search for thq with h b b · christian boser search for thq with h¨! b b 22.09.2014 6/14. analysis...

1 INSTITUT FÜR EXPERIMENTELLE KERNPHYSIK

Search for tHq with H → bb̄A test of Higgs couplings

Christian Böser1 | GK-Workshop Bad Liebenzell | 22.09.2014

KIT – University of the State of Baden-Wuerttemberg and

National Laboratory of the Helmholtz Association

www.kit.edu

http://www.kit.edu

Introduction to pp → tHqInvestigate coupling of the Higgs boson to fermions→AtHq ∝ (κV − κf )

q q′

b t

HW

W

q q′

b t

HW

Destructive interference in SM⇒ cross-section is 18.3 fbWith κ†t = −1

σtHq = 234 fb13 times enhanced comparedto SM case

† κt = Yt/YSMt

[GeV]HM120 122 124 126 128 130 132

H +

X)

[pb]

→(p

p σ

-210

-110

1

10

H (NNLO + NNLL QCD + NLO EBW)→pp

qqH (NNLO QCD + NLO EW)→pp

WH (NNLO QCD + NLO EW)→pp

ZH (NNLO QCD + NLO EW)→pp

ttH (NLO QCD)→pp

= 8 TeVs

tHq (NLO QCD)→pp

M. Farina et al., JHEP 1305 (2013) 022

LHC

HIG

GS

XS

WG

2014

Christian Böser Search for tHq with H → bb̄ 22.09.2014 2/14

MotivationDiscover such a unique final state!Help excluding κt = −1 scenario

Moriond 2014

Vκ0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6

Fκ

-2

-1

0

1

2

3

4

bb→H bb→H

ττ →H ττ →H

4l→H 4l→H

νlν l→H νlν l→H

γγ →H γγ →H

bb→H ττ →H 4l→H νlν l→H

γγ →H CombinedSM Best Fit-1Ldt = 20.3 fb∫ = 8 TeV s

-1Ldt = 4.6-4.8 fb∫ = 7 TeV sATLAS Preliminary

ATL

AS

CO

NF-

2014

-009

Vκ0 0.5 1 1.5

fκ

-2

-1

0

1

2

95%

C.L.

bb→H

ττ→H

ZZ→H

WW→H γγ→

H

Preliminary CMS (7 TeV)-1 (8 TeV) + 5.1 fb-119.7 fb

Observed SM Higgs

CM

SPA

SH

IG-1

4-00

9

Coupling constraints from ATLAS + CMS disfavor κt = −1,under assumption of only SM contributions to the total widthBSM contributions to the loops in the Hγγ and Hgg couplings areallowed→ κt = −1 still tolerated


http://cds.cern.ch/record/1670012http://cds.cern.ch/record/1728249

Direct search for tHq and H→ bb̄ with κt = −1CMS-HIG-14-015

Approval scheduled for this Friday


The CMS detector

CMS


Analysis strategy

Challenging multijet final state

1 forward light jet

3 or 4 b jets (spectator b outside oftracker acceptance in ∼ 30% of thecases)

1 isolated charged lepton and ET/

Expected yields

S/B3 tag region ∼ 0.7% (13/1900 events)4 tag region ∼ 2.1% (1.4/66 events)

q

g

q′

b̄

b

b̄

b

`+

ν

W

b

tH

W


Analysis strategy

Challenging multijet final state

1 forward light jet

3 or 4 b jets (spectator b outside oftracker acceptance in ∼ 30% of thecases)

1 isolated charged lepton and ET/

Expected yields

S/B3 tag region ∼ 0.7% (13/1900 events)4 tag region ∼ 2.1% (1.4/66 events)

t̄t production dominant background

Validation done in t̄t control region

Signal enriched phase space

tH reconstruction ttbar reconstruction

fit on MVA output

final MVA discriminatortH vs background

tH variables ttbar variables global variables

MVAs for reconstruction andclassification


Reconstruction of tHq

Find correct jet assignment to final state quarks is combinatorial issue

Reconstruct all possible hypotheses per eventAdditional constraints to reduce combinatorics

Correct interpretation =̂ all four quarksmatched to a jet within ∆R < 0.3Wrong interpretations: all other possible jetassignments

Train MVA to discriminate between correct andwrong hypotheses

Kinematic variablesb-tagging informationAngular correlations

j1 j2

j3

j4

j5 j6

j7 jn

𝑗1,𝐻?

𝑗2,𝐻?

𝑗𝑡?

𝑗𝑞?

Application on data: Take hypothesis with largest MVA response


MVA Response of tHq Reconstruction

Good discrimination betweencorrect and wrong hypothesesfound

Data/MC comparisons done int̄t control region

MVA response-0.5 0.0 0.5 1.0

Arb

itrar

y un

its

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14sgn exambkg examsgn trainbkg train

Simulation Preliminary CMS 8 TeV

Performance Validation

MVA output of random tHq hypothesis

-0.5 0 0.5 1

Eve

nts

0

1000

2000

3000

Data=-1

ttHq, y

+bttb+btt

+1,2ctt+LFtttt/

ttHEWK

PreliminaryCMS

(8 TeV)-120 fb

Muon channel CRtt

MVA output of random tHq hypothesis-0.5 0 0.5 1

MC

Dat

a-M

C

-0.50

0.5


Additional reconstruction under t̄t hypothesis

Good discrimination of processes tHq and t̄t is crucial for the analysis

Idea to get better separation power:Perform additionally reconstructionunder t̄t production hypothesisUse similar set of variables as for tHqreconstructionTake this as input for classification

j1 j2

j3

j4

j5 j6

j7 jn

𝑗1,𝑊?

𝑗2,𝑊?

𝑗𝑡 ?

𝑗𝑡?

Again: Take correct assignment for training and take hypothesis withlargest MVA response for application


MVA Response of t̄t Reconstruction

Good discrimination betweencorrect and wrong hypothesesfound

Data/MC comparisons done int̄t control region

MVA response-0.5 0.0 0.5 1.0 1.5

Arb

itrar

y un

its

0.00

0.05

0.10

0.15



Performance Validation

hypothesistMVA output of random t

-0.5 0 0.5 1

Eve

nts

0

1000

2000

3000

Data=-1

ttHq, y

+bttb+btt

+1,2ctt+LFtttt/

ttHEWK

PreliminaryCMS

(8 TeV)-120 fb

Muon channel CRtt

hypothesistMVA output of random t-0.5 0 0.5 1

MC

Dat

a-M

C

-0.50

0.5


ClassificationTake all information after reconstructionOptimize classification MVA to separate signal from background

4 tHq variables

(light jet)|η|0 1 2 3 4 5

Eve

nts

/ Bin

wid

th

0

500

1000Data

=-1t

tHq, y+btt

b+btt+1,2ctt+LFtttt/

ttHEWK50x tHq

PreliminaryCMS

(8 TeV)-120 fb

Muon channel3T region

=125 GeVHm

(light jet)|η|0 1 2 3 4 5

MC

Dat

a-M

C

-0.50

0.5

3 t̄t variables

hadtlog m

4.5 5 5.5 6

Eve

nts

0

100

200

300

Data=-1

ttHq, y

+bttb+btt

+1,2ctt+LFtttt/

ttHEWK50x tHq

PreliminaryCMS

(8 TeV)-120 fb


=125 GeVHm

hadtlog m

4.5 5 5.5 6

MC

Dat

a-M

C

-0.50

0.5

1 global variable

lepton charge

-1 0 1

Eve

nts

0

500

1000

Data=-1

ttHq, y

+bttb+btt

+1,2ctt+LFtttt/

ttHEWK50x tHq

PreliminaryCMS

(8 TeV)-120 fb


=125 GeVHm

lepton charge-1 0 1

MC

Dat

a-M

C

-0.50

0.5

MVA response-0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2

Arb

itrar

y un

its

0.00

0.02

0.04

0.06

0.08




Fit on MVA output distributions

0 0.5 1

Eve

nts

0

100

200

Data=-1

ttHq, y+btt

b+btt+1,2ctt+LFtt

tSingle t/ttHDibosonW+Jetsstat.+sys.50x tHq

PreliminaryCMS

(8 TeV)-120 fb

Electron channel3T region

=125 GeVHm

MVA output0 0.5 1

MC

Dat

a-M

C

-0.50

0.5stat.+sys.

3Tre

gion

Electron channel

0 0.5 1

Eve

nts

0

5

10

Data=-1

ttHq, y+btt

b+btt+1,2ctt+LFtt


PreliminaryCMS

(8 TeV)-120 fb

Electron channel4T region

=125 GeVHm

MVA output0 0.5 1

MC

Dat

a-M

C

-101

stat.+sys.

4Tre

gion

0 0.5 1

Eve

nts

0

100

200

300

Data=-1

ttHq, y+btt

b+btt+1,2ctt+LFtt


PreliminaryCMS

(8 TeV)-120 fb


=125 GeVHm

MVA output0 0.5 1

MC

Dat

a-M

C

-0.50

0.5stat.+sys.

Muon channel

0 0.5 1

Eve

nts

0

5

10

Data=-1

ttHq, y+btt

b+btt+1,2ctt+LFtt


PreliminaryCMS

(8 TeV)-120 fb


=125 GeVHm

MVA output0 0.5 1

MC

Dat

a-M

C

-101

stat.+sys.


Results

95% Upper Limit on σ/σκt=−1

Expected Observed

MC-driven 5.14+2.14−1.44 7.57

data-driven 6.24+2.26−1.71 6.95

Observations in goodagreement with SM

Cross-check analysis withsimilar result

Combination of all MVAdistributions into a single plot

Events in all channelssorted in bins of similarexpected S/B ratio

(S/B)10

log-7 -6 -5 -4 -3 -2 -1

Eve

nts

-110

1

10

210

310

410

(8 TeV)-120 fb

=125 GeVH

)q, mbtH(b

Preliminary

)µ3T+4T regions (e+

CMS

Data=-1ttHq, y

All bkgs.Bkg. unc.

(S/B)10

log-7 -6 -5 -4 -3 -2 -1

Bkg

.D

ata-

Bkg

.

-1

0

1 = 0.992χKS test = 1.00


Summary

Investigation of tHq is an active field with a lot of potential

Search for tHq with H→ bb̄ is challengingUpper limits on κt = −1 caseCMS-PAS-HIG-14-015 will be(hopefully)

.. approved this Friday

.. shown at TOP 2014 in Cannesfor the first time

Combination paper with allavailable channels planned

Stay tuned!

]-1Luminosity [fb0 20 40 60 80 100

= -

1tκσ/σ

lim

it on

S

95%

CL

0

2

4

6

8

= 14 TeVs

tHq production→pp b b→, H ν bl→t

= 125 GeVH

= -1, mtκ

expectedsCL

10 to 30% improvement

Private work

σNLO(pp)→ tHq [fb]κF = 1 κF = −1

8 TeV 18.3 23414 TeV 88.2 982


BACKUP


Cross sections

Cross section is challengingly smallThe main background is t t̄ ; its cross section is provided for comparison

Cross-section 8 TeV 14 TeV

tHq, yt = +1 (SM) 18.3± 0.4 fb 88.2+1.7−0.0 fb

tHq, yt = −1 233.8+4.6−0.0 fb 980+30−0 fb

t t̄ 245+9−10 pb 950+40−30 pb

tHq cross sections are cited according to M. Farina et al., JHEP 1305 (2013) 022

[arXiv:1211.3736]. Cross-sections for t t̄ are calculated in M. Czakon, P. Fiedler, Phys. Rev.

Lett. 110 (2013) 252004 [arXiv:1303.6254]. Uncertainties are combined following R. Barlow,

arXiv:physics/0306138


http://arxiv.org/abs/1211.3736http://arxiv.org/abs/1303.6254http://arxiv.org/abs/physics/0306138

Flavor Scheme Comparison

m(b) > m(p)→ b quark is no protonconstituent,

b quarks can only be pair-produced in highQ2 production

b quarks in initial state, i.e. inside the proton

Additional b comes through parton shower


CMS-PAS-HIG-14-015

Baseline Selection

Logical OR of single lepton triggers hlt isomu24 / hlt ele27 wp80Exactly 1 reconstructed isolated lepton

Muons: pT > 26 GeV, |η| < 2.1, reliso < 0.12Electrons: pT > 30 GeV, |η| < 2.5, reliso < 0.10

3 tag Region

#jets30 ≥ 4#jetsCSVT = 3

4 tag Region

#jets30 ≥ 5#jetsCSVT = 4


Data-driven QCD estimation

Apply ABCD methodRequires two selection cuts which are assumed to be uncorrelatedfor QCD events

A region: ordinarily-defined signal or control region used in thisanalysis.B region: as A, but the lepton cuts are inverted; ie. the event isrequired to have exactly one lepton failing the “tight” definition butpassing the “loose” definition.C region: as A, but the pfMET cuts are inverted; in the electronchannel the cut is ET/ < 45 GeV and in the muon channelET/ < 35 GeV.D region: both cuts described above are inverted.

NA,QCD =(NB,data − NB,non-QCD) · (NC,data − NC,non-QCD)

ND,data − ND,non-QCD(1)


Data-driven QCD estimation

number of prim. vertex candidates

0 10 20 300

1000

2000

data)btH(b

+bttb+btt

+cctt+lftt

tSingle t/ttHElectroweakMC stat.

=8 TeVs -1L=20 fb

Region A

number of prim. vertex candidates0 10 20 30

data

(dat

a-m

c)

-0.5

0

0.5 MC stat.


0 10 20 300

50

100

150

=8 TeVs -1L=20 fb

Region B


data

(dat

a-m

c)

-0.5

0

0.5 MC stat.


0 10 20 300

500

1000

1500

2000

=8 TeVs -1L=20 fb

Region C


data

(dat

a-m

c)

-1

-0.5

0

0.5

1MC stat.


0 10 20 300

100

200

300

=8 TeVs -1L=20 fb

Region D


data

(dat

a-m

c)

-0.5

0

0.5 MC stat.


CMS-PAS-HIG-14-015

Table : Input Variables of tHq Reconstruction

Variable DescriptionCharge BTop Electric charge of b-quark jet from decay of top quark, multiplied by lepton’s charge;Cos LepRecoil TH Cosine of the angle between momenta of lepton and recoil jet in the rest frame of t + H

systemDeltaR BJetsHiggs ∆R between the two jets from decay of Higgs bosonDeltaR BTopW ∆R between b-quark jet and W boson from decay t →WDeltaR TopHiggs ∆R between reconstructed top quark and Higgs bosonabs(Eta Recoil) Pseudorapidity of recoil jetlog(Mass BTopLep) Invariant mass of b-quark jet from decay of top quark and charged leptonlog(Mass Higgs) Mass of reconstructed Higgs bosonabs(MaxEta BHiggs) Pseudorapidity of the most forward jet from decay of Hlog(MinPt BHiggs) Transverse momentum of the softest jet from decay of HNumBTag Higgs Number of b-tagged jets among the two jets from decay of HPassBTag BTop Equals 1 if the b-quark jet from decay of t is b-tagged, 0 otherwisePassBTag Recoil Equals 1 if the recoil jet is b-tagged, 0 otherwiseRelHt Relative HT , (pT(t) + pT(H))/HT


CMS-PAS-HIG-14-015

Table : Input Variables of t t̄ Reconstruction

Variable DescriptionCharge BTopHad -

Charge BTopLep

Difference of electric charges of b-quark jets from decays of thad and tlep , multi-plied by lepton’s charge

DeltaR Light ∆R between the two light-flavor jets from decay of thadDeltaR BTopHadWHad ∆R between b-quark jet and W boson from decay thad →WDeltaR BTopHadWLep ∆R between b-quark jet and W boson from decay tlep →Wlog(DMass TopHadWHad) Difference between masses of thad and W from decay of thadabs(Eta TopHad) Pseudorapidity of thadlog(Mass BTopLepLep) Invariant mass of b-quark jet from decay of tlep and charged leptonlog(Mass WHad) Mass of W from decay of thadNumBTag Light Number of b-tagged jets among the two light-flavor jets from decay of thadPassBTag BTopHad Equals 1 if the b-quark jet from decay of thad is b-tagged, 0 otherwisePassBTag BTopLep Equals 1 if the b-quark jet from decay of tlep is b-tagged, 0 otherwiselog(Pt TopHad) Transverse momentum of thadlog(Pt TopLep) Transverse momentum of tlepRelHt Relative HT , (pT(thad) + pT(tlep))/HTSumCharge Light Sum of electric charges of the two light-flavor jets from decay of thad , multiplied by

lepton’s charge


CMS-PAS-HIG-14-015

Table : Input variables for the classification MVA

Variable Descriptionglb Charge Lep Electric charge of the leptonabs(thq Eta Recoil) Pseudorapidity of the recoil jetthq NumBTag Higgs Number of b-tagged jets among the two jets from the

Higgs boson decaylog(thq Pt Higgs) Transverse momentum of the Higgs bosonlog(thq Pt Recoil) Transverse momentum of the recoil jettt DeltaR Light ∆R between the two light-flavor jets from the decay of

thadlog(tt Mass TopHad) Mass of thadtt NumPassBTag Light Number of b-tagged jets among the two light-flavor jets

from the decay of thad


CMS-PAS-HIG-14-015

Table : Impact of systematic uncertaintiesOmitted systematic impact on limit [%]JES −1.3JER −0.2BTag light flavor −0.7BTag heavy flavor 2.1Pile up 0.1Unclustered energy −0.6Lepton efficiency −2.5Luminosity −2.0Cross section (PDF) −2.2Cross section (Scale) −2.4MC Bin-by-Bin unc. −3.3Q2 scale (tHq + t t̄) −9.1Matching −1.4Top pT reweighting −3.2t t̄ HF rates (b) −1.2t t̄ HF rates (bb̄) −0.9t t̄ HF rates (c / cc̄) −3.8t t̄ HF rates (total) −7.9

only Systematic impact on limit [%]JES −1.8JER 1.4BTag light flavor 13.3BTag heavy flavor 1.9Pile up −0.9Unclustered energy −4.0Lepton efficiency −0.9Luminosity −1.0Cross section (PDF) 17.5Cross section (Scale) 18.9MC Bin-by-Bin unc. 1.8Q2 scale (tHq + t t̄) 33.7Matching −2.9Top pT reweighting 43.8t t̄ HF rates (b) 29.4t t̄ HF rates (bb̄) 33.6t t̄ HF rates (c / cc̄) 23.0t t̄ HF rates (total) 32.3


IntroductionCoupling exclusion

CMS-PAS-HIG-14-015

search for thq with h b b · christian boser search for thq with h¨! b b 22.09.2014 6/14. analysis...

Documents