First Look at H→ WW → lepton + Ω Analysis Ajay Kumar ...indiacms/indiacms-meetings/december-2011/ajay...First Look at H→ WW * → lepton + Ω h Analysis Ajay Kumar , Arun Kumar,

First Look at H→ WW*→ lepton + h

Analysis

Ajay Kumar , Arun Kumar, Dr. Kirti Ranjan

University of Delhi

India CMS Meeting, 22-23 December 2011

@ University of Delhi, New Delhi 1

Overview

Motivation

Introduction

Samples and Software

Event Selection

Data/MC comparison

First Look at Boosted Decision Tree (BDT)

Summary and Future Plan

2

Motivation

*H →WW*

2l2 is well explored channel at CMS in the final states ee,

,e+MET in the mass range 110 to 600GeV. *Highest BR above higgs mass of 125GeV

*Relatively good S/B Ratio

* More sensitivity to di-leptonic channels by studying l+h final state.

* More challenging because of tau & MET in final state. * In addition to H-> & H-> ZZ->4l channels, CMS has seen modest excess of events in region 115 -127GeV from H-> , H->WW->2l2 and H ->bbbar. Additionally, ATLAS has also seen 1.4 in H->WW->2l2nu at Mh=126 GeV.

3

Introduction: (1/3) Higgs Production at LHC

4

Decay modes & BR (2/3)

5

2/3

1/3 Low Mass region Mh<130 GeV H , H , H bb are explored. Mass Region M->115-130GeV : needs to explored extensively now.

6

Introduction (3/3)

Signal: H→WW*→ l+ h where l is either e or µ and decays

hadronically. At present, considered only electron with gluon-gluon

fusion production at Higgs Mass of 160 GeV

Muon, VBF mode will be added.

whole mass range will be added.

Backgrounds:

W+Jets : Most dominant background ,due to jet faking lepton

TTJets : Second dominant

Diboson a.) WW : Irreducible

b.) WZ : when one lepton from Z is lost

c.) ZZ : minimal contribution, lepton faking as tau

Drell-Yan : fake MET

7

Samples and Software

Data:

/TauPlusX/Run2011A-May10ReReco-v1/AOD /TauPlusX/Run2011A-PromptReco-v4/AOD /TauPlusX/Run2011A-05Aug2011-v1/AOD /TauPlusX/Run2011A-PromptReco-v6/AOD

JSON: Cert_160404-173692_7TeV_PromptReco_Collisions11_JSON.txt (~2.1 fb-1)

S/W : For Data : CMSSW_4_2_5 For MC : CMSSW_4_1_8

Physics Analysis Toolkit (PAT) objects used.

Spring 11 MC samples are used.

8

MC samples (details)

Samples (σ) * (B.R) (pb) Processed Events

Skimmed

Events

Expected

Events** at

L = 2.036 fb-1

Scaling

factor at

L = 2.036 fb-1

ggH->WW->e+ 0.1029 1.1E+5 7.2E+4 138 0.00191

W+Jets 31314 (NNLO) 2.1E+6 7.8E+5 2.4E+7 30.67

TTJets 157.5(NLO &

NNLL) 1.1E+6 9.0E+5 27.0E+4 0.30

WW 43 (NNLO) 2.1E+6 1.2E+6 5.1E+4 0.04246

WZ 18.2 (NLO) 2.1E+6 1.1E+6 2.0E+4 0.01757

ZZ 5.9 (NLO) 2.1E+6 4.3E+5 2.5E+3 0.00570

Drell Yan 3048 (NNLO) 1.5E+6 7.2E+5 3.0E+6 4.14

** skimming: at least one electron and one tau in |h| < 2.5

9

Event Selection (1/2) Trigger

Lepton +Tau Lowest threshold unprescaled triggers have been used. Same as used in H→ analysis.

HLT_Ele15_CaloIdVT_CaloIsoT_TrkIdT_TrkIsoT_LooseIsoPFTau15_v* HLT_Ele15_CaloIdVT_CaloIsoT_TrkIdT_TrkIsoT_LooseIsoPFTau20_v* HLT_Ele18_CaloIdVT_CaloIsoT_TrkIdT_TrkIsoT_LooseIsoPFTau20_v* HLT_Ele18_CaloIdVT_CaloIsoT_TrkIdT_TrkIsoT_MediumIsoPFTau20_v* HLT_Ele15_CaloIdVT_CaloIsoT_TrkIdT_TrkIsoT_TightIsoPFTau20_v*

No Trigger in MC till now. Primary vertex

Primary Vertices reconstructed with DA Algorithm . Ndof >= 4.0 && Rho < 2.0mm && abs(z) < 24cm

Tau Selection

PFTaus reconstructed with HPS algorithm has been used. All prong decays considered. |η| < 2.5 and pT > 20 GeV AgainstElectronTight (mva based discriminator) AgainstMuonLoose (tau lead track not matched to chamber hits) LooseIsolation

10

Event Selection (2/2)

Electron Selection

|η| < 2.5 , pT > 20 GeV , 1.4442 < |h| < 1.566 excluded Electron ID VBTF WP80 has been used (details in back up)

Relative Combined Isolation < 0.1 (fast-jet corrected)

Vertexing cuts : |dZ|< 0.1cm and |dxy|<0.02cm

Further cuts :

Events having exactly 1 lepton + 1 tau has been selected after applying this

selection on leptons and taus.

R (e, tau) > 0.7 , opposite charge

PF Jets with L1 and L2 Corrections are selected with pT > 10GeV and dR

from lepton and tau > 0.5

Projected MET > 20 GeV

Pre-selection = Skimming + Tau Selection + Electron Selection + Further cuts

Pile Up Interaction Distributions

11

Red: MC Black: Data

Red: MC Black: Data

Without Pileup Reweighting With Pileup Reweighting

No. of PU interactions No. of PU interactions

PU-Reweighting has been applied for in-time PU only using the official recipe mentioned on twiki : https://twiki.cern.ch/twiki/bin/viewauth/CMS/PileupMCReweightingUtilities

12

DATA/MC Comparison Plots (1/3)

Electron PT Electron Eta Electron Phi

Red : W+Jets Violet : DYJets Cyan : TTJets Yellow : WW Blue : WZ Green : ZZ Black line : Signal (glu-glu) Black Dots : Data

13


Tau PT Tau Eta Tau Phi

Red : W+Jets Violet : DYJets Cyan : TTJets Yellow : WW Blue : WZ Green : ZZ Black line : Signal (glu-glu) Black Dots : Data

14


Dphi (e,tau)

Dphi (e,MET) Dphi (tau,MET) Minimum Transverse Mass of W (GeV)

Projected MET (GeV) PF MET (GeV)

15

Boosted Decision Tree (TMVA based)

Cuts Applied: Ele PT > 10 GeV ,Tau PT > 10 GeV, ProjMET >10GeV

Inclusive jets input variables Distribution

Blue - Signal Red – combined Background

Input variables

chosen for

training, testing

Ele PT

Min(MT)

of W Proj_MET

dPhi (e,)

Tau PT

16

BDT output

alljets

= 0 jet = 1jet > 1 jet

17

Zero jet One jet >one jet

Signal (S) .004 .015 .006

Total bkg (B) .043 .043 .603

S/B .089 .36 .010

S/sqrt(S+B) .018 .064 .007

Signal and Background

After putting BDT cut

No optimization done

18 18

Summary

☺ Had first look at H→WW→ l+ analysis. ☺ Data/MC comparison plots are shown for various kinematic and other discriminating variables. No major discrepancy seen. ☺ TMVA BDT was applied on 5 input variables

Future Plan

1. Try to optimize our selection. More variables will be added to BDT -- include more discriminating variables like b-tagging etc to reduce tt-jets contamination.

2. Include VBF signal and also Mu+tau to increase sensitivity.

3. For estimating backgrounds we will use data-driven techniques.

4. Need to switch to Summer 11 MC and ultimately to Fall 11 MC with whole 2011 data.

5. Volunteered for Trigger and DQM online shifts for 2012 Run.

19

20

21

WP80

22


Invariant Mass of (e,tau) Transverse Mass of Higgs

23 23

TMVA (Toolkit for Multivariate Data

Analysis)

BDT (boosted decision tree) method chosen

Pile UP Reweighted

Signal to background ratio maximization

Input variables chosen for training, testing

Electron Transverse Momentum (PT)

Tau Transverse Momentum

Azimuthal angular separation of lepton and tau (dPhi)

Minimum Transverse Mass of W boson

Projected Missing Transverse Energy

Sensitivity study in following Jet bins:

All jets

Zero jet

One jet

More than one jets

24

BDT Working

25

Color scheme for overlapping plots :

Color scheme for stack plots:

S/B maximization in different jet bins

with different cuts

Note: Signal significance = s/sqrt(signal+background) extreme signal significance = signal/sqrt(background)

26

(c) (b) (a)

a.) Ele PT>10, Tau PT>10, ProjMET >10GeV b.) Ele PT>20, Tau PT>20, ProjMET >10GeV c.) Ele PT>20, Tau PT>20, ProjMET >20GeV

S/B maximization for inclusive jets with

different cuts

27

S/B maximization for 0-jet with

different cuts


(a) (b) (c)

28

S/B maximization for 1-jet with

different cuts


(a) (b) (c)

29

S/B maximization with > 1- jet with

different cuts


(a) (b) (c)

30

Maximum a b c

signal(s) .015 .006 .029

bkg(b) .12 .085 3.7

s/b .127 .067 .008

s/sqrt(s+b) .042 .019 .015

Tables for various jet-bins Inclusive Jets

Maximum a b c

signal(s) .004 .004 .006

bkg(b) .043 .043 .043

s/b .089 .089 .135

s/sqrt(s+b) .018 .018 .026

Zero Jets

Maximum a b c

signal(s) .015 .011 .002

bkg(b) .043 .361 .06

s/b .360 .032 .032

s/sqrt(s+b) .064 .019 .008

One Jet

Maximum a b c

signal(s) .006 .012 .010

bkg(b) .603 1.53 5.19

s/b .010 .008 .002

s/sqrt(s+b) .007 .009 .004

More than One Jets

Correlation Matrix

32

Correlation Matrix

33

Training and Testing Plot

34

BDT Training, Testing, Correlation scatter plots

35

BDT Control Plots

Anatomy of the excess Best fit /sm of the various channels

36

CMS New Limits to Higgs Mass

The presence of the SM Higgs boson below 127GeV has not been excluded because of a modest excess of events in the region between 115 and 127GeV. The excess at low mass is produced by a broad excess driven by the low resolution channels (H2TT, H2WW, H2BB), modulated by the localized excesses seen by the high resolution channels (H2GG and H2ZZ). CMS has established a new 95% CL exclusion limits: 127GeV-600GeV. The excess is most compatible with a SM Higgs hypothesis in the vicinity of 124 GeV and below, but the statistical significance (2.6σ local and 1.9σ global after correcting for the LEE in the low mass region) is not large enough to say anything conclusive.

37

38

39

We cannot exclude the presence of the SM Higgs boson below 127GeV because of a modest excess of events in the region between 115 and 127GeV. The excess at low mass is produced by a broad excess driven by the low resolution channels (H2TT, H2WW, H2BB, center), modulated by the localized excesses seen by the high resolution channels (H2GG and H2ZZ, right).

Limits on /sm (CLs method)

95%CL: obs 127-600, exp:117-543 GeV

40

41

ATLAS With current data set ATLAS has excluded 112.7-115.5GeV and 131-453GeV (except for 237-251GeV) ATLAS is now competing with LEP's low mass results! There is a large deviation in p0-values at 126GeV. 1.9e-4, or an excess of 3.6sigma (gamma gamma 2.8 has sigma, ZZ* has 2.1, WW* has 1.4sigma)

CMS CMS has established a new 95% CL exclusion limits: 127GeV-600GeV. The excess is most compatible with a SM Higgs hypothesis in the vicinity of 124 GeV and below, but the statistical significance (2.6σ local and 1.9σ global after correcting for the LEE in the low mass region) is not large enough to say anything conclusive.

Current Status of higgs searches

42

43

Measure Background process

43

Drell yan process

ttbar process

W+Jets process

44

HLT_IsoMu12_LooseIsoPFTau10_v* HLT_IsoMu15_LooseIsoPFTau15_v*

For Muons+Tau:

|η| < 2.4, pT > 20GeV

Number of Hits > 10

Pixel hits >= 1

Matched stations >= 2

(PT)/PT < 0.1

2 < 10

Isolation(rho-corrected) < 0.15

Muon Selection:

45

CaloID_VT: H/E hh TrkId_T | h | CaloIso_T EcaiIso/ET HcaiIso/ET TrackIso_T TrkIso/ET

46

IsoECAL : The sum of electromagnetic transverse energy of the crystals lying in a cone of ΔR = 0.4, centred around the super-cluster with a veto cone (ΔRi = 3.5 crystals) and eta-slice (Δη = 2.5 crystals) should be less than the threshold value. IsoHCAL : The sum of hadronic transverse energy of all the particles in the HCAL towers in a hollow cone with an inner radius of ΔRi = 0.15 and an outer radius of ΔRo = 0.4 centred around the super-cluster should be less than a threshold value. IsoTrk : The sum of transverse momenta of all the tracks in a full cone (ΔR = 0.4) centered around line joining the primary vertex to the cluster should be less than a threshold value. H/E : - The fraction of hadronic energy to the total electromagnetic energy inside a cone. Σiηiη : - The transverse shape of the electromagnetic cluster.

Some definitions

47

PFTaus reconstructed with HPS algorithm has been used. All prong decays considered. |η| < 2.5 and pT > 20 GeV DecayMode Finding AgainstElectronTight: PFemva <-0.1, tau is not in ECAL crack and a.)if tau has no EM constituents H/P >0.08 B.)if tau has EM constituents fbrem<0.99ORmt>0.55 AgainstMuonLoose LooseIsolation

48

PAT summarizes information for you: • The reco::Candidate is a base class common to all kind of “particles” • It has a lot of information from different subdetectors and reconstruction algorithms Facilitated Access to Event Information • PAT objects summarize this information which is distributed over different collections • When you are using PAT it is just calling a member function to get this information!

PAT

First Look at H→ WW → lepton + Ω Analysis Ajay Kumar ...indiacms/indiacms-meetings/december-2011/ajay...First Look at H→ WW * → lepton + Ω h Analysis Ajay Kumar , Arun Kumar,

Documents