Marco De Mattia, Tommaso Dorigo, Mia Tosi, Pietro Vischia

Universita’ degli Studi di Padova & INFN

October 18th, 2010

Status Report:Status Report:HHZZZZ(*)(*)qq(bb) via VBFqq(bb) via VBF

• studies @7TeV• pre-selection• tag-jets selection BDT• high-mass Higgs selection

• leptonic Z selection• hadronic Z selection b-tagging

• conclusion&plans

2

analysis strategy

high-mass Higgs both Zs are• on-shell• boosted

look for• a Z [ kinematics]• a Zbbbar [b-tag+jet kinematics]

high-mass Higgs both Zs are• on-shell• boosted

look for• a Z [ kinematics]• a Zbbbar [b-tag+jet kinematics]

low-mass Higgsone Z is on-shell and @restwhile, the other is off-shell

look for• an on-shell Z [ kinematics]• an off-shell Zbbbar [b-tag]

low-mass Higgsone Z is on-shell and @restwhile, the other is off-shell

look for• an on-shell Z [ kinematics]• an off-shell Zbbbar [b-tag]

0-order pre-selection[physics objects and topology]

0-order pre-selection[physics objects and topology]

tag-jets[signal kinematics by BDT]

tag-jets[signal kinematics by BDT]

jetZ1

jetZ2

jetT2

jetT1

longitudinal view

muon1

muon2

transversl view

3

samples Spring10• produced in CMSSW_3_5_6• generated @s=7TeV w/o pile-up• fully simulated w/ IDEAL geometry• ntplized

signal datasetNLO (fb)

H (ZZqq)

NLO (fb)

H (ZZbb+eebb)

mH=130GeV/c2

~2.16 ~0.93

mH=150GeV/c2

~3.86 ~1.67

mH=500GeV/c2

~1.07 ~0.46

signal samples

•VBF [PYTHIA6]130,150,500 GeV/c2

+200,300 GeV/c2 [private]

number of expected eventsin Lint=1fb-1 is ~0 !!!

[integrated luminosity in 2010+2011]

number of expected eventsin Lint=1fb-1 is ~0 !!!

[integrated luminosity in 2010+2011]

•Z+jets - SHERPA [M10to50 + M50to10000] [MADGRAPH one has cuts @GEN]•Zbb - MADGRAPH VQQ•ttbar - MADGRAPH TTjets•VV - PYTHIA6 ZZincl - PYTHIA6 WZincl - MADGRAPH VVJets

bkg samples (official)

• only limits on the Higgs mass• only limits on the Higgs mass

4

pre-selections

@ 0-order ask for

at least 2 muons w/• pT 15 (10) GeV/c• q1 q2 < 0• m> 12.GeV/c2

at least 4 (PF)jets w/• pT 15 (30) GeV/c in the barrel (forward) region

sample # of eventsefficiency

(%)

HZZqqll (130)

12.3k/90.1k 13.65±0.11

HZZqqll (150)

16.6k/91.8k 18.10±0.13

HZZqqll (500)

23k/84k 27.37±0.15

Z+jets 12k/2.1M 0.01±0.01

VQQ+jets 37k/2.9M 1.25±0.01

TTbar+jets 31k/1.4M 2.23±0.01

WZ 869/118k 0.74±0.02

ZZ 12k/140k 1.59±0.03

all backgroun

ds93k/6.6M 1.41±0.01

5

tag-jets selection: BDT

known the right combination [jet-parton matching w/in R0.5]now, it is possible to

exploit the kinematics and choose the tag-jets pair applying a multivariate method [BDT]

trained to distinguish between two different categories:

• signal-like events: all matched jets to tag-partons

• background-like events: wrong [random] choice

|max| jj1x2 mjj

6

tag-jets selection: BDT

known the right combination [jet-parton matching w/in R0.5]now, it is possible to

exploit the kinematics and choose the tag-jets pair applying a multivariate method [BDT]

trained to distinguish between two different categories:

• signal-like events: all matched jets to tag-partons

• background-like events: wrong [random] choice

BDT output@70%

of signal efficiency

mH=130GeV/c2 mH=500GeV/c2

7

tag-jets selection: BDT cut [mH=500GeV/c2]

sample# of

eventsefficiency

(%)

HZZqqll (500)

5.4k/23k(84k)

77.75± 0.75(4.99±0.

75)

Z+jets192/12k(2.1M)

0.01±0.01(0.01±0.01)

VQQ+jets797/37k(2.9M)

2.17±0.08(0.03±0.01)

TTbar+jets1074/31k

(1.4M)3.46±0.10

(0.08±0.01)

WZ21/869(118k)

2.42±0.52(0.02±0.01)

ZZ33/2k(140k)

1.48±0.26(0.01±0.01)

all backgroun

ds

2117/83k(6.8M)

2.54±0.05(0.03±0.01)

signal efficiency (%)

8

Z candidates selection

muon ID - identified as both Global and Tracker Muon - 2/ndof < 10 [global fit] - # tracker Hits > 10 - # muon-detector Hits > 0 - dxy < 0.2 cm

Zselection - opposite charge muons - pT(1) > 15 GeV/c - pT(2) > 10 GeV/c - || < 2.4 - mclosest to mZ - m≥ 12 GeV/c2

L=30fb-1 L=30fb-1

L=30fb-1 L=30fb-1

L=30fb-1

after pre-selection+BDT+2b-tags

9

Z selection: mass window optimization

after (2)b-tag cut,no sufficient statistics for optimizing the Z mass window

scale distributions before this cut to statistics after it

after (2)b-tag cut,no sufficient statistics for optimizing the Z mass window

scale distributions before this cut to statistics after it

L=30fb-1

after pre-selection+BDT

10

Z selection: mass window optimization

q-value:= S/√B

1.70

signal efficiency (%)

81

background suppression

(%)77

[83;99][83;99]

L=30fb-1

scaled to after b-tag cut

11

Z selection: mass window optimization

sample# of

eventsefficiency

(%)

HZZqqll (500)

4319/5375(84k)

80.35± 0.54(5.16±0.

76)

Z+jets143/192(2.1M)

74.48±3.15(0.01±0.01)

VQQ+jets587/797(2.9M)

73.65±1.56(0.02±0.01)

TTbar+jets79/1074(1.4M)

7.36±0.80(0.01±0.01)

WZ17/21(118k)

80.95±8.57(0.01±0.01)

ZZ21/33(140k)

63.64±8.38(0.01±0.01)

all backgroun

ds

847/2.1k(6.8M)

40.01±1.06(0.01±0.01)

L=30fb-1

scaled to after b-tag cut

[83;99][83;99]

≥ Nb-tag

events w/

#b-quarks ≥ 1

12

b-tagging

CMS develops several b-tagging algorithms….. but, WHICH ONE TO CHOOSE ?!?CMS develops several b-tagging algorithms….. but, WHICH ONE TO CHOOSE ?!?

considering as discriminator cut [for each algo]

the value|b-tagging 60% [on signal],

look@ different performance in terms of S/N, paying attention on the b-jet multiplicity

≥ Nb-tag

CombinedSecondary

Vertex

all events

≥ Nb-tag

13

b-tagging efficiency

≥1 b-tags ≥2 b-tags

sample# of

eventsefficiency

(%)# of

eventsefficiency

(%)

HZZqqll (500)

860/4319(84k)

19.91±0.61

(1.03±0.03)

276/4319(84k)

6.39±0.37(0.33±0.0

2)

Z+jets5/143(2.1M)

3.50±1.54(0.01±0.0

1)

1/143(2.1M)

0.69±0.69(0.01±0.0

1)

VQQ+jets139/587(2.9M)

23.68±1.76

(0.01±0.01)

9/587(2.9M)

1.53±0.51(0.01±0.0

1)

TTbar+jets43/79(1.4M)

54.43±5.60

(0.01±0.01)

7/79(1.4M)

8.86±3.20(0.01±0.0

1)

WZ1/17

(118k)

5.88±5.71(0.01±0.0

1)

0/17(118k)

0.00±0.01(0.00±0.0

1)

ZZ3/21

(140k)

14.29±7.64

(0.01±0.01)

0/21(140k)

0.00±0.01(0.00±0.0

1)

all backgroun

ds

191/847(6.8M)

22.55±1.46

(0.01±0.01)

17/847(6.8M)

2.01±0.48(0.01±0.0

1)

L=30fb-1

14

Zqq selection

jet ID - R(j,Z) ≥ 0.5 - pT(j1) ≥ 50 GeV/c - pT(j2) ≥ 25 GeV/c

Zqq selection

CENHIGHPT VBF BTAG ZKIN ZZKIN

highest-pT jetsmin

(j1,j2)max

(j1j2, 12)w/ ||≤2.5between tag-jets

b-tagged

15

Zqq mass reconstruction

L=30fb-1

CENHIGHPTVBF

BTAG

ZKIN

ZZKIN

16

4-bodies mass reconstruction

L=30fb-1

CENHIGHPTVBF

BTAG

ZKIN

ZZKIN

17

4-bodies mass reconstruction

L=30fb-1

requesting

(2)b-tagrequesting

(2)b-tag

CENHIGHPTVBF

BTAG

ZKIN

ZZKIN

18

addendum

the search ofa high-mass Higgs boson into ZZ*in its semi-leptonic final state w/ spass through

an almost standard Z-> extraction looking @ events

• w/ high multiplicity jets [>=4]• w/ b-tagged jets -> tracker ;)• w/o MET -> commissioning PAS [JINST]

19

BACKUP

20

ntuplizer

produce light-ntuplew/ the most complete set of information• event and trigger• MCTruth• Muons• Jets [different clustering algorithms (AK,IC)] both Calo and PF• Electrons• MET• Vertices

•event and trigger•genPartciles

•GenJets•GenMet•CaloJets•PFJets

•globalMuons•CaloMet

•GfsElectron•PrimaryVertex

simpleNtple

21

tag-jets selection: BDT cut [mH=130GeV/c2]

sample# of

eventsefficiency

(%)

HZZqqll (130)

/()

±(±)

Z+jets/()

±(±)

VQQ+jets/()

±(±)

TTbar+jets/()

±(±)

WZ/()

±(±)

ZZ/()

±(±)