Jet vetoes and tagging in Higgs-boson
studies at CMS
Tom Cornelis
for the CMS Collaboration
Jet Vetoes and Jet Multiplicity Observables at the LHC, Durham, July 19th 2013
Higgs production modes
t̄
t
g
q
q
g
V
H
t
q
gluon-gluon fusion
V
V
t̄
t
V
q
H
q
Higgs strahlung
q
gvector boson fusiong
g
q
H
t̄
H
t
gt t̄ fusiong
Introduction 2/16
Higgs analyses with jets
I
Higgs analyses are divided in sub-categories, which are mostly
based on jet multiplicities
I
This improves sensitivity and helps to identify production modes,
useful to measure couplings
I
The 2-jet category is specific to the vector boson fusion (VBF)
process
I
VBF topologies could be used in this categorie to discriminate
between signal and background
Introduction 3/16
Vector boson fusion
q
q
W
q
H
W
q
I
Higgs boson is produced in the central rapidity region
I
Two energetic quark jets at large rapidities
I
Jets are color connected with the proton remnant
I
Expect no (or small) hadronic activity between the two tagging jets
Introduction 4/16
Tagging jets
q
q
W
q
H
W
q
I
The VBF tagging jets are found at large rapidities
I
These jets originate from the hadronization of light quarks while
the backgrounds will also contain jets originating from gluons
Tagging jets 5/16
Tagging jets
The forward - backward jet signature is mostly exploited by
I
a large rapidity separation
I
a large invariant dijet mass
s = 8 TeV
CMS Preliminary
Presel. & Trigger
Data (19.0 fb-1)
QCD (× 1.31)
4
10
Z+jets
tt
single top
103
W+jets
VBF H(125)→ bb
2
10
GF H(125)→ bb
Events / 160 GeV
Events / 0.20
CMS Preliminary
105
Data (19.0 fb-1)
QCD (× 1.31)
Z+jets
10
tt
single top
103
W+jets
VBF H(125)→ bb
102
GF H(125)→ bb
10
1
1
2
1.5
Data / MC
Data / MC
Presel. & Trigger
105
4
10
1
0.5
0
s = 8 TeV
106
3
4
5
6
7
8
|∆ηqq|
2
1.5
1
0.5
0
500
1000
1500
2000
2500
3000
3500
4000
Mqq (GeV)
Tagging jets 6/16
Quark - gluon jet discrimination
I
Quarks and gluons have different color interaction
I
This will mirror in different hadronization
At a given energy a gluon jet will, on average:
I
I
I
I
have a higher multiplicity
be angularly wider
have a more uniform energy fragmentation
quark jet
gluon jet
Tagging jets 7/16
Quark - gluon jet discrimination
Likelihood based discrimination on properties of the jet constituents:
I Angular spread in the (η,φ) plane
I
I
I
major axis
minor axis
Multiplicities
I
I
central region: charged multiplicity with PV association
transition and forward region: charged + neutral multiplicity
(pT > 1 GeV)
I
Asymmetry
I
I
Jet Pull:
X i 2
~
pT | ri | ~
r
t = i
jet
i∈jet pT
Jet R: Energy fraction carried by the leading constituent
i )
max(pT
i )
sum(pT
(only for charged in central region)
Tagging jets 8/16
Quark - gluon jet discrimination
Already applied in the VBF H → b b̄ analysis:
Events / 0.04
s = 8 TeV
CMS Preliminary
Presel. & Trigger
105
Data (19.0 fb-1)
QCD (× 1.31)
104
Z+jets
tt
single top
103
W+jets
Events / 0.04
CMS Preliminary
106
Z+jets
4
10
GF H(125)→ bb
tt
single top
103
W+jets
102
GF H(125)→ bb
VBF H(125)→ bb
10
1
1
2
1.5
Data / MC
Data / MC
Data (19.0 fb-1)
QCD (× 1.31)
10
10
1
0.5
0
Presel. & Trigger
5
VBF H(125)→ bb
102
s = 8 TeV
106
0
0.2
0.4
0.6
0.8
1
3rd jet (btag ordering) quark-gluon discriminant
quarks
gluons
2
1.5
1
0.5
0
0
0.2
0.4
0.6
0.8
1
4th jet (btag ordering) quark-gluon discriminant
quarks
gluons
These QG discrimination values were used as inputs to the neural
network
Slightly modified QG discriminator under construction for future
analyses
Tagging jets 9/16
Pile-up jet discrimination
I
PU jets could cause events to migrate between the different jet
categories
I
The forward backward jets in the VBF selections suffer most from
pile-up jet backgrounds
I
Therefore we try to identify these pile-up jets
Tradionally done by cuts on the most sensitive variables:
I
I
P
2 2
i ∆R pT (i)
P
2 (i)
p
i T
P
charged
P candidates∈other PV
charged candidates
I
h∆R 2 i =
I
β∗ =
More recently, a boosted decision tree is constructed using
I
I
I
Track/vertex variables
Shape variables
Multiplicities
Pile-up jet discrimination 10/16
Central jet veto
No (or small) hadronic activity in the rapidity region between the
tagging jets ⇒ Apply central jet veto
ϕ
j
H
j
η
no jets in central rapidity region
CMS uses the central jet veto of 30 GeV in:
I
H → WW
I
H → ττ
Central jet veto 11/16
Zeppenfeld variable
The topology of VBF events is also reflected in the Zeppenfeld variable,
which places an observed pseudorapidity with respect to the tagging
jets:
η(j1 ) + η(j2 )
(1)
Z = η |obs −
2
The Zeppenfeld variable can be measured for
I
the Higgs candidate, e.g. η |obs = η(H), the VBF signal will be
more strongly peaked around 0 with respect to the backgrounds
This is exploited in the VBF H → γγ analysis with a cut on
Z < 2.6
I
a 3rd jet in the event, e.g. η |obs = η(3rd jet), the VBF signal will
show a double peak, contrasting with the peak around 0 in the
backgrounds
Variables sensitive for VBF 12/16
∆φH−jj
I
The VBF H → γγ search uses a cut on the azimutal angle
between the dijet and diphoton systems:
∆φγγ−jj > 2.6
I
If the event only contains the diphoton system and two jets,
momentum conservation leads to ∆φγγ−jj ≈ π
I
Thus, this variable is sensitive to remove events with additional
radiation or PU jets
Variables sensitive for VBF 13/16
Soft hadronic activity
A way to measure central hadronic activity with track jets:
I
Select charged high-purity tracks associated with the primary
vertex
I
Exclude tracks associated with the tagging jets and Higgs decay
products
I
Cluster these tracks into soft track jets with the anti-kT algorithm
I
Only keep track jets between the tagging jets
Soft hadronic activity 14/16
Soft hadronic activity
The central soft hadronic is measured in the VBF H → bb analysis:
CMS Preliminary
Events / 10 GeV
106
s = 8 TeV
Presel. & Trigger
105
Data (19.0 fb-1)
QCD (× 1.31)
Z+jets
4
10
tt
single top
103
W+jets
VBF H(125)→ bb
102
GF H(125)→ bb
10
Data / MC
1
2
1.5
1
0.5
0
0
20
40
60
80
100
120
140
160
180
200
Soft HT (GeV)
The soft hadronic activity was also used as one of the inputs of the
neural network to achieve maximal separation between signal and
background
Soft hadronic activity 15/16
Summary
I
CMS is using VBF topologies in analyses with the 2-jet category
I
A quark-gluon jet discrimination is successfully applied
I
Pile-up jets could be tagged and rejected from the event
I
The low hadronic activity in VBF events is exploited by:
I
I
I
The central jet veto in H → WW and H → τ τ
The ∆φγγ−jj variable in H → γγ
The soft hadronic activity in H → b b̄
Summary 16/16
Back-up
back-up 17/16
References
I
CMS Collaboration, ”Search for the standard model Higgs boson
produced in vector boson fusion, and decaying to bottom quarks”,
CMS PAS HIG-13-011
I
CMS Collaboration, ”Updated measurements of the Higgs boson at
125 GeV in the two photon decay channel”, CMS PAS HIG-13-001
I
CMS Collaboration, ”Search for the Standard-Model Higgs boson
√
decaying to tau pairs in proton-proton collisions at s = 7 and 8
TeV”, CMS PAS HIG-13-004
back-up 18/16
PU jet ID variables
A boosted decision tree is constructed using
I Track/vertex variables:
P
charged candidates∈PV
P
P charged candidates
charged
P candidates∈other PV
I β∗ =
charged candidates
I
β=
I
dZ : distance in the z-axis of the highest pT charged candidate wrt
the PV
nvertices
I
I
Shape variables:
P
2 2
i ∆R pT (i)
P
2 (i)
p
iP
T
I p A (∆R) = 1
T
i∈A<∆R<A+0.1
p
T
P 2
i pT (i)
I pD =
T
pT
I
I
h∆R 2 i =
pTi
Multiplicities:
I
I
Ncharged
Nneutral
back-up 19/16
CMS σ/σSM results
back-up 20/16