Search for the Standard Model Higgs Boson in the
Z Gamma Channel
by
Kevin Singh
Submitted to the Department of Physics in partial fulfillment of the
Requirements for the Degree of
BACHELOR OF SCIENCE
at the
ARCHNES
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
OF TECHNOLOGY
June 2013
© 2013 KEVIN SINGH.
All Rights Reserved
L BRARIES
The author hereby grants to MIT permission to reproduce and to
distribute publicly paper and electronic copies of this thesis document
in whole or in part.
Signature of Author
d2
Certified by
Accepted by
Department of Physics
May 10, 2013
It)
/Professor
Christoph Paus
Thesis Supervisor, Department of Physics
-
Professor Nergis Mavalvala
Senior Thesis Coordinator, Department of Physics
2
Search for the Standard Model Higgs Boson in the Z
Gamma Channel
by
Kevin Singh
Submitted to the Department of Physics
on May 10, 2013, in partial fulfillment of the
Requirements for the Degree of
BACHELOR OF SCIENCE
Abstract
The Higgs decay into a photon and a Z boson, with the Z boson decaying into an
electron-positron pair (electron channel) or muon-antimuon pair (muon channel), allows for accurate reconstructions of the Higgs boson mass and measurement of the
Higgs to Z-y coupling. We explore selection criteria for the photon and the two leptons
and provide preliminary observed and expected limits for the Higgs boson production
cross section in the mass range from 120 GeV to 150 GeV. The data used in this
analysis was collected with the CMS detector and corresponds to 5 fb- 1 and 19 fb-1
at center-of-mass energies of 7 TeV and 8 TeV, respectively.
Thesis Supervisor: Professor Christoph Paus
Title: Professor of Physics
3
4
Acknowledgments
I would like to thank CERN Scientists Fabian Stoeckli and Gerry Bauer, Northwestern
Professor Kristian Hahn and MIT Professors Christoph Paus, Markus Klute and Steve
Nahn for their guidance throughout the analysis. I would also like to thank Max
Goncharov for introducing me to the wonderful world of particle physics.
5
6
Contents
1
Introduction
2
Experimental Setup
11
2.1
The Large Hadron Collider (LHC) . . . . . . . . . . . . . . . . . . . .
11
2.2
The CMS Detector - . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12
2.2.1
Silicon Tracker
. . . . . . . . . . . . . . . . . . . . . . . . . .
13
2.2.2
Electromagnetic Calorimeter (ECAL) . . . . . . . . . . . . . .
14
2.2.3
Hadron Calorimeter . . . . . . . . . . . . . . . . . . . . . . . .
15
2.2.4
Muon Chambers
16
9
. . . . . . . . . . . . . . . . . . . . . . . . .
3
Data and Monte Carlo Samples
19
4
Basic Object Selection
23
4.1
Vertex Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23
4.2
T riggers
24
4.3
Photon Selection
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25
4.4
Electron Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
26
4.5
M uon Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
28
5
6
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Higgs Reconstruction
31
5.1
32
Data and background Monte Carlo comparison after full event selection
Energy Corrections
33
6.1
33
Electron Energy Corrections . . . . . . . . . . . . . . . . . . . . . . .
7
6.2
7
Photon Energy Corrections . . . . . . . . . . . . . . . . . . . . . . . .
34
Background and Signal Modeling
37
7.1
C ategories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
37
7.2
Background Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . .
39
7.3
Signal Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
39
8 Systematic Uncertainties
45
9 Results
47
9.1
Limit Setting Procedures . . . . . . . . . . . . . . . . . . . . . . . . .
47
9.2
Observed Limits.
. . . . . . . . . . . . . . . . . .. . . . . . . . . . . .
47
9.3
Comparison to the Approved Analysis
. . . . . . . . . . . . . . . . .
48
10 Summary
53
A Signal Model Fits
55
B Isolation Computation
89
B.1 Photon Isolation
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
89
B.2 Electron Isolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
90
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
91
B.3 Muon Isolation
8
Chapter 1
Introduction
The standard model (SM) of particle physics has been extraordinarily successful in
modeling experimental data [1, 2]. Only one fundamental particle in the standard
model, the Higgs boson, remains to be experimentally confirmed. The Higgs boson is
postulated to be responsible for the electroweak symmetry breaking that gives mass
to the W and Z bosons [3].
Recent searches for a SM Higgs boson at the LHC have led to an observation of a
Higgs-like mass peak in the 125 to 126 GeV mass region [4]. This observation along
with data from previous accelerators suggest the existence of a low mass Higgs boson.
The next step in the analysis process is to measure the properties of the mass
resonance, like spin and parity, and to perform measurements that give insight into
the dynamics of the resonance. In the standard model, fermions and bosons acquire
their mass via interaction with the Higgs field. The strength of the coupling to the
Higgs field is determined by the mass of the particle. The dynamics of the Higgs-like
resonance is determined by measuring the branching ratios of the resonance to its
decay products. Furthermore, the spin [5] (expected to be zero for the SM Higgs
boson) and parity (expected to be even for the SM Higgs boson) of the Higgs-like
resonance is established from the angular correlations of the Higgs decay products,
notably the ZZ -+ 41, Z-y and -/ decay modes.
The H -+ Zy
-±
f+-
channel is notable for several reasons. The HZy coupling
is induced by loops of heavy charged particles (like the top quark and W boson)
9
[6].
Therefore, the channel is sensitive to heavy charged particles. It also provides
a clean final-state topology where the Higgs mass is accurately reconstructed from
the three body mass of the photon and the two leptons, allowing for measurement
of the branching ratio. In addition, the spin of a mass signal can be determined
by analyzing the angular correlations of the final state particles. Although the Higgs
standard model branching ratio into the Zy channel varies between 0.111% to 0.246%
in the range between 120
<
mH
<
140 GeV, the CMS experiment should be sensitive
to this channel in the coming years.
This paper describes a search for a Higgs boson decaying into a photon and a Z
boson, with the Z boson decaying into either an electron-positron pair or a muonantimuon pair. The search was conducted using data taken from the CMS detector
in 2011 and 2012. This analysis uses a cut-based selection
[7],
with the requirement
of a reconstructed Z boson and an isolated photon. Exclusion limits are set on the
cross section of a SM Higgs boson in the 120 < Mffy < 150 GeV mass window.
Additionally, the results of this analysis will be compared to the H -+ Zy analysis
approved by CMS. The official analysis. hereafter referred to as the approved analysis,
is found in Reference [8].
10
Chapter 2
Experimental Setup
2.1
The Large Hadron Collider (LHC)
The Large Hadron Collider (LHC) is a proton-proton collider located at the European
Organization for Nuclear Research (CERN) in Geneva Switzerland.
It was built
from 1998 to 2008 with the goal of testing different theories in high-energy physics
and proving or disproving the existence of the Higgs boson and many new theorized
particles. The LHC tunnel, located 45-170 m beneath the surface, measures 26.7
km in circumference and straddles the border between France and Switzerland. The
collider tunnel contains two adjacent beam pipes that each contain a proton beam
travelling in opposite directions. There are 1,232 dipole magnets used to keep the
proton beams on a circular path, while 392 quadrupole magnets are used to keep the
beams focused.
Protons are injected into the LHC beampipe after being accelerated through a
series of systems to increase their energy. Protons are first accelerated to 50 MeV
in the linear particle accelerator LINAC 2. The protons are then transferred to the
Proton Synchrotron Booster (PSB), where they are accelerated to 1.4 GeV. Afterwards, the protons are injected into the Proton Synchrotron (PS) and accelerated to
26 GeV. Finally, the protons then enter the Super Proton Synchrotron (SPS) and are
accelerated to 450 GeV prior to entering the main ring of the LHC.
There are four intersection points in the main ring of the LHC where four main
11
experiments are located. There are two general purpose detectors: the A Toroidal
LHC Apparatus (ATLAS) experiment and the Compact Muon Solenoid (CMS) experiment. These two experiments represent the energy frontier in the particle physics
field and among other goals are used to confirm or deny the existence of the Higgs
boson.
The other two main detectors are The A Large Ion Collider Experiment
(ALICE), used to study heavy ion collisions, and the Large Hadron Collider beauty
(LHCb) experiment, used to study b physics with a focus on physics with b hadrons.
2.2
The CMS Detector
The CMS detector consists of a superconducting solenoid 13m in length and 6m in
diameter that produces an axial magnetic field of 3.8 T [9]. Figure 2-1 shows a sketch
of a transverse slice of the detector. A silicon tracker, electromagnetic calorimeter
(ECAL) and a hadron calorimeter (HCAL) are contained within the bore of the
solenoid. Outside of the solenoid are muon chambers and iron return yokes for the
magnetic field. The iron return yokes are used to collapse the fringe magnetic field
and to ensure a strong enough magnetic field outside of the solenoid to measure the
curvature of muons in the muon chambers. The barrel of the detector is closed off
with two endcaps.
In order to describe collisions, a coordinate system is defined with the origin
located at the interaction point along the beampipe. The z-axis points along the
beampipe direction. The x-axis is located in the horizontal plane and points radially
towards the center of the main ring. The y-axis points vertically upwards. Polar
coordinates are more often used in the analysis, where the angle
# is
the azimuthal
angle in the x-y plane measured counterclockwise from the from the positive z-axis
and the polar angle 6 is measure from the positive z-axis. However, the polar angle is
usually expressed in terms of the pseudorapidity, 1 = - ln tan 6/2, because differences
in q are invariant under boosts along the z-axis. Additionally, the analysis often uses
transverse momentum, which is the momentum measured in the x-y plane.
Electrons and photons are detected in the electromagnetic calorimeter (ECAL).
12
Orn
-
IM
2M
3m
4m
SM
6M
7M
Charged Hadron (e.g. Pion)
Neutral Hadron le.g. Neutron)
-
-----Photon
~~-TI
Electromnagnetkc
Cakornmeter
Hadron
Super~ondticting
Iron return yoke Interspersed
Transverse slice
through CMS
with Muon cham s
Figure 2-1: Transverse slice through the Compact Muon Solenoid (CMS) detector
When electrons and photons enter the ECAL, their energy is deposited into a cluster of
crystals collectively called a supercluster. The electron will also leave charged particle
tracks in the silicon pixel and strip tracker. Muons are detected in the gas detectors
located in the iron return yoke placed outside of the superconducting solenoid. A
complete description of the CMS detector can be found in Reference [10].
2.2.1
Silicon Tracker
The silicon tracker [11] is used to measure the trajectories of charged particles. The
momentum of charged particles is determined by measuring the curvature of the
trajectory of charged particles in the magnetic field [12]. The tracker is composed of
silicon pixels and silicon microstrip detectors. As a charged particle travels through
the tracker, the silicon pixels and microstrips produce small electrical signals that are
amplified and detected.
The silicon pixel detector is the closest detector to the beampipe. It consists of
65 million pixels and is composed of cylindrical layers at 4, 7 and 11 cm with disks
at each end. Each cylindrical layer is composed of numerous tiles, 100 Jim by 150
pum. When a charge particle goes through a tile, it forces an electron to be released
13
from the silicon atoms. Each pixel uses an electric current to collect these ejected
electrons on the surface. Each silicon tile is associated with an electronic silicon chip
that takes the small electronic signal as input and outputs an amplified signal.
There are ten layers of silicon strip detectors that surround the silicon pixel detector. The tracker silicon strip detector consists of 10 million detector strips read
by 80,000 microelectronic chips. Just as with the silicon pixel detector, charged particles knock out electrons in the silicon sensors. An applied electric field moves these
knocked out electrons to create a small electrical pulse that enters the microelectronic
chips. The chips then send infrared pulses through fiber optic cables to record the
particle hit.
2.2.2
Electromagnetic Calorimeter (ECAL)
The purpose of the electromagnetic calorimeter (ECAL) [13] is to determine the
energies of electrons and photons. The ECAL is made up of a barrel section and two
endcaps and lies between the silicon tracker and the hadron calorimeter. The barrel
is composed of 61,200 lead tungstate crystals organized in 36 modules. The endcaps
consists of roughly 15,000 crystals. When electrons and photons enter the ECAL,
their energy is deposited into a cluster of crystals collectively called a supercluster.
The transition region between the barrel and the endcaps occurs between 1.444 <
rj| < 1.566.
When high-energy photons or electrons hit the heavy nuclei of the ECAL crystals,
a shower of electrons, positrons and photons is produced. As this shower proceeds
through the ECAL, the electrons in the lead tungstate crystals take energy away
from the shower of particles and become excited. The crystal electrons then release a
short burst of scintillation photons when they relax back to their former energy state.
Each crystal is associated with a photodetector. The burst of scintillation photons
hits the silicon of the photodetector and knocks off an electron. This electron is then
accelerated through the electric field of the photodetector, colliding with other nuclei
and forcing out a shower of electrons. The photodetector then reads this shower of
electrons as an electrical signal and amplifies it. The amplified signal is then digitized
14
and sent away by fiber optic cables.
The ECAL also contains a preshower used to distinguish high-energy photons
from neutral pions. The preshower detector is located in the endcap regions of the
ECAL where the angle between two emerging photons from the decay of a neutral
pion is too small to distinguish the two decay photons from a high-energy photon.
The preshower detector is made of two planes of lead followed by silicon sensors. The
crystals in the preshower are much finer than those in the ECAL, making it possible
to resolve pion-produced photons as a photon pair. When a high-energy photon hits
the lead layer, it creates an electromagnetic shower which the silicon sensors detect
and measure. The two layers of silicon allow for measurement of the photon energy
and position. When high-energy photons in the ECAL, their paths can be retraced
to look for "hits" in the preshower, making it possible to deduce whether the highenergy photon was indeed a high-energy photon or a photon pair. The total energy
of the photon is determined by summing the energy deposited in the preshower and
the ECAL.
2.2.3
Hadron Calorimeter
The purpose of the hadron calorimeter (HCAL) [9, 13] is to measure the energy of
hadrons, particles made of quarks and gluons. The HCAL is organized into barrel
(HB, HO), endcap (HE) and forward (HF) sections. The HE is the portion of the
barrel located inside the magnetic coils and the HB is the portion of the barrel located
outside of the magnetic coils to ensure that no particle ensures escape through the
ends of the solenoid magnet. The HF calorimeters are located at either end of the
detector close to the beampipe and are used to pick up energies of particles that
scatter with angles close to the beampipe.
The HB, HO and HE are made of several layers of dense absorber and tiles of
plastic scintillators. When a hadronic particle hits the dense absorber, either brass
or steel, numerous secondary particles are produced.
These particles then shower
and avalanche through successive layers of absorbers. The tiles of plastic scintillators
count the light produced in order to measure the energy of the shower as it proceeds
15
through the HCAL by sending blue-violet light through tiny optical fibers located in
each tile. The different layers of tiles are organized into geometrical arrangements
called towers. The energy of the shower is determined by summing the amount of
light in a tower.
2.2.4
Muon Chambers
The muon chambers, as one would guess, are used to detect muons [14].
Muons
are minimally ionizing particles and penetrate deep into the detector. They are not
stopped by the calorimeters.
There are four muon stations that record "hits" of
muons as they travel through the muon chambers. The powerful CMS magnet bends
the trajectory of the muons as it travels through the trackers and the muon chambers,
allowing for measurement of the momenta.
The muon chambers are composed of 250 drift tubes (DTs), 540 cathode strip
chambers (CSCs) that track the muon positions and 610 resistive plate chambers
(RPCs) that measure timing and position.
The drift tubes are 4 cm wide tubes filled with gas and contain a positively charged
stretched wire. When a muon (or any charged particle) passes through the gas volume,
electrons are knocked off the atoms of the gas. An electric field forces these electrons
to drift towards the positively charged stretched wire. The position of the muon is
determined by the location of where the electrons hit the wire and the time taken for
the electrons to drift to the wire. Each DT consists of 12 aluminum layers in three
groups of four. The middle group measures the position coordinate parallel to the
beampipe and the two outside groups measure the perpendicular position coordinate.
The cathode strip chambers are used in the endcaps and consist of arrays of
positively-charged wires crossed with negatively-charged copper cathode strips in a
gas volume. As with the drift tubes, incoming muons force electrons off the atoms
in the gas volume. These electrons collect on the positively-charged wires. Positively
charged ions move away from the wire and towards the copper cathode which then
induces a charge on the copper cathode strips. The wires and strips are perpendicular,
thus allowing for measurement of two position coordinates.
16
The RPCs consist of two parallel plates (an anode and a cathode) made of high
resistivity plastic material. The plates are separated by a gas volume. When a muon
strikes an electron off the atoms in the gas volume, the electron hits other atoms and
creates a shower of electrons. Metallic strips pick up the signal from the electron
shower. The trajectory of hits in the RPCs allows for fast measurement of the muon
momentum.
This quick momentum measurement is used by the CMS trigger to
determine whether or not the data are worth storing.
17
18
Chapter 3
Data and Monte Carlo Samples
Data for this analysis were collected during the 2011 and 2012 runs of the CMS
experiment. Yearly datasets are divided into sections. The 2011 data is divided into
sections A and B. The 2012 data is divided into sections A, B, C and D. Additionally,
2011 data were taken at a center-of-mass energy of 7 TeV and 2012 data were taken
at a center-of-mass energy of 8 TeV. The specific datasets are shown in Table 3.1
along with their corresponding official good run list files in JSON file format. Good
run list files are JSON files published by CMS and are used to specify which run
and luminosity periods during the data taking should be considered for analysis. The
total integrated luminosity for 2011 data is 5.05 fb-1. The total integrated luminosity
for 2012 data is 19.40 fb-1.
Dataset
JSON
Dataset
JSON
Dataset
JSON
Dataset
JSON
Dataset
JSON
Dataset
JSON
2011 A and B dielectron datasets: r11a-del-j16-v1-bp, rllb-del-j16-v1-bp
Cert-160404-180252_7TeVReRecoNov08_Collisions 11 _JSONv2.txt
2011 A and B dimuon datasets: rlla-dmu-j16-vl-bp, rllb-dmu-j16-vl-bp
Cert_160404-180252-7TeVReRecoNov08-Collisions 11 _JSONv2.txt
2012 A and B dielectron datasets: r12a-del-j13-vI, rl2b-del-j13-vl
Cert_190456-19653l8TeV_13Jul2Ol2ReRecoCollisionsl2_JSON-v2.txt
2012 A and B dimuon datasets: rl2a-dmu-j13-vl, rl2b-dmu-j13-vl
Cert_190456-196531_8TeV_13Jul2Ol2ReRecoCollisionsl2-JSON-v2.txt
2012 C and D dielectron datasets: rl2c-del-pr-v2, rl2d-del-pr-vl
Cert-190456-208686_8TeVPromptRecoCollisionsl2-JSON.txt
2012 C and D dimuon datasets: rl2c-dmu-pr-v2, rl2d-dmu-pr-vl
Cert_190456-208686_8TeV_PromptRecoCollisions 12_JSON.txt
Table 3.1: Table of dataset names and corresponding JSON good run list files
19
The description of the Higgs boson signal used in this analysis is obtained using
Monte Carlo simulation. The SM Higgs boson is produced via four main mechanisms:
gluon fusion, vector boson fusion, associated production with a W or Z boson and
associated production with a top quark pair. Monte Carlo simulations of the gluon
fusion and vector boson fusion Higgs production mechanisms are used to produce
Higgs mass signals from 120 Gev to 150 Gev in 5 GeV intervals. These signal samples
are used to determine the signal efficiency and the acceptance. The signal samples are
fit to signal models. The shapes of these signal models are used to extract exclusion
limits on the cross section of a standard model Higgs boson. The signal samples are
generated with POWHEG [15] at next to leading order (NLO) for the gluon fusion
[16] and vector boson fusion [17] production processes and the associated production
process was simulated with PYTHIA [18] at leading order (LO). Table 3.2 shows
the expected cross-sections and branching ratios outlined by the LHC Cross-Section
Working Group [19] for all four production processes at 7 TeV.
20
Higgs mass (GeV)
Production Process
ggH
Cross-Section (pb)
16.65
Branching Ratio
120
VBF
WH/ZH
ttH
ggH
1.27
1.02
0.10
15.32
0.00111
125
VBF
WH/ZH
ttH
ggH
1.21
0.89
0.09
14.16
0.00154
130
VBF
WH/ZH
ttH
ggH
1.15
0.78
0.07
13.11
0.00195
135
VBF
WH/ZH
ttH
ggH
1.10
0.68
0.07
12.18
0.00227
140
VBF
WH/ZH
ttH
ggH
1.05
0.60
0.06
11.33
0.00246
145
VBF
WH/ZH
ttH
ggH
1.02
0.53
0.05
10.58
0.00248
150
VBF
WH/ZH
0.98
0.47
0.00231
ttH
0.05
Table 3.2: SM Higgs boson cross sections and branching ratios (for H
TeV
21
-+
Zy) at 7
22
Chapter 4
Basic Object Selection
The Higgs mass is reconstructed using the three body mass of the photon and two
leptons. The photons, electrons and muons first have to pass cut-based selection
requirements. A cut-based selection means the kinematic and topological variables
of the particles have to be higher or lower than certain thresholds (cut values). After
selection, the Z- three body mass is reconstructed with additional cuts placed on the
photons, electrons and muons. This section will explain the process to preselect the
photons, electrons and muons and then reconstruct the Z-y mass.
4.1
Vertex Selection
The primary vertex (PV) of the collision is reconstructed using the so called Deterministic Annealing (DA) clustering algorithm [20]. This algorithm ensures that the
reconstructed primary vertex has a distance to the interaction point less than 24 cm
in z and less than 2 cm in the transverse plane. Of the selected vertices, the vertex
with the largest sum squared of the momentum tracks associated with the vertex is
chosen as the event vertex. This vertex has the highest probability of being the vertex
of the relevant interaction.
23
4.2
Triggers
Events in the CMS detector are accepted and saved if they pass certain triggers. CMS
uses a tiered trigger system with two levels. The Level-i (LI) trigger is implemented
in hardware and firmware, and the High Level Trigger (HLT) is implemented in the
software. The Li triggers process detector information from every single LHC bunch
crossing at a rate of 40 MHz. It uses coarse granularity information from the ECAL,
HCAL and muon chambers to make a decision on whether to keep the data from the
event. Currently, the Li trigger does not use information from the tracking detectors
to make its decision.
If an event passes the Li trigger, it is then sent to the HLT. Full event information
is available to the HLT. When events fire a specific trigger in the HLT, they are sent
to storage and labeled with the corresponding HLT trigger name. Data is collected
for this analysis by collecting events with the HLT trigger names listed in Table
reftriggers. The dielectron triggers require two electrons with one electron having a
transverse momentum (pt) greater than 17 GeV and the other having a pt greater
than 8 GeV in addition to loose selection requirements on the isolation and track (in
the silicon tracker) of the two electrons. The dimuon triggers require varying cuts on
the transverse momenta of the muons and loose identifications on the isolation and
track of the muons.
H
-±
e+e-y Decay Mode Triggers
HLTEle17_CaloldT-TrkIdVLCaloIsoVLTrkIsoVLEle8_CaloIdT-TrkIdVL-CaloIsoVLTrklsoVL-v*
HLTEle17_CaloldT-CaloIsoVL-TrkldVLTrklsoVLEle8_CaloldT-CalolsoVLTrkIdVL-TrkIsoVL-v*
H
Decay Mode Triggers
HLT-Mu17_Mu8-v*
HLTMu13_Mu8_v*
HLTDoubleMu7v*
-± [ptj--y
Table 4.1: A list of the HLT trigger names used in the analysis
24
4.3
Photon Selection
Photons are required to have a transverse momentum (pt) greater than 15 GeV and
have a supercluster pseudorapidity r,| less than 2.5. Additionally, photons in the
barrel-to-endcap transition region are rejected. Photons are considered to be in the
barrel if 1rj < 1.442 and in the endcap if 1.566 < 1rj < 2.5. Cuts are applied on the
following discriminating variables for candidate photons:
" SigmalEtalEta: The energy weighted (single crystal energy over the supercluster energy) standard deviation of a single crystal q in an ECAL supercluster
within a 5 by 5 crystal block centered on the crystal with the highest energy.
SigmalEtalEta characterizes the shower shape of the photon. This variable is
required to be less than 0.011 if the photon is in the barrel and less than 0.033
if the photon is in the endcaps. SigmalEtalEta is required to be small in order
to distinguish photons from a jet showering into the ECAL.
" Single Tower H/E: Ratio of the energy in the HCAL tower behind the supercluster to the energy in the ECAL supercluster. This ratio is required to be less
than 0.05. This ratio is small because no to very little energy is deposited into
the HCAL for detected photons.
" Isolation: Sum of the transverse momentum of particles within a cone around
the photon direction. Isolation is computed separately for charged hadron, photon and neutral hadron candidate particles. The isolation values are corrected
using a pile up energy subtraction.
Further explanation and cut values for
isolation are available in Appendix B.
Isolation for photons was summed over candidate particles within a AR cone of 0.3,
where AR is defined to be the angular displacement using the pseudorapidity, r;,
and the azimuthal angle, q>, as coordinates. Isolation is used to reject non-prompt
background resulting from jets.
Additionally, the photon is required to pass an electron veto. The electron veto
ensures that the photon supercluster did not match the supercluster for an electron
25
that both returns zero hits in the inner silicon tracker and successfully passes a conversion filter. The electron conversion filter is explained in the next section. The
selection criteria for the photon is summarized in Table 4.2
Variable Name
oq
Single Tower H/E
Charged Hadron Isolation
Neutral Hadron Isolation
Photon Isolation
Barrel Cut
< 0.011
< 0.05
< 1.5
< 1.0 + 0.04 x pt
< 0.7 + 0.005x pt
Endcap Cut
< 0.033
< 0.05
< 1.2
< 1.5 + 0.04 x pt
< 1.0 + 0.005x pt
Table 4.2: Summary of the selection criteria for the photons
4.4
Electron Selection
Electrons are selected by applying a loose cut-based electron identification approved
by the E/Gamma Particle Object Group (POG) [21]. Preselection for electrons require the electron transverse momentum to be greater than 10 GeV and the electron
171 to be less than 2.5. Electrons with rj| < 1.566 are defined to be in the barrel and
electrons with 1.566 < iJ| < 2.5 are defined to be in the endcaps. Cuts are placed on
the following variables:
" sigmalEtalEta: This variable is required to be less than 0.01 for barrel electrons
and less than 0.03 for endcap electrons.
" H/E: The ratio of the energy in the HCAL behind the supercluster to the energy
of the supercluster. This ratio is required to be less than 0.12 for barrel electrons
and 0.10 for endcap electrons.
" dEtaln: The supercluster q minus the track 71 at the point of closest approach
to the supercluster. The difference is required to be less than 0.007 for barrel
electrons and less than 0.009 for endcap electrons.
" dPhiln: The supercluster
# minus the
track
# at the point
of closest approach
to the supercluster. The difference is required to be less than 0.15 for barrel
electrons and less than 0.10 for endcap electrons.
26
" dO: Defined to be the displacement of the electron track from the primary event
vertex in terms of the x and y coordinates. This variable is required to be less
than 0.2 for both barrel and endcap electrons.
" dZ: Defined to be the displacement of the electron track from the primary event
vertex in terms of the longitudinal distance. This variable is required to be less
than 0.05 for both barrel and endcap electrons.
" 1/E - 1/p: One over the energy in the ECAL minus one over the momentum of
the track at the point of closest approach to the beam spot. 1/E characterizes
the energy measured in the ECAL and 1/p characterizes the energy measured
in the tracker. For electrons, the difference should ideally be zero. This variable
is required to be less than 0.05 for both barrel and endcap electrons.
Isolation for electrons is computed in a AR cone of 0.4 and required to be less than
0.4. Further explanation of the electron isolation and cut values are given in Appendix
B.
There is a significant probability that photons convert into e+e- pairs inside the
silicon pixel and strip tracker detectors. Therefore it is necessary to perform a conversion filter to make sure that an electron did not originate from a photon. This is
done be performing a vertex fit on the common vertex of the e+e- pair. If the X2
probability of the fit was greater than 10-5, then the e+e- pair was considered to
Variable Name
oginq
H/E
dEtaIn
dPhiln
dO
dZ
1/E - 1/p
Conversion Veto: Fit Probability
Conversion Veto: Missed Hits
Isolation/pt
Barrel Cut
< 0.01
< 0.12
< 0.007
< 0.15
< 0.2
< 0.05
< 0.05
< le-6
< 1
< 0.4
Endcap Cut
< 0.03
< 0.10
< 0.009
< 0.10
< 0.2
< 0.05
< 0.05
< le-6
< 1
< 0.4
Table 4.3: Summary of the selection criteria for the electrons
27
have originated from a photon and neither of the electrons passed the preselection.
The selection criteria for the electrons is summarized in Table 4.3. A more detailed
description of photon conversions and the conversion filter are given in Reference [22].
Muon Selection
4.5
A tight cut-based identification approved by the Muon POG [23] is used to select
muons. Preselection for muons requires the muon transverse momentum to be greater
than 10 GeV and the muon 1rj1 to be less than 2.4. Additionally, the muon must be
a global muon, meaning that the muon must return a track in the muon chambers
that also matches a track in the silicon tracker using a muon track fit. In order to
suppress particles punching through the HCAL into the muon chambers or detecting
muons resulting from particle decays in flight, the following cuts are placed on the
muons:
" The reduced chi-square of the global muon track fit is required to be less than
10.
" At least one muon chamber must return a valid hit in the global muon track fit.
* At least two muon stations in the muon chambers must return a muon track
segment.
" The track of the muon in the silicon tracker must have a transverse impact
parameter (d)
less than 2 mm with respect to the primary vertex.
" The longitudinal distance (dZ) of the track of the muon in the silicon tracker
with respect to the primary vertex must be less than 5 mm (only for 2012 data).
" The number of pixel hits in the silicon detector assigned to the track must be
greater than 0.
" The number of tracker layers with hits must have be greater than eight for 2011
data and greater than nine for 2012 data. This ensures accurate measurement
of the muon transverse momentum.
28
For 2012 data, the muon is also required to be a particle flow muon object. Particle
flow objects are a list of all particles in an event that have been reconstructed using
an optimized combination of subdetector information.
Isolation for muons is computed in a AR cone of 0.4 and required to be less than
0.12. A summary of the muon selection criteria is shown in Table 4.4. A description
of the isolation cut values and how the muon isolation is computed are given in
Appendix B.
Variable Name
GlobalMuon
2011 Cut
Yes
2012 Cut
Yes
PF Muon
-
x /ndof
Number of Valid Hits
< 10
> 0
Yes
< 10
> 0
Number of Matched Stations
dO
dZ
Number of Pixel Hits
Number of Tracker Layers
> 1
< 0.2
> 0
> 8
> 1
< 0.2
< 0.5
> 0
> 5
Isolation/pt
< 0.12
< 0.12
Table 4.4: Summary of the selection criteria for the muons
29
30
Chapter 5
Higgs Reconstruction
The Z-y three body mass is reconstructed after the photons, electrons and muons are
selected. Each Zy event is required to have at least one good primary vertex. The Z
candidate is chosen by taking the two oppositely charged leptons closest to the Z mass.
The highest pt photon is chosen for the three body mass reconstruction. The photon
is rejected if it was within AR < 0.4 of a lepton. The transverse momenta of the
leading lepton is required to be greater than 20 GeV and the transverse momentum of
the trailing lepton is required to be greater than 10 GeV. The ratio of the transverse
momentum of the photon to the three body mass is required to be greater than
15/110. The mass window for the mass is chosen to be 115 < m
< 180 GeV. This
window allows the final state radiation (FSR) events (located near the Z mass value)
to be cut away. FSR events are events where a Z boson decays into an electron or
muon pair, with one of the electrons or muons then radiating off a photon. In these
events, the Z-j three body mass sums to be the Z mass.
Additionally, the dilepton mass is required to be greater than 50 GeV. This cut
is placed in order to reject events from H
and produces a lepton pair.
31
-
y-, where one of the photons converts
5.1
Data and background Monte Carlo comparison after full event selection
This section shows a comparison of data and background Monte Carlo simulation
after full event selection for 7 TeV data. The background simulations include both
the irreducible Z-y process and the Drell-Yan process. A comparison of the three
body mass to Monte Carlo for electrons using 2011 (7 TeV) data is shown in Figure
5-1. The data and the background Monte Carlo agree.
Three Body Mass (eey)
80
-4-- Data
70
DY Jets
60
L
SM Zy
III
50
C
40
kiII}EI II
30
;T
4 11,
W
4
20
10
0
I
120
I
1
1,
130
1
1
a
a
I
140
I
I
150
Mass (GeV)
1
0
1
a
I
160
170
180
Figure 5-1: Data and MC comparison for the three body mass in the electron channel
is shown. The Drell Yan (DY Jets) Monte Carlo simulates the Drell Yan process.
The standard model Zy Monte Carlo simulates events where the Z boson decays into
two leptons (with one of the leptons possibly radiating a photon).
32
Chapter 6
Energy Corrections
6.1
Electron Energy Corrections
The electron energy scale and resolution is corrected by examining the Z -* ee mass
peak because the Z boson peak is well-known and well-defined and comparing it
to background Monte Carlo. Plots of the uncorrected Z peak for both data and
Monte Carlo are shown in Figure 6-1. Both of the electrons are required to pass
the selection criteria as stated in the previous chapter. Additionally, corrections are
made separately for electrons entering either the barrel or the endcap.
Z Mass (ee)
40000
--
Data
DY Jets
35000
SM
30000
Zy
25000
W20000
15000
10000
5000
0
82
84
86
88
90
92
94
96
98
Mass (GeV)
Figure 6-1: Uncorrected data and MC comparison for the Z-peak
33
100
4X
Mass Corrected (ee)
1Z
3.5 -
]DY Jets
Data
3
2.5-
1.5-
0.5-
82
86
84
86
88
90
Mass (GeV)
92
94
96
98
100
Figure 6-2: Corrected data and MC comparison for the Z-peak. There still remains
small disagreement between the data and Monte Carlo. This disagreement is taken
into account as an uncertainty.
The figure clearly shows differences between the data and Monte Carlo. In order to
correct the data, the official CERN EGamma electron scale and resolution corrections
are applied. These official corrections perform run dependent scale shifts on the data
and smear the Monte Carlo. Figure 6-2 shows the corrected Z peak for both data
and monte carlo. The errors from the scale and resolution corrections are propagated
and taken to be 1% for the electron energy scale correction and 2% for the energy
resolution correction.
6.2
Photon Energy Corrections
The photon energy scale and resolution are corrected using the VGamma Photon
Energy Scale and Photon Energy Resolution (PHOSPHOR) software package. The
width and position of the Z boson are used to measure the photon energy resolution
and energy scale. For the photon corrections, the Z -+ ppy process provides a clean
source of high energy photons. The photon energy is corrected by correcting the three
body mass to the Z boson mass. Additionally, the mass is required to be within 30
34
GeV of the Z boson mass to reduce background from jets.
The PHOSPHOR program is used to fit the three body mass with the photon energy scale and resolution as parameters. These parameters provide the reconstructed
photon energy scale and resolution. The measured photon energy scale and resolution
percentage corrections in the barrel and the endcap for different ranges of the photon
transverse momentum are shown in Table 6.1.
Barrel Scale, R9 > 0.94
MC 2011
MC 2012
Data 2011
Data 2012
Barrel Scale, R 9 < 0.94
MC 2011
MC 2012
Data 2011
Data 2012
Barrel Res., R 9 > 0.94
MC 2011
MC 2012
Data 2011
Data 2012
Barrel Res., R 9 < 0.94
MC 2011
MC 2012
Data 2011
Data 2012
Endcap Scale
MC 2011
MC 2012
Data 2011
Data 2012
Endcap Res.
MC 2011
MC 2012
Data 2011
Data 2012
10 < PT< 12
0.8 ± 0.0
0.46 ± 0.10
-0.89 ± 0.45
2.18
0.46
10 < PT< 12
2.15 ± 0.27
1.32 t 0.21
-0.25 ± 0.43
0.95 0.42
10 < PT< 12
2.69 t 0.0
1.93
0.03
5.03 ± 0.61
3.01 ± 0.69
10 < PT< 12
8.2 t 0.08
6.16 ± 0.07
15.69 ± 0.29
9.24 ± 0.59
10 < PT< 12
3.0 ± 0.32
1.82 ± 0.36
1.41
0.58
0.9 ± 0.73
10 < PT< 12
7.48 ± 0.1
6.17 ± 0.1
15.16 ± 0.48
10.65 ± 0.93
12 < PT< 15
0.79 ± 0.10
0.29 ± 0.08
-0.06 t 0.36
0.64 ± 0.37
12 < PT< 15
1.81 ± 0.21
0.81 ± 0.16
0.28 ± 0.33
1.15 ± 0.32
1 2 < PT< 15
2.56 ± 0.03
1.8 ± 0.02
3.75 ± 0.53
3.95 ± 0.61
12 < PT< 15
7.31 ± 0.06
5.92 ± 0.54
11.13 ± 0.54
8.87 ± 0.5
12 < PT< 15
2.33 ± 0.24
1.3 t 0.25
0.07 ± 0.48
-0.66 ± 0.51
12 < PT< 15
6.05 ± 0.07
5.2
0.08
13.46 t 0.73
8.86 ± 0.66
15 < PT< 20
20 < PT
0.45 ± 0.07
0.45 ± 0.04
0.28 ± 0.06
0.11 ± 0.04
-0.06 ± 0.28
1.1 ± 0.12
1.78 ± 0.22
1.15 ± 0.13
1 5 < PT< 20
2 0 < PT
1.48 ± 0.16
0.8 ± 0.09
0.64 ± 0.12
0.62 ± 0.07
0.24 ± 0.25
-0.36 t 0.16
0.28 ± 0.25
0.39 t 0.15
2 0 < PT
15 < PT< 20
2.1 ± 0.02
1.47 ± 0.01
1.51 ± 0.01
1.12 ± 0.01
3.59 t 0.38
2.18 ± 0.16
1.91 ± 0.26
2.29 ± 0.22
15 < PT< 20
20 < PT
5.82 ± 0.05
3.94 ± 0.03
4.7 ± 0.05
2.77 ± 0.02
7.54
0.34
5.36 ± 0.23
7.58 ± 0.31
3.58 ± 0.24
2 0 < PT
15 < PT< 20
1.64 ± 0.18
0.96 ± 0.11
1.14 ± 0.19
0.62 ± 0.12
0.58 ± 0.37
0.45 ± 0.22
1.14 ± 0.0
1.24 ± 0.24
2 0 < PT
15 < PT< 20
4.93 ± 0.06
3.39 ± 0.03
4.28 ± 0.05
2.93 ± 0.03
10.68 ± 0.5
6.21 ± 0.4
4.28 ± 0.0
5.62 ± 0.43
Table 6.1: Summary of the measured photon energy scale and resolution corrections
in percentages
35
36
Chapter 7
Background and Signal Modeling
In order to statistically interpret the selected events, maximum likelihood fits for the
Zy three body mass distribution are needed. Results are extracted using unbinned
maximum likelihood fits of the data using different background only and signal plus
background hypotheses in order to set confidence limits on model parameters [24]
like the Higgs cross section. This is popularly referred to as the "CLs method". In
order to produce these confidence limits, probability density functions (PDFs) of the
background and the expected signal are needed.
This section outlines the process
used to obtain these background and signal models.
7.1
Categories
The search is made more sensitive by dividing the selected events into categories based
on the expected mass resolution and signal-to-background ratio [25].
Probability
density functions for the background and the expected signal are determined for each
individual category and then combined together.
Higher energy events are more likely to spray into the barrel of the detector while
lower energy events (and events with more background) tend to shower in the forward
direction along the beampipe. Additionally, Irj squeezes due to the forward relativistic
boost as it approaches the beampipe, making lower energy events more difficult to
resolve. This means the most valuable signal events are expected to have both the
37
leptons and the photon in the barrel. Moreover, events with unconverted photons have
higher energy resolution and less background. For these reasons, the selected events
are divided into event categories based on the pseudorapidities of the leptons and
photons and the R9 value of the photon. R 9 is a topological variable that corresponds
to the ratio of the energy in a 3 x 3 crystal matrix in the electromagnetic calorimeter
to the supercluster energy. The R9 variable is used to differentiate between converted
and unconverted photons
Four event classes are used in both the electron and muon analyses. The definitions
of the classes are slightly different for electrons and muons due to the geometry of
the detector. The definition of the classes are shown in Table 7.1. The best signalto-background ratio occurs for Event Class 1, where both leptons and the photon are
in the barrel of the detector.
In the transition from 7 TeV data to 8 TeV data, the acceptance for the high R9
class drops and the low R 9 class increases due to the increase in the number of pileup
collisions. This class migration is taken into account as a systematic uncertainty.
Event Class 1
(Category 0)
Photon: 0 < Jr| < 1.442
Photon: 0 < 1,/ < 1.4442
Both leptons: 0 < i1 < 1.442
Both leptons: 0 < iq < 2.1
and one lepton: 0 < Iq| < 0.9
R9 > 0.94
R 9 > 0.94
Event Class 2
(Category 1)
Photon: 0 < i) < 1.442
Both leptons: 0 < i1 < 1.442
Photon: 0 < i1 < 1.4442
Both leptons: 0 < 17| < 2.1
Event Class 3
(Category 2)
R9 < 0.94
Photon: 0 < 177 < 1.442
At least one lepton: 1.566 < IJq < 2.5
and one lepton: 0 < Iq| < 0.9
R 9 < 0.94
Photon: 0 < lql < 1.4442
Both leptons: Ig| > 0.9
and none between 1.442 < iq < 1.566
or one lepton: 2.1 < Ig| < 2.4
No R9 requirement
Photon: 1.566 < Jil < 2.5
Both leptons: 0 < JqJ < 2.5
and none between 1.442 < 177 < 1.566
No R9 requirement
No R9 requirement
Photon: 1.566 < Inj < 2.5
Both leptons: 0 < 17| < 2.4
Event Class 4
(Category 3)
No R 9 requirement
Table 7.1: Definition of the event categories for both the muon and electron analyses
38
7.2
Background Modeling
The main backgrounds in the analysis come from initial state radiation (ISR) of a
photon, shown in Figure 7-1(a), and final state radiation (FSR) from Drell-Yan Z
boson production, shown in Figure 7-1(b) [26]. In FSR, the three body Zy mass is
close to the Z boson mass. Therefore, the FSR background is removed by looking in
a three body mass window above 100 GeV.
Zq
Z
(a)
(b)
Figure 7-1: Feynman diagrams for ISR (a) and FSR (b)
This analysis did not use Monte Carlo simulation of background processes. Instead, a model for the background is obtained by fitting the three body mass distributions [27]. Bias studies by the approved analysis have shown that a 5th-order
Bernstein polynomial for event classes 2, 3 and 4 and a 4th-order Bernstein polynomial for Event Class 1 provide an acceptably small bias with a reliable background
shape. These Bernstein polynomial fits are used for both 7 TeV and 8 TeV and are
discussed in the following pages.
7.3
Signal Modeling
The description of the Higgs boson signal used in this analysis is obtained using
Monte Carlo simulation. Monte Carlo simulations of the gluon fusion, the vector
boson fusion, the associated production with a W or Z boson and the associated
production with a top quark pair Higgs production mechanisms are used to produce
39
Higgs samples at masses of 120, 125, 130, 135, 140, 145 and 150 GeV. The signal
models specify the efficiency times acceptance and the shape of the signal in each of
the event classes.
The signals are modeled using the addition of a Crystal Ball (CB) function and
a Gaussian function. The free parameters for the fit are the means of the CB and
Gaussian (1 parameter), the width of the CB and Gaussian (2 parameters), the
standard CB parameters a and n and the ratio between the CB and Gaussian (1
parameter) for a total of six parameters. Example signal model fits for the gluon
fusion (ggH) and vector boson fusion (VBF) Higgs production mechanisms are shown
in Appendix
A. Once the full set of signal models at each Monte Carlo set are
determined, a continuous signal model as a function of the Higgs mass is obtained by
using a linear interpolation of the fit parameters. This interpolation is then used to
determine continuous exclusion limits on the Higgs production cross section.
40
CMS preliminary
f- = 7.0 TeV L 5.05 fb-
30
C)
eey Cat 0
25
35
+
Data
Model
-Bkg
±1 a
0±2(y
0
30
: CMS preliminary
1
:s = 7.0 TeV L =5.05 fb
+ Data
Model
-Bkg
eey Cat 1
±2
25
w
w
20
20
15
15
-
10
10
5
5
p
120
140
160
meey
(a) 7
120
180
(GeV)
0
20
-
CMS preliminary
~ = 7.0 TeV L = 5.05 fb-
+ Data
-
eey Cat 2
160
180
m.,, (GeV)
(b) 7 TeV Electrons Event Class 2 Background
Fit
TeV Electrons Event Class 1 Background
Fit
25
140
;7
Bkg Model
3
30
-
CMS preliminary
+ Data
1
Bkg Model
F = 7.0 TeV L = 5.05 fb~±1(3F
eey Cat 3
12,a
0i2o
25-
w
w
20
15
15
10
-
10-
5
5 -120
140
160
180
m.,y (GeV)
120
140
160
m..ee
(c) 7 TeV Electrons Event Class 3 Background
Fit
180
(GeV)
(d) 7 TeV Electrons Event Class 4 Background
Fit
Figure 7-2: Background model fits for the four event classes in the ee-y channel using
7 TeV data
41
CMS preliminary
1
= 7.0 TeV L = 5.04 fb
0
25
ppyCat 0
w
+
-
CMS preliminary
CMS preliminary
_ = 7.0 TeV L = 5.04 fb"
40
Data
Bkg Model
CO,
*±2a
35
ggy Cat 1
30
20
w
15
-4- Data
- Bkg Model
±a
m ±2
25
20
-H
I
4
15
10
10
5
5
0
120
180
m.,y (GeV)
160
140
120
(b) 7 TeV Muons Event Class 2 Background Fit
(a) 7 TeV Muons Event Class 1 Background Fit
25
-
CMS preliminary
1
= 7.0 TeV L = 5.04 fb
w
Data
Bkg Model
0
1i2o
20
w
15
30 . CMS preliminary
fs = 7.0 TeV L =5.04 fb'
25
yCat 2
C
+
-
180
m,. (GeV)
160
140
-
.y yCat 3
+ Data
Model
-Bkg
M 2a
20
15
1
10
5
0
2
120
a
140
.
I
-.
10
5
160
'
180
mr,, (GeV)
120
140
160
"
" 180
M,,, (GeV)
(d) 7 TeV Muons Event Class 4 Background Fit
(c) 7 TeV Muons Event Class 3 Background Fit
Figure 7-3: Background model fits for the four event classes in the
7 TeV data
42
[ppy
channel using
tIA 1.
CMS preliminary
f = 8.0 TeV L = 19.40
0
50
-4- Data
fb
CMS preliminary
-L = 8.0 TeV L = 19.40 fb
0
Bkg Model
-
60
Data
Bkg Model
+1t
+
-
S+O±10
ee7 Cat 0
i
-±2
w 40
: eey Catl1
50
±2a
w
40
30
30
-
20
.
20
10
10
'
0
120
~~
~
140
' ~ 'IT
' ~'
'1 '
160
mee,
u
180
(GeV)
(a) 8 TeV Electrons Event Class 1 Background
Fit
0
-
50
40
-
+
CMS preliminary
i = 8.0 TeV L = 19.40 fb'
-
ery Cat 2
M±
120
140
160
180
m.,y (GeV)
(b) 8 TeV Electrons Event Class 2 Background
Fit
Data
Bkg Model
±10
0
U9
CD
w
81
CMS preliminary
1
- = 8.0 TeV L =19.40 fb
70
eey Cat 3
+ Data
Model
-Bkg
60
50
30
40
20
30
20
10
10
120
140
160
meey
180
(GeV)
120
140
160
m.ey
(c) 8 TeV Electrons Event Class 3 Background
Fit
180
(GeV)
(d) 8 TeV Electrons Event Class 4 Background
Fit
Figure 7-4: Background model fits for the four event classes in the eey channel using
8 TeV data
43
60
CMS preliminary
-s = 8.0 TeV L=19.40
50
pgy Cat 0
a)
1
fb
ou
+
Data
Model
-Bkg
M±2o
w
w
40
70- L
CMS preliminary
= 8.0 TeV L = 19.40 fb'
60-
LpyCat 1
50
4
40
30
30
44
20
12-4
6
20
8
10
10
140
S120
160
.
180
m,,, (GeV)
(a) 8 FeV Muons Event Class 1 Background Fit
a)
+ Data
Bkg Model
±10
± o
-
50 _ CMS preliminary
i = 8.0 TeV L = 19.40 fb1
40 -9Y
-
140
160
180
m,,, (GeV)
(b) 8 TeV Muons Event Class 2 Background Fit
90
+ Data
Bkg Model
2
120
80
1at
w
30
a)
70
w
60
-
MS preliminary
1
L =19.40 fb
fs = 8.0 TeV
-4- Data
-
Bkg Model
i1ac
0±2o
p Cat 3
50
40
20
30
20
10
10
120
140
160
120
180
m,,Y (GeV)
140
160
180
M,,, (GeV)
(d) 8 TeV Muons Event Class 4 Background Fit
(c) 8 TeV Muons Event Class 3 Background Fit
Figure 7-5: Background model fits for the four event classes in the ppyy channel using
8 TeV data
44
Chapter 8
Systematic Uncertainties
Table 8.1 lists the systematic uncertainties considered in the analysis. Systematic
uncertainties come from the uncertainties on the luminosity measurement, the trigger efficiencies, the functions used to model the Higgs signal, the Higgs branching
fractions, the photon and lepton selections and the migration of events between the
different classes.
An overall 2.2% uncertainty is applied to the luminosity in 2011 and an overall 4.4% uncertainty for the 2012 luminosity. The background contribution to the
uncertainty is estimated from the three body fy mass.
The trigger uncertainties have been set by the CERN HZ-y Group. Additionally,
the uncertainties on the object selections have been established by the approved analysis using tag-and-probe methods to test for uncertainties in the object identification
and isolation procedures, as well as by using guidelines by the CERN Physics Object
Groups (POG).
45
7 TeV (%)_8
Systematic Variable
Integrated Luminosity
Theory
Gluon-gluon fusion cross section (scale)
Gluon-gluon fusion cross section (PDF)
Vector boson fusion cross section (scale)
Vector boson fusion cross section (PDF)
W associate production (scale)
W associate production (PDF)
Z associate production (scale)
Z associate production (PDF)
Top pair associate production (scale)
Top pair associate production (PDF)
Branching fraction
Trigger
Electron
Muon
Selection
Photon Barrel
Photon Endcap
Electron
Muon
Signal Model Parameters
Mean
Sigma
Event Migration
Tev(%)
2.2
4.4
+12.5, -8.2
+7.0, -7.7
+0.5, -0.3
+2.7, -2.1
+0.7, -0.8
+3.5, -3.5
+1.7, -1.6
+3.7, -3.7
+3.4, -9.4
+8.5, -8.5
6.7, 9.4, -6.7, -9.3
+7.6, -8.2
+7.6, -7.0
+0.3, -0.8
+2.8, -2.6
+0.2, -0.7
+3.5, -3.5
+1.9, -1.7
+3.9, -9.7
+3.9, -9.3
+7.9, -7.9
6.7, 9.4, -6.7, -9.3
0.5
0.5
0.5
1.5
0.5
1.0
0.8
0.7
0.5
1.1
0.8
1.4
1.0
5.0
5.0
1.0
5.0
5.0
Table 8.1: The systematic uncertainties used in the analysis
46
Chapter 9
Results
9.1
Limit Setting Procedures
The expected limits are calculated using the CLs modified frequentist method [28]
which uses the profile likelihood as a test statistic. The likelihood test statistic is
evaluated using the signal and background models described in Chapter 7. After
combining the electron and muon results, Figure 9-1 shows the 95% confidence level
observed limits for the exclusion of the Higgs production cross section (o-) separately
for 7 TeV and 8 TeV data. The expected limits are calculated by throwing 1000 toy
background events for each Higgs mass point.
9.2
Observed Limits
The observed combined exclusion limit on the cross section of a SM Higgs boson
decaying into a Z boson and a photon as a function of the Higgs boson mass is shown
in Figure 9-2. The observed limits are between 10 and 38 times the Higgs standard
model cross section. No excess of events is observed.
47
9.3
Comparison to the Approved Analysis
Figure 9-3 shows the observed combined exclusion limit plot created by the approved
analysis on May 3rd, 2013. The observed limits are between 4 and 25 times the
Higgs standard model cross section. The search methods performed by the approved
analysis differ from the ones conducted in this thesis. The approved analysis uses
a Multivariate Analysis Tool
[29]
to optimize selection of the electrons, whereas a
cut-based approach is used in this thesis. Additionally, the approved analysis uses
an additional Event Class for VBF dijet tagged events, which increases the search
sensitivity.
48
=100.; z100 .
90
ObservedCM
m
-+
80
7 TeV L
+ la Expected
.
S
= 7 TeV..L .. 5,042.b (PA).
Expected
--
-
70
I
CMS preliminary
Median Expected
-.....
..
. .....-
...
...
50
~----
-
--
-----------
- - - -.----..-.-.-
1 0
lXGSM
125
120
21'
0
0
135
Observed
Median Expected
j± 1a Expected
0+± 2(y Expected
-........
4+ -is
1
%J 0
3
. .1 .9
....-............----..
..-.-.-.----
-
- - - -- - -
-
- - - - - - --- --
---
145
= 8 Te
-----
.... -...--
0
11 0
mH (GeV/c 2)
140
CMS preliminary
= 8 TeV L 19 96 fb (ee)
........--..
-..
0
=L
130
b
.s......
--.---- ---- .-- .--- .
-,*,* " ", 11 "***1 11 *** ****"** *, "*
- - - - - -- - - - -
0
=L 2 0
+.
.
1)
0
..............
- -.......
--..
----.. ........
0
.
120
.
.
.
...
-..
.
|1XGSM
-
125
130
135
140
145
150
mH (GeV/c 2)
Figure 9-1: Observed limit plots separately for 7 Tev (top) and 8 TeV (bottom) data
49
M100
90
80
Observed
-.....-- Median Expected
+ la Expected
±
2a Expected
70
60
50
- - - --
CMS preliminary
= 8 TeV L= 19.396 fb (ee)
s
=8
f.
-----
TeV.L:. 19395..
(
)
7 TeV L = 5.048 fb (ee)
7 TeV -Lt 5-.1042-fb l! (
.....)...-
--.
------.-.--.-.-.---.--.-.----...-.-.-.-.--...-.-.-........
.........
....
--.
... ...
.....
...
.
--. -----.--.-.---.-.-.--.-.--.
............
......
--.--.
..... ...
.....
... ...
.....
...
40
30
y 20
.........
---.
...-.
..-..----.---.----..-...
.....
.....
... ..
-.
......
- .-.
-~amLLLamd.~
10
920
...----. -
xOSM
125
130
135
140
145
15 0
mH (GeV/c 2)
Figure 9-2: Combined 7 TeV and 8 TeV observed limit plot
50
s 40
s=7 TeV, L=5.0 fb
CMS
r,
is=8 TeV, L = 19.6 fb"
,,, ,
H-+Zy
-o
30!-
E
25.
-J
o
Observed
Expected * Ilo
--- Expected t 2 a
20
-
00
O 15
CD
920
125
130
135
140
145
150
155
160
mH (GeV)
Figure 9-3: Combined 7 TeV and 8 TeV observed limit plot published by the approved
analysis and based on 5.0 fb-r of data taken at 7 TeV and 19.6 fb-1 of data taken at
8 TeV
51
52
Chapter 10
Summary
A search for the SM Higgs boson decaying into a Z-boson and a photon has been
performed based on 5.05 fb-
of data taken at 7 TeV and 19.4 fb-
of data taken at
8 TeV. No excess of events has been observed for a Higgs boson decaying into a Z
boson and photon. Exclusion limits on the standard model Higgs cross section have
been set and are shown in Figure 10-1.
c10o
-
Observed
90 -- ----- Median Expected
L
l Expected
S8 -~ +2a Expected
S80:
70
-..
~.~..-
CMS prelimiinary
= 8 TeVL
=9396 fb0'(ee)
=8 TWL.L19,395J ....(sp)......
i70 7 TeV L:= 5.048 fb1 (ee)
-- --. =- 7 TeV-L --5.-042-fb - ---..------
-..-.-.-.---..-.-.-..-.--.--..-.--.--
60
40
40
30 --
2
'6 20--;
-
10
20
125
20
125
.
0
135
140
.45
130
135
140
145
.
1XcYSM
150
1
mH (GeV/c)
Figure 10-1: Combined 7 TeV and 8 TeV observed limit plot
53
54
Appendix A
Signal Model Fits
55
> 3 50-
400
0
V3
00-
350
50 -
300
U,
w
250
2 00-
200
50150
14
00-
100
50-
50
00
110
120
130
141
m, (GeV)
mf
(a) 120 GeV ggH signal model
(b) 130 GeV ggH signal model
00-
4
(GeV)
4501
0)
3 50-
4
0
350300-
2 50-
250-
20
2001 50
50
00
00
500-
50
120
130
140
150
20
rr, (GeV)
(c) 135 GeV ggH signal model
4
130
140
m4, (GeV)
0
(d) 140 GeV ggH signal model
00-
0
U')
3
50
3 50
300
Z 3
w
a)
250 2
5050 2 00-
200
00-
1~
50-
1 50-_
1' 00
50-J
0
130
140
150
:0
160
m (GeV)
(e) 145 GeV ggH signal model
~
U
140
14
150
~
4
10
7
10
m,,, (GeV)
(f) 150 GeV ggH signal model
Figure A-1: Event Class 1 example ggH signal model fits in the eey channel for 7
TeV data.
56
450
50
0-
40
0
0
0
(0
CO 400350
U)J
30030 0-
250200-
20 0150
100
10 i0-
50 00
130
120
110
140
rr, (GeV)
0
120
130
500-
5C
400
4 )0w
300-
30 0
200
2C)0-
100
10
0 -4
120
130
150
m,, (GeV)
(b) 130 GeV VBF signal model
(a) 120 GeV VBF signal model
140
140
0
1
150
m1 , (GeV)
q, (GeV)
(d) 140 GeV VBF signal model
(c) 135 GeV VBF signal model
500
50 0
400-
0
40
0
300
30 020020 0100-
10 0
13
130
140
150
:30
160
m, (GeV)
L4
140
5
5
150
160
/
17
m,
1 (GeV)
(f) 150 GeV VBF signal model
(e) 145 GeV ggH signal model
Figure A-2: Event Class 1 example VBF signal model fits in the ee-/ channel for 7
TeV data.
57
>6 006
50 0
0
U)
5 00
40 0
> 4
w
00
30 0
3
00-
20 021
00
10
1I
0 -
100
120
110
130
140
10
m, (GeV)
120
130
140
150
m1, (GeV)
(a) 120 GeV ggH signal model
(b) 130 GeV ggH signal model
00
CD
0 60
00 -
t5
50
>4
w 00-
40
300-
30
200-
20
1 00-
10
120
130
140
15
m, (GeV)
10
(c) 135 GeV ggH signal model
130
140
150
160
mr (GeV)
(d) 140 GeV ggH signal model
00-
6
C,
CD 5 00
U,
6
500-
4 00
4
00
C,
w
3 00-
00
3
2
00-
2 00-
00-
1 00
1
130
140
150
0
160
mrr
(GeV)
(e) 145 GeV ggH signal model
140
150
160
17 '0
m,,, (GeV)
(f) 150 GeV ggH signal model
Figure A-3: Event Class 1 example ggH signal niodel fits in the pp-y channel for 7
TeV data.
58
700
0
60
600-
6
0
C
U')
LAI
500_
0-
50
400
40
300-
30
200 -
20
100
10
0-
00-
-
110
120
130
001"0
140
m. (GeV)
(a) 120 GeV VBF signal model
120
130
140
m,
150
(GeV)
(b) 130 GeV VBF signal model
70
W)
6, 6 0-
600500
500
400
40 0
300-
30 0-
200
20 0-
100-
10
120
130
140
0
0
1
M, (GeV)
(c) 135 GeV VBF signal model
130
140
150
160
m, (GeV)
(d) 140 GeV VBF signal model
700 7 0600
U)
6(
50050
400
40 0-
300-
30
0-
200-
20 0
-
100-
10
M
-n
130
140
150
160
130
m, (GeV)
(e) 145 GeV ggH signal model
1W
140
150
160
170
m, (GeV)
(f) 150 GeV VBF signal model
Figure A-4: Event Class 1 example VBF signal model fits in the ppy channel for 7
TeV data.
59
2 50-
0
25
25 0 -
"
2 00C2C
w
15
1!50
1
0
1! 00
50 -
50-
110
120
130
910
140
m% (GeV)
120
130
140
150
my (GeV)
'b) 130 GeV ggH signal model
(a) 120 GeV ggH signal model
in
C!3 2 50
250-
1
U)
6
2
00-
2
0-
in
w
1!50-
50 -
100-
0 -
50-
120
130
140
?20 130
1 0
rr (GeV)
(c) 135 GeV ggH signal model
140
150
160
m1, (GeV)
'd) 140 GeV ggH signal model
> 2 50-
2 50
i)
0
in
2
2
1
50-
1n
10 -
1
050
1
C0-
io
50
130
140
150
160
30
rr (GeV)
(e) 145 GeV ggH signal model
140
150
16
m,
17
(GeV)
(f) 150 GeV ggH signal model
Figure A-5: Event Class 2 example ggH signal model fits in the eey channel for 7
TeV data.
60
30
0-
25
0-
20
0
15
0-
10
0
0
0
250-
U')
,
200-
w
w
150-
100-
50
50-
900
0
130
120
110
140
10
mr (GeV)
~
120
130
140
10
m (GeV)
(b) 130 GeV VBF signal model
(a) 120 GeV VBF signal model
00
LO2500
0
4)
6 250
0j
200-
0200
150150
100-
100
50
50
120
130
140
20
rrm (GeV)
140
1
0
16C
rr
(GeV)
(d) 140 GeV VBF signal model
c) 135 GeV VBF signal model
00 -
3
3
0
0
250
6 250-
0L
130
200-
w
200-
50-
150
00-
100
50 -
50-
130
140
150
30
160
m, (GeV)
140
15
160
170
m1,, (GeV)
(f) 150 GeV VBF signal model
(e) 145 GeV ggH signal model
Figure A-6: Event Class 2 example VBF signal model fits in the ee-y channel for 7
TeV data.
61
40 0
400
0
UO
0
35 0-
o 350
30 0
300
25 0 -
W
250
20 0
200
150 -
150
100
100
S0
50
u0
120
110
130
1
140
mr, (GeV)
(a) 120 GeV ggH signal model
120
130
140
150
m1, (GeV)
(b) 130 GeV ggH signal model
00-
00
4
4
a)
350-
350
LU
300
3
00
250-
2
50
200-
200
50
50
1'
00
0
50-
50
0-
120
130
140
T2- 130
158
m (GeV)
(c) 135 GeV ggH signal model
140
150
160
m1, (GeV)
(d) 140 GeV ggH signal model
00-
4
0)
300
350
300-
250
250-
200
2 00
150
1. 50
100
1i 0
50
50
C
130
140
150
rr
160
130
(GeV)
(e) 145 GeV ggH signal model
140
.I
150
160
170
m,,, (GeV)
(f) 150 GeV ggH signal model
Figure A-7: Event Class 2 example ggH signal model fits in the prny channel for 7
TeV data.
62
450
a)
400
400
U)
6
a)
w
350
350
300-
300
w
250
250
200-
200
150
150
100
100
50
50
T00
110
120
130
110
140
mr (GeV)
(a) 120 GeV VBF signal model
0
40
130
140
15
mr (GeV)
b) 130 GeV VBF signal model
0-
> 4
0-
0
50
a)
35 0-
n35
1
0-
3
30 0-
w
120
12
25 0-
0501c )0 50-
20 0-
2
15
15
100-
50
120
130
140
20
10
m, (GeV)
(c) 135 GeV VBF signal model
130
140
15 mn (GeV)
16
d) 140 GeV VBF signal model
40 0-
400
35 07
350
30 0 -
300
a)
C
0
a)
w 25 0-
W
250
20 0-
200
15 0-
150
10 0
100
50
50
130
140
150
13
m, (GeV)
-
140
150
160
170
m,, (GeV)
-
(e) 145 GeV ggH signal model
(f) 150 GeV VBF signal model
Figure A-8: Event Class 2 example VBI signal model fits in the p/yy channel for 7
TeV data.
63
1,
01
40
140-
20
120-
01 100
100-
80-
80
60
60-
40-
40-
20-
20-
120
110
130
S10
14,
m%, (GeV)
120
130
14
15(
mn, (GeV)
(b) 130 GeV ggH signal model
(a) 120 GeV ggH signal model
160
14A
140-
120
120-
100
0
w 100
80
8060
6040
40-
20
20:0
120
130
140
150
0
m (GeV)
(c) 135 GeV ggH signal model
0
130
140
150
16(
mr (GeV)
(d) 140 GeV ggH signal model
0
60_
40-
(1
11
20-
1) 40:LU
20
00
1l
00-
80
80 60
40-
40-
20-
20130
140
150
f30
160
mr (GeV)
(e) 145 GeV ggH signal model
130
140
150
16
140
150
1630
170
170
m1, (GeV)
(f) 150 GeV ggH signal model
Figure A-9: Event Class 3 example ggHEsignal model fits in the ee-y channel for 7
TeV data.
64
>)
0
4)
w
160
012 010 080604020-
0
14
140
120-
w 10080
6040
20
00
110
130
120
r
120
?10
140
(GeV)
(a) 120 GeV VBF signal model
130
140
151
m1, (GeV)
(b) 130 GeV VBF signal model
160-
14 0-
0
0O
6) 12 0 -
1406
120-
100-
0d
100
8080
60-
60-
40
40-
20
20
120
140
130
150
20
mg (GeV)
a)
16 0-
140
ISO
160
m, (Ge V)
0
1
0
U-)
U)
{
6 14 0
12
130
(d) 140 GeV VBF signal model
(c) 135 GeV VBF signal model
0
1
0
6
14 00
12
0
0
w 100
10
80-
80
60-
60-
40
4 0-
20
2
130
140
150
lb
0
L
T36-
mrr (G eV)
140
1
1
150
160
170
10
m,, (GeV)
(f) 150 GeV VBF signal model
(e) 145 GeV ggH signal model
Figure A-10: Event Class 3 example VBF signal model fits in the ee-y channel for 7
TeV data.
65
4)180
13 600
160
40-
1i
140
I2J
20-
W 120
13
60-
100
80
60
40-
40
20
20
0
110
120
130
10
140
m (GeV)
(a) 120 GeV ggH signal model
120
130
140
150
m,, (GeV)
(b) 130 GeV ggH signal model
200
200
a)
CD
180
U..
180
160
160
140
140
120
120
100
100-
80
80
60
40
40
20
20120
130
140
120
1 0
rr (GeV)
(c) 135 GeV ggH signal model
130
140
150
160
m1, (GeV)
(d) 140 GeV ggH signal model
''
2 20:-
200
-
180
2 30
13 80_
-
160
13 60
c
140-
40
W
13 20-
120
100-
0
80
80
60-
60-
40
40-
20
20
130
140
150
0
10
m (GeV)
(e) 145 GeV ggH signal model
140
150
160
170
140
150
160
170
M, 3 (GeV)
(f) 150 GeV ggH signal model
Figure A-11: Event Class 3 example ggH signal model fits in the rppy channel for 7
TeV data.
66
W)
0
18 0-
0a) 180-
U)
160-
160U)I
140
140
a)
w 12 0-
120
10 0-
100-
80-
80
6
60
40-
40
2 0 -.
0
20-
00
110
120
130
?10
140
M. (GeV)
a)
U)
o
a)
16
14
0
0-
146
154
m1, (GeV)
0 180
U)
160
140
12 0-
w
1i
130
(b) 130 GeV VBF signal model
(a) 120 GeV VBF signal model
0
120
W
0-
120
100
80-
13-4
5
6
80
60-
60
40-
40
20
20
120
130
140
T20
1
rm, (GeV)
(c) 135 GeV VBF signal model
130
140
150
161
m1, (GeV)
(d) 140 GeV VBF signal model
200
0
U)
0
180
180
U)
160
160
140
6L 140
120
120
100
100
80
80
60
60
40
40
20
"""0 130
I
140
I
150
20
166
M, (GeV)
140
150
166
170
m,, (GeV)
(f) 150 GeV VBF signal model
(e) 145 GeV ggH signal model
Figure A-12: Event Class 3 example VB F signal model fits in the sply channel for 7
TeV data.
67
100-
a)
80
L0
20
1
008060-
4040 20
20-
00
110
0
1 0
120
120
?
14
130
140
mr , (GeV)
(a) 120 GeV ggH signal model
150
moy (GeV)
(b) 130 GeV ggH signal model
140
120-
C5
1; 20-
.4
100-
6L 1i 30
-
8060-
60-
40-
40
20-
20-
120
130
140
20
mr (GeV)
(c) 135 GeV ggH signal model
130
140
150
160
mry (GeV)
(d) 140 GeV ggH signal model
140(0
(0
120-
Ii,
6
100-
100
80
w
80
60
6040
4020
20-
130
140
150
rr
160
(GeV)
3
(e) 145 GeV ggH signal model
140
150
160
140
150
16
170
170
m1, (GeV)
(f) 150 GeV ggH signal model
Figure A-13: Event Class 4 ggH signal model fits in the ee-y channel for 7 TeV data.
68
a) 100
0
(0
120
(0
U)
0
0
80
0
0
w
w
100
80
60
60
40
40
20
20
10
120
120
4(
m (GeV)
a) 120 GeV VBF signal model
1,6
rm, (GeV)
0
U)
20-
I
0080-
1
0
140
(b) 130 GeV VBF signal model
14001~
130
w
80-
60-
604040-
20
20-
30
120
130
140
20
10
mr (GeV)
c) 135 GeV VBF signal model
150
163
m1, (GeV)
40-
2020
0
1! 000
140
(d) 140 GeV VBF signal model
-~1
0
130
-
-1 Do
3-
80 -
60
60-
40
40-
20
20-
0
130
140
150
160
30
mr (GeV)
(e) 145 GeV ggH signal model
140
150
160
170
m1 , (GeV)
(f) 150 GeV VBF signal model
Figure A-14: Event Class 4 VBF signal model fits in the ee-y channel for 7 TeV data.
69
1600)
0
140-
U,
0
160140-
120
0)
100
w
80-
120100
80-
60-
60-
40-
40
20-
20300
120
110
13
14(
mr (GeV)
?10
(a) 120 GeV ggH signal model
160
140
It)
C
180
160
140
120
100
100
80
80-
60
60-
40
40-
20
20
140
150
rm, (GeV)
200 -
0)
130
140
0)
0
w 120
120
130
(b) 130 GeV ggH signal model
4
180
120
20
rr (GeV)
(c) 135 GeV ggH signal model
130
140
150
160
mr (GeV)
(d) 140 GeV ggH signal model
180-
180
0
160
LO
-
160-
140-
140-
120-
120-
100
100-
80-
80
60-
60:-
40 L
40-
20-
20130
140
150
160
30
m, (GeV)
(e) 145 GeV ggH signal model
140
150
160
170
m,, (GeV)
(f) 150 GeV ggH signal model
Figure A-15: Event Class 4 ggH signal model fits in the ppwy channel for 7 TeV data.
70
16
0
14
0-
0
0-
160-
010 0-
140-
12
w
180-
w 120
100-
80
6
80-
0-
60
4
40
2
20-
i0o
120
110
130
140
0
rr, (GeV)
(a) 120 GeV VBF signal model
120
130
140
10
m1, (GeV)
(b) 130 GeV VBF signal model
180Ln
>
160-
0
wo
LU
140120
18 016 014 012 0
100:
100
80-
8
60 -
060-
40-
40
20-
2
0
12
130
140
2
mg (GeV)
(c) 135 GeV VBF signal model
130
140
150
m
160
(GeV)
(d) 140 GeV VBF signal model
200-
20 0-
180-
(D 18 0U)
160-
16 0
140-
14 0
0
120
12 0-
100-
10 0-
80-
80
60-
60
40-
40
20-
20
130
S
140
150
160
U30
m, (GeV)
(e) 145 GeV ggH signal model
140
150
160
170
m,1 (GeV)
(f) 150 GeV VBF signal model
Figure A-16: Event Class 4 VBF signal model fits in the ppy channel for 7 TeV data.
71
180
0
LO
U)
-
220
(0 200
160-
LO
6
140-
180
160-
w
120-
140
100-
120
80-
100
80
60-
60404020-
20-
00
110
120
130
144
?10
mr (GeV)
(a) 120 GeV ggH signal model
120
130
140
154
m
4, (GeV)
(b) 130 GeV ggH signal model
200-
> 200
4)
(0 180-
180
9n
6
160
160140
140
C)I
120
120-
100
100
80
80
60
60-
40
40-
20
20
120
130
14
20
mr (GeV)
(c) 135 GeV ggH signal model
130
140
150
16
rr,, (GeV)
(d) 140 GeV ggH signal model
250-
(0
2 02
0
Go -
200-
1480
U)
1460
w
150
40
1420
00-
100
80-
6050-
40
20
130
140
15o
rr
1
9:30
(GeV)
(e) 145 GeV ggH signal model
140
150
160
1'
m, (GeV)
0
(f) 150 GeV ggH signal model
Figure A-17: Event Class 1 example gg i signal model fits in the ee-y channel for 8
TeV data.
72
240a)
0
CD
a)
U,
200C)
25 0-
CD
220-
200-
180a)
160-
w
140-
0
15
120
1007
10 0
8060-
50-
4020
0000
120
110
130
Tio
140
nr, (GeV)
130
140
150
m,,, (GeV)
(b) 130 GeV VBF signal model
(a) 120 GeV VBF signal model
a)
120
250-
25
0
200-
0-
20
w
150-
15
100-
10
0
10
0-
50-
120
130
140
10
150
mr, (GeV)
130
140
150
160
m, (Ge V)
(d) 140 GeV VBF signal model
(c) 135 GeV VBF signal model
0
30
a)
L0
1
250-
o
200-
0
S25
w 200-
uLJ
150-
0-
15
100-
10
50
5
00
0
130
140
150
rr
10
160
(GeV)
140
15h
100
1
M,, (GeV)
(f) 150 GeV VBF signal model
(e) 145 GeV ggH signal model
Figure A-18: Event Class 1 example VBF signal model fits in the eey channel for 8
TeV data.
73
0
CO
U,
0
0
LU
30
3 50
0-
3 00-
25 0-
2 506 1
2 )0-
20 0
150
50
11
00
100
L
50
100
110
120
130
140
m (GeV)
?10
(a) 120 GeV ggH signal model
130
140
1Sf
mr (GeV)
(b) 130 GeV ggH signal model
350
0
0
LO
6,
120
3
00
CO
300
.4
c 2 50
250T
LU
2
0
200
1I 50
150100-
1i 00-
50
50
120
130
140
rr
9:-0
(GeV)
(c) 135 GeV ggH signal model
130
140
1
if1 0
m~ (GeV)
(d) 140 GeV ggH signal model
350
3 50
0
Ui
300
0
250
30
2 50
L2
200
21 00
150
100
50
11
00
50
50-
0
130
140
15
930
16
m
(GeV)
(e) 145 GeV ggH signal model
140
150
160
1
m,, (GeV)
0
(f) 150 GeV ggH signal model
Figure A-19: Event Class 1 example ggH signal model fits in the
TeV data.
74
p-y channel for 8
4bU
0
0
30 0
U)
6
25 0
0
w
350300
w
20 0-
250
15 0
200150
0-
10
400-
10050C, 100
110
120
130
140
rmg (GeV)
1
(a) 120 GeV VBF signal model
120
130
140
150
m1, (GeV)
(b) 130 GeV VBF signal model
400-
40 0
0
0
350-
0
35
U)
U)
300-
0-
250
0-
200-
25
ic
20
0-
150-
15
)0-_
100-
10
050
5012
120
30
10
130
14
5
20
1
m, (GeV)
(c) 135 GeV VBF signal model
140
150
160
m (GeV)
(d) 140 GeV VBF signal model
0
40
130
0
(±3 400U)
3Cio -
350-
30
w
0
0-
w
25
2C0
-
250
200
0-
150
0-
100-
50
50
15
1
300
130~
140
150
160
?:0
140
150
160
1
mn
q, (GeV)
(e) 145 GeV ggH signal model
(GeV)
0
(f) 150 GeV VBF signal model
Figure A-20: Event Class 1 example VBF signal model fits in the ppy channel for 8
TeV data.
75
0
CD 1400
120
U)
0
w 100
0L
0
15 50-
4o
0)
2000-
80
80
60
60
40
40
20
20 110
?00
120
130
910
14(
ng (GeV)
(a) 120 GeV ggH signal model
120
130
140
mn
150
(GeV)
(b) 130 GeV ggH signal model
1800
160
0
140-
Z- i 20
0 120
w
100-
40
> 1
w
0
80
40I0
60
40-
40-
20 -L
20
120
130
140
n
0
(GeV)
(c) 135 GeV ggH signal model
130
140
1
60
mr, (GeV)
(d) 140 GeV ggH signal model
>1
0 160
C;
1)
1 40
140
0
I~2
U)
20 120
00-
100
80
80
60
60
40
40
1
20
.
130
.
140
20
15
(Ge)
rT)% (GeV)
130
(e) 145 GeV ggH signal model
140
150
160
170
M,, (GeV)
(f) 150 GeV ggH signal model
Figure A-21: Event Class 2 example ggH signal model fits in the eey channel for 8
TeV data.
76
0)
16
0-
0
U)
6
0)
180
UO
14 012
w
C; 160
0-
140
0
120
10
100-
80
80
6060
1
40-
40-
20 --
20
100
110
120
130
1
140
m (GeV)
(a) 120 GeV VBF signal model
18
120
130
140
150
m,, (GeV)
(b) 130 GeV VBF signal model
0-
0)
U) 16 0
6
14 0-
0)
180
160
140-
12 0w 120100-
10080-
80
60-
60
40-
40-
20
C
20
120
130
140
20
150
rm) (GeV)
200180-
>0)
200-
U)
180
160-
160-
140-
140
120-
120
100-
100
80-
80
60-
60-
40-
40
20-
20
0
130
140
150
1
0
(d) 140 GeV VBF signal model
0
w
140
m1, (GeV)
(c) 135 GeV VBF signal model
U)
130
160
30
m, (GeV)
140
150
160
10
mg,,
(e) 145 GeV ggH signal model
(GeV)
(f) 150 GeV VBF signal model
Figure A-22: Event Class 2 example VB F signal model fits in the eey channel for 8
TeV data.
77
25 0-
> 2
0
20
0
0
w
0
CDJ 200-
0-
1s
0-
15 0
Ic
10 0-
)0 5
50-
110
i'0
110
120
130
140
11
m (GeV)
1
120
130
140
150
m, (GeV)
(b) 130 GeV ggH signal model
(a) 120 GeV ggH signal model
o2350-
0
0-
2 50
2 00
w
13 50-
150-
1 50-
100
sO -
50-
12
130
140
M20
1
mTg (GeV)
130
140
10
160
m,, (GeV)
(d) 140 GeV ggH signal model
(c) 135 GeV ggH signal model
2A40
0
2 20
2 50
0
LO2 0
2
13 80
Ui
13 60-
w
40 -
50-
13 200
1300-
50
I0
50-
40
o
130
140
150
1
30
m (GeV)
(e) 145 GeV ggH signal model
140
150
1
li
mo, (GeV)
(f) 150 GeV ggH signal model
Figure A-23: Event Class 2 example ggH signal model fits in the pp-y channel for 8
TeV data.
78
G)
3C
LO
U)
25 0
C
0
25
20 02C
w
150
15 010
100
50
50
0itiL
10
110
110
I340
mn,, (GeV)
120
(a) 120 GeV VBF signal model
120
130
140
150
mr (GeV)
(b) 130 GeV VBF signal model
300LO
2
C;
U)
250
2 10
w
200-
io -
150-
)0-
100-
010
50~
64,
120
C.l
~
120
13-10
130
~
5
140
?20
m. (GeV)
130
140
160
m,1 (GeV)
(c) 135 GeV VBF signal model
(d) 140 GeV VBF signal model
300-
30 00
0)
250-
o 25
20
200
W 2C 0w
w
15
)0 -
150
10
100-
0-
50-
50
1/
1
0
140
150
3.0
100
my (GeV)
140
150
0
M,, (GeV)
0
(f) 150 GeV VBF signal model
(e) 145 GeV ggH signal model
Figure A-24: Event Class 2 example VB F signal model fits in the pptjy channel for 8
TeV data.
79
80CD
80
70 -
70-
C
In
10
60-
w
50-
60-
w
50
40
30
20-
10-
10
00
110
120
130
10
14
120
130
140
mrg (GeV)
150
m, (GeV)
(b) 130 GeV ggH signal model
(a) 120 GeV ggH signal model
40L
30-
10
C
01
10J
1
50
1
0
10
70-
1
w
20-
60
5040
30
20
10
20
rr (GeV)
130
140
150
m1 , (GeV)
(c) 135 GeV ggH signal model
(d) 140 GeV ggH signal model
40-
>)
>)
20 -
00-
0
cL
10
CD
Cq
90
80
70
0
80-
m(GeV)
4)
4)
60
50
60
40
30
20
10
1
~Tr
10
140
(
130
(e) 145 GeV ggH signal model
140
150
160
""
-
m1, (GeV)
0
(f) 150 GeV ggH signal model
Figure A-25: Event Class 3 example gg H signal model fits in the ee-y channel for 8
TeV data.
80
80-
0
0
0O
Uf)
70
C0
0;
90
8
0-
7
60-
0-
6
50
50 40
40
0
30 0
200
1
302
10
100
120
110
130
110
140
ri, (GeV)
14
80
70
130
140
150
m,, (GeV)
(b) 130 GeV VBF signal model
(a) 120 GeV VBF signal model
90
120
0
120
C,
U)
6
0
60
w
50
100
80
60
40
40
30
20
20
10
0-012T
130
140
TO
15W
m, (GeV)
13
140
1~~
m,, (GeV)
(d) 140 GeV VBF signal model
,c) 135 GeV VBF signal model
100U)
I
0
80-
wO
80-
60
60
40
4020 -
20-
130
140
150
30
160
q,, (GeV)
140
150
160
1 0
M,,, (GeV)
(f) 150 GeV VBF signal model
(e) 145 GeV ggH signal model
Figure A-26: Event Class 3 example VBF signal model fits in the ee-y channel for 8
TeV data.
81
120-
a)
0
CD 14
CD
Ii)
0
100a)
w
80-
12
0-
10
00
8
60-
0-
6
40
4
Qf)
20-
0
2
0-
900
110
120
130
140
10
mg (GeV)
120
130
140
150
m, (GeV)
w
(a) 120 GeV ggH signal model
(b) 130 GeV ggH signal model
140-
a)
CD
0-
14
ID
6
120100-
a)
w
12 0100-
0
80-
8
60-
6
40-
4
20-
2
0
0-
00
-
1
130
14
150
mr
20
(GeV)
(c) 135 GeV ggH signal model
130
140
1
10
m,,, (GeV)
(d) 140 GeV ggH signal model
140a)
CD
It)
0
120
120
C
100
100
a)
w
w
80
80
60
60
40
40 -3
20-
20
1
140
150
160
3
M,, (GeV)
(e) 145 GeV ggH signal model
4
5
6
14
5
A6
7 0
m,, (GeV)
(f) 150 GeV ggH signal model
Figure A-27: Event Class 3 example ggH signal model fits in the ppy channel for 8
TeV data.
82
6
0
10
0
0 12
0-
0-
U,
6
10)0-
0-
8
0
w
w
0-
80 -
6
60 -
0
4
0-
4
0-
2
100
110
120
130
O
10
140
rr (GeV)
(a) 120 GeV VBF signal model
0
(1,
In
0
0
w
120
I
l
130
140
150
m, (GeV)
(b) 130 GeV VBF signal model
012 0
10080604020
14
-L
-
20
0
0q 12
0
6i
0-
10
0-
8
6040204
12I3
120
130
140
5
;0
150
M. (GeV)
(c) 135 GeV VBF signal model
130
140
m,
1 (GeV)
(d) 140 GeV VBF signal model
0-
14 0-
16
0
0
14 0
0-
12 0
100-
w
w
100
8080
60
60
40
4
20-
0m-
13
140
150
1 0-
160
n (GeV)
-140
150
16
1%0
M,, (GeV)
(f) 150 GeV VBF signal model
(e) 145 GeV ggH signal model
Figure A-28: Event Class 3 example VBF signal model fits in the ppy channel for 8
TeV data.
83
10
0
U)
0
0
70
U,
0
80-
60
0
w
0
50
w
6
0-
40
30
40
20
o
10
110
6
120
.11I
130
3
4
.
L
14(
m (GeV)
1
m1, (GeV)
(a) 120 GeV ggH signal model
(b) 130 GeV ggH signal model
0
0
U)
2
C,
1
80-
U)
0
100r
0
80
w
60-
60
40-
40
20-
20
120
130
140
1
U
20
m (GeV)
130
140
150
m1 ,
(c) 135 GeV ggH signal model
160
(GeV)
(d) 140 GeV ggH signal model
00
0
0
1
00-
C
80-
U)
w
60
60
4040-
10
2020
130
140
150
160
M3
mTe (GeV)
(e) 145 GeV ggH signal model
140
150
160
m
170
(GeV)
(f) 150 GeV ggH signal model
Figure A-29: Event Class 4 ggH signal model fits in the ee-y channel for 8 TeV data.
84
80
0
100
70
1~1
80
60
w
50
60
40
40
30
20
20
10
110
120
130
14,
ma(GeV)
mt (GeV)
(a) 120 GeV VBF signal model
(b) 130 GeV VBF signal model
1000
>±3
00
1
80-
o
80-
606040-
40
20-
20-
1
130
140
150
?20
m, (GeV)
130
140
16
m
(c) 135 GeV VBF signal model
16
(GeV)
d) 140 GeV VBF signal model
100 o 100
I0
8000
w
60-
60
40-
4
5
6
140
150
160
7
40
20-
0
-3
20
130
140
150
#30
160
m, (GeV)
(e) 145 GeV ggH signal model
170
rt, (GeV)
(f) 150 GeV VBF signal model
Figure A-30: Event Class 4 VBF signal model fits in the eey channel for 8 TeV data.
85
160
w
0 140
140
U.)
6
120
120
w
w 100
10080-
s0
60
60-
40
40-
20
20-
120
S0
4
mr (GeV)
18
G1
10
130
mr, (GeV)
(b) 130 GeV ggH signal model
(a) 120 GeV ggH signal model
0
120
80-
0
-
160
60a)
01
120
20
100
00
80-
80-
60
60-
40-
40
20
20-
10
130
140
150
m, (GeV)
2-2
(c) 135 GeV ggH signal model
>1
130
140
1
160
mq (GeV)
(d) 140 GeV ggH signal model
180
60-
0
6
140-
6
40-
0 160-
40
In
U')
20
140-
6
120
1U
00
1008080-
60-
60
40-
40
20-
20
130
140
150
160
m, (GeV)
?30
140
150
160
170
m,, (GeV)
(f) 150 GeV ggH signal model
(e) 145 GeV ggH signal model
Figure A-31: Event Class 4 ggH signal model fits in the ppy channel for 8 TeV data.
86
1460
120
0
40
wi
100
1420-
0
4)
80
wi
00
80-
60
60 40
4020
1
20
1 o'0
120
130
90
14'
rr, (GeV)
(a) 120 GeV VBF signal model
120
130
0
1
m1, (GeV)
(b) 130 GeV VBF signal model
180
0)
160-
14 60
0
40-
14014
120100-
00
80-
80-
60-
60-
40-
40-
20-
20-
0
12
130
140
0
150
M, (GeV)
(c) 135 GeV VBF signal model
130
1
61
0
m (GeV)
140
(d) 140 GeV VBF signal model
160-
0)
0
14 60-
U)
6
140120-
4)
w
100
80-
40
20
-
40
40
60
0
40
0
20-
210
130
140
150
160
M. (GeV)
f3
(e) 145 GeV ggH signal model
14
1
140
150
160
170
M, (GeV)
(f) 150 GeV VBF signal model
Figure A-32: Event Class 4 VBF signal model fits in the pp-y channel for 8 TeV data.
87
88
Appendix B
Isolation Computation
Isolation is computed by summing the transverse momentum of all particle flow (PF)
candidate particles within a AR cone of 0.3 for photons and 0.4 for electrons and
muons. Particle flow particles are a list of all particles in an event that have been
reconstructed from the detector. Electron and muon type PF candidates are not
counted in the isolation cones. Additionally, charged hadrons that did not originate
from the primary vertex associated with the two leptons are not counted in the
isolation sums.
B.1
Photon Isolation
Isolation sums for photons are computed separately for charged hadron, neutral
hadron and photon type PF candidates.
When calculating ArI and AR, the di-
rection of the photon is defined as the line connecting the vertex of the PF candidate
to the supercluster position of the photon. If the photon is detected in the barrel,
charged hadron PF candidates within a AR cone of 0.02 and photon PF candidates
within ATI of 0.015 are not included in the isolation sum. If the photon is detected
in the endcap, charged hadron PF candidates within a AR cone of 0.02 and photon
PF candidates within a AR cone of 0.00864 x Isinh(supercluster I) are not included
in the isolation sum. The photon PF candidate rejection in the endcaps utilizes the
crystal-size (determined from the supercluster 7j) in order to make the number of
89
crystals in the veto cone the same in all y regions.
Pileup corrections are applied to the isolation sums by subtracting the product of
the pile-up energy density in the event and the corresponding effective area for either
charged hadron, neutral hadron or photon type PF candidates [?]. The effective areas
are listed in Table B.1 and depend on the q position of the photon. Cuts are then
placed on the corrected isolation sums. These cuts are shown in Table B.2. The cut
values depend on whether the photon is detected in the barrel or the endcap.
1771
0.0 < ig < 1.0
1.0 < 17| < 1.479
1.479 <
'Jq
Charged
0.012
0.010
Neutral
0.030
0.057
Photons
0.148
0.130
0.014
0.039
0.112
0.012
0.016
0.020
0.012
0.015
0.024
0.039
0.072
0.216
0.262
0.260
0.266
< 2.0
2.0 < In| < 2.2
2.2 < Inj < 2.3
2.3 < i7 < 2.4
2.4 < I
Table B.1: Effective areas for charged hadron, neutral hadron and photon PF candidates used to compute the pileup energy density corrections for the photon isolation
PF Type
Charged Hadrons
Neutral Hadrons
Photons
Barrel
1.5
1.0 + 0.04 x pt
0.7 + 0.005 x pt
Endcaps
1.2
1.5 + 0.04 x pt
1.0 + 0.005 x pt
Table B.2: The isolation sums for the photons are required to be less than the values
listed above. pt is the transverse momentum of the photon. In this analysis, no cut
is placed on the photon type PF candidates if the signal photon is in the endcap.
B.2
Electron Isolation
Charged hadron PF candidates are not counted in the electron isolation sum if they
are within a AR cone of 0.015 when the supercluster IqJ for the electron is greater
than 1.479. Photon PF candidates are not counted in the electron isolation sum if
their supercluster matches the supercluster of the electron. Photon PF candidates
are also rejected from the isolation sum if they are within a AR, cone of 0.08 when
the
T
of the supercluster for the electron is greater than 1.479.
90
Electron 177
0.0 < 177 < 1.0
1.0 < lql < 1.479
1.479 < Jll < 2.0
2.0 < Jll < 2.2
2.2
Jill <2.3
2.3 < Inj < 2.4
2.4 < Jil
EA 2011
0.180
0.200
0.150
0.190
0.210
0.220
0.290
EA 2012
0.190
0.250
0.120
0.210
0.270
0.440
0.520
Table B.3: Effective areas used for the pileup energy density corrections for electrons
The final isolation sums for charged hadron, neutral hadron and photon type PF
candidates are combined together. This combined isolation sum is then corrected for
pileup by subtracting the product of the pile-up energy density in the event and the
corresponding effective area. The electron effective areas are the sums of the effective
areas for neutral hadron and photon type PF candidates. The effective area for the
charged hadron type PF candidates is not needed because charged hadron type PF
candidates that are not associated with the primary vertex of the electron are not
included in the isolation sum. The effectives areas for 2011 and 2012, shown in Table
B.3, depend on the electron q. The corrected combined isolation sum is divided by
the electron pt and required to be less than 0.4.
B.3
Muon Isolation
Charged hadron PF candidates are not counted in the muon isolation sum if their
track in the silicon tracker matched that of the muon. Photon PF candidates are not
counted in the muon isolation sum if the photon PF candidate had a pt less than 0.5
GeV and is within a AR cone of 0.01. Neutral hadron PF candidates are not counted
in the isolation sum if their pt is less than 0.5 GeV.
The final isolation sums for charged hadron, neutral hadron and photon type PF
candidates are combined together. This combined isolation sum is then corrected for
pileup by subtracting the product of the pile-up energy density in the event and the
corresponding effective area. The muon effective areas are the sums of the effective
areas for neutral hadron and photon type PF candidates.
91
The effective area for
the charged hadron type PF candidates is not needed because charged hadron type
PF candidates that are not associated with the primary vertex of the muon are not
included in the isolation sum. The effective areas for 2011 and 2012, shown in Table
B.4, depend on the muon
,i. The
corrected combined isolation sum is divided by the
muon pt and required to be less than 0.12.
Muon 171
0.0 < jyj < 1.0
1.0 < 0IJ < 1.5(1.479)
1.5(1.479) < TIJ < 2.0
2.0 < 1rj < 2.2
2011 EA
0.132
0.120
0.114
0.139
2012 EA
0.674
0.565
0.442
0.515
2.2 < yj < 2.3
2.3 < 177
0.168
0.821
0.189
0.660
Table B.4: Effective areas used for the pile-up energy density corrections for muons.
The ji| in parenthesis indicates the values used for 2012 data.
92
Bibliography
[1] S. Weinberg. A Model of Leptons. Phys. Rev.Lett., 19:1264, 1967.
[2] F.Englert and R. Brout. Broken symmetries and the mass of gauge bosons. Phys.
Rev.Lett., 13:321, 1964.
[3] P.W. Higgs. Broken symmetry and the mass of gauge vector mesons. Phys.
Rev.Lett., 13:508, 1964.
[4] L. Taylor. Observation of a new boson at a mass of 125 GeV with the CMS
experiment at the LHC. 2012.
[5] S. Choi, M. Muhlleitner, and P. Zerwas. Theoretical Basis of Higgs-Spin Analysis
in H -4 yy and Zy Decays. arxiv:1209.5268.
[6] Z. Kunszt. Associated Production of Heavy Higgs Boson with Top Quarks. Nucl.
Phys, B247:330, 1984.
[7] Procedurefor the LHC Higgs boson search combination in summer 2011. Technical 302 Report ATL-PHYS-PUB-2011-011,, 2011.
[8] CMS Collaboration. Search for the standard model Higgs boson in the Z boson
plus a photon channel in pp collisions at V7 = 7 and 8 TeV. CMS PAPER
HIG-13-006, 2013.
[9] A. Read. The CMS hadron calorimeter project:
oai:cds.cern.ch:357152, 1991.
Technical Design Report.
[10] G. Aad et al. CMS Detector Description. Phys. Lett B, 705:452, 2011.
[11] V Karimki. The CMS Tracker System Project:
oai:cds.cern.ch:369412., 1997.
Technical Design Report.
[12] W. Adam, B. Mangano, T. Speer, and T. Todorov. Track reconstruction in the
CMS tracker. 2005.
[13] The CMS Electromagnetic Calorimeter Project:
oai:cds.cern.ch:349375., 1997.
Technical Design Report.
[14] The CMS muon project: Technical Design Report. oai:cds.cern.ch:343814., 1997.
93
[15]
Paolo Nason, Carlo Oleari, Simone Alioli, and Emanuele Re. A general framework for implementing NLO calculations in shower Monte Carlo programs: the
POWHEG BOX. JHEP, 1006:043, 2010.
[16] Paolo Nason, Carlo Oleari, Simone Alioli, and Emanuele Re. NLO Higgs boson
production via gluon fusion matched with shower in POWHEG. JHEP, 0904:002,
2009.
[17] Paolo Nason and Carlo Oleari. NLO Higgs boson production via vector-boson
fusion matched with shower in POWHEG. JHEP, 1002:037, 2010.
[18] Torbojorn Sjostrand, Stephen Mrenna, and Peter Z. Skands.
Physics and Manual. JHEP, 0605.026, 2006.
[19] LHC Higgs Cross Section Working Group.
PYTHIA 6.4
https://twiki.cern.ch/twiki/
bin/view/LHCPhysics/CrossSections, May 2013.
[20] E. Chabanat and N. Estre. Deterministic Annealing for Vertex Finding at CMS.
2003.
[21] V. Beaudette et al. E/Gamma Physics Object Group. https: //twiki. cern.
ch/twiki/bin/viewauth/CMS/EgammaPOG, July 2012. Accessed: 23/08/2012.
[22] CMS Collaboration. Search for a Standard Model Higgs boson decaying into two
photons employing multivariate methods. CMS AN AN-12-048, 2012.
[23] I. Mikulec et al. Muon Physics Object Group. https: //twiki. cern. ch/twiki/
bin/viewauth/CMS/MuonPOG, July 2012. Accessed: 23/08/2012.
[24] F. James and M. Roos. Minuit. A system for Function Minimization and Analysis
of the Parameter Errors and Correlations. Comput.Phys.Commun, 10:343-367,
1975.
[25] Roger J. Barlow. EVENT CLASSIFICATION USING WEIGHTING METHODS. J.Comput.Phys, 72:202, 1987.
[26] J. Gainer, Wai-Yee Keung, I. Low, and P. Schwaller. Looking for a light Higgs
boson in the overlooked channel. Phys.Rev, D86 033010, 2012.
[27] Paul Avery. Applied Fitting Theory VI: Formulas for Kinematic Fitting.
http://www.phys.ufl.edu/ avery/fitting/kinematic.pdf, 1999.
[28] A. Read. Modified frequentist analysis of search results (the CLs method). Technical Report CERN-OPEN-2000-005, CERN, 2000.
[29] Andreas Hoecker, Peter Speckmayer, Joerg Stelzer, Jan Therhaag, Eckhard von
Toerne, and Helge Voss. TMVA: Toolkit for Multivariate Data Analysis. PoS,
ACAT:040, 2007.
94