Tau Physics at CMS - phipsi11.inp.nsk.suphipsi11.inp.nsk.su/talks/FromPhiToPsi_.pdf · Intro...

Tau Physics at CMS

Vladimir CherepanovOn behalf of the CMS Collaboration

III.Phys.Institut B, RWTH Aachen

20.09.2011

Intro Particle Flow TauRec Z→ ττ measurement and tau-ID W→ τν observation Summary

LHC

1 pp collisions at√

s =7 TeV

2 in september 2011 ≈2.5 fb−1 delivered

3 ≈ 107 taus (Z, Wsource)

2 / 30


CMS

3 / 30


Tau lepton

τ properties:mτ = 1.776 GeV

cτ = 87.11µm

leptonic decay: 35.2%

hadronic 1 and 3-prong: 63.1%

significant part of τ momentum escapeswith neutrino.

Tau lepton is a very suitable tool for studyingmany physical processes at LHC

Decay Mode Resonance Mass, MeV/c2 Branching ratio, %

τ → h−ντ 11.6%τ → h−π0ντ ρ 770 26.0%τ → h−π0π0ντ a1 1200 10.8%τ → h−h+h−ντ a1 1200 9.8%τ → h−h+h−π0ντ 4.6%

Total 63.1%

Other hadronic decays 1.7%4 / 30


Particle Flow Algorithm

Particle Flow (PFlow) combines and links signals coming from differentCMS sub-detectors to provide a complete event description

As an output PFlow gives a list of reconstructed particle candidates(e, µ, γ, h±, h0)

Tau lepton reconstruction algorithms start from PFlow candidates

5 / 30


Particle Flow Algorithm (CMS PAS PFT-10-001)

Example of links between tracks and ECAL/HCAL clusters

Unlinked Ecal cluster (photon)

Cluster linked to track

Tracker hit

The link distance is defined as ∆R in the (η,φ) plane between objectspositions. ∆R =

p

∆φ2 + ∆η2

6 / 30

η = −ln(tg θ2)


Particle Flow Performance

Simulated QCD-multijet events:PFlow jet energy response (prec

T - pgenT )/pgen

T and resolution

Mass of PF photon pair. Agreementfor mass within ± 2% of the PDGvalue

see CMS PAS PFT-10-0017 / 30


Tau Reconstruction and identification

The main challenge in tau reconstruction is to discriminate between τhad and QCDjets.

The experimental signatures ofhadronically-decaying taus:

collimated jet

low multiplicity (up to three chargedhadrons and up to two π0’s)

decay products are isolated (requirelow detector activity around tau-jetdirection)

The best performance in terms ofefficiency&fake-rate is achieved byanalyzing PFlow jet constituents andbuilding individual decay modes.

Two main alghorithms are developed.Presented today: Hadron Plus StripsAlgorithm.

8 / 30


Hadron Plus Strips Algorithm (HPS) CMS PAS TAU-11-001

1 Cluster photons within the PFJet intostrips accounting for possible broadeningdue to photon conversions

2 Combine charged particles in the jet withstrips and reconstruct individual τh decaymode: π±ντ ,π±π0ντ , (πππ)±ντ

3 Require charged and strips to be containedwithin a cone ∆R = 2.8/Pτ

T

4 Most isolated decay “hypothesis” withcompatible visible mass is given preference

9 / 30


Hadron Plus Strips Algorithm (HPS)

10 / 30

Expected reconstruction efficiencyfor three “working-points” definedby isolation tresholds

loose

medium

tight

Matching efficiency betweengenerated taus and reconstructed byHPS loose working point ( >80% forall three channels )


Hadron Plus Strips Algorithm (HPS)

The tau fake-rate from jets in threedifferent samples.

QCDj: jet pT > 15GeV, |η| < 2.5

QCDµ: non-iso µ, MET < 40 GeV

Wj: iso µ, MET > 50 GeV

HPS “loose” efficiency ≈ 50% at 1%fake-rate

11 / 30


Z→ ττ and tau-ID estimation

two nearlyback-to-back taus

tag&probe with 4combinations:

τµτhad

τeτhad

τeτµ

τµτµ

τhadτhad difficultbut doable

12 / 30


Z→ ττ and tau-ID estimation (CMS PAS TAU-11-001)

The reco and ID efficiency of tau algorithm is obtained from data using a tag&probeapproach.

Preselection

only one iso µ with pT > 15 GeV,|η| < 2.1

iso tau-jet candidate with pT > 20GeV, |η| < 2.3

leading track within tau-jet with pT

> 5 GeV

MT < 40 GeV

MT (µ, EmissT

) =q

2pµT

EmissT

(1 − cos∆φ)

Efficiency

apply TauID

fit visible µτjet invariant mass usingsignal and background templates

ǫ =NZ→ττ

pass

NZ→ττpass +NZ→ττ

fail

13 / 30

FAILPASS


Z→ ττ and tau-ID estimation

The reco and ID efficiency of tau algorithm is obtained from data using a tag&probeapproach.

14 / 30

FAILPASS


Z→ ττ cross section measurement (CMS-EWK-10-013)

Measurement of Z→ ττ cross section wtih 36pb−1 combining 4 channels:τµτhad channel, τeτhad channel, τeτµ channel and τµτµ channel.

Selection of τlτhad

single lepton trigger

iso µ or e with pT > 15 GeV and|η| < 2.1

associated τ oppositely charged

pτT

> 20 GeV and |η| < 2.3

EmissT

and MT (l, EmissT

) < 40 GeV

lepton isolation:

sum of transverse momentum of tracks,photons and neutrals within a cone of ∆ R= 0.4/0.3 divided by lepton pT is requiredto be less that 0.1/0.08.

15 / 30


Z→ ττ cross section measurement

QCD background estimation from data

ABCD method OS/SS and lepton isolation

C

BA

D

OS SS

Iso

Non-Iso NA = NBNCND

EWK and tt̄ fromMC

W + jets contribution:

extrapolated from W dominatedregion (MT > 60 GeV)

shape from MC

Results for Z→ τlτhad : visible invariant mass for τµτhad and τeτhad channels.

16 / 30



Selection for τµτµ:

two oppositely charged muons withpT > 19 GeV and pT >10 GeVwithin |η| < 2.1

EmissT

< 50 GeV

∆φµµ > 2

multivariate likelihood to suppressDY.

Selection for τeτµ:

pe,µT

> 15 GeV with |η| < 2.1

opposite charge

MT (e, EmissT

) < 50 GeV

MT (µ, EmissT

) < 50 GeV

17 / 30



Total cross section:

σ(pp → ZX ) × B(Z → τ+τ−) = N

AǫB′L

– A geometrical acceptance– N = Nfit(1 − fout) corrected fraction of signal events outsidethe generator-level mass window

Measured values:

Main systematicuncertainties

– Tau ID 24%– A 3.5%

18 / 30



combined result:

σCMScomb

(pp → ZX ) ×B(Z → ττ) = 1.00 ± 0.05(stat) ± 0.08(syst) ± 0.04(lumi) nb

σtheor = 0.972 ± 0.042 nb

Total cross section of Z→ ττ in consistent with theoretical expectation andwith result recently published by the ATLAS Collaboration:σATLAS

comb(pp → ZX ) ×B(Z → ττ) = 0.97 ± 0.07(stat) ± 0.06(syst) ± 0.03(lumi) nb

(arxiv: 1108.2016v1)19 / 30


W→ τν observation

W→ τν candidate:– isolated jet

– large EmissT

20 / 30



Selection for analysis with 18.4 pb−1

τh + Emiss trigger

τh with pτT > 30 GeV + plead.track

T

> 15 GeV

e/µ veto

EmissT > 35 GeV

RHT = pτT /ΣjetpT > 0.65

Background estimate:

EWK from simulation

QCD estimated from data.ABCD method withRHT and Emiss

T

TmissE

HTR

AB

C D

Signal Region

0.65

35 GeV

> 0.65HTR

< 35 GeVTmissE

< 0.65HTR < 0.65HTR

> 35 GeVTmissE < 35 GeVT

missE

21 / 30



Selection for analysis with 18.4 pb−1

τh + Emiss trigger

τh with pτT > 30 GeV + plead.track

T

> 15 GeV

e/µ veto

EmissT > 35 GeV

RHT = pτT /ΣjetpT > 0.65

Background estimate:

EWK from simulation

QCD estimated from data.ABCD method withRHT and Emiss

T

Observation:

372 events in data selected174 ± 3 expected signal from MC217 ± 20 observed signal with nosystematics assessed!

CMS PAS EWK-11-002

22 / 30


Summary I

Advanced tau reconstruction algorithm has been developed andsuccesfully commisioned with data collected

HPS shows a good performance in terms of efficiency vs. fake-rate.With an efficiency of ≈ 50% only a few percent fake-rate from QCDjets

Tau ID efficiency is measured with a precision of 24% (today 6%)

The Z→ ττ cross section is measured in four channelsσcomb(pp → ZX ) × B(Z → τ

+τ−) = 1.00 ± 0.05(stat) ± 0.08

(syst) ± 0.04(lumi) nbin good agreement with SM expectation as well as with Z→ ll crosssection

Clear signal is observed for W→ τhadν in 18pb−1

Many other interesting results with tau in a wide range of physics arealready available

23 / 30


Summary II

— Search for Neutral Higgs Bosons Decaying to Tau Pairs with 1.6 fb−1

(CMS PAS HIG-11-020)

— First measurement of the tt̄ production cross section in the dilepton channelwith 1.09 fb−1

(CMS PAS TOP-11-006)

— Measurement of the WW, WZ and ZZ cross section with 1.1 fb−1

(CMS PAS EWK-11-010)

— Search for new physics with same-sign isolated dilepton with 0.98 fb−1

(CMS PAS SUS-11-010)

— Search for New Ditau Resonances with 36 pb−1

(CMS PAS EXO-10-022)

24 / 30


BACKUP

25 / 30


Tau Neural Classifier Algorithm (TaNC)

1 signal objects from “schrinking-cone” taualgorithm

2 consider gammas pair with mass closest tomπ0 as π0 candidate

3 ensemble of neural networks to decide oneof the five decay modes

26 / 30



Systematic uncertainties:

27 / 30



fit visible mass distributions using shapes from MC or data

background normalizations correspond to estimated:

28 / 30



Likelihood contours for the join parameter estimation of the cross section adnthe τ ID.

DATA/MC correction factor for tau ID efficiency: 0.93 ± 0.09.

29 / 30


H → ττ (CMS PAS HIG-11-020)

Observed and expected limit on σHB(H→ ττ ) normalized to SM expectation.

30 / 30

Tau Physics at CMS - phipsi11.inp.nsk.suphipsi11.inp.nsk.su/talks/FromPhiToPsi_.pdf · Intro...

Documents

Transcript of Tau Physics at CMS - phipsi11.inp.nsk.suphipsi11.inp.nsk.su/talks/FromPhiToPsi_.pdf · Intro...