measurement of the jet mass distribution in boosted top ... · measurement of the jet mass...

Measurement of the jet mass distribution in boosted top quark decays

at 8 TeV in CMS

Torben Dreyer, Johannes Haller, Roman Kogler

International School of Subnuclear Physics 2016

itroduction

jet mass mjet

: invariant mass of all stable particles in a jet

boosted top quarks:

● decay products merge

● reconstruction of the full top quark in one jet

→ mjet

~ mtop

analysis aim: measurement of the differential tt

cross section as a function of mjet

on particle level

motivation:

● mjet

used to identify boosted top quarks

e.g. used by the CMS top-tagger

● test the modeling of jets in simulation

● extraction of the top quark mass[CMS-PAS-JME-13-007]

Torben Dreyer 1International School of Subnuclear Physics 2016

top quark plays an important role in the Standard Model (SM)

top mass important for consistency checks of the SM

LHC allows precise top mass measurements

→

(world combination 2014)

● measurements based on reconstructed

decay products

→ measurement of mass parameter

in simulation ( )

→ theoretically not well defined

● connection to well defined mass uncertainty ~ 1 GeV→


top quark mass measurements

t

W b

t

Wl

ν

b

[Phys. Rev. D93 (2016) 072004]

measurements of well defined top quark masses:

e. g. measurement of the pole mass from the inclusive tt cross section

● measurement of the inclusive tt cross

section in data

● comparison to theory predictions

→

● dependence on not large

→ sensitivity limited

● measured cross section depends on

→ additional uncertainty

→ dependence on smaller compared to direct measurements


top quark mass measurements

[arXiv:1603.02303]

alternative method

measure cross section as a function of a sensitive observable

top quark jet mass calculated in:

● boosted top quarks in lepton-lepton collisions

● all decay products in one hemisphere (“jet”)

→ calculation in an effective field theory

→ study of and its relation to well

defined mass schemes

● no LHC calculations available yet

interesting to compare to data

→ top quark mass in a well defined renormalization scheme

● proof of principle with 8 TeV data (low statistics)


[S. Fleming, A. H. Hoang et al., Phys. Rev. D77 (2008) 074010]

[A. H. Hoang, A. Pathak et al., JHEP 12 (2015) 059]

.

.

.

goal: phase space calculable theoretically and measurable experimentally

theory constraints:

all decay products in the jet

→ large pT

→ veto on additional jets

experimental constraints:

enough statistics

→ pT not too large

large jets

small background

=> measurement in lepton + jets channel

Cambridge/Aachen (CA) jets with R = 1.2 and pT > 400 GeV

Torben Dreyer 5

phase space definition


[S. Fleming, A. H. Hoang et al., Phys. Rev. D77 (2008) 074010]

measurement of the jet mass of the leading jet in tt-decays

electron/muon+jets channel

selection on particle level

● >= 1 jet with pT > 400 GeV , |ƞ| < 2.5

● >= 2 jets with pT > 150 GeV, |ƞ| < 2.5

● veto on additional jets with pT > 150 GeV

● 1 electron/muon pT > 45 GeV, |ƞ| < 2.5

particle level!

Torben Dreyer 6

mjet

[GeV]

work in progress



simulated tt events

measurement of the jet mass of the leading jet in tt-decays

electron/muon+jets channel


● >= 1 jet with pT > 400 GeV , |ƞ| < 2.5

● >= 2 jets with pT > 150 GeV, |ƞ| < 2.5

● veto on additional jets with pT > 150 GeV

● 1 electron/muon pT > 45 GeV, |ƞ| < 2.5

● ΔR(second jet, lepton) < 1.2

● mleading jet

> m2nd jet+lepton

particle level!

Torben Dreyer 7



mjet

[GeV]

work in progress

simulated tt events

reconstruction and selection based on [Phys.Rev. D93 (2016) 1, 012001]

small distance between lepton and b-jet → non isolated leptons

one b-tag

similar selection as on particle level

2012 CMS dataset with 19.7 fb-1 at 8 TeV

Torben Dreyer 8

CA 1.2

leading jet mass spectrum m

jet [GeV]

selection on detector level

work in progress


particle level distribution x folded with detector effects

→ detector effects described by

a response matrix A

unfolding: obtain true distribution x from

the detector level distribution y

response matrix determined in simulation

use regularized unfolding method

(TUnfold)

[S.Schmitt, JINST 7 (2012) T10003]


unfolding

response matrix (POWHEG+PYTHIA):

measurement phase space divided into

two pT bins

● 400 GeV < pT < 500 GeV

● pT > 500 GeV

additional sideband regions:

→ more information on migrations

→ reduction of model dep. effects

Torben Dreyer 10

300 < p T,1 < 400 GeV

100 < p T,2 < 150 GeV

150 < p T,3 < 200 GeV

measurement phase space

work in progress

10-3

10-2

10-1

response matrix


● slight overestimation of cross section in simulation

● consistent with other cross section measurements in boosted events

[Phys. Rev. D93 (2016) 032009 (ATLAS), arXiv:1605.00116 (CMS)]

Torben Dreyer 11

e/μ+jets

pT > 400 GeV

differential cross section

work in progress


● no LHC calculations available yet

● extraction of the top quark mass from

simulated templates

● just to estimate the uncertainty of

the method on 8 TeV

● only shape of mjet

is sensitive to mtop

→ use normalized cross section


top quark mass extraction

work in progress

calculate for every template:

perform a fit to the distribution

minimum best fit value for →

result:

→ large statistical and model uncertainties with 8 TeV data

→ expect large improvement with more data on 13 TeV


top quark mass extraction

work in progress

data bins simulation bins covariance matrix

mtop

[GeV]

measurement of the differential tt production cross section as a

function of mjet

at 8 TeV on particle level

sensitivity to mtop

calculable on particle level

→ independent determination of the

top quark pole mass

large uncertainties on mtop

with the 8 TeV data

8 TeV: statistical uncertainty dominant

→ expected improvement with 13 TeV data

Torben Dreyer 14

conclusion


work in progress

Back Up

response matrix from simulation:

different MC generators

→ different modeling of tt production

→ difference in the mjet

spectrum

unfolding should be independent of such

differences

→ needs to be tested

Torben Dreyer 16

variation of the Q scalem

jet [GeV]

pT, leading jet

[GeV]

unfolding bias tests

work in progress

work in progress


side

bin

influence of the simulation model:

response matrix: POWHEG

pseudo data: MADGRAPH

more bias tests:

variation of mtop

variation of the Q scale

variation of the shower model

→ differences included as a model uncertainty

Torben Dreyer 17

unfolding bias tests

work in progress


e/μ+jets

mjet

[GeV]

Torben Dreyer 18

e/μ+jetse/μ+jetsmodel uncertainties systematic uncertainties

systematic uncertainties

work in progresswork in progress


→ statistical uncertainties dominant on 8 TeV

→ expect improvement with more statistics with the 13 TeV data

The CMS detector

Compact Muon Solenoid (CMS)

● general purpose detector

● usage of the Particle Flow algorithm


true distribution x folded with detector effects

→ detector effects described by a response matrix A

→ regularized unfolding method:

→ obtain true distribution x from the

reconstructed distribution y by a maximum

likelihood fit

→ amplification of statistical fluctuations

→ suppress with a regularization term

[S.Schmitt, JINST 7 (2012) T10003]


unfolding

measurement of the jet mass distribution in boosted top ... · measurement of the jet mass...

Documents