STUDY OF CHARGED HIGGS BOSONS SEARCH AT …epx.phys.tohoku.ac.jp/eeweb/meeting/2017_LCWS_christian.pdfJet pairing • Jet pairing with chi squared minimization • j 1, j 2, j 3and

High Energy Accelerator Experiments Group Tohoku University

Presentation by Christian Drews

Academic advisors: Professor Hitoshi Yamamoto (Tohoku Uni.)Professor Arno Straessner (TU Dresden)

STUDY OF CHARGED HIGGS BOSONS SEARCH AT THE ILC FOR A COLLISION ENERGY OF 1 TEV

Overview

• Full simulation study of ILC/ILD• mH± = 350 GeV cross section = 9 fb BR(H±→bt) = 90%

• E+e- -> H+H- -> tb tb -> Wbb Wbb 8 jets (hadronic)

Wbb Wbb 6 jets + lepton (semi-lep.)• Major background:• ttH/ttZ/ttg -> ttbb• tt -> bWbW• HA -> bbbb (SUSY) • H/A -> tt at resonance• Ignoring SUSY background

• Goal: mH± meassurment• Samples with 340, 346, 348, 350, 352, 354, 360 GeV mass

17.05.2017 ChristianDrews’jobinterviewpresentation

W!2 jets��!W!2 jets��!W!⌫l

Cross section Allowed parameter space

24.10.17 StudyofchargedHiggsbosonssearchattheILCforacollisionenergyof1TeV 3

Source: A.Arbey etal.“StatusoftheChargedHiggsBosoninTwoHiggsDoubletModels.”In:(2017).arXiv:1706.07414[hep-ph].

ps = 1000GeVps = 800GeV

• 𝜎 ≈ 9 fb with P = (-80%,20%) 10.4 fb• 𝓛 = 1000 1/fb• Hadronic: 5100 events• Semileptonic: 3200 events

Source: ChargedHiggsBosonproductionatee collidersinthecomplexMSSM:afullone-loopanalysis Heinemeyer,S.andSchappacher,C. Eur.Phys.J. (2016)

• MSSM excluded

• Brought parameter space allowed

/ GeVH+m200 250 300 350 400 450 500

even

ts /

bin

0

50

100

150

200

250

300

/ GeVH-m200 250 300 350 400 450 500

even

ts /

bin

0

50

100

150

200

250 w/o overlay

w/o correction

R = 1.0

R = 1.3

R = 1.5

4

ReconstructedH+ andH- masswithrealisticclusteringandpairingwithgeneratorinformation

mH+mH�

Beam background reduction with kt-Algorithm

Rchosentobe1.3

Jet pairing

• Jet pairing with chi squared minimization• j1, j2, j3 and j4 – jets with

highest b-tag• For semi-leptonic: j7 and j8

are neutrino and lepton


�2 =

��(mj1j2j3j4)

2 � (mj5j6j7j8)2

2�2H+

��+✓mj2j3j4 �Mt

�t

◆2

+

✓mj6j7j8 �Mt

�t

◆2

+

✓mj3j4 �MW

�W

◆2

+

✓mj7j8 �MW

�W

◆2

w/ooverlay

R:1.3 withoverlay

B-tagefficiency 44.6 42.5 38.0 the4b-jetshavehighestb-tagintheevent

Clusteringworkswell 50.7 49.4 40.2 Foreverycolorsingletthereare 2jetswithamajorfractionfromthissinglet

Pairingworks 27.8 25.0 17.2 Jetpairingagreeswithmajorcolorsingletfractioninjet

• Missing-Energy-Method (MEM)

• Standard for most studies• Missing-Direction-Method (MDM)

• Using the Direction of Missing-Energy-Method and calculation the Enerty by fixing W-Mass

ChristianDrews StudyofchargedHiggsbosonssearchattheILCforacollisionenergyof1TeV 6

Neutrino four momentum

p⌫,MEM = (pCMS � pvis)

pCMS = (1000, 0, 0, 1000 · sin(0.014/2))pvis =NPFOX

i=1

pi

↵ =~p⌫,MEM · ~pl|~p⌫,MEM||~pl|

p⌫,NDM = (E⌫,NDM, E⌫,NDM~p⌫,MEM

|~p⌫,MEM| )

E⌫,NDM =m2

W

El(1� ↵)


• Missing Momentum Method (MMM)• Using momentum form MEM for

energy estimation

• Missing Transversal Momentum Method (MTMM)• Using only the momentum in transversal momentum

has two solutions

Neutrino four momentum

Neutrino reconstruction


gen)/Egen-Ereco(E1− 0.8− 0.6− 0.4− 0.2− 0 0.2 0.4 0.6 0.8 1

even

ts /

bin

0

1

2

3

4

5

6

Neutrino energy

MTMM

MDM

MEM

MMM

z,gen)/pz,gen-pz,reco(p1− 0.8− 0.6− 0.4− 0.2− 0 0.2 0.4 0.6 0.8 1

even

ts /

bin

0.5

1

1.5

2

2.5 MTMM

MDM

MEM

MMM

Neutrino momentum in z-direction

MTMM Transversal momentum

MDM Direction

MEM Standardwithmissing energy

MMM Only usingmomentum

Event selection


• Different event selections were used• Static cuts• BDT

• Optimizations• Signal significance• Significance for correctly paired signal (BDT + separate BDTG)

Mass measurement

• Template method

• Compare mass distribution shape -> calculation Chi^2

• Get uncertainty from Chi^2 parabola • Shape method

• Get shape of BG, correctly paired signal and false paired signal

• Calibrate fitted mean to Higgs mass• Combined method

• Reduce fitting variables to Higgs mass from cor. and false paired signal


Higgs mass / GeV260 280 300 320 340 360 380 400 420

Even

ts /

( 3 )

0

50

100

150

200

250

300

350

= 340 GeV±H

Template fit for m

Higgs mass / GeV260 280 300 320 340 360 380 400 420

Even

ts /

( 3 )

0

50

100

150

200

250

300

350

= 346 GeV±H

Template fit for m

Higgs mass / GeV260 280 300 320 340 360 380 400 420

Even

ts /

( 3 )

0

50

100

150

200

250

300

350

= 348 GeV±H

Template fit for m

Higgs mass / GeV260 280 300 320 340 360 380 400 420

Even

ts /

( 3 )

0

50

100

150

200

250

300

350

= 350 GeV±H

Template fit for m

Higgs mass / GeV260 280 300 320 340 360 380 400 420

Even

ts /

( 3 )

0

50

100

150

200

250

300

350

= 352 GeV±H

Template fit for m

Higgs mass / GeV260 280 300 320 340 360 380 400 420

Even

ts /

( 3 )

0

50

100

150

200

250

300

350

= 354 GeV±H

Template fit for m

Higgs mass / GeV260 280 300 320 340 360 380 400 420

Even

ts /

( 3 )

0

50

100

150

200

250

300

350

= 360 GeV±H

Template fit for m


Template method

�2 =NX

i=0

Ti � SipSi

Shape method


Higgs mass / GeV250 300 350 400 450 500

Even

ts /

( 7.1

4286

)

0

50

100

150

200

250

300

350

400

450

Correctly paired signal

Higgs mass / GeV250 300 350 400 450 500

Even

ts /

( 7.1

4286

)

0

10

20

30

40

50

60

Failed clustering

Higgs mass / GeV250 300 350 400 450 500

Even

ts /

( 7.1

4286

)

0

100

200

300

400

500

600

700

800

900

Total signal

Static cuts (exp.) = 5491SN

Correctly paired: (24.5%):

: 352.1 GeVµ

: 24.8 GeVLσ

: 12.8 GeVRσ

Wrong paired (75.5%):

: 355.1 GeVµ

: 45.6 GeVLσ

: 71.5 GeVRσ

= 2.052χred.

• Get signal shape (Static Cuts)

Shape method


Higgs mass / GeV250 300 350 400 450 500

Even

ts /

( 7.1

4286

)

0

50

100

150

200

250

300

Background fit to BiGauss (original)

Higgs mass / GeV250 300 350 400 450 500

Even

ts /

( 7.1

4286

)

0

20

40

60

80

100

120

140

160

180

200

220

240

Background p.d.f. and generated data set

Static cuts (exp.) = 2418BN = 373.8 GeVµ

: 69.2 GeVLσ

: 74.8 GeVRσ

(original BG) = 3.062χred. (gen BG) = 0.962χred.

• Get background shape (Static Cuts)

Added fit


• Added fit for different event selections for shape method

• All fit parameter fixed except mean of signal distributions

• Background• Signal (solid)• Wrong paired signal (dashed)• Correctly paired signal

Shape method – mass estimation (BDT – correctly paired)


(Generated mass - 350) / GeV10− 5− 0 5 10

352

.1/ G

eV)

cµM

axim

um o

f cor

r. pa

ired

dist

ribut

ion

(

10−

5−

0

5

10

/ ndf 2χ 1.369 / 5

Prob 0.9277

p0 0.2589±0.03791 −

p1 0.04322± 1.054

/ ndf 2χ 1.369 / 5

Prob 0.9277

p0 0.2589±0.03791 −

p1 0.04322± 1.054

Linear regression for corr. paired signal

(Generated mass - 350) / GeV10− 5− 0 5 10

- 33

9.1

/ GeV

)wµ

Max

imum

of w

rong

pai

red

dist

ribut

ion

(

8−

6−

4−

2−

0

2

4

6

8 / ndf 2χ 0.5786 / 5

Prob 0.989

p0 0.5018± 0.3524

p1 0.08468± 0.5363

/ ndf 2χ 0.5786 / 5

Prob 0.989

p0 0.5018± 0.3524

p1 0.08468± 0.5363

Linear regression for wrong paired signal

BDT correctly paired

0.68 (Corr. pairing)±/GeV = 349.74 c,Hm

2.59 (Wrong pairing)±/GeV = 350.16 w,Hm

0.66 (combined)±/GeV = 349.77 a,Hm

0.46 (reduced)±/GeV = 349.83 r,Hm

: 0.2456c,const∆

: -0.012c,linear∆

: 0.6319c,fit∆

0.26)± + (352.1 cµ 0.04) ±/GeV = (1.05 c,Hm

0.5)± + (339.09 wµ 0.08) ±/GeV = (0.54 w,Hm

• Linear regression for signal distribution maximum and generated mass

• Test data set (blue)

MC toy with varied cross section (BDT – correctly paired)


cross section / fb2 4 6 8 10 12 14 16 18 20

(m -3

50) /

GeV

0.2−

0

0.2

0.4

0.6

0.8c,Hmw,Hma,Hmt,Hmr,Hm

Mass deviation with TMVA correctly paired optimized

• Result from 10 000 toy MC

cross section / fb2 4 6 8 10 12 14 16 18 20

m /

GeV

∆

0.5

1

1.5

2

2.5

3

c,H m∆w,H m∆a,H m∆t,H m∆r,H m∆

Mass uncertainty with TMVA correctly paired optimized

MC toy with varied cross section (BDT – correctly paired)


• Result from 10 000 toy MC samples

• 0.5 GeV mass precision

• Optimization for correctly paired signal beneficial

cross section / fb2 4 6 8 10 12 14 16 18 20

m /

GeV

∆

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6 stat. cut

stat. cut - corr. paired

BDT

BDT - corr. paired

BDT - corr. paired (w/o BG)

BDT - corr. paired (h)

BDT - corr. paired (sl)

Mass uncertainty with reduced shape method

Result


• 0.5 GeV mass precision • Neutrino four momentum reconstruction with Missing Momentum Method

was used• Missing Transversal Momentum Method has great potential

ChristianDrews StudyofchargedHiggsbosonssearchattheILCforacollisionenergyof1TeV 19

Backup

Cross section


Source: ChargedHiggsBosonproductionatee collidersinthecomplexMSSM:afullone-loopanalysisHeinemeyer,S.andSchappacher,C. Eur.Phys.J. (2016)

ps = 1000GeVps = 800GeV

• 𝜎 ≈ 9 fb with P = (-80%,20%) 10.4 fb• 𝓛 = 1000 1/fb• N = 10400 H± events• Assuming BR(H±->tb) = 90%• BR(t -> bW) = 100%• BR(W -> 2jets) = 67.6%• BR(W –> eν) = 10.75• BR(W –> eν) = 10.57• Hadronic: 5100 events• Semileptonic: 3200 events

Allowed parameter space


Source: A.Arbey etal.“StatusoftheChargedHiggsBosoninTwoHiggsDoubletModels.”In:(2017).arXiv:1706.07414[hep-ph].

Analysis Overview

• Isolated Lepton selection• Reduce beam background by kt-Algorithm• Jet-clustering and flavor tagging (LCFIplus)• Calculating neutrino four-momentum (only semi-leptonic)• Jet-pairing• Extracting signal and background mass shape• Added fit to find Higgs-mass

24.10.17 StudyofchargedHiggsbosonssearchattheILCforacollisionenergyof1TeV

23

ReconstructedH+ andH- masswithrealisticclusteringandpairingwithgeneratorinformation

Find R for kt-Algorithm

mH+ mH�

R:0.1R:0.3R:0.6R:0.8R:1.0R:1.1R:1.2R:1.3R:1.4R:1.5

Chi2 - Jet Pairing (hadronic)

24

w/ooverlay

R:1.3 withoverlay

B-tagefficiency 44.6 42.5 38.0 the4b-jetshavehighestb-tagintheevent

Clusteringworkswell 50.7 49.4 40.2 Foreverycolorsingletthereare 2jetswithamajorfractionfromthissinglet

Pairingworks 27.8 25.0 17.2 Jetpairingagreeswithmajorcolorsingletfractioninjet

�2 =

��(mj1j2j3j4)

2 � (mj5j6j7j8)2

2�2H+

��+✓mj2j3j4 �Mt

�t

◆2

+

✓mj6j7j8 �Mt

�t

◆2

+

✓mj3j4 �MW

�W

◆2

+

✓mj7j8 �MW

�W

◆2

Lepton Selection

• Using the IsolatedLeptonTaggingProcessor• From MarilnReco• Based on MVA

• Open task: reduce false Lepton Tag in hadronic Channel • With event shape or b-tag• But actually the pairing efficiency is not

effected

24.10.17 StudyofchargedHiggsbosonssearchattheILCforacollisionenergyof1TeV

Total(%) w/o tau(%)

LeptonTag 60.3 90.4

Correct Tag 60.0 90.0

FalseLeptonTagged

0.3 0.4

Electron 29.5 89.4(w/o tauandmyon)

Myon 30.3 90.5(w/o tauandelectron)

FalseLeptonTaginhadronic

2.1

Documents

STUDY OF CHARGED HIGGS BOSONS SEARCH AT …epx.phys.tohoku.ac.jp/eeweb/meeting/2017_LCWS_christian.pdfJet pairing • Jet pairing with chi squared minimization • j 1, j 2, j 3and