PT of 4th jet
Lowering PT requirement on 4th jet to 20 GeV
increase in significance top signal by factor > 2
40 GeV % 30 GeV % 20 GeV %
PT cut Fraction
Top signal
Jet 4
42.6 %
42.6 %
Cut on 4th jet does not bring any advantage
Actually … it cuts away important information: large fraction of jets associated to quarks from W decay
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

18. Gaussian Ideogram, full scan ideogram
CMS NOTE
Build event by event from the covariance matrices of the kinematics of the 3 fitted jets,
the relative compatibility of the reconstructed kinematics of the event with the hypothesis of the a heavy object of mass mt decays into 3 jets.
This is usually called the ideogram of the event or resolution function of the event
In the full scan the top mass is not fixed but varies 125<mt<225 GeV
To estimate the top mass, the ideogram is convoluted with the Theorectical expected probability density function and after combining all likelihoods from all events a maximum likelihood method is applied to get mt.
Sources
(5% On-Off)
(1.5%)
(2%)
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

19. Mtop other 2 methods Exclusive (J/ℓℓ) decays: detect clean high mass
products, easy to reconstruct and little background
2) via cross-section: very much sensitive to the top mass: 5% on  gives mt~2 GeV/c2, error luminosity and the pdfs
different systematic contributions: top mass determined
via m(3l):
without using the b-tagging !
almost ignoring jet reconstruction !
 large improvement in combination with direct reconstruction
Challenging because of the extremely low branching ratio:
BR = 0.810-4 after selection+trigger efficiency
CMS NOTE
m(3l)
Sensitive
to mt
(stat)
mt (GeV/c2)
(syst. instr.)
(syst. th.)
mt(GeV/c2)
tot
(GeV/c2)
exclusive J/ dec
~0.5
~1.4
~1.5
via cross-section
~0.1
~0.7
~4.0
~4.1
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

20. Di-lepton event selection
CMS NOTE
Cut based selection:
Single or di-lepton trigger
Two isolated oppositely charged
leptons with ET>20 GeV,<2.5
Missing ET>40 GeV
2 jets with ET>20 GeV, <2.5
2 b-tagged jets
Main background represented by Z+jets when no b-tagging is present ( efficiency 15% )
With b-tagging, efficiency about 5% with excellent
background reduction S/B~5 (B mainly from leptonic  decays)
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

21. Systematic effects for top physics
Almost all SM measurements at LHC dominated by systematic errors.
Can be divided into instrumental and from theory/modeling
Dominant instrumental uncertainties for top physics:
Luminosity:
Reasonable goal is 3-5%
 measure number of interactions/bunch crossing (HF) and (pp) (TOTEM)
Reconstruction related:
Jet energy scale
need calibrated calorimetry (beam tests, MB, single particles, Z, W…)
need jet energy calibration to a few % (with Z()+jet)
need excellent energy flow (association tracker+calo+muon system)
b-tagging efficiency+fake rate
use tt for calibration: to 4-5% with 10fb-1
Lepton identification and energy scale
use Z, other mesons. Less crucial than for the W mass measurement
Theory related systematics are as important as instrumental ones !
Pamela Ferrari
Rencontres de Moriond 2007 QCD session

22. Theoretical systematic uncertainties
ISR/FSR:
vary QCD and Q2max from low to high radiation
light jets Fragmentation: vary only the fragmentation parameters within
errors from LEP/SLD tunings.
“b jets” Fragmentation:
error estimated changing the Peterson parameter ( ) within its
theoretical uncertainty (0.0025)
Minimum bias and underlying event:
extrapolate from low energy UA5/Tevatron and change main pT cut-off
parameter
PDF parametrization: CTEQ6M, contrain using LHC data
Hard process scale: switch among reasonable definition of the Q2 scale
Pamela Ferrari
Rencontres de Moriond 2007 QCD session