Precision Measurement of the Top Quark Mass at CDF

1
Precision Measurement of the Top Quark Mass at
CDF
Un-ki Yang University of Chicago
HEP Seminar at Argonne Lab, October 12, 2005
2
Our World View

• Standard Model very successful in describing
  matter to very high precision
• 3 leptons and quarks families
• Bosons carry fundamental interaction
  (electromagnetic, weak and strong )
• Building blocks almost complete!!!
• Top quark discovery in 1995
• One missing Higgs boson, responsible for particle
  mass
• Surprise in the neutrino sector
• (non-zero neutrino mass)

Higgs
Standard Model
(gravity not included)
3
Top quark Physics

• The Discovery of the top quark in 1995 was
• no big surprise. What was surprising is that
• its huge mass (0.35M times larger than
• electron mass)
• Why is top so heavy?
• (really SM-like particle?)
• Is top involved with electroweak symmetry
  breaking?
• Is top connected to new physics ?

• SM can only say top is just heavy, even with
  all its great success in the past three decades.
• For experimentalist, top quark is a good clue to
  starts with!!!

4
Why is Top Mass so special?

• Top mass is a fundamental SM parameter, the
  largest of all known particles
• Large coupling to the Higgs
• Large radiative corrections

Run I
DMh a Mt2

• Constrain Higgs mass together with W mass, and
  look
• for a deviation from the other EWK results?
• Constrain new physics (SUSY) with precise
  knowledge of top and Higgs masses.

5
For High Energy Frontiertools at Fermilab
Record 1.5 ? 1032

• The Tevatron collider produces millions of events
  per second.
• The place to do top physics (3k evts for 2fb-1)
  untill LHC turns on
• ( LHC is a top factory, 1Hz)

6
CDF at Tevatron

• Multi-purpose detector precision
• meas. search for new physics

• Silicon detector (SVX)
• top event b-tag 55
• COT drift chamber
• Coverage hlt1
• sPt / Pt 0.15 PT
• Calorimeters
• Central, wall, plug
• Coverage hlt3.6
• EM sE / E 14 /ÖE
• HAD sE / E 80 /ÖE
• Muon scintillatorchamber
• muon ID up-to h1.5

7
CDF Data Taking

• Delivered luminosity gt 1.3 fb-1
• Recorded luminosity gt 1.0 fb-1
• Data taking efficiency 80-90

Analyses for todays talk based on 318 pb-1
(Physics quality data until Sep. 2004)
8
Production of the Top quark at the Tevatron
Top quarks are primarily produced in pairs
(s7pb), via qq (85) and gg(15), LHC via gg
(90) ttop 4 x 10 -25 s (due to large mass) Top
decays as free quark (L-1 (200MeV) -1 10 -23
s)

• Dilepton (5, small background)
• 2 high-PT leptons(e/m), 2 b jets,
  large missing ET
• LeptonJet (30, manageable bkgd)
• 1 high-PT lepton(e/m), 4 jets (2 b jets),
  large missing ET(30)
• All-hadronic (44, large background)
• 6 jets (44)

9
Summary of Run I Measurements

• Mtop meas in Run I avg(100pb-1)
• Higgs mass fit
• Run II goals (based on RunI)
• .
• Consistency among diff. channels (same top?
  non-SM decays like t-gtbH)
• Consistency with Xsection (non-SM X0, t
  contributions)

10
Many challenges

• Not a just calculation of the invariant mass of
  the W and b objects
• At least six different objects to figure out with
  correct assignments
• Undetected neutrino
• Quarks have showered, hadronized, then clustered
  into jets, need to go back to the quark-level
• Assignments of jets to quark? poor jet energy
  scale and resolution makes it worse.
• Even, more extra jets to figure out from
  initial/final state gluons only 50 of the time,
  leading 4 jets correspond to 4 partons (qqbb)
• Dont even mention top-like backgrounds

LeptonJets channel
Observed Final state Complicated final state
to reconstruct Mtop
Full detector knowledge (including every cracks)
on particle responses is essential jet energy
scale resolution, and b-tagging of course,
good algorithm to reconstruct Mtop
11
Event Selection

• Final State (LeptonJet channel) lepton,
  neutrino plus 4jets
• Datasets (Mar, 2002 Aug. 2004)
• 318pb-1 for High-pt central electron/muon
  triggers
• High pt electron or muon with Pt gt 20 GeV
• Isolated
• Electron EM cluster in calorimeter with
  matched track
• Muon track matched to hits in muon
  chambers,
• MIP ionizing energy in
  calorimeter
• Large missing Et gt20 GeV
• Leading 4 jets
• Reconstructed with cone algorithm (0.4) using
  calorimeter towers
• hlt2.0

12
B-tagging and Sample Division

• Signal/Background improvement
• (only 1-2 of Wjets contains heavy flavor)
• Big help to remove wrong choice in jet-parton
  assignment for Mtop
• Samples are divided to get the best sensitivity
  on top mass

Top Event Tag Efficiency 55 False Tag Rate (per
jet) 0.5
13
Jet Energy Correction
Determine true parton E from measured jet E in
a cone 0.4
Correction to central region using dijet balance
to make response uniform in ?
Correction to particle jets using dijet MC tuned
for single particle E/P, material, and
fragmentations due to non-linear and
non-compensating cal.
Out-of-Cone correction to parton top-specific
correction to light quark jets and b-jets
separately
14
Jet Energy Uncertainties
Total fractional uncertainty on jet Pt
Fractional uncertainty on the relative
Run I
Run I
Run II 2005
Run II 2005
Run II 2004
Central h region
Central region
A lot of work has been done to reduce the syst.
from jet-energy scale (a factor of two
improvement compared to last year). The new Run
II uncertainties are at the same level or better
than Run I.
15
Top Mass using the template method

• c2 mass fitter
• Finds top mass that fits event best
• One number per event
• Reject badly reconstructed event

Data
Wbb MC
Massfitter
tt MC
Signals/background templates
Datasets
Data
Likelihoodfit
Result
Likelihood fit Best
signal bkgd templates to fit datawith
constraint on background normalization
16
Mass Fitter (event by event)

• Try all jet-parton assignments with kinematic
  constraints, but assign b-tagged jets to
  b-partons
• Select the rec. mass Mt from the choice of lowest
  c2
• Badly reconstructed Mt (c2 gt 9 ) is removed

• c2 kinematic fitter

Top mass isfree parameter
Fully reco. tt system, powerful tool to
study kinematic properties!!!
All jets are allowed to be float according to
their resolutions to satisfy that
M(W)M(W-)80.4 GeV, M(t)M(t)
17
Signal Templates
Signal Templates Analytical functions (2
Gaussian gamma) of reconstructed mass, Mt
as a function of true mass, Mtop
Reconstructed Top Mass Dist. at Mtop 178 GeV
18
Backgrounds
gt1-btag

• Wheavy flavor jets(bb,cc,c)
• HF fraction from ALPGEN MC
• Normalized to data
• Wjets(mistag)
• Use measured mistag rate, applied to the data
• Multijetfake-W (jet-gte, track-gtm)
• Estimated from data
• Single top, dibson (WW,WZ)
• Estimated from MC

control
signal
19
Background templates
Mt(GeV/)
Mt(GeV)

• 0tag Wjets
• Tagged WHF, Mistag, fake-W, Single-top
• Shape mostly by ALPGEN MC, cross-check with data

20
Extracting Mtop

• Unbinned Likelihood Fit

Min vs Mout
Mtop pull width

• No bias appear. Pull width 1.03 is due to
  non-Gaussian tail of the Likelihood

21
Results on the data
Comb -Log Likelihood
22
Systematic Errors
Though we find that 70 of JES uncertainty comes
from b-jet, B-jet uncertainties mainly due to
generic jet corrections.
23
Jet Energy Scale and W?jj

• Use W -gtjj to calibrate JES (fully in situ ).
  This scale is applied to light-quark jets and
  b-jets
• Simultaneous fit to mjj mt 2-D templates mt (
  true Mtop, JES),
• mjj ( true Mtop, JES)
• With combined information on JES from W?jj and
  standard calibration, we improve top mass
  resolution
• JES from W?jj is mostly statistical, scale with
  luminosity!
• More promising at LHC

24
Mt templates (true Mtop, JES)
PDFs ( Mt JES, Mtop180)1tagT
PDFs ( Mt Mtop, Mtop180)1tagT
Mtop
Rec Mt
Rec Mt

• Mt strongly depends on JES and Mtop

25
JES and Mtop Fit

• Mtop and JES are simultaneously determined in
  likelihood fit using shape comparisons of Mt and
  , Mjj distributions, taking correlations between
  them
• Mjj sensitive to JES, but mostly independent of
  Mtop
• Mt sensitive to both JES and Mtop
• For optimal performance use combined information
  on JES from W-gtjj and standard calibration
• A Gaussian constraint on JES from the standard
  calibration is included in likelihood as a priori
  
• All other things are same as the 1-D template
  analysis

26
2-D results on the data

• Best single top mass measurement in the world

27
Various Cross-Checks

• All consistent!!!

28
Matrix Element Method
Calculate likelihood as a function of mtop
according to Matrix Element for each event. (all
combinations used)
Excellent agreement with Template method
29
Dilepton Analyses

• Unconstrained problem
• 2 neutrinos, 1 missing ET observables
• Template assume neutrino ? or PZ((tt),
  reconstruct mt template as a function of assumed
  quantity
• Matrix Element calculate event probability
  using Matrix Element (unknown quantities
  integrated) expected to be more powerful than
  leptonjet channel
• Dominated by the statistics, but less dependence
  on the JES, good for high Lum.

Matrix Element Method
30
Comparisons of the syst. Errors.
No additional syst. due to the NLO effect
(covered by ISR/FSR syst) Template vs ME in
leptonjets very similar
31
How to control ISR?

• In Run I, switch ISR on/off
• using PYTHIA, dMtop 1.3GeV
• In Run II systematic approach
• ISR/FSR effects are governed
• by DGALP evolution eq.
• ltPtgt of the DY(ll) as a function of Q2

(2Mt)2
ISR syst 0.4 GeV
log(M2)
32
Summary of Run II Measurements

• All channels consistent? Hadronic channel is
  coming
• RunII combined (dMtlt3 GeV)

33
Projections
dilepton

• Realistic to achieve dMt2 GeV _at_ 4fb-1 in
  leptonjet channel,
• combined with D0, dMt1.5 GeV _at_ 4fb-1
• Dilepton not yet systematic limited,
  dMt(stat)2.5 GeV _at_ 4fb-1
• consistency between each channel will be
  interesting too.

34
Implication for Higgs and SUSY

• A Precision EWK Fit
• Direct search(LEP) MH gt 114 GeV
• New result favors SUSY over SM, light SUSY

MSSM mHlt140 GeV Light SUSY mass close to exp.
lower limit, Heavy SUSY2 TeV scale (Heinemeyer
et al.)
35
Looking for new physics signals?using top mass
tool results

• Top Mass fitter provides Mtop and fully rec. tt
  system.
• Mtop leptonic vs hadronic side in leptonjet
  channel
• top vs anti-top
• top mass as a function of new
  physics sensitive variables
• tt kinematics non-standard model contributions
  (shown below)

W mass in hadronic system search for
t-gtbH(-gtcsbar)
tt resonance
Top mass vs s(tt)
But s(tt) based on counting appears to be higher
by 20
36
Prospects at LHC

• LHC is a place not only for discovery of new
  particles but also for detailed studies of new
  particles.
• Even if new physics signal come, there could be
  many challenges to figure it out!
• Top pair events are good objects to study in
  early stage.
• Why? top is produced at the factory level.
• Background for many new physics processes
• Excellent place to study any anomalous top
  production and irare decays
• Top mass analysis ( dMt1 GeV _at_ 10fb-1)
• A good way to establish sharp tools (detector
  calibration and algorithms)
• Provides opportunity to study all kinematic
  distributions (better understanding on tt events
  search for non-SM properties )

37
Conclusions

• Top Mass is now in the phase of precision
  measurements at the Tevatron one of the highest
  priority
• We have reached to dMtlt3 GeV (world average)
  and possible to achieve
  dMt1.5 GeV _at_ 4fb-1
• - CDF publications coming soon.
• 2 PRDs / 1 PRL in Leptonjets (paper seminars
  scheduled tomorrow!)
• 2 PRDs/ 1 PRL in Dilepton
• Possibly to over-constrain the SM (Higgs) and to
  pin down new physics (SUSY) with other EWK
  measurements.
• Top mass measurements for many different samples
  (as a function of new physics
  sensitive variables) will come.
• Once Higgs is discovered, the precise top mass
  will be very powerful in pinning down a true
  theory among many attractive theories.

38
Comparison of c2 distributions
39
Combined Likelihood
40
Pt of tt NLO vs ISR syst.
Pt(ttbar) for 1tagT2tag
41
Njets (Etgt12 GeV at HEPG particles)NLO vs ISR
syst.
42
(No Transcript)