Title: LHC School Lecce
1Top Physics at CDF
- Sandra Leone
- INFN Pisa
- Italo-Hellenic School of Physics 2004
- The Physics of LHC Theoretical Tools and
Experimental Challenges - Lecce
- May 20-24, 2004
2Outline
- Motivations for studying top
- A brief hystory
- Top production and decay
- The Tevatron CDF
- Detector issues
- Identification of final states
- Cross section measurement
- Single top production
- Mass determination
- Study of top properties
- Whats next?
3Motivations for Studying Top
- Only known fermion with a mass at the natural
electroweak scale. - Similar mass to tungsten atomic 74
35 times heavier than b quark. - Why is Top so heavy?
- Is top involved in EWSB?
- (Does (2 ? 2 GF)-1/2 ? Mtop mean anything?)
- Special role in precision electroweak physics?
- Is top, or the third generation, special?
- New physics may appear in production (e.g.
topcolor) or in decay (e.g. Charged Higgs). - The Fermilab Tevatron has been the only place,
and will be until the LHC turns on, to study the
top quark.
4A Brief History of Top
- Top quark was expected in the Standard Model (SM)
of electroweak interactions as a partner of
b-quark in SU(2) doublet of weak isospin for the
third family of quarks - (weak isospin of b can be inferred from the
forward-backward asymmetry in ee- ? bb) - Anomaly free SM requires the sum of the family
charges to be zero given the b (and the tau
lepton) there should be a 2/3 charge quark
5A Brief History of Top
- 1977-1994 increasing lower top mass limits
- First top evidence in 1994 in CDF data, 19 pb-1,
15 events on a background of 6, - 2.8 s excess, not enough to claim discovery
- Confirmed in 1995 by CDF and D0 in first 70
pb-1 of run 1 data (4.8 s excess). - Final run 1 top analyses based on 110 pb-1.
- Production cross sections in many channels.
- Mass 174.3 ? 5.1 GeV (CDF/DØ combined).
- Study of several aspects of event kinematics.
- Limits on single top production, rare/non-SM
decays. - Overall consistency with the standard model.
- But only 100 analyzable top events
- analyses statistics-limited.
6A more cultural perspective of Top
About 5 orders of magnitude range in quark masses!
7Production and Decay Basics
SM predicts BR( t ? Wb) 100
At the Tevatron top quarks are mainly pair
produced
b
Event topology determined by the decay modes of
the 2 Ws in final state
t
W
85
t
W
15
stheory 7 pb
b
NB qq, gg fractions reversed at LHC
b-jet identify via secondary vertex or soft
lepton tag
8Top Decay add. Motivation for Studying Top
- In the SM, assuming V-A coupling with a CKM
matrix parameter Vtb 1 for the t ? bW
decay vertex, one gets (LO) - G(t ? bW ) ? 175 MeV (MT/MW)2 (MT,MW gtgt Mb)
- ? G(t ? bW ) ? 1.5 GeV ? t(top) ? 4 x 10-25 s
- Non-perturbative QCD hadronization takes place in
a time of order
L-1QCD ? (100 MeV)-1 ?
10-23 s - top decays before hadronizing, as free quark
(no top hadrons, no toponium spectroscopy) - the top quark provides the first opportunity to
study the decay characteristics of a bare
quark. - t ? Ws, t ? Wd allowed but suppressed by factors
of ? 10-3 and ? 5 x 10-5 respectively
9t-tbar Final States
- Dilepton (ee, µµ, eµ)
- BR 5
- 2 high-PT leptons 2 b-jets large missing-ET
- Lepton (e or µ) jets
- BR 30
- single lepton 4 jets (2 from bs) missing-ET
- All-hadronic
- BR 44
- six jets, no missing-ET
- thad X
- BR 23
Most favorable channels for top physics
More challenging backgrounds, but measurements
still possible
10The Tevatron
Run II ?s 1.96 TeV Started in spring
2001 After a commissioning period, data good for
physics since February 2002
11 Accelerator Improvements for Run II
- Energy upgrade 1.8 ? 1.96 TeV
- 30-40 increase in top cross section
- New Main Injector
- Improve p-bar production
- Recycler ring (commissioning)
- Accumulate p-bars
- Luminosity upgrades factor of 3-4 so far
- Increased from 6x6 bunches with 3.5ms between
bunch crossing in Run I to 36 p and pbar bunches
?396 ns between bunch crossing
12Tevatron Peak Luminosity
7 x 1031
Typical recent stores 5 6 x1031
13Integrated Luminosity
delivered 500 pb-1
on tape 400 pb-1
Start of physics-quality data
Taking data with gt 85 efficiency
Running stably since Feb. 02
Analyses use data till 10/2003 Results from
first 200 pb-1 ( 2 x Run 1) presented here.
14The CDF Run II Detector
- new front-end,DAQ
- trigger
- new L1 track trigger
- new L2 secondary vertex tracker (SVT)
- new silicon tracker
- 7 layers hlt2
- 3-D reconstruction
- new central tracker
- NstereoNaxial48
- Dpt/ptlt0.001pt
- Time of flight
- new forward calorimeter
- m,e id to h2
- Inherited from Run I
- Central Calorimeter
- Muon System (some new)
- Solenoid
15CDF Run II Detector
CentralPlug Calorimetry ??? ? 3.6
Muon Chambers ??? ? 1.5
10m
Central tracking ??? ? 1.0
Silicon tracking ??? ? 2.0
16Physics at a hadron collider
High pT lepton High ET jet, photon High Missing
ET (MET)
- Is all about the trigger!
Examine each pp collision (1.7 MHz) Select few
interesting events (lt70 Hz) Store for further
offline analysis
Keep 1 out of 25,000
Process Cross-section Event Rate
Inelastic pp 60 mb 6 MHz
pp?bb (b pTgt6 GeV, ?lt1) 10 µb 1 kHz
pp?WX?l?X 5 nb 0.4 Hz
pp?ZX?llX 0.5 nb 0.04 Hz
pp?tt?WWbb?l?bbX 2 pb 0.0002 Hz
pp?WH?l?bb (if MH120GeV) 15 fb 0.0000015 Hz
Assume L 100x1030 cm-2s-1, lelectron or muon
17Comparison of Cross Sections
- In Run 1, over 5 x 1012 total collisions, one in
every 1010 producing a ttbar event (c.f. one
every 2.5 X 106 producing a W event) - In Run 1 About 500 ttbar events produced per
experiment - About 50 remaining after all cuts, per experiment
- Run 1 uncertainties dominated by lack of
statistics
Note LHC will be a top factory, producing ? 2800
ttbar events per hour at low luminosity (1033)
18Principles for Particle Identification
Beam direction
19Principles for Particle Identification
Electrons (h lt 2.8) Muons (h lt 1)
A track in the Central Outer Chamber A track in the Central Outer Chamber
Corresponding energy deposition in EM calorimeter Small Had/EM shower shape consistent with that expected for an electron Corresponding energy deposition in calorimeter consistent with MIP Extrapolated track matching signal in muon chambers
20Electron identification in W events
- Energetic transverse energy ET gt 20 GeV
- Track pointing to EM pT gt 10 GeV/c
- and matching EM energy ET/pT lt 2.0 when ET lt
50 GeV - Well contained into detector Track z0 lt 60 cm
- Shower profile consistent with electron
- Signal in strip chamber embedded in EM at shower
max. - Shower contained into EM
EHAD/EEM lt 0.055 0.00045 E - Track-to-shower match ? 3 cm
- Topological W cut? the electron is isolated
Fractional calorimeter energy isolation lt 0.1
21Muon identification in W events
- Energetic Track pT gt 20 GeV/c
- Well contained into detector Track z0 lt 60 cm
- Track match to a muon chamber stub 3, 5, and 6
cm for CMU, CMP, and CMX, respectively - Cosmic ray veto
- Small energy deposition in calorimeter,
consistent with MIP - EEM lt 2 max0, 0.0115 (p - 100) GeV
- EHAD lt 6 max0, 0.0280 (p - 100) GeV
- Topological W cut? the muon is isolated
Fractional calorimeter energy isolation lt 0.1
22Principles for Particle Identification
Neutrino Jet Photon
No interaction seen in the detector Energy deposition in em and had calorimeter Energy deposition in em calorimeter
Missing Transverse energy ET Si ETi ? ni ET - ET Projective tower geometry. Fixed cone algorithm in h-f, DR 0.4 No associated track
23Method to identify a top signal
- Start from counting events passing cuts in all
decay channels - Optimization of signal region with respect to SM
background processes (control region) - Background dominates the production of ttbar
pairs by several orders of magnitude - How to separate signal from background
- Top events have very distinctive signatures
- Decay products (leptons, neutrinos, jets) have
large pTs - Event topology central and spherical
- Heavy flavor content always 2 b jets in the
final state!
24Dilepton Channel
Relatively clean Top and a small amount of SM
bkgds Down side is small event samples
- Signature
- Two high pT Isolated leptons opposite sign
- Veto Z, cosmic, conversion
- ??(ET,l/j)gt20o, or ETgt50 GeV
- ET gt 25
- Two jets with ETgt10 GeV
- Total ET gt 200 GeV (HT Scalar summed ET of
jets, leptons, and ET )
Expect S/B 9 Dominant backgrounds
Drell-Yan, Wjets (fakes)
25Dilepton Channel ee, em, mm
HT
Jet Multiplicity
13 events on 2.4 0.7 background 1 ee, 3 ??, 9 e?
ttbar signal bin
26Dilepton Channel lepton track
- Signature looser requirements for second lepton
- 1 lepton1 isolated track opposite sign
- pTgt20 GeV
- h lt 1
- Et gt 25 GeV
- 2 central jets (h lt 2, ETgt20)
- Sensitive also to t lepton final states (20
from t)
Jet Multiplicity
19 events on 7.1 1.2 background 11 e-track, 8
?-track
ttbar signal bin
27Dilepton background overview
- Instrumental backgrounds
- Drell-Yan (ee, mm)
- False ET from mismeasured leptons, jets
- Fake leptons
- Wjets with jet misidentified as lepton
- Use data whenever possible
- Physics backgrounds
- Diboson (WW/WZ/ZZ) and Z ? tt
- Real leptons, ET, jets
- Evaluate using MC
- Determine bkgd in 0j, 1j bins to give confidence
in signal bin prediction
28Dilepton DY (ee, mm) background
- Large cross section but no intrinsic ET
- False Et
- Detector coverage isnt 4p
- Reconstruction isnt perfect
- Tails of ET resolution critical
- Simulation doesnt accurately model this
- Extra cuts in Z window
- Estimate residual contamination
- Use loose Z data to normalize (subtract expected
non-Zs) - Use MC to distribute inside -outside Z window,
across jet bins
ET gt 25
29Dilepton Fake lepton backgrounds
- Determine lepton fake rate from jets in j50
sample - Cross-check fake rates in other samples with jets
- Apply fake rates to jets in Wjets data sample
Jet20
observed (predicted) fakes
LTRK Obsv Pred
j20 74 70 /- 14
j100 384 304 /- 77
inc lep 231 189 /- 37
DIL Obsv Pred
j20 51 49 /- 6
j70 75 65 /- 9
j100 69 114 /- 31
Jet70
ET of observed (predicted) fake tracks in green
(black)
30DileptonZ -gt tt, diboson bkgds
- In both cases
- Real missing energy
- Jets from decays or initial/final state radiation
- Estimates derived using PYTHIA, ALPGENHERWIG MC,
normalized to theoretical xsecs - Correct for underestimation of extra jets in MC
- Determine jet bin reweighting factors for Z ? tt
from Z?ee, mm data - Reweight WW similarly
31Why Measure the ttbar Cross Section?
- Basic engineering number, absolute measurement (?
very difficult!) , starting point for all top
physics. - Requires detailed understanding of backgrounds
and selection efficiencies. - Test of SM
- Latest calculations NNLO NNNLL
- Departures from QCD prediction could indicate
nonstandard production mechanisms, i.e.
production through decays of SUSY states.
32Dilepton tt acceptance
- Determine from PYTHIA MC (mt175 GeV)
- Apply trigger efficiencies, lepton ID MC
correction factors, luminosity weights for
different detector categories - DIL (0.62 /- 0.09)
- LTRK (0.88 /- 0.14)
33DileptonLepton ID, track efficiencies
- Lepton efficiencies
- Determined using second leg of Zs (see Sidotis
talk) - Get efficiency for data and MC, estimate
difference and derive scale factor (0.95 is
typical) - Lepton track efficiency
- Use W sample selected w/o tracking requirement
- Compare track efficiency with efficiency for W
track in top Monte Carlo simulation
34DileptonSignal acceptance systematics (I)
Must take into account many effects potentially
contributing to the acceptance uncertainty
Systematic LTRK() DIL()
Lepton ID efficiency - Variation of data/MC SF with isolation 5 5
Track-lepton efficiency - iso efficiency difference btw W2j data and ttbar MC 6 -
Jet Energy Scale - vary jet corrections by 1s, for evts w/2 jets 6 5
35DileptonSignal acceptance systematics (II)
Signal Systematics Continued LTRK() DIL()
Initial- and Final-state radiation - ISR difference from no-ISR sample - FSR parton-matching method, different PYTHIA tune 7 2
Parton Distribution Functions (PDFs) - default CTEQ5L vs MRST PDFs, different as samples 6 6
Monte Carlo Generators - compare acceptance of PYTHIA to HERWIG 5 6
36Dilepton Systematic Uncertainty on Background
Estimate
Systematic LTRK() DIL()
Lepton (track) efficiency same as signal 5 (6) 5(-)
Jet Energy Scale same procedure on bkgnd acc. 10 18-29
WW, WZ, ZZ estimate - Compare WW0p(njet scaling) to WW2p 20 20
Drell-Yan Estimate - Absolute scale (data driven), Monte Carlo shape 30 51
Fake Estimate - J20, J50, J70, J100 x-check - DIL shape of HT, effect of MET cut for fake rates 12 41
37DileptonEvents expected, observed
Compare number of events observed with total
expected from tt ( s 6.7 pb) and backgrounds
LTRK
DIL
Good agreement in 0j, 1j bins (all bkgd)
s(tt) measured here
38DileptonCross section results
DIL
LTRK
SM s 6.7 pb (mt 175 GeV/c2)
39DileptonCross checks
- Identical analysis techniques reproduce the
expected W and Z cross sections - Number of like-sign events (fakes-dominated
sample) observed agrees with prediction in all
jet bins - Number of events containing b-quark evidence
(DIL 7, LTRK 10) consistent with that expected
from top - Cross section stable over a range of jet, lepton
ET thresholds - Measure cross section with tight-tight subset
of LTRK
Good agreement with DIL, LTRK
40Dilepton Combining the cross sections
- The two analyses are complementary
- Combining them reduces the largest uncertainty
(statistics) - Strategy divide signal, bkgd into three disjoint
regions
41DileptonCombination technique (I)
- Need to divide into 3 subgroups
- Acceptance
- Background
- MC used to determine acceptance overlap
- event by event weighting
- Distribute bkgd according to expected S/B in 3
regions, vary for systematic - We are NOT averaging two analyses, we are
combining three distinct analyses
42DileptonCombination technique (II)
- For three regions, minimize combined c2 or
maximize combined likelihood (i.e. lower green
bar until 68.28 is included in area ) -gt
identical results - Be conservative with systematics between regions
- Treat as 100 correlated, distribute to give
largest total systematic
12 reduction in stat error w/r/t LTRK
43Dilepton Kinematics lepton track
RunI had seen hints of discrepancy in kinematic
distributions
HtScalar summed ET of jets, leptons, and missing
ET
Missing ET
Leptons transverse momentum
With higher statistics in Run II Data follow SM
expected distribution of top bkgd
44DileptonFlavor distribution
Every lepton in DIL is an electron or muon
channel Expected (scaled to 13 total obsvd) Observed
ee 3.3/- 0.5 1
mm 2.8/- 0.5 3
em 6.8/- 0.8 9
? Flavor distribution is consistent with
expectation
45Dilepton event display
- 2 electrons (ET173 GeV, PT263 GeV)
- Missing ET 59 GeV
- 2 central jets 1 forward jet
jet
jet
e1
n
e2
e1
e2
e1
e1
jet
jet
46tt Dilepton decays with thad
- We look for anomalously large or small t lepton
rate in top decay - One W decays into t , the other into m or e
- t decay mode semileptonic, 1 charged hadron
(50) or 3 charged hadrons (15) - Signature high PT e or m, large ET, ? 2 jets
and a object passing t identification cuts - PTt gt 15 GeV/c
- Isolated S PT lt 1 GeV/c
- electron and muon removal
- HTgt205 GeV is requested to reduce the
background. - Use W-gt t n to understand t ID
47tt Dilepton decays with thad
Results
Sample Ele thad Muon thad
Bckground 0.77 0.12 0.13 0.53 0.08 0.08
ttbar 0.59 0.05 0.10 0.47 0.04 0.07
Obs. Data 2 0
In 193 pb -1 expect 1.03 signal events.
Background dominated by jets faking ts.
Test for anomalous t contribution in ttbar decays
wrt SM. rt 1 if SM is correct
48Single lepton channel (Lepton jets)
- Signature
- One high pT Isolated lepton
- Veto Z, cosmic, conversion, dilepton
- ET gt 20 GeV
- 3 or more jets with ETgt15 GeV h lt 2.0
- S/B 1/6
W 1 jet W 2 jets W 3 jets Wgt4 jets
Secondary vertex sample (162 pb -1) 15314 2448 387 107
Soft lepton sample (126 pb-1) 11780 1867 295 84
49Single lepton channel counting events
- Require 1 b-tagged jets to reduce background
- Motivation of using b-tags
- Reduce the backgrounds, especially W light
flavor jets events while keeping ttbar signal
efficiently - B tag improves S/B from 1/6 ?3/1
- Count ttbar candidate events
- Predict rates for SM non-top processes in
tagged Wjets, excess in 3 jets is top - Use data as much as possible to determine
background contamination (non-W QCD, fake tags) - Use MC when necessary (diboson, Wheavy flavor)
50Lepton jets Channel HT cut
- HT is the scalar sum of ET of jets, leptons, and
ET - HT is a powerful discriminator for Top signal
- HT gt 200 GeV keep 96 of signal and reject 38 of
background
An HT cut increases the systematics due to Top
mass dependence, Energy scale and Heavy flavor
fraction, but still small compared with other
syst (SF, lum)
51Tagging Tools Vertexing and Soft Muons
B hadrons in top signal events
are long-lived and massive
may decay semileptonically
Identify low-pt muon from decay
Vertex of displaced tracks
55 0.5
Top Event Tag Efficiency False Tag Rate (QCD jets)
15 3.6
52Double b-tagged LepTrk event at CDF
53Vertex Tagging Algorithm
- Take advantage of the long lifetime of B hadrons
t(b) ? 1 ps (ct ? 450 mm) ? B hadrons travel
Lxy ? 3mm before decay - Select good quality tracks with large impact
parameter. - Try to reconstruct a vertex with ? 2 traks
- first pass searches for at least 3 tracks with
loose kin and Pt gt2 GeV/c - second pass looks for 2 tracks vertices with
tighter cuts on tracks quality and Ptgt 3GeV/c - A jet containing a vertex is considered b-tagged
if has large (positive) decay length
significance Lxy/sLxy gt 3 (typically sLxy ?
150 mm)
54Vertex b-Tagging Efficiency
- Use events with back-to-back jets, non-isolated
electron, require away jet to be tagged (enriched
in heavy flavor) - The efficiency of b-tagging determined by the
ratio of the number of double tagged to single
tagged events, in data and MC - Define a Scale Factor between data and MC tagging
efficiency (usually eDATA lt e MC) SF 0.81 ?
0.07 - SF lt 1 due to of good vertex tracks higher in
MC - Check in generic jets the ET dependence of the
b-tagging efficiency
- Efficiency for tagging at least one jet in a
ttbar event (Lgt3 jets, including data-MC
scaling) - e 53 ? 4
55Vertex Tagging Background
- Most from Wheavy flavor and Wmistags
- In generic jets heavy flavor pairs are produced
by both direct production and gluon splitting. In
Wjets by gluon splitting only. - The fake background is measured with inclusive
jet data using negative decay-length tags - Assume that positive mistag rates are well
described by negative (Lxy lt0) tag rates - The fraction of Wbb, Wcc events is determined
from MC and scaled to the observed number of W
events in each jet multiplicity bin - The (smaller) QCD (non-W) background is evaluated
from data lepton isolation vs Missing ET method
(standard), see Sidotis talk
56Lepton jets with 1 vertex tag HT
HT distribution of W ? 3 jets with ? 1 b vertex
tag with expected ttbar and background
contributions
572nd Part
58Outline
- Motivations for studying top
- A brief hystory
- Top production and decay
- The Tevatron CDF
- Detector issues
- Identification of final states
- Cross section measurement
- Single top production
- Mass determination
- Study of top properties
- Whats next?
Yesterday Introductory Part
Dilepton channel
Single Lepton channel
59t-tbar Final States
- Dilepton (ee, µµ, eµ)
- BR 5
- 2 high-PT leptons 2 b-jets large missing-ET
- Lepton (e or µ) jets
- BR 30
- single lepton 4 jets (2 from bs) missing-ET
- All-hadronic
- BR 44
- six jets, no missing-ET
- thad X
- BR 23
Most favorable channels for top physics
More challenging backgrounds, but measurements
still possible
60Important Background issue fakes
- Dilepton channel
- Fake electron prob lt 0.1 per jet
- A jet with track multiplicity 1 and a energy
deposition consistent with that of an electron
(I.e. hadrons that shower early in the detector) - Low Hadronic energy
- In the tail of the jet fragmentation function
- Fake muon
- A track part of a jet, being mismesured in h,
although has the same f of the jet can appear to
be isolated because of the error in h - V ery rare, but can happen
61You cannot trust MC to predict detector effects
involving distribution tails
Is our detector/simulation particularly bad?
No! You always have at some level this kind of
effects
We are looking for very rare events
We are afflicted by very rare backgrounds
What can we do to avoid this?
Optimize our cuts to reduce the bckgnd
contamination, optimize with available data the
simulation, but in the end well have to wait
for data in order to study all the instrumental
effects
62L ? 3jets (? 1 b-tag)
?
- Fake tags
- The probability to have a negative Lxy is
obtained from generic jet data ( 0.5 per jet) - it is parametrized as a function of the jet ET,
jet silicon track multiplicity, h - it is used to estimate the mistag rate
- Negative- non-physical tags positive mistags
- This probability matrix is then applied to Wjets
events to obtain the background estimate in our
sample
63Lepton jets with ? 1 vertex tag results
Vertex tag
Mis tag from Generic jet data
Method II based on MC
MET vs Isolation method data driven
top signal region
Theoretical value
SVX 57 events on 23.4 3.0 bkgnd, before HT
cut
Number of jets per event
64Summary of secondary vertex counting
65Summary of secondary vertex counting
Require HT gt 200 GeV for 3, ? 4 jets bins only
W 1 jet W 2 jets W 3 jets W ? 4 jets
Events before tagging 15414 2448 179 91
Total background 141.8 ? 18.9 66.0 ? 8.9 13.8 ? 2.0 13.8 ? 2.0
Observed positive tags 160 73 21 27
66Tagged Jets Properties
The tagged events contain b quarks, as seen by
the decay length Lxy distribution
67Tagged Jets Properties
The tagged jets are kinematically consistent with
the expectation from top, as seen by their ET
distribution
68Additional cross check HT in 1 and 2 jet bins
Expected ttbar and background contributions are
shown
69Additional cross check Z jets
Results from the secondary vertex -tagged Zjets
selection (no HT requirement). Background is
estimated from the positive tag rate of generic
jet sample applied to the Zjets pre-tag sample
70Ljets cross section counting
- High pt lepton plus missing ET
- N(jet) 3
- 1 jet SEC(ondary) VTX
- HTgt200 GeV
48 events
13.8 2.0 events
A ttbar acceptance b-jet eff. for ttbar
3.80 0.03(stat.) 0.45(sys)
5.6
1.2 1.0 - 1.1 - 0.7
pb
(stat.)
(sys.)
71Ljets with ? 2 vertex tags
Low statistics but low background 8 events
observed, 0.99 0.26 bkgd expected
b-tagging efficiency and its uncertainty enters
twice here
top signal region
2 jet bin 8 events observed, 2.360.64 bkg.
1.3 ttbar
Number of jets per event
72Summary of double tag counting
73Soft Lepton Tagging
- Leptons from semi-leptonic decays of B have a
softer pT spectrum than W/Z leptons and are less
isolated - Soft Muon Tagger based on a Global ?2 to
identify low-pt muons - Global ?2 combines information from muon
matching variables - There is no calorimetry or isolation
requirements - b-tagging with SLTm
- Track with dR slt-jetlt0.6,
dZ
slt-Zvtxlt5cm pTgt3 GeV/c
are considered by
the SLTm tagger - An event is tagged if at least
one SLTm tag is
found
74Lepton jets Soft Lepton Tag counting
Alternative way to secondary vertex b-tag (b???X,
b?c ? ??X)
SLT 18 events on 12.7 2.5 bkgnd
Predicted by fake rate matrix made by photon
jets sample.
MET vs isolation method
Predicted by data in/out of Z-mass window.
From MC and theory
top signal region
Number of jets per event
75Lepton Jets Soft Lepton Tag Results
76Soft Lepton Tag Properties
Comparison of the SLTm pT distribution in W ? 3
jets for SLtagged events and for expectations
from background and ttbar (scaled to the measured
cross section of 4.1 pb).
Jet ET distribution in W ? 3 jets for SLtagged
events and expectations from background and ttbar
(scaled to the measured cross section of 4.1 pb).
77Top cross section summary
vs-Dependence
- Main data driven systematics (jet energy
scale, ISR, ebtag) scale with 1/?N - RunII(2fb-1) dstt/stt lt10
78Lepton jets event display
Vertex display of double tagged event with an
electron plus 3 jets
Lego plot
79All Hadronic channel
- Signature
- 6 or more jets
- kinematical selection
- ? 1 vertex tag.
- Background estimate based on data.
Observe 62 tags excess over 264 background in
signal region.
80Top Mass Measurement
Mtop is a precision electroweak parameter that
helps constrain the mass of the Higgs.
81Top Mass Measurement Challenges
Many combinations of leptons and jets which one
is correct?
- Choose assignment kinematically most consistent
with top - Use all combinations, but weight them
- Link observables to parton-level energies
- go through generation-physics effects-detector
effects - Largest uncertainties come from this difficult
relation
Final likelihood fit methods
- Derive mass templates from top MC and fit data to
most likely template - Add event kinematic information, possibly
including matrix element
82Jet Energy Corrections
- Jets may be mismeasured due to a variety of
effects - Calorimeter nonlinearities
- Curvature of low-momentum charged particles by
the magnetic field - Reduced cal. response at boundaries between
modules and cal. subsystems - Contributions from underlying event
- Out-of-cone losses (due to fragmentation or final
state radiation) - Undetected energy carried by muons or neutrinos
- The correction factor depends on jet ET and h and
is meant to reproduce the average jet ET
correctly, (not to reduce the jet fluctuations
around this mean) - A set of corrections was developed for generic
jets. - Absolute corrections (gamma-jet balancing)
- Relative corrections (central-forward
calorimeters, dijet balancing) - A set of top-specific corrections is also applied
(to account, for instance, for the presence of
neutrinos in b-jets)
83Lepton ? 4 jets with sec. vertex b-tag
template method
c2 mass fitter?Finds top mass that fits event
best? All event info into one number ? 12
parton/jet matching assignments possible, 2
longit. neutrino possible, use b-tag to reduce
permutations ? test for consistency with
top using kinematic constraints
? choose combination with lowest ?2
Massfitter
Templates
Likelihoodfit
Result
Likelihood fit
? fit resulting mass distribution to MC
background top signal templates at different
values of Mtop
? Best signal bkgd templates to fit
data? Constraint on background normalization
84Top mass Lepton 4 jets with SVX tag
28 vertex-tagged events
85Signal Templates
140
150
160
170
190
180
200
210
220
ParametrizationBuild signal probability d.f. as
a function of generated mass.
86Top Mass Uncertainties, L 4jets
CDF Run II Preliminary
Source Uncertainty (GeV/c2)
Statistical 7.1-7.7
Jet energy scale 6.3
Final State Radiation 0.9
Parton Dis. Functions 0.2
Initial State Rad. 0.4
Other MC modeling 0.7
Generator 0.4
Backgrounds 0.8
b-tagging 0.1
Total systematic 6.5
Dominated by calorimeter energy scale
We estimate all systematics using large
ensembles of pseudoexperiments in Monte Carlo.
87Uncertainty due to Jet Energy Scale
CDF Run II Preliminary
Source Uncertainty (GeV/c2)
Relative to central (h response) 3.0
Central Calorimeter Response (g-jet) 4.6
Abs. Scale (non-linear.,fragm,underl) 2.2
Corrections to partons (out-of-cone) 2.3
Total 6.3
88Top Mass dilepton channel
- 6 dilepton cand. in 125.8 pb-1
- system is underconstrained
- scan over ? directions, weight
- by pT of ttbar
- For each lepton-b pair assignment
- Calculate best raw mass, from most probable
combination - Likelihood fit to signal and background templates
Likelihood fit result
89Top Mass Uncertainties, dilepton
Source Uncertainty (GeV/c2)
Jet Energy Scale 7.1
GeneratorsPDFISRFSR 3.5
Background 1.3
Total 7.9
Dominated by calorimeter energy scale
90How to improve JES
- Most improvements of jet energy scale using W ?
jj in tt and Z?bb
Run 1 Results
The best sample to measure the top mass may
become the double b-tagged lepton jets sample
or the tagged dilepton sample ( S/B 10)
M(W) ? jj
M(Z)?bb
? Some improvement from the sample where a jet
recoils against a reconstructed Z (now
we use jet photon)
91Other methods to measure Mtop Dynamical
likelihood
- Use the leading order ttbar matrix element,
convoluted with transfer functions, to make an
event by event likelihood distribution as a
function of top mass - Combines all combinations and events according to
their power to distinguish the top mass - Transfer functions given a jet, return the
probability that it came from partons of various
energies - Models the shape of the response curve, not just
the mean - Mapping function account for background and for
the dependence of the transfer functions on top
mass - Similar to the recent ljets analysis from D0,
but a CDF original measurement - Proposed in 1988 by K. Kondo (J. Phys. Soc. 56,
4126)
b jets
w jets
Transfer functions for various ET bins
92Dynamical Likelihood Method Results
Official current CDF Run 2 top mass value Best
precision, Systematics dominated!!!
Mtop 177.8 4.5 -5.0 (stat.) 6.2(syst.)
GeV/c2
93Other methods to measure MtopMultivariate
Templates
- Reduce jet systematics (while increasing stat.
error) by calibrating jet energy scale event by
event with W mass - Improve signal/background separation by using
other kinematic variables (sum of four leading
jet PTs) in addition to reconstructed top mass - Estimate the probability to pick correct
combination event-by-event and reweight events. - Use nonparametric techniques (kernel density
estimation) to make multivariate templates - Fit background fraction in data
94Multivariate Templates Method Results
Fitted background fraction 34 14
Mtop 179.6 6.4 -6.3 (stat.) 6.8(syst.)
GeV/c2
95Single top physics (EW Production)
The dominant single-top EW production processes
are
s-channelW process s(W?tbX) 0.88 pb
- Cross section smaller than top pair production,
but - can provide a direct, independent
measurement of the Wtb vertex (s a Vtb2) - sensitive to new physics
- t-channelanomalous couplings, FCNC
- s-channel new charged gauge
- bosons
t-channel W-gluon fusion process s(Wg?tbX)
1.98 pb
96Single top search
- Single Top Signature
- High-pT electron or muon
- Missing transverse energy ET
- 2 jets
- t-channel
- 1 b-jet 1 light-quark jet 1 soft b-jet (from
gluon splitting) which is rarely seen - s-channel 2 b-jets
- Final state is W 2 jets
- Strategy
- isolate W exactly 2 jets and tag one jet
- likelihood Fit to Qh (t-channel)
- Q charge of lepton, h pseudorapidity of
forward jet - likelihood Fit to Ht (combined)
97Single top search
Fit to the data (combined search)
t-channel search
Using 162 pb-1 of data
st(t-channel)lt8.5pb _at_95 C.L. (th. 2 pb)
st(combined)lt13.7pb _at_95C.L.(th. 2.9 pb)
98Measurement of Top Quark Production and Decay
Properties
- Once a top signal has been re-established the
natural path will be to measure its properties in
order to confirm SM or find deviations from it - Is top quark adequately described by the
Standard Model?
99Test for new physics in tt production Search for
X -gt ttbar
- Studies of the ttbar invariant mass spectrum
provide a general search for heavy objects
decaying to top pairs - Are dynamical ttbar condensates in models such
as Topcolor responsible for EWSB? We can look for
the predicted resonances in the Mttbar spectrum.
Topcolor is needed in technicolor theories to
explain the large t b mass difference. - The particle manifestation of Topcolor are
- the topgluon, gT a vector boson that
preferentially couples to the third generation - the topcolor Z a neutral gauge boson resulting
from the additional U(1) symmetry needed to keep
the large mass split
Hill, Phis Lett B 345, 483 (1995) Hill,Parke,
Phys. Rev. D 49, 4454 (1994) Harris et al,
FNAL-FN-687 (1999)
100Test for new physics in tt production
ttbar invariant mass spectrum can be used to set
limits on X -gt ttbar
Model independent search for a narrow resonance
X?ttbar excludes a narrow, leptophobic X boson
with mX lt 560 GeV/c2 (CDF)
N.B. LHC will be relatively insensitive to new
colour singlet gauge bosons (such as Z) because
gluon fusion will be the dominant production
mechanism
101W Helicity Measurement in top decay
- Top decays before it can hadronize, because width
Gt 1.4 GeV gt ?QCD. - Decay products preserve information about the
underlying Lagrangian. - Unique opportunity to study the weak interactions
of a bare quark, with a mass at the natural
electroweak scale! - SM Prediction
- W helicity in top decays is fixed by Mtop, MW,
and V-A structure of the tWb vertex. - W helicity reflected in kinematics W decay
products
102W Helicity Measurement, contd.
The angular dependence of the semileptonic decay
in the W rest frame is given by
right
left
long.
SM (V-A) predictions (for mb0)
where ? MW2/Mtop2
VA 70 long., 30 r.h.
103Helicity Measurement dilepton channel
Helicity affects lepton PT in lab frame
F0 lt 0.52 _at_ 95 CL
104Helicity measurement L jets channel
F0 0.880.12-0.47 (stat. syst.) F0 gt 0.24 _at_
95 CL
105Helicity measurement combined
Combined Ljets and dilepton channels F0
0.270.35-0.21 (stat. syst.) F0 lt 0.88 _at_ 95 CL
106Top Branching fractions I
- Dilepton and lepton jets cross-section values
assume t?Wb always, never t?Xb. - Ratio of measured ss (R??ll/?l) tests that
assumption - Should be unity
- Ratios are good
- Many systematic uncertainties cancel
- Independent of theory value of s (PDFs, mt)
- Sensitivity to non-SM decays of top, e.g. t ?
Hb, different tan ? gives different mix of ll/lj
107? ratio, non-SM limits
- Create probability distribution for R?
- R? 1.450.83-0.55,
- 0.46ltR?lt4.45 _at_ 95CL
- Set limits on non-SM BRs of top assuming
simple model detect non-SM decays with same
efficiency as SM - BR lt 0.46 for additional t?Xb all-hadronic decay
Preliminary
w/o cancellation
w/cancelllation
108 Top Branching fractions II
- Cross section measurements assume t?Wb always,
never t?Wq. - b-tag rates in tt events test that assumption,
depend on B(t?Wb) b, tag efficiency ? - Measure from the ratio of tag rates the product
b? assume ? and measure b, or vice-versa.
109Measuring b, ?
- Find most likely b ? in SVX-tagged sample
- b ? 0.250.22-0.18,
- ? b 0.540.49-0.39,
- ? consistent with measurements in calibration
samples - b gt 0.12 _at_ 95 CL
110Search for rare decays
- Look for FCNC decays t ? gq and t ? Zq (qc,u)
(BR ? 10-10 10-12) - Any sign of such decays would indicate new
physics - Run 1 Result assume that one of the 2 top quarks
decays according to SM t -gt Wb - t ? gq look for leptong2jetsEt and g4jets
1 event seen, Wjets background ?
1 - BR (t ? gq ) lt 3.2 (95CL)
- t ? Zq look for ll- 4jets and 3 leptons
2jets - 1 event seen, expected background ? 1
- BR (t ? Zq) lt 33 (95CL)
111Summary of top measurements and future
expectations
Top quark Property Current best measurement Precision Precision Precision
Top quark Property Current best measurement Run 1 Exp. Run 2 LHC
Mass (CDFD0) 178.0 4.3 (run 1) 2.4 1.2 1
stt 6.5 1.7-1.4 25 10 5
W helicity 0.91 0.37 0.13 0.4 0.09 0.01
Vtb (run 1) 0.960.16-0.12 (3gen) gt0.78 at 95 CL 15 2.5 1.5
s(single top) Vtb lt 13.7 pb at 95 CL - 26 14 5 2
BR(t-gtgq) BR(t-gtZq) lt0.03 lt0.30 0.03 0.30 2x10-3 0.02 2x10-5 2x10-4
112Conclusions and Outlook
- The top quark is back!
- First Run II measurements of cross section mass
are available and will improve rapidly. - Other analyses (W helicity, single top) are
making excellent progress. - It is the start of a program of precision top
physicsand hopefully top surprisesat the
Tevatron. - We still expect 50x more data compared to Run I
in hand before LHC starts!
DØ / CDF Run 2a Goal
- Run II with first 2 fb-1 will provide a
chance to - Measure Mw to lt 40 MeV/c2
- Measure Mtop to lt 3 GeV/c2
113The Road Ahead
- Search for top ? H
- Precise measurement of Mtop
- Single top production, measure Vtb
- ttbar resonant production, strong EWSB
- Searches for rare decays
- Is top the connection to new physics?
- you are invited to join CDF to experience the
real data before LHC turns on!