Z0+- Method to determine hadronic tau efficiencyZtautau W->taunu W->munu W->enu Z->mumu Z->ee...

Z0 → τ+τ− → Method to determinehadronic tau efficiency

Gordon Fischer

Deutsches Elektronen Synchrotron - DESY

Summer school Rathen 2009

Gordon Fischer (DESY) Z0 → τ+τ− 21.09.2009 1 / 22

Schedule

What do I have to do?

Efficiency determination

Embedding method

Summary and Outlook




"....determine the hadronic tau selection efficiencyfrom channel Z0 → τ+τ− → had ` in first data atATLAS...."

I Why?

I How?



Why Tau Leptons?



Why Tau Leptons? Higgs search!

I Higgs boson couples tomass favouring thedecay to the heaviertau-leptons

I for light Higgs masses→ τ τ final statebecomes important

I bb̄ is difficult to measure

I third generation is veryuseful for new physics! Figure: BR for different possible Higgs

masses



Why Tau Leptons? SUSY search!

I τ leptons often finalstate in SUSY models

I τ̃L , τ̃R mixing → τ̃1 , τ̃2

→ τ̃1 and τ productionenhanced

I BR for χ̃02 → τ̃1 τ larger

than for e or µ decays

I τ final states provideinformation for e.g. τ̃masses

I tau decay offersopportunity to measuretau polarization →χ̃0

1,χ̃02, τ̃1 coupling

Figure: typical SUSY chain with τ ’s in thefinal state



Motivation for the channel Z0 → τ+τ− → had `

I in first data Z0 → τ+τ− → had ` is useful (τ ) channel for (τ )trigger studies, with hadronic channel trigger studies notpossible

I most important control channel (together with W → τνand t → τνb) to control the tau reconstruction

I Comparison with Z0 → µ+µ− or Z0 → e+e− allowsefficiency determination using Z-Resonance



different τ decay channels → what does that mean?



important detector moduls for hadronic τ identification

I to select 1 prong and3 prong decays →good tracking system

I τ Jets have a smallerEM-radius than QCDDi-Jets→ for τ Jets and QCDDi-Jets separationgood calorimetersystem required

I QCD background is abig challenge!



Selection cutsI Event cuts:

- Missing ET and Sum ET

(QCD BG reduction)I Particle cuts:

- trigger e10, mu10i- PT for Tau-Jet or Leptoncandidate- |Charge| and η- mT

- different identification cutsI for τ ,` combinations:

- |∆ΦTauJet,Lepton|- invariant Mass window

- opposite minus samesign charge (OS-SS)Assumption: forBackground OS-SS ≈ 0 (forstatistical reasons)⇒ good backgroundrejection

Electron m_T [GeV]0 20 40 60 80 100 120

# el

ectr

on

s

0

500

1000

1500

2000

2500

Electron m_T [GeV]0 20 40 60 80 100 120

# el

ectr

on

s

0

500

1000

1500

2000

2500

Electron mT distribution

Muon m_T [GeV]0 20 40 60 80 100 120

# m

uo

ns

0

500

1000

1500

2000

2500

3000

Muon m_T [GeV]0 20 40 60 80 100 120

# m

uo

ns

0

500

1000

1500

2000

2500

3000

Muon mT distribution

Tau m_T [GeV]0 20 40 60 80 100 120

# ta

us

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

20000

22000

Tau m_T [GeV]0 20 40 60 80 100 120

# ta

us

0

2000

4000

6000

8000

10000

12000

14000

16000

18000

20000

22000Taus mT distribution

Ztautau

W->taunu

W->munu

W->enu

Z->mumu

Z->ee

Figure: mT distribution for final state leptons



τ -efficiencyhadronic τ efficiency is defined as:

ετ→had =Nsel

Z0→τ+τ−

N iniZ0→τ+τ− ·BRτ→had ·BRτ→lep · εkin

Z0→τ+τ− · ετ→lep(1)

possible to substitute N iniZ0→τ+τ−

with

NselZ0→µ+µ− ·

BRZ0→µ+µ−

BRZ0→τ+τ−

εkinZ0→µ+µ− · ε2µ

(2)

Finally for efficiency calculation:

ετ→had=εkinZ0→µ+µ−

εkinZ0→τ+τ−

·Nsel

Z0→τ+τ− ·BRZ0→µ+µ− · εµ

NselZ0→µ+µ− ·BRτ→lep ·BRτ→had ·BRZ0→τ+τ−

(3)



Parameter for L = 100 pb−1

I all BR available from PDG

I kinematic efficiencies (systematic studies for uncertainties of kin.efficiencies upcoming! Assume error O(10 %) just in order to checkwhether this would lead to acceptable order of magnitude of overallerror)

εkinZ0→µ+µ− = 0.26 ± 0.026

εkinZ0→τ+τ− = 0.0230 ± 0.002

I selected number of events

NselZ0→τ+τ− = 219 ± 14

NselZ0→µ+µ− = 21517 ± 147

I Muon efficiency (uncertainty from ATLAS Note)

εµ =

√√√√ NselZ0→µ+µ−

σZ0→µ+µ−·100pb−1

εkinZ0→µ+µ−

= 0.85 ± 0.05



Current values for hadronic efficiency

I Tau-Jet,Tau-Lepton combinations for L = 100 pb−1

Signal OS SS OS-SSselected taus 208 11 197

selected taus truthmatched 201 2 199Background OS SS OS-SSselected taus 96 91 5

I hadronic efficiency εhad for (OS-SS) (uncertainty with errorpropagation)εhad (Signal) = 0.537 ± 0.0642εhad (Signal (truth)) = 0.542 ± 0.0654εhad (Signal + BG) = 0.550 ± 0.073

I effects from selection uncertainties have to be studied in detail


Embedding method

Challenge: determine kinematic efficiency for Z0 → τ+τ− and

Z0 → µ+µ−

I don’t know all detector (and trigger) effects in detail → challenge to

determineεkinZ0→µ+µ−

εkinZ0→τ+τ−

I Idea 1: select µ pairs from Z decay (in real data)

I kinematic behaviour of µ and τ (before decaying) is the same →allowed

I replace µ with τ

I let τ decay and run full reconstruction chain again

I compare number of (kinematic) selected τ and µ → no more initialinformation needed

1Markus Schumacher, Nicolas Möser (Uni Bonn), Martin Schmitz (UniBonn) et al.


Embedding method

recipe for kinematic efficiency determination from dataI select µ pairs from Z0 → µ+µ− with pT > 5 GeV, opposite charge

I determine correction factor for lepton pT −→ ξ = E2−mτ

|p|2 and replace µ

with τ

I let τ decay (TAUOLA) → digitisation → reconstruction → embedding→ Z0 → τ+τ−

I Embedding → ∆R matching of τ to corresponding µ

I all track segments in muon spectrometer are deleted → trackinformation from Z → ττ events

I Track particles and tracks of µ are deleted and replecad by tracksfrom Z → ττ events

I all higher-level objects and missing et will be reconstructed (Offlinereconstruction (e.g. MuTag or STACO)

I run the kinematic selection and ratioεkinZ0→µ+µ−original

εkinZ0→τ+τ−replaced

yields the

required efficiency!


Embedding method

Performance plots

ET for visible τ vs. Full detector energy on:

I case 1: truth level

I case 2: reconstructionlevelwith truth match

I case 3: reconstructionlevel

without truth match

0

2

4

6

8

10

12

14

16

18

20

22

full detector transv. energy [GeV]0 100 200 300 400 500 600

visi

ble

tru

th t

au p

air

tran

sv. e

ner

gy

[GeV

]

0

20

40

60

80

100

120

140

160

180

200Energy Tau vs. Full Energy

Energy_NewTaus_10TeVEntries 79681Mean x 259.6Mean y 47.33RMS x 99.95RMS y 25.25

Energy Tau vs. Full Energy

0

2

4

6

8

10

12

14

16

18


visi

ble

rec

o t

au p

air

ener

gy

mat

ched

[GeV

]

0

20

40

60

80

100

120

140

160

180

200Energy Reco-Tau vs. Full Energy

EnergyReco_NewTaus_10TeVEntries 49479Mean x 268.7Mean y 42.29RMS x 99.93RMS y 26.92

Energy Reco-Tau vs. Full Energy

0

2

4

6

8

10

12

14


visi

ble

rec

o t

au p

air

ener

gy

[GeV

]

0

20

40

60

80

100

120

140

160

180

200Energy Reco-Tau vs. Full Energy without Truth Match

EnergyReal_NewTaus_10TeVEntries 78787Mean x 249.7Mean y 87.77RMS x 92.2RMS y 43.33

Energy Reco-Tau vs. Full Energy without Truth Match


Embedding method

case 2: reconstructed (and truth matched) τ ’sI =⇒ τ pair energy contribution relative small

0

1

2

3

4

5

6

7

8

9


visi

ble

rec

o t

au p

air

ener

gy

mat

ched

[GeV

]

0

20

40

60

80

100

120

140

160

180

200

Energy Reco-Tau vs. Full EnergyEnergyReco_Taus_10TeV

Entries 13153Mean x 268.7Mean y 41.65RMS x 99.74RMS y 26.48

Energy Reco-Tau vs. Full Energy


Embedding method

Performance plots → Embedding

I choose Z0 → µ+µ− →full embedding chain

I compare with regularZ0 → τ+τ− sample(same production)

I plot pT fromreconstructed τ

I compare R = #τ`#µµ for

truth MC andEmbedding:

htempEntries 705Mean 2.064e+04RMS 1.552e+04

Tau_p_T0 20 40 60 80 100 120 140

310×0

10

20

30

40

50

60

70

80htemp

Entries 705Mean 2.064e+04RMS 1.552e+04

Tau_p_T {Tau_p_T>1000 && Tau_p_T<200000}



I

Sample events R ∆R/RMC truth 271±16

2131±46∼ 0.120 ± 0.01 ∼ 0.08

Embedding 13±4122±11

∼ 0.106 ± 0,04 ∼ 0.40


Summary and Outlook

Summary and Outlook

Summary:

I τ and µ kinematic well understood

I technical problems solved

I embedding method works fine (on this level)

Outlook:

I of course more statistic needed

I disussion of further uncertainties (e.g. µ selection, TAUOLA, effectsfrom underlying events..)

I more detailed studies of embedding tool will be done

I .....


Summary and Outlook

Backup


Summary and Outlook

Embedding method in more detailI µµ declared as ττ → four vectors together with four vector of Z boson

are written in HEPEVT formatI ASCII file read in into TAUOLA (→ lets τ decay) and PHOTOS (→

final state radiation)I fed into ATLAS detector simulation, digitisation, reconstructionI Embedding → ∆R matching of τ to corresponding µ

I all track segments in muon spectrometer are deleted → trackinformation from Z → ττ events

I Track particles and tracks of µ are deleted and replecad by tracksfrom Z → ττ events

I all higher-level objects and missing et will be reconstructed (Offline

reconstruction (e.g. MuTag or STACO)


Summary and Outlook

Embedding flow

ATL-COM-PHYS-2009-446.pdf


Z0+- Method to determine hadronic tau efficiencyZtautau W->taunu W->munu W->enu Z->mumu Z->ee...

Documents

Transcript of Z0+- Method to determine hadronic tau efficiencyZtautau W->taunu W->munu W->enu Z->mumu Z->ee...