Top quark mass measurements at CMS Minsuk Kim Helsinki Institute of Physics 23 November 2018 Particle Physics day
Top quark mass measurements at CMS
Minsuk KimHelsinki Institute of Physics
23 November 2018Particle Physics day
Overview• Motivation
• Top Production and Mass Definition
• Object Definitions & Event Selection
• Kinematic Fit and Mass Extraction
• Potential Improvements
• CMS mt Measurements
• Summary
2
Motivation
• Its precise value impacts directly on key predictions of the SM
• Production rates at LHC, size of quantum correction to electroweak processes, and coupling strength of top quark with Higgs boson
• Its value leads to a significant constraint on stability of the EW vacuum
• Theoretical predictions use the top-quark pole mass
• We need a good understanding of the measured top-quark mass w.r.t. the pole mass or a mass with a well defined renormalization scheme
• The mass of the top quark (mt) is one of the fundamental parameters of the Standard Model
!3
Motivation
4
• Strong coupling 𝛼s and top-quark mass mt are two of the most fundamental ones
• Using the measured values of mt and mH, it seems that the current electroweak vacuum is meta-stable
• Quartic Higgs self-coupling (𝜆) depends heavily on mt, may turn negative at a scale lower than the Planck scale and induce vacuum instability
• Hope for greater accuracy in the experimental measurements to establish whether we are in the region that would be meta/unstable within the SM
• LHC jet measurements are key input, and Jet Energy Correction (JEC) is their fundamental uncertainty
Stability of SM vacuum
EPJC 77 (2017) 746
• We want ∆mt < 0.2 GeV for 3𝜎 confidence on vacuum meta-stability
• As the experimental sensitivity is entering the sub-GeV range, issues of theoretical interpretation become important
• To get there
• Assume theories can reduce mt(MC) vs. mt(pole) uncertainty (e.g. Hoang, Nason)
• Perform ultra-legacy re-reco and time-dependent MC to understand data
• Factorise main physics effects and constrain them with control regions
• Question
• How much data we need to constrain the dominant systematic uncertainties related to jet energy scale (JES) calibration and QCD modeling?
To get ultimate precision
5
Top Quark Production
!6
• Top quark is the heaviest known elementary particle
• It decays before hadronization (lifetime τ = ~5×10-25 s)
• Main top decay: t → Wb
• Top quark pair production has 3 decay channels
• All-jets (46.2%) W → qq
• Lepton+jets (43.5%) W → 𝑙𝜈 or qq
• Dilepton (10.3%) W → 𝑙𝜈
• LHC is a top quark factory: 𝜎tt(14 TeV) = 800 pb, 2 ttbar events/sec
• Total cross section a factor of 100 larger at LHC than at Tevatron
Top quark mass definition• Decay before hadronizing ⇒ measure mt directly from decay products
• Different definitions of mt
• Most precise top mass measurements use kinematic reconstruction methods, determining the top mass parameter of a Monte Carlo event generator, mtMC, however mtMC (direct measurement) ≠ mtpole (indirect measurement)
• The uncertainty on the translation from the mtMC definition to a theoretically well defined short distance mass definition at low scale is currently estimated to be of the order of 1 GeV (arXiv:1405.4781, arXiv:1408.6080, Nucl. Phys. Proc. Suppl. 185 (2008) 220)
• Due to hadronization and parton shower dynamics, relating mtMC to a theory mass is difficult, but perhaps < 100 MeV (arXiv:1712.02796, arXiv:1608.01318)
!7
• mtMC (Monte Carlo mass) parameter measured from comparison to MC events with top-quark decay products• mtpole (pole mass) parameter is the classic rest mass entering the
top propagator (the pole is fixed order by order)• mtMS (running mass) parameter defined in a low-scale short
distance scheme (the pole is shifted at each order)
Object definitions
!8
Lepton + jets Channel
Estimated composition: 92.5% tt, 2% W+jets, 3.5% single top, 2% other!9
Precision measurement based on kinematic reconstruction where double-b-tagging and leptonic W decay reduce combinatorial background and allow to select ttbar events with high purity
Selection:
• Exactly 1 isolated e (𝜇) with pT > 34
(26) GeV, |𝜂| < 2.1 (2.4) and veto
additional e, 𝜇
• ≥ 4 jets with pT > 30 GeV, |𝜂| < 2.4
• 2 b-tagged jets
Kinematic fitSplit 3 permutation classes:• correct• wrong: flipped b-quarks, mistags• unmatched
Kinematic fit with constraints:• two untagged jets
mjj = 80.4 GeV• lepton and neutrino (MET)
m𝑙𝜈 = 80.4 GeV
• Combine with two b-tagged jetsm(jj,b1) = m(𝑙𝜈,b2)
!10
Constraining mjj=mW effectively removes light-quark-jet uncertainties, leaving b-jet corrections as the limiting uncertainties
Mass extraction: ideogram method• Ideogram method: modification of template
method using multiple permutations with different weights (suited with large datasets)
• To extract the maximum amount of mass information out of a tt candidate event
• Realized by constructing 2D likelihoods for each event (ideograms)
• To determine simultaneously mt and jet energy scale (JES) as the JES was found to result in the dominating systematic uncertainty in the previous top mass measurements (CMS-TOP-10-009)
• All selected permutations are taken into account and weighted by their fit probability
!11
High Luminosity2011 2012 2015-2018 nominal HL-LHC
CM energy (TeV) 7 8 13 14Cross section (pb) 167 246 806 951Luminosity (fb-1) 5 20 3/36/40/60 300 3000
<Pileup> 10 21 13/27/38/37 ~40 ~100
300 fb-1we are here!
!12Precision achievable with 300-3000 fb-1 ?
Possible Improvementsin the understanding of systematic uncertainties for standard techniques
unc
erta
inty
[GeV
]to
pm
0
0.2
0.4
0.6
0.8
1
1.2 Present -1 30 fb 13 TeV
-1 300 fb 14 TeV
-1 3000 fb 14 TeV
CMS preliminary projectionTotal Stat+iJESb-JES UE,CRdJES others
,ME-PS2Q
Std. methods
full NLO
move to 3D fit
dedicated UE studies
differential measurements
by using lepton+jets measurement at 7 TeV with 5 fb-1, JHEP 12 (2012) 105, mt = 173.49 ± 0.43 (stat) ± 0.98 (syst) GeV (results are comparable in precision to those from Tevatron)
0.2 GeV
• CMS estimated ultimate precision achievable with 30 fb-1, 300 fb-1 & 3000 fb-1 of data at 13-14 TeV
• Done in 2013 based on optimistic assumptions with large data
• Promising methods to reduce errors
• Particle-level studies with data: improve tuning of the UE event
• 3D fitting methods: fit both the light-jet and b-jet energy scale (JES, bJES) in-situ, with mt
• Differential mt studies: compare different CR models and check for different MC generators
• Full NLO tools: improve description of ISR and FSR
5 fb-1 7 TeV
Extrapolation to high lumi. (FTR-13-017)
13
CMS mt Measurementsin the understanding of systematic uncertainties for standard techniques
unc
erta
inty
[GeV
]to
pm
0
0.2
0.4
0.6
0.8
1
1.2 Present -1 30 fb 13 TeV
-1 300 fb 14 TeV
-1 3000 fb 14 TeV
CMS preliminary projectionTotal Stat+iJESb-JES UE,CRdJES others
,ME-PS2Q
Std. methods
0.48 GeV (Run1)
5 fb-1 7 TeV
EPJC 78 (2018) 891
14
Lepton + jets channel mt (GeV)2012 8 TeV 19.7 fb-1 172.35 ± 0.16 ± 0.48 2017 13 TeV 35.9 fb-1 172.25 ± 0.08 ± 0.62
0.2 GeV
Possible with ~150 fb-1 of
data in Run2 ?
First goal!
Extrapolation to high lumi. (FTR-13-017)
JEC calibrations
!15
Slide from M. Voutilainen
Wishlist for Run 2• Improvements to analysis:
• NLL with nuisances to make analysis more similar to others
• get optimal hybrid weight automatically
• Improve selection and/or split in categories
• Add more observables, measurements to reduce Flavor uncertainties
• External improvements:
• High statistics MC samples
• As low JEC/JER uncertainties as possible allowing to focus on modelling
• Second Parton Shower generator
• More CR models to understand spread
• Understand ME generator difference, more gen studies
• Supporting measurements to constrain models (e.g. b fragmentation)!16
Slide from H. Stadie
Summary• Precise measurements of mt are fundamental to provide inputs to test
the self-consistency of the SM
• Presented prospect and latest CMS results of mt
• For the first time mt measured at √s = 13 TeV at CMS
• With ultimate precision achievable at LHC, the relation between mt definition of the experimental analysis and mtpole is becoming relevant
• mt measurements dominated by systematic uncertainties
• First goal with the challenge to bring systematic uncertainties down
• Better than Run 1 with new 2016 JEC/flavor uncertainty or 2017
Backup
ATLAS+CMS Preliminary
ATLAS: 172.51 ± 0.27 ± 0.42
CMS: 172.44 ± 0.13 ± 0.47
!19
Mass extraction: ideogram method
!20
Slide from LHCtopWG open meeting, 20.11.2018
Calibration
!21
Slide from LHCtopWG open meeting, 20.11.2018
Mass extraction: hybrid approach
Slide from LHCtopWG open meeting, 20.11.2018
Systematic uncertainties
!23
Lepton + jets channel
Slide from LHCtopWG open meeting, 20.11.2018
Systematic uncertaintiesAll-jets channel
Slide from LHCtopWG open meeting, 20.11.2018
Color reconnection in ttbar• CR affects the reconstruction of the top quark system
!25
JEC uncertainties
!26
Slide from M. Voutilainen
!27
Slide from M. Voutilainen
JEC: current status
!28
JEC: current statusSlide from M. Voutilainen