Precision Measurements of the Top Quark Mass at the Tevatron

1
Precision Measurements of the Top Quark Mass at
the Tevatron

• Aspen Winter Conference
• New Physics at the Electroweak Scale and
• New Signals at Hadron Colliders
• Rainer Wallny
• University of California, Los Angeles
• On behalf of the
• CDF and Dø Collaborations

2
The Tevatron Accelerator

• Worlds highest energy collider (until 2007)
• Proton-antiproton Synchrotron
• Experiments CDF and DØ
• Run I (1992-1996)
• ?s 1.8 TeV
• 100 pb-1 int. luminosity
• Major upgrade to accelerator complex
• Main Injector (x5)
• Pbar Recycler (x2)
• Run II (2001-200x)
• ?s 1.96 TeV
• Current peak luminosity 25.0 x 1031 cm-2s-1
  
  
• Both experiments have now gt 1.7 fb-1 on tape.
• Aim for 4-9 fb-1 int. luminosity in Run II
• Only place in the world to produce top quarks.

3
The CDF and DØ Detectors
Both experiments New tracking systems Upgraded
electronics, trigger, DAQ
All crucial for top physics!
Excellent muon coverage New pre-shower
detectors New magnet
Excellent tracking New forward calorimeters
4
The Top Quark in the Standard Model

• The top quark was discovered only 10 years ago
• Existence is required by the Standard Model, but
  striking property its mass is surprisingly large
• Higgs was not yet seen it is required by the
  Standard Model

The Standard Model
CDF Dø Run I combined mtop 178 ? 4.3 GeV/c2
5
Precision Top Quark Measurements

• Allows for prediction of the mass of the Higgs
  boson
• Constraint on Higgs can point to physics beyond
  the standard model
• Consistency check of the Standard Model

6
Top quark production and decay

• Mainly pair produced by strong interaction
• Electroweak decay before hadronizing
• Different channels with different sensitivity and
  challenges

Br(t ?Wb) 100
7
Challenges of Top Physics

• A needle in a haystack!
• 1 top pair each 1010 inelastic collisions at
  ?s1.96 TeV
• We detect missing transverse energy ET not
  neutrinos
• Pz of neutrinos unknown
• We measure jets not quarks
• Measured energy has to be corrected back to
  parton-level
• Many possible jet-parton assignments
• B-tag reduces ambiguities, but also statistics
• Utilizes full detector capabilities
• Well-identified electron(s) or muon(s)
• Large missing transverse energy ET
• Several jets identified in calorimeters

8
Top Mass Measurement Overview

• Robust program of complementary
• measurements
• Many measurements in all the different channels
  gt consistency
• Different methods of extraction with different
  sensitivity gt confidence
• Combine all channels and all methods
  gt precision
• As I cannot cover them all, I will
• present the most recent and the
• most precise.

9
The Jet Energy Scale (JES)

• Determine the energy of the
• quarks produced in the hard
• scattering
• Correct for hadronization, calorimeter
  non-linearity and non-compensation,
  multiple-interactions, underlying event,
  algorithm effects .
• Derived from Data and MC
• Jet energy scale uncertainties known to 3 for
  Mtop range of jet energies
• gt Largest uncertainty in Mtop

10
LeptonJets Channel
Final State from Leading Order Diagram
What we measure

• Branching fraction 30 (lepton e or ?)
• Final state 2 light quark jets, 2 b quarks, 1
  lepton, 1 neutrino
• SB 14 to 111 depending on the b-tagging
  requirement
• Jet-parton assignment ambiguity 12 (0 b tag), 6
  (1 b tag), and 2 (2 b tags)
• Can infer neutrino energy from ET
• Provides an in-situ calibration candle W?qq to
  determine JES
• Most precise Mtop measurements

11
Extraction Techniques

• Need to extract Mtop from imprecise measurements
  (jets) and non-measured (neutrino) quantities
• Template analyses
• Evaluate a variable strongly correlated with Mtop
  
• Obtain Mtop comparing data to Monte Carlo with
  different Mtop input
• Matrix Element analyses
• Evaluate ttbar and background probability
  densities as a function of Mtop
• Obtain Mtop multiplying event probability
  densities

12
Matrix Element Method

• Most precise measurements by CDF and DØ use the
  Matrix Element method in the leptonsjets channel
  with in-situ determination of the jet energy
  scale
• Define event likelihood using signal Pttbar and
  background PWjets probability density
• Use maximum likelihood to fit simultaneously
  Mtop, JES, and signal fraction, Cs

13
LeptonJets Matrix Element Results
1 fb-1

• CDF has used 940 pb-1 and measured with 166
  candidates with at least one b-tagged jet
• DØ has used 380 pb-1 and measure with 175
  candidates with and without b-tagging
  requirement
• Update from DØ coming soon.

1.5
Worlds most precise measurement!
2.6
14
Dilepton Channel
Final State from Leading Order Diagram
What we measure

• Branching fraction 5 (lepton e or ?)
• Final state 2 leptons, 2 b quarks, 2 neutrinos
• SB 21 and 201 requiring ?1 identified b tag
  
• Combinatorial ambiguity 2 b-jet assignments
• under-constrained kinematics due to 2 final
  state neutrinos
• gt complicated to solve for Mtop

15
Neutrino Weighting Method (DØ )

• Solve kinematical ambiguityby assuming a top
  mass andassign neutrino momentaconsistent with
  measured event
• Up to 4 solutions
• Assign each solution a weight(using ET or El and
  knowledge from Matrix Element or MC)
• DØ result (370 pb-1)
• New preliminary results (835 pb-1, e? channel
  only)

Mtop178.1 6.7(stat.) 4.3(JES)
2.1(syst.) GeV/c2
4.6
Mtop171.6 7.9(stat.)5.1-4.0(syst.) GeV/c2
16
Dilepton Matrix Element Method (CDF)
1 fb-1

• Probability density calculated for ttbar and 3 of
  the major backgrounds
• Using 1030 pb-1 and 78 candidates CDF measures
• Cross-check result requiring b-tagging

3.3
Mtop (GeV/c2)
17
All Hadronic Channel
Final State from Leading Order Diagram
What we measure

• Branching fraction 44
• Huge amount of background SB 18 after
  requiring at least 1 b-tag jet
• Combinatorial ambiguity 90 combinations
• Backgrounds mainly from multi-jet QCD production
• Use W?qq to dermine JES

18
All-Hadronic Top Mass Measurement (CDF)
Single Tags (red model black data)

• Event Selection
• Select exactly 6 jets with ?lt2 ETgt15 GeV
• Cuts on jet shapes (aplanarity and centrality)
• ? ET gt280 GeV/c2
• Require ?1 b-tags
• Impose ttbar-ME Likelihood cut
• Use data driven background model
• Use pre-tag data without ME likelihood cut (S/B
  1/25)
• Randomize b-tags from mistag expectations
• Validate background modelin various side bands
  ALPGEN MC

ET
?
vertices
?
SumEt
SumEt3
(ALPGEN)
Signal region
Centrality
Aplanarity
19
All-Hadronic Top Mass Measurement (CDF) (contd)

• 2D Template analysis (Mtop and JES)
• Signal templates derived froma matrix element
  calculation
• Background templates from datadriven background
  model
• Mtop171.1 3.7(stat.JES) 1.9(syst.)GeV/c2
• JES 0.5 /- 0.9
• Supersedes previous 1D template result on 1 fb-1
  (ICHEP06)
• Mtop174.0 2.2(stat.) 4.8(syst.)GeV/c2
• 2D All-hadronic channel measurement now becomes
  competitive to leptonjets measurements

2.4
1 fb-1
20
Tevatron Combination

• Excellent results in each channel
• Combine them to improve precision
• Include Run-I results
• Account for correlations
• Use Best Linear Unbiased Estimator (NIM A270 110,
  A500 391)
• We reached a precision of of 1.2 in Mtop
• Note Latest all-hadronic CDF result not yet
  included!

1.2
Mtop171.4 1.2(stat.) 1.4(JES)
1.0(syst.)GeV/c2
21
Comparison

• Are the channels consistent?
• We compare them taking into account their
  correlated systematic uncertainties
• Determination of Mtop from the 3 different
  channels is consistent with one another
• () not including latest CDF 2D all-hadronic
  result

Mtop(All Jets) () 173.4 4.3 GeV/c2
Mtop(Dilepton) 167.0 4.3 GeV/c2
Mtop(LeptonJets) 171.3 2.2 GeV/c2
22
The Future of Mtop Measurements
All-hadronic not included

• New results are better than our predictions 6
  months ago
• Add JES to all-hadronic channel makes sensitivity
  comparable to leptonjets
• DØ has similar sensitivity (new results with 1
  fb-1 coming soon)
• We expect to achieve an uncertainty of lt1 in the
  next years

23
Conclusions and Outlook

• New more precise measurements in every channel
  from CDF and DØ
• All-hadronic channel becomes competitive
• New world average
• Present uncertainties on Mtop
• (and MW, new results to be presented at this
  conference!) help constrain MHiggs to about
  35 ?MHiggs/ MHiggs
• Tevatron should reach a precision of ltlt1 with
  the full Run II data set

Summer 06
1.2
Mtop171.4 2.1 GeV/c2
24

• Here be BACKUP
• slides

25
Tevatron performance
Both CDF and DØ now have gt1.7 fb-1 on tape
26
Production and Decay

• At Tevatron, mainly produced in pairs via the
  strong interaction
• - One top pair each 1010 inelastic
  collisions at ?s 1.96 TeV
• Decay via the electroweak interactions
• - Final state is characterized by the decay of
  the W boson
• Dilepton
• LeptonJets

85

15
Br(t ?Wb) 100
Different sensitivity and challenges in each
channel

27
CDF Matrix Element Systematic Uncertainties
Source of uncertainty CDF Magnitude (GeV/c2)
b-JES 0.6
Signal (Initial and final state radiation, parton distribution functions) 1.1
Background (composition and shape) 0.2
Fit (Method, Monte Carlo statistics) 0.4
Monte Carlo (Modeling of ttbar) 0.2
Total 1.4
28
Jet permutations

• Different ways to assign jets to partons 90
  all-jets, 12 leptonjets, 2 dilepton
• Combinatorial background degrades the precision
  in Mtop
• B-tagging helps reducing numberof jet
  permutations and other background contamination
  

29
CDF Matrix Element Analysis

• Good agreement between data and Monte Carlo
  simulation

30
Matrix Element Technique

• Determine mass of the top quark evaluating a
  probability density using all the variables in
  the event, integrate over all unknowns
• Probability for event to originate from signal or
  background is calculated event by event

31
Mtop Extraction Techniques

• Template Method 1-D binned Likelihood Method
• Advantages Computationally simple
• Disadvantages Only one variable, all events
  have equal weight in the likelihood
• Matrix Element n-D unbinned Likelihood Method
• Advantages Uses all variables, each event enters
  in the likelihood weighted by how well it was
  measuredCorrect jet-parton combination is always
  included
• Disadvantages CPU intensive

Reconstruct Mtopreco for the best jet-parton
assignment and Pz?
Perform a maximum likelihood fit between data and
templates
Form signal and background templates of Mtopreco
Form a per-event likelihood using matrix elements
Evaluate the likelihood of each event as a
function of mtop
Multiply each event likelihood into one combined
likelihood