h → zz → a promising new channel for high higgs mass sara bolognesi – torino infn and...

H → ZZ →

A promising new channel for high Higgs mass

Sara Bolognesi – Torino INFN and University

Higgs meeting 23 Sept. 2008 – CMS Week

Introduction

≈10×BR(H→ZZ→4), same final state of H→WW→

Analysis code/workflow similar to H→VV• re-use the code (MET corrections)

• use the skimmed sample of H→WW

(CMSSW_1_6_9)

Results up to preselection already shown inHiggs meeting on 9th May:

• resolution and efficiency on signal

• signal and background kinematic

• definition of preselection cuts

High background:

• inclusive ZZ ≈ 15 pb (Pythia)

• inclusive WW ≈ 114 pb (Pythia)

• ttbar ≈ 840 pb (MC@NLO)• Z→ ≈ 830 pb (Pythia) • Z+jets ≈ 5777 pb (Alpgen Chowder)

• ttbar+jets ≈ 836.5 pb (Alpgen Chowder)• W+jets ≈ 58155 pb (Alpgen Chowder)

compared also with

New channel ( × BR ≈ 50 fb) along the line of H→VV:

http://indico.cern.ch/materialDisplay.py?contribId=3&materialId=slides&confId=32886

S. Bolognesi (INFN To) Higgs meeting – 23 September 2008 2

Analysis workflow HLT: 1 or 2 muons isolated or not

Skimming: at least 2 global muons with pT > 20, 10 GeV and ||<2.4

Preselection: • exactly two muons with opposite charge

• 80 GeV < M() < 100 GeV

• MET > 25 GeV after Type 1 and 2 corrections (i.e. Jet Energy Scale and muons corrections)

Selection cuts studied:

• Central Jet Veto or Track Counting Veto

• muon isolation and impact parameter cut

• kinematic selection

Final distributions to search signal excess


(no Higgs mass peak → alternative variable like (ll) in H→WW→ll)

Event yields after preselection

(9.54 ± 0.03)%

(0.203 ± 0.004)%

(1.40 ± 0.03)%

(0.071 ± 0.003)%

(49.7 ± 0.3)%


Track counting veto

Similar results from Central Jet Veto with calorimetric jets

VBF events in signal (20% for mh500) have 2 jets

sign > √back

gain in significance

• supposed to be at high • but in highly boosted events (43% of events has pZ

H>200GeV) one jet can be central(only is boost invariant)

No CUT on TCV


UE activity increases with hard scattering energy scale

• bigger for high mass signal

Muon isolation

Good rejection against W+jets and tt+jets

2 1.5cone cone cone

T T Ttracks ECAL HCAL

p E E in cone (R=0.3) around the second muon (with lower pT)

Very low signal rejection: • VBF jets far away from muons (high (jet-))

• UE recoils with respect to Higgs


Muon tranverse impact parameter

no gain in significance

Same efficiency on signal and Z+jets (which is 97% of the background after preselection)


i.e. |d0/(d0)| < cut

Kinematic

• Signal kinematic dependent on Higgs mass: low mass similar to background

• No samples available for optimization of the cuts as a function of the Higgs mass(however premature at this stage)


All these variables are correlated → not so powerful!

Kinematic: pT()/MET

• scalar Higgs (isotropic H→ZZ decay) → pT()/MET ≈ pTZ1/pT

Z2 ≈ 1 for any mass

Strong rejection (47%) of Z+jets: • high Z pT (Z recoling w.r.t. jets and UE)

• low MET because no neutrinos

• difference between mh200 and mh500 due to MET measurement error

→ related with hard scattering scale (i.e. Higgs mass)→ related with UE

Good behaviour on signal:


Results (1 fb-1)

150-250 GeV

signal 200

signal 500

Z+jets

tt+jets

W+jets

WW

ZZ

significance

MH200: likelihood from kinematic variables → significance 0.2

MH500: 12.5 fb-1 (35 fb-1) for Higgs evidence (discovery)

events (1 fb-1)

Higgs transverse mass • includes all the previous kinematic variables• simple dependence on the Higgs mass

to be used as final reference distribution to search for signal excess

> 400 GeV

22.4 ± 0.2

24003 ± 165

202 ± 7

33 ± 7

51 ± 3

78 ± 3

--- 4.96 ± 0.08

31 ± 5

0.6 ± 0.4

0

0.4 ± 0.2

2.6 ± 0.5

---

0.143 ± 0.001 0.85 ± 0.06


Systematics

• MET: 5% from Z/W comparison* → propagated into analysis 2%

• Muon efficiency (trigger, reconstruction, isolation):measured from single Z data* with negligible uncertainty

• Muon momentum scalemeasured and calibrated from single Z data* → residual systematics due to magnetic field and misalignement is negligible (<1% on Z xsec)

• Jet energy scale and jet reconstruction efficiencyno impact (no CJV used)

Systematics on background yield (luminosity, PDF, NLO) can be avoided with background normalization from data

Experimental systematics

* Towards a measurements of the inclusive W→ and Z→ cross sections in pp collisions at √s = 14 TeV CMS AN 2007/031, PAS 2007/002


Z+jets normalization

Control region: all analysis cuts but MET<25 GeV

has negligible uncertainty: • pollution from other processes 0.02%

• statistical uncertainty 0.2% with 1fb-1

MC extrapolation dominated by MET uncertainty →

3% final uncertainty on Z+jets yield in signal region

signal regionsignal region control region

control regionMC

dataMC

BB B

B

control regiondataB

signal region

control regionMC

MC

B

B2% in each region


ZZ normalization No region where ZZ dominates because Z always overwhelming

→ use single Z process as reference

control region: all analysis cuts on muons, MET not considered

Extrapolation to signal region (= control region + MET related cuts)

dominated by MET uncertainty

→ 4% on

control region dataMC

ZZ ZZZZ Z MC

Z Z

N N

uncertainty negligible:main effects (muon trigger, reco, iso efficiency), which are however small, mostly cancel out

MCZZMCZ

main effect (PDF) mostly cancel out uncertainty negligible:ZZ

Z

uncertainty very small:• 0.2% pollution from other processes

• 0.15% statistical uncertainty with 1fb-1

dataZN

signal regionsignal region control region

control regionMC

ZZ ZZMC

NN N

N signal region

MCN


Conclusion

15.5 fb-1 (43 fb-1) needed for Higgs evidence (discovery) with mass 500 GeV.


Sources signal MH500 background

luminosity

PDF

MET modeling

normalization

MC statistics

Total

10%

10%

0.3%

---

1.6%

14.2% 15.4%

2%

---

15%

---

3% CLB 22.34%

0.76

Open issues Too small MC background samples available: background MC statistics is the main systematics

Samples with different Higgs mass needed to assess the discovery power for different masses (possibly optimizing cut VS mass)

Z inclusive pT (and MET in Z+jets events) must be measured from data to tune MC!!

MC level Z pT (after HLT+skimming) reco Higgs transverse mass(after full analysis selection)


Back-up

Higgs transverse mass


Two possible definitions:

• using only muon info

• using also MET ≈ pT() (energy and direction: px,py)

2 22T TM Z p M

2 22 2 2 2 2

1 2,T T T T TM Z Z p M p M p p

2 22 2 2 2 2

1 2,T T T T TM Z Z p M p M p p

mh200 mh500

MH200 likelihood


Signal reconstruction resolution

Z → : pT resolution

M() resolution

mean = 0.12%

std.dev. = 1.6%

mean = 0.19%

std.dev. = 1.3%

• resolution: gaussian with underestimated mean (recoMET < genMET)

+ not gaussian queue with recoMET > genMET

• importance of the JES and muon corrections

Z → (MET):

MET

MET resolution

= -10%

= 33%

S. Bolognesi (INFN To) Higgs meeting - 23 September 2008 19

Signal VS backgroundsAfter HLT (all muon paths) and skimming (at least 2 mu with pT > 20, 10 GeV and ||<2.4)

M()

signalZ → ttbar inclusiveZZ inclusiveWW inclusive

All histograms normalized to 1

biggest muon pT

2° muon pT

MET (i.e. Z pT)

number of muons

charge of muons

important for Zbb e ZW

HLT, skimming, preselection efficiency on signal

HLT: all muon paths

Skimming: at least muons with pT > 20, 10 GeV and ||<2.4

exactly two muons with opposite charge muon pT > 20, 40 GeV 80 GeV < M() < 100 GeV corrected MET > 25 GeV

Preselection:

acceptance and reco eff. for each 86.0 ± 0.1 %

HLT 91.5 ± 0.2 %

HLT + skimming 71.5 ± 0.3 %HLT+skim

accept×accept( )96.7 ± 0.3 %

HLT + skimming + preselection

40.1 ± 0.3 %

HLT+skim+presel HLT+skim

( )56.1 ± 0.3 %

Signal efficiencies:


Background yield

0.05 0.05 0.1

(only poissonian error = √N)

Additional cuts needed to increase the significance


(CJV, mu isolation and IP, kinematic cuts)

Main backgrounds Z→ and ttbar

preselection cuts efficiency (after HLT+skim)Significance:

signalZ → ttbar inclusiveZZ inclusiveWW inclusive

0.06

isolation efficiency

reconstruction efficiency

trigger (L1×HLT) efficiency

Muon efficiencies from tag & probe


MET systematics

Muon scale calibration


h → zz → a promising new channel for high higgs mass sara bolognesi – torino infn and...

Documents