Document 7661984

Download Report

Transcript Document 7661984

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
0
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
1
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
2
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  lb jjb selection with MC@NLO 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 = 0.937 ± 0.004, b = 1.47 ± 0.05
Statistics limited. Unknown syst limit< 0.5% from
combinatorial backg. and templates shape
a=1
@ 1.3 fb-1
Ej/Eq
« Data »
JES as a function of energy ( n energy
bins and nxn matrix template)
±2%
•b-jets
•Light jets
Best fit
Pamela Ferrari
Rencontres de Moriond 2007 QCD session
Eq
3
Calibrating b-tagging
 Using semi-leptonic (and fully leptonic) tt events
 Optimize the jet pairing efficiency via mass
constraints in kinematic fits and likelihoods.
W CANDIDATE
TOP CANDIDATE
 Only one jet is tagged as b-jet (on Whad side)
CMS NOTE 2006-013
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
4
Day one: can we see the top?
We will have a non perfect detector:
Let’s apply a simple selection
W =2 jets maximising pT W in jjj rest frame
 No
No b-tag
b-tag
 relaxing cut on 4th jet: pT>20 GeV:
doubles signal significance!
Hadronic top=3 jets
maximising pT top
4 jets pT> 40 GeV
Isolated lepton
pT> 20 GeV
600 pb-1
|mjj-mW| < 20 GeV
-1-1
100pb
100 pb
Siginficance (s)
ETmiss > 20 GeV
Combinatorial
only with 100 pb-1
W+jets (few days)
Pamela Ferrari
Rencontres de Moriond 2007 QCD session
Luminosity (pb-1)
5
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
total w/o lumi
total w lumi
esel ~ 6.3%
S/B~26.7
@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 =
4
p
jet 1
T
Df(lep,)
 KTmin=min D(h,f) between 2 jets

Pamela Ferrari
Not very useful
to separate
from W+jets
after selection
1fb-1
Rencontres de Moriond 2007 QCD session
6
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 2006-064/ 2006-077
Dstt/stt
syst (%)
10fb-1
SemiLeptonic
9.7
Dstt/stt
stat (%)
0.4
Dstt/stt
lumi (%)
Main
syst (%)
Eff
(%)
S/B
7
3.4
3.2
tt
W+j
6.3
26.7
3
Btag
PDF
PileUp
5
4
4
ttll with
(Wtt,
tl)
5
5.5
11
10
QCD
1.6
1/9
10fb-1
Dilepton
11
0.9
3
PDF
Btag
JES
1fb-1
hadronic
20
3
5
JES
PileUp
Pamela Ferrari
Main
bkg
Rencontres de Moriond 2007 QCD session
7
Top mass measurement in lepton + jets channel
1) minimization of c2  reconstruct mW hadronic & jet E rescaling (a1,a2)
χ2
(M




jj
(α1,α2 )  MW )2
Γw2
2
 Ej1 (1  α1 ) 
 E (1  α2 ) 
   j2






σ j1
σ j2




2
keep W’s if |mW - 80.4 GeV| < 2GmW
chose b-jet that maximises top pT
W purity 56.5%, top purity 45%, e=1.1%
10fb-1
MC@NLO
Fullsim
2) Kinematic fit:
CMS NOTE 2006-066
mtop
Kinematic fit to reconstruct entire tt final state:
PYTHIA
Fullsim


•

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(c2 = 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
Rencontres de Moriond 2007 QCD session
8

Pamela Ferrari
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
Comparing the 3 methods
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
0.2
b-jet energy scale (1 %)
0.7
0.7
b-quark fragmentation
0.1
0.1
0.3
ISR
0.1
0.1
0.1
FSR
1.
0.5
0.1
0.1
0.1
Combinatorial background
Mass rescaling
0.9
UE estimate (± 10 %)
1.3
Total
Statistical error @10fb-1
Pamela Ferrari
1.3
0.9
1.6
0.05
0.1
0.2
Rencontres de Moriond 2007 QCD session
9
Di-lepton channel and Hadronic channels
Dilepton channel: clean channel but need to reconstruct
2 ’s. Reconstruction via 0C fit assuming mW and 2 equal
masses for top mt1=mt2 (6 eq. ,6 unknowns)


S/B = 12
The different  solutions are weighted using the
SM prediction for the  and  E spectra
The neutrino solution with the highest weight is
chosen mtop
PYTHIA
1fb-1
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.
Pamela Ferrari
CMS NOTE 2006-077
(stat)
mt (GeV/c2)
(syst)
mt (GeV/c2)
@1fb-1 dilepton
~1.5
~4.2
@1fb-1 hadronic
~0.6
~4.2
Rencontres de Moriond 2007 QCD session
10
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
5fb-1
giving best top mass is retained
• |mjj-mW|  20 GeV
3xZ’ signal
• b-jet associated with hadronic top is the
one maximising pTtop
• |mbjj-mT|  40 GeV
 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~0.06-0.08)
Pamela Ferrari
@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
~3
Rencontres de Moriond 2007 QCD session
11
Flavour Changing Neutral Currents
u (c,t)
No FCNC at tree level in SM:
SM
10-14-10-12
2HDM
10-7-10-4
MSSM
10-6-10-5
CMS NOTE 2006/093
5σ sensitivity
u
Z/γ
Look for FCNC in top decays:
tqZ
(2jets+3l+missing)
tqg
(+1l+missing)
tqg
(+1l+1g+missing)
SN-ATLAS-2007-059
t
@ 10 fb-1 2 orders of magnitude better than Tevatron/LEP/HERA
Pamela Ferrari
Rencontres de Moriond 2007 QCD session
12
Top spin correlations
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
Fit to double differential distribution
Eur.Phys.J.C44S2 2005 13-33
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)
Pamela Ferrari
A=
s(tLtL) + s(tRtR) - s(tLtR) - s(tRtL)
s(tLtL) + s(tRtR) + s(tLtR) + s(tRtL)
d 2N
1
 (1  A1 2 cos1cos2 )
N dcos1dcos2 4
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
+0.055
(syst)
-0.096
+0.026
0.021(stat) -0.055
(syst)
Abt,lt=0.375 ±0.014(stat)
Aqt,lt=0.346 ±
13
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



Early top signal will help



Understand the detectors and control systematics
Complete study using full simulations and NLO generators
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
14
BACK-UP TRANSPARENCIES
Pamela Ferrari
Rencontres de Moriond 2007 QCD session
15
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!
Example from SUSY analysis
Events
t
Miscalibrated detector or
escaping ‘new’ particle
Perfect detector
Missing ET (GeV)
Pamela Ferrari
t
Range: 50 < pT < 200 GeV
Rencontres de Moriond 2007 QCD session
16
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
Top signal
PT cut
40 GeV
Fraction
Jet 4
42.6 %
30 GeV
68.1 %
20 GeV
87.7 %
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
17
CMS NOTE 2006-066
Gaussian Ideogram, full scan ideogram
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
18
Mtop other 2 methods
1)
Exclusive (J/ℓℓ) decays: detect clean high mass
products, easy to reconstruct and little background
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 2006-058
Sensitive
to mt
m(3l)
2) via cross-section: very much sensitive to the top mass: 5% on s gives
mt~2 GeV/c2, error luminosity and the pdfs
Pamela Ferrari
(stat)
mt (GeV/c2)
(syst. instr.)
mt (GeV/c2)
(syst. th.)
mt(GeV/c2)
tot
(GeV/c2)
exclusive
J/ dec
~0.5
~0.5
~1.4
~1.5
via crosssection
~0.1
~0.7
~4.0
~4.1
Rencontres de Moriond 2007 QCD session
19
Di-lepton event selection
CMS NOTE 2006-077
Cut based selection:
 Single or di-lepton trigger
 Two isolated oppositely charged
leptons with ET>20 GeV,h<2.5
 Missing ET>40 GeV
 2 jets with ET>20 GeV, h<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 t decays)
Pamela Ferrari
Rencontres de Moriond 2007 QCD session
20
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 s(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(g)+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
21
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 (-0.006 ) 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
22