Higgs Discovery

Topics in Contemporary Physics
Discovery of the
Higgs Boson
Luis Roberto Flores Castillo
Chinese University of Hong Kong
Hong Kong SAR

February 25, 2015
PART 1
Brief history
Basic concepts
Colliders & detectors
From Collisions to
papers
5
S
ATLA
(*)
s=8
TeV:
/ GeV
-1
t = 4.8
V: Ld
Ldt =
fb
-1
5.8 fb
s=7
2000
1800
TeV,
Ldt =
-1
5.9 fb
1200
1000
800
600
10
nary
ATLAS
400
5
150
0
TeV,
s=8
1600
1400
100
Prelimi
200
250
0
]
[GeV
200 m100
4l
- Bkg
15
Te
s=7
GeV)
sample
2
126.5
oton
and 201 fit (m H =
2011
usive
Data
Bkg incl
-1
Sig +
nomial
4.8 fb
er poly
Ldt =
4th ord
d diph
Selecte
2400
2200
Events
Unc.
Syst.
20
4l
Data
V
ts/5 Ge
Even
The Higgs discovery
HZZ
(*)
Data
ZZ
round
s, tt
Backg
Z+jet
round
V)
Backg
25 Ge
l (m H=1
Signa
25
150
140
-100
100
160
V]
m [Ge
130
0
120
110
BSM
MVA Techniques
The future
L. R. Flores Castillo
CUHK
February 25, 2015
last time:
From proposals to publications
Statistical tools II
The Monte Carlo Method
Event generators
Detector simulation
Testing significance / goodness of fit
p-value
Significance
Limits
Quick reminders:
Homework today in the mailbox for the course
Midterm on Friday, February 27
Announcement:
Undergraduate student helper for webpage development
CUHK
February 25, 2015
Reminder: interactions
QED:
QCD:
SM Particle Content
Cabibbo-Kobayashi-Maskawa matrix
Weak:
NO Flavor-Changing-Neutral-Currents
W/Z:
W/Z/:
CUHK
February 25, 2015
Reminder: Relativistic kinematics

Maxwell
equations
"
v %
t ' = $t 2 x '
# c &
x ' = (x vt)
y' = y
z' = z
Lorentz
transformations
c for all observers
x 0 = ct,
1
x = x,
x 2 = y,
x3 = z
x 0 ' = (x 0 x1 )
Four-vector
x1 ' = (x1 x 0 )
x ' = x
x2 ' = x2
x3 ' = x3
time-position:
proper velocity:
energy-momentum:
v
c
E = mc 2 = mc 2 + 12 mv 2 + 83 m cv4 +...
4
contravariant
covariant
I (x 0 )2 (x1 )2 (x 2 )2 (x 3 )2
I=x x
is Lorentz-invariant
Energy-momentum
p p = (m )(m ) = m 2 c 2
a b a b
2 4
2 2
E = mc
E = m c +p c
E2
p p = 2 p 2
c
Scalar product:
Useful:
p = mv
2
For v=c,
CUHK
x = (ct, x, y, z)
=dx/d = (c, vx, vy, vz)
p = m = (E/c, px, py, pz)
February 25, 2015
p / E = v / c2
v = pc 2 / E
E = hv
5
Symmetry, conservation laws, groups

1917: Emmy Noethers theorem:
Every symmetry yields a conservation law
Conversely, every conservation law reflects
an underlying symmetry
A symmetry is an operation on a system that leaves it invariant.

i.e., it transforms it into a configuration indistinguishable from the
original one.
The set of all symmetry operations on a given system forms a group:
Closure: If a and b in the set, so is ab
Identity: there is an element I s.t. aI = Ia = a for all elements a.
Inverse: For every element a there is an inverse, a-1, such that aa-1 = a-1a = I
Associativity: a(bc) = (ab)c
if commutative, the group is called Abelian
CUHK
February 25, 2015
Angular Momentum
Classically, orbital (rmv), spin (I) not different in essence.
In QM,
Spin interpretation no longer valid
All 3 components cannot be measured simultaneously; and most we can measure:
the magnitude of L ( L2 = L L ).
Allowed values: j(j+1)2
one component (usually labeled z)
Allowed values: -j,,j in integer steps
2j+1 possibilities
Differences:
Orbital angular momentum (l)
Allowed values
integer
integer or half integer
any (integer) value
fixed
For each particle type
Ket notation: l ml ,
s ms ,
Spin angular momentum (s)
j mj
A particle with spin 1 :

a particle with s=1
simple label, not the magnitude of its spin angular momentum:
S 2 = s(s +1) !
notice that the magnitude is always a bit larger than the maximum z component
Spin-statistics connection: half-integer spin: fermions; integer spin: bosons.

CUHK
February 25, 2015
Addition of Angular Momenta

Besides total angular momentum, sometimes we need the specific states:
j1+ j2
j1 m1 j2 m2 =
Cmj jm1 1j2m2 j m , with m = m1 + m2
j= j1 j2
2 1
Clebsch-Gordan
coefficients
1
2
1
2
State of the
separate systems
CUHK
2 5
5 2
(Particle Physics Booklet,

internet, books, etc.)
12
3 3
5 2
12
Linear combination of angular

momentum eigenstates of the
combined system
February 25, 2015
Spin
Spinors: two-component column vectors:
ms= (spin up, ), ms= (spin down, )
General state of a
spin particle:
! $
!
$ !
$
##
&& = # 1 & + # 0 &
" 0 % " 1 %
" %
Operators for the three

spin projections:
or in terms of the
Pauli spin matrices:
1
2
! 1 $
=#
&
0
"
%
and complex, and
1
2
1
2
" 0 %
=$
'
1
#
&
+ =1
! ! 0 1 $ ! ! 0 i $ ! ! 1 0 $
Sx = #
&, Sy = #
&, Sz = #
&
2" 1 0 %
2" i 0 %
2 " 0 1 %
S =
! 0 1 $
! 0 i $
! 1 0 $
x =#
&, y = #
&, z = #
&
" 1 0 %
" i 0 %
" 0 1 %
( )
!
2
Probability of possible
outcomes in a QM system:
Rotating a spinor:
! ' $
! $
#
&
#
&
# ' & = U ( ) # &
"
%
"
%
1
2
CUHK
Construct matrix representing observable A

Allowed values of A are the eigenvalues of
Write state as linear combination of these eigenvectors
The probability to measure ei is |ci|2
where
U ( ) = e i( )/2
is a vector pointing along the axis of rotation,
and its magnitude is the angle of rotation.
e A = 1+ A + 12 A2 + 3!1 A3 + ...
February 25, 2015
Flavor Symmetries
Neutron & proton very similar wrt strong interaction
mp = 938.28 MeV/c2; mn = 939.57 MeV/c2.
! $
Two states of the same particle?
&
N = ##
&
Heisenberg introduced Isospin I.
" %
Borrowing angular momentum machinery:
!
$
!
$
p = ## 1 && , n = ## 0 &&
" 0 %
" 1 %
Nucleon carries isospin

Third component, I3 (from an abstract space), has eigenvalue + , -.
Strong interactions are invariant under rotations in isospin space.

by Noethers theorem, isospin is conserved in strong interactions
= 1 1 ,
0 = 1 0 ,
+ = 1 1
CUHK
3
2
23 ,
0 =
3
2
12 ,
+ =
February 25, 2015
3
2
1
2
++ =
3
2
3
2
10
Discrete Symmetries. Parity

Nature is not invariant under mirror reflection (parity)
1956, Lee&Yang: no exptl constrain on Weak force. C.S. Wu
carried out the test that they proposed.
Maximal P violation:
all neutrinos are left-handed
all anti neutrinos are left-handed
Strong and EM forces do respect Parity; Weak force does not
scalar
P(s) = s
time, m, E, charge density
pseudoscalar
P(p) = -p
helicity, magnetic flux
vector (polar vector)
P(v) = -v
x, v, a, p, F, electric field,
psudovector (axial v)
P(a) = a
angular momentum, mag. field,
Eigenvalues +1 and -1
Hadrons (baryons, mesons) are eigenstates of P, correspond to either P=+1 or P=-1
Fermions: P(particle) = - P(antiparticle)
Bosons: P(particle) = P(antiparticle)
Quarks: positive parity
Photons: intrinsic parity -1
Composite system: product of parities of its constituents (in its ground state)
Excited states of two particles: additional factor (-1)l where l is the orbital angular momentum.
CUHK
February 25, 2015
11
From Proposal to Publications

Energy: known energy target? promising models? interesting new effect to study?
Luminosity: How much needed for new knowledge? 1032,34 cm-2s-1? is it achievable?
Investment: ~10 B USD for the tunnel + detectors, ~1 B USD/yr for operations
~ 15 years for government approval, 10-15 years for construction
Detector and capability design

expected future capabilities in accelerator, detector, computing technologies
Accelerator, detectors
collisions, hits in the detectors
Triggers
event streams / datasets (leptons, jets, MET,, combs)
Reconstruction software physics objects: identification, obtain their 4-momenta
Computing, networking
storage & processing capabilities
Grid processing
search/measurement specific; selection and treatment
Theoretical models
expected physics phenomena
MonteCarlo simuation expected observation
Statistical analysis
physics results
Collaboration-wide review publication
CUHK
February 25, 2015
12
Trigger systems
Systems that select which collision events
to store or throw away
Real trigger rate = R/(1 + R) per s
Rate in Hz Dead time ms. Live time % Trigger rate Hz

10
1000
10
10
91
9.1
9.1
91
To reduce dead time: multi-level systems

Detector output usually pipelined to allow
more decision time at Level 1
High-level trigger these days use PC
farms to increase decision time per event
One event per computing node
Level
decision
time up to
few s
ASICs ,FPGAs
Pipelines to reduce deadtime
Low granularity / precision
100 ms
Processor farms (1 ev/proc)

Use of hi-speed networks
3D reco, fine-grain calo, tracking,

Sub-detectors processed in parallel
seconds
Proc. farms (1 ev/proc)

Standard server PCs
Event building (data from all detectors)

Fulll or partial event reconstruction
Technology
CUHK
Comments
February 25, 2015
13
Statistics tools
Goal: measure properties in data and compare with theoretical expectation
Several sources of uncertainty (QM, measurement uncertainties, ) should be modeled
Random variable: a numerical characteristic assigned to an element of the sample
space; can be discrete or continuous.
f(x): probability density function (pdf)
Histograms and pdfs
Often several variables are involved: joint pdf, marginal pdf, conditional pdf,
x1, x2 independent if
Expectation value (E), variance (V), standard deviation ():
CUHK
February 25, 2015
14
Some pdfs, The MC Method

Binomial
N independent experiments, each with success probability p
RV: number of successes
Poisson
Limit of the Binomial for large N and small p, but Np v
Uniform
It can be used to produce rv with other pdfs

The Monte Caro Method
Numerical technique to calculate probabilities and related quantities using sequences
of random numbers.
CUHK
February 25, 2015
15
The Monte Carlo Method

Technique to calculate probabilities & related quantities using random numbers sequences.
(1) Generate sequence r1, r2, ..., rm uniform in [0, 1]
(2) Transform it into a sequence x1, x2, ..., xn distributed according to the pdf of interest, f(x)
(3) Use the x values to estimate some property of f(x), e.g., fraction of x values a < x < b
MC-generated values are often used as simulated data
used to test or improve statistical procedures
MC event generators [ Available: PYTHIA, HERWIG, ISAJET, ]
Production of four-momenta of true particles.
Output: events (simulated collision results); each has a list of particles & their momenta.
MC detector simulation [ Available: GEANT ]
Input: events from MC generator; simulates detector response
multiple Coulomb scattering (generate scattering angle)
electromagnetic, hadronic showers
particle decays (generate lifetime)
production of signals
ionization energy loss (generate E)
electronics response , ...
Output: simulated raw data; input to reconstruction software (track finding, fitting, etc.)
Goals
Obtain expected detector level results for a given generator level hypothesis
Compare expected results w/data
Estimate efficiencies = # events found / # events generated.
CUHK
February 25, 2015
16
Hypothesis tests
Hypothesis pdf, experiment observations
Statement about H?
p-value: probability, under assumption of H, to observe data with equal
or lesser compatibility with H ( than in the data we got).
This is not the probability that H is true!
We frame it so that a very small value of p allows us to reject H.

Significance (Z): number of standard deviations in a Gaussian
distribution that leave an area in the tail equal to the p-value.
Z = 5 (a 5 sigma effect) means p = 2.9 10-7
Significance of a peak
Poisson-distributed bins
p value: P(# in data or more)
p-value Z
Setting limits:
If no clear signal, what values of s (expected # signal events) can we rule out?
95% CL: find the values of s for which the p-value is below 5%
CUHK
February 25, 2015
17
The Discovery of the Higgs boson
18
CERNs Large Hadron Collider

Geneva
Airport
27km circumference, 50-150m below ground, across French-Swiss border

pp collisions @ s = 7 TeV in 2011 and 8 TeV now
Each beam ~ 1400 proton bunches, each bunch ~ 1.151011 protons
On average, ~ 20 pp collisions per bunch crossing
times 20 MHz bunch crossing rate = 400 M pp collisions per second
Reduced by trigger systems to ~400 events stored per second
CUHK
February 25, 2015
19
High luminosity
Cross sections: barns (area: 10-28 m2 ~ U nucleus): fb, pb
Amount of data: fb-1, pb-1. Data Cross-section = # events (dimensionless)
1 fb-1 ~ 71013 collisions

(2011 conditions)
6.63 fb-1 delivered

6.28 fb-1 recorded
(efficiency 94.7%)
5.86 fb-1 after data
quality requirements
(available for physics
analysis; efficiency
93.3% )
LHC produced pp collisions until Dec. 17, 2012 before a 2 year shut down
Obtained a total ~25 fb-1 at 8 TeV in 2012
Bunch spacing : 50 ns
CUHK
February 25, 2015
20
at a price
Z event from 2012 data with 25 reconstructed vertices
Many pp interactions per bunch crossing (in-time pile-up) and remnants

of the detector response to the previous interaction (out-of-time pile-up)
Pile-up-robust algorithms developed (for both trigger & offline analyses)
Reconstruction and identification of physics objects (e, , , , jet, ETmiss)
optimized to be robust against pile-up
CUHK
February 25, 2015
21
Particle identification
Measure energy and momentum
Measure energy
photon
muon
electron
CMS
~ 3300 physicists (~1500 students)
179 institutes, 41 countries
CUHK
February 25, 2015
22
Particle identification
Measure energy and momentum
Measure energy
photon
electron
CMS
Reconstruction and identification are

not perfect:
A very narrow jet of particles can look
like an electron
A high energy pion may show up in the
muon detectors (mis-idd as a muon)
If a track for an electron is not found, it
will look like a photon
Good precision when measuring e,
Jets are messier and very frequent
muon
~ 3300 physicists (~1500 students)

179 institutes, 41 countries
CUHK
February 25, 2015
23
Coordinate system
LHC
y
x
proton
ATLAS detector
n
o
t
o
r
p
Right handed coordinate system

z along beam pipe, x to center of LHC, y points upwards
Transverse plane: polar coordinates (r,)

azimutal angle, around the beampipe
Pseudorapidity: = - ln[tan(/2)], is the polar angle

CUHK
February 25, 2015
24
102
(pp H+X) [pb]
SM Higgs production
Gluon fusion
(dominant
at LHC)
pp
s= 8 TeV
H(
NN
LO
+N
NL
LQ
CD
10
Vector
boson
fusion
Associated
production
with Z/W
pp
10-1
pp
pp
ZH
qqH
(NN
LO
(N
+N
QC
LO
D+
(N
EW
)
NLO
NL NL
O
pp
O
Q
Q
CD
C
ttH
D
+N + N
(N
LO
LO LO
QC
EW E
W
D)
)
)
LHC HIGGS XS WG 2012
SM Higgs production
EW
)
(Higgs-strahlung)
Associated
production
with top
10-2
80 100
CUHK
200
300
February 25, 2015
1000
MH [GeV]
25
Gluon fusion
(dominant
at LHC)
~2000 fb at mH=125 GeV

In 5fb-1: ~100k sproduced!
= 8 TeV
102
(pp H+X) [pb]
SM Higgs production
pp
H(
NN
LO
+N
NL
LQ
CD
10
Vector
boson
fusion
Associated
production
with Z/W
pp
10-1
pp
pp
ZH
qqH
(NN
LO
(N
+N
QC
LO
D+
(N
EW
)
NLO
NL NL
O
pp
O
Q
Q
CD
C
ttH
D
+N + N
(N
LO
LO LO
QC
EW E
W
D)
)
)
SM Higgs production
EW
)
(Higgs-strahlung)
Associated
production
with top
10-2
80 100
CUHK
200
300
February 25, 2015
1000
MH [GeV]
26
Branching ratios
(WW, ZZ, tt, bb, )
bb
WW
ZZ
10-1
tt
gg
Direct coupling to
massive particles
Higgs BR + Total Uncert
SM Higgs decays
cc
10-2
Through a triangle loop

to massless ones
10-3
100
200
300
400 500
1000
MH [GeV]
(gg, )
CUHK
February 25, 2015
27
Branching ratios
(WW, ZZ, tt, bb, )
bb
Direct coupling to
massive particles
Higgs BR + Total Uncert
SM Higgs decays
WW
ZZ
10-1
tt
gg
cc
10-2
Through a triangle loop

to massless ones
10-3
100
200
300
400 500
1000
MH [GeV]
100,000 Higgs bosons produced times ~0.002

BR to , ~ 200 Higgs to be found via search
(gg, )
CUHK
February 25, 2015
28
Introduction
10
s = 8TeV
+-
WW l qq
+ -
VBF H
WW l+l
10-1
ZZ l+l qq
-
ZZ l+l
-2
10
ZZ l+l l+l
-3
10
WH l bb
ttH ttbb
ZH l+l bb
10-4 100
200
l = e,
= e,,
q = udscb
300
400
MH [GeV]
BR [pb]
Cross section times BR
WW, ZZ split into decay modes

Targeting production modes
can improve sensitivity
Not yet the full story!
Missing: triggers, efficiencies,
resolutions, background cross
sections, rejection for each, etc.
Low mH: is largest (cons:
detection and backgrounds)
High mH: llvv most sensitive
Experimentally, 100<mH<200 is
accessible in the most ways
All modes labeled in the plot
(and more) have been studied;
here, well focus on three
https://twiki.cern.ch/twiki/bin/view/LHCPhysics/CrossSectionsFigures
CUHK
February 25, 2015
29
Introduction
10
s = 8TeV
+-
WW l qq
+ -
VBF H
WW l+l
10-1
ZZ l+l qq
-
ZZ l+l
-2
10
ZZ l+l l+l
-3
10
WH l bb
ttH ttbb
ZH l+l bb
10-4 100
200
l = e,
= e,,
q = udscb
300
400
MH [GeV]
BR [pb]
Cross section times BR
WW, ZZ split into decay modes

Targeting production modes
can improve sensitivity
Not yet the full story!
Missing: triggers, efficiencies,
resolutions, background cross
sections, rejection for each, etc.
Low mH: is largest (cons:
detection and backgrounds)
High mH: llvv most sensitive
Experimentally, 100<mH<200 is
accessible in the most ways
All modes labeled in the plot
(and more) have been studied;
here, well focus on three
https://twiki.cern.ch/twiki/bin/view/LHCPhysics/CrossSectionsFigures
CUHK
February 25, 2015
30
Limits and p0 plots
Always a probabilistic
statement
102
95% CL limit on /SM
Null search results do provide

valuable information:
What signal sizes can be ruled out?
Need reliable background
estimations
Observed CLs
ATLAS
Expected CLs
(*)
H ZZ 4l
s=7 TeV: Ldt =4.8 fb-1
10
Need to state the CL (95%)
Being a random process,

uncertainty bands are needed
Expected: median of limits if the
signal does not exits
Observed: from the actual dataset
CUHK
10-1
110
200
300
400
500
600
mH [GeV]
Too few events strong limit

Too many events weak limit
February 25, 2015
31
Limits and p0 plots

Too many events may also,
instead, represent a signal
do they?
We quantify it by the probability
that background alone would
produce an excess as large as
observed (or larger)
Local p0
Instead of quoting p0, we refer
to it using the number of
sigmas that it would represent
in a Gaussian tail.
January 2012 PRL Publication
Expected from SM
Higgs at given mH
1 sigma p0 = 16%
3 sigma p0 = 0.13%
5 sigma p0 = 2.910-7
CUHK
February 25, 2015
32
General comments
The discovery was achieved through the statistical
combination of several channels
The main ones:
H
HZZ4l
HWWlvlv
Each one has its own strategies, backgrounds and

challenges
The LHC detectors were designed w/these searches in
mind
Refinements for over 20 years
Full simulation, test beams data, refined with data
Large teams for each channel
CUHK
February 25, 2015
33
CUHK
February 25, 2015
34
Overview
Search for a narrow m peak

110 < m < 150 GeV
Tens of signal events over a
relatively large background
x BR ~50 fb @ mH ~126 GeV
Events / 1 GeV
800
Inclusive diphoton sample

Data 2011
Background model
SM Higgs boson mH = 120 GeV (MC)
700
600
s = 7 TeV,
500
Ldt = 4.9 fb
-1
400
300
Simple topology:
2 high-pT isolated photons
ET(1, 2) > 40, 30 GeV
100
Data - Bkg model
Main backgrounds:
continuum (irreducible)
-jet, jet-jet (reducible)
200
0
100
100
50
0
-50
-100
100
ATLAS Preliminary
110
120
130
140
150
160
110
120
130
140
150
160
m [GeV]
[Above: from CERN Council

meeting, on Dec 13, 2011]
CUHK
February 25, 2015
35
Background composition
The , -jet and jet-jet contributions can be decomposed using control

samples defined by photon identification and isolation.
~ 75-80%
-jet ~ 20%
jet-jet~ 2%
Considerable effort made in background modeling to avoid biases.

Various background models considered, different for different categories.
CUHK
February 25, 2015
36
Mass resolution for signal

Mass resolution not
affected by pile-up
Mass resolution of
inclusive sample:
1.6 GeV
CUHK
February 25, 2015
37
Event selection
Neural-net based photon ID for 2011 data

Re-optimized cut-based photon ID for 2012, stable with high pileup
New 2-jets category for enhanced VBF sensitivity
Events divided in 10 categories based on:

pseudorapidity
whether is converted/unconverted
pTt (pT perpendicular to thrust axis);
2 jets (VBF-like)
thrust axis
CUHK
February 25, 2015
38
Weighted mass distribution
Weighted invariant mass distribution of selected events combining

2012 (35k events) and 2011(24k) data
From July 31, 2012 publication
For mH=126.5 GeV

BR = 39 fb at 7 TeV
BR = 50 fb at 8 TeV
Full results obtained by splitting
data into 10 categories, fitting
mass distributions separately.
Signal
Yield
Observed
Expected
2011
146.9
79.6
2012
205.5
110.5
Weight for category i: ln(1+Si/Bi)

CUHK
February 25, 2015
39
Results
p0: probability for background

to fluctuate up to observed
data (or higher)
Expected from SM
Higgs at given mH
4.5
Data
mH at max deviation
Local p0
Local significance
Expected
2011
126 GeV
310-4
3.4
1.6
2012
127 GeV
710-4
3.2
1.9
Both
126.5 GeV
310-6
4.5
2.5
Global 2011+2012 (including LEE over 110-150 GeV range): 3.6

CUHK
February 25, 2015
40
Signal strength
Br
( Br )SM
Signal strength () =
(signal rate from fit to data) / (expected SM signal rate at given mH)
Fitted signal strength: =1.90.5 at mH=126.5 GeV
About twice the value expected from the SM Higgs cross section!
CUHK
February 25, 2015
41
HZZ(*)4l
4 candidate with m4=125.1 GeV

pT (muons)= 36.1, 47.5, 26.4, 71 .7 GeV m12= 86.3 GeV, m34= 31.6 GeV. 15 reconstructed vertices
CUHK
February 25, 2015
42
Search for a narrow m4l peak

110 < m < 600 GeV
A few signal events over a small
background; S/B varies with m4l
x BR ~5 fb @ mH ~125 GeV
Overview
Events/5 GeV
HZZ(*)4l
10
8
Preliminary
(*)
The golden channel:

4 isolated leptons from IP
mass resolution ~2% @130 GeV
DATA
ATLAS
Background
Signal (m =125 GeV)
H
Signal (m =150 GeV)
H
Signal (m =190 GeV)
HZZ 4l
-1
Ldt = 4.8 fb
s = 7 TeV
4
2
At low mH:
one Z on-shell 0
(m12), the other 100
is off-shell (m34)
Four final states: 4e, 4, 2e2, 22e

CUHK
150
200
250
m4l [GeV]
[Above: from CERN Council

meeting, Dec 13, 2011]
February 25, 2015
43
HZZ(*)4l
Background rejection
Other processes can end up with four leptons

t tbar : each to Wb, Wlv, bjet with a lepton
Zbb : Zll, each b to a jet with a lepton
Z+light jets: Zll, jets misreconstructed as electrons
ZZ production (no Higgs)
In tt, Z+jets and Zbb (reducible):

There are other particles around the leptons
Leptons from b decays come from secondary vertices
Reduced by requiring that leptons are isolated and non-displaced
ZZ production is similar to the signal (Irreducible)
CUHK
February 25, 2015
44
HZZ(*)4l
Event counts in
120<m4l<130 GeV
7+8 TeV
2e2 + 22e
4e
Sum
Background
1.3 0.1
2.1 0.2
1.5 0.2
4.9 0.3
mH=125 GeV
2.1 0.3
2.3 0.3
0.9 0.1
5.3 0.4
13
1.6
1.1
0.6
1.1
Observed
S/B
Low mass region
CUHK
February 25, 2015
45
HZZ(*)4l
Low mass region
Adding both
years data:
CUHK
February 25, 2015
46
95% CL limit on /SM
10
Observed CLs
ATLAS
Expected CLs
(*)
H ZZ 4l
10
1
2
Exclusion limits
10
95% CL limit on /SM
HZZ(*)4l
Observed CLs
ATLAS
Expected CLs
(*)
H ZZ 4l
10
10-1
110 120 130 140 150 160 170 180
mH [GeV]
10-1
110
200
300
400
500
600
mH [GeV]
Exclusion at 95% C.L. :

Expected:
124 < mH < 164 GeV and 176 < mH < 500 GeV
Observed:
131 < mH < 162 GeV and 170 < mH < 460 GeV
For mH ~120-130 GeV weaker limit than expected in background-only hypothesis

CUHK
February 25, 2015
47
HZZ(*)4l
Significance
102
10
Obs. combined
Exp. combined
Obs. 2011
Exp. 2011
Obs. 2012
Exp. 2012
ATLAS
(*)
H ZZ 4l
1
10-1
-2
10
(*)
H ZZ 4l
Best Fit
-2 ln () < 1
3.5
3
2.5
ATLAS
s = 7 TeV: Ldt = 4.8 fb-1
s = 8 TeV: Ldt = 5.8 fb-1
2
1.5
1
10-3
10-4
Signal strength ()
Local p
Best-fit value at 125 GeV: =1.4 0.6
0.5
0
3.6
10-5
110 120 130 140 150 160 170 180
mH [GeV]
Data sample
2011
2012
2011+2012
mH at max deviation
125 GeV
125.5 GeV
125 GeV
-0.5
110 120 130 140 150 160 170 180
local p-value
1.1 %
0.4 %
0.03 %
mH [GeV]
local significance
2.3
2.7
3.4
expected
1.5
2.1
2.6
Global 2011+2012 (including LEE over full 110-141 GeV range): 2.5
CUHK
February 25, 2015
48
HWW(*)lvlv
CUHK
February 25, 2015
49
HWW(*)4l
Overview
W+
0.18 pb at 7 TeV ~ 900 in 5 fb-1

l+
W- _
0.23 pb at 8 TeV ~ 1150 in 5 fb-1
l-
Large cross section, but the neutrinos cannot be detected

Detectors cover almost all 4 of solid angle, so we can
calculate the missing momentum
but only the transverse component is constrained
Limited mass information can be recovered:
Transverse mass: mT =
(E
ll
T
miss 2
T
+E
)p
ll
T
miss 2
T
+E
ll
T
with E =
ll 2
T
+ mll2
Rather than a peak, a signal produces a broad distribution

CUHK
February 25, 2015
50
Data
WW
ATLAS
60
s = 8 TeV, Ldt = 5.8 fb-1
50
tt
Z+jets
(*)
HWW e + 0 jets
mT distributions
Events / 10 GeV
Events / 10 GeV
HWW(*)4l
BG (sys stat)
WZ/ZZ/W
Single Top
W+jets
H [125 GeV]
e, 0-jet
signal is stacked
40
30
60
Data
WW
ATLAS
50
s = 8 TeV, Ldt = 5.8 fb-1
tt
Z+jets
(*)
HWW e + 0 jets
H [125 GeV]
40
e, 0-jet
signal is stacked
30
20
20
10
10
0
50
100
150
200
250
0
50
300
100
150
200
50
ATLAS
s = 8 TeV, Ldt = 5.8 fb
Bkg. subtracted Data

-1
H [125 GeV]
(*)
HWW e / e + 0/1 jets
40
Bkg-subtracted
data, 2012 only.
0/1 jets
30
20
Events / 10 GeV
60
100
80
60
0
80 100 120 140 160 180 200 220 240
ATLAS
s = 7 TeV, Ldt = 4.7 fb
s = 8 TeV, Ldt = 5.8 fb
CUHK
H [125 GeV]
-1
HWW ll + 0/1 jets
Bkg-subtracted
data, 2011 + 2012
0/1 jets
-20
60
80 100 120 140 160 180 200 220 240
mT [GeV]
Bkg. subtracted Data

-1
40
20
60
300
(*)
10
-10
250
mT [GeV]
mT [GeV]
Events / 10 GeV
BG (sys stat)
WZ/ZZ/W
Single Top
W+jets
mT [GeV]
February 25, 2015
51
ATLAS
103
Observed
102
Expected
10
1
10-1
10-2
10-3
10-4
10-5
10-6
10-7
110 120 130 140
2011+2012 data
(*)
HWW ll
s = 7 TeV: Ldt = 4.7 fb-1
s = 8 TeV: Ldt = 5.8 fb-1
0
1
2
3
4
5
150
160
170
180 190
mH [GeV]
Observed
significance
310-3
Expected
significance
2.8
Signal strength ():
(*)
ATLAS
HWW ll
Best fit
-2 ln () < 1
Exp. mH = 126 GeV
-2 ln () < 1
s = 7 TeV: Ldt = 4.7 fb-1

s = 8 TeV: Ldt = 5.8 fb-1
4
3
2
1
0
115
120
125
130
135
140
145 150
mH [GeV]
Event yields in 2012:
mH = 125 GeV:
p0
Signal strength ()
Local p
HWW(*)lvlv
2.3
= 1.4 0.5
Signal
Background
Observed
0-jet
1-jet
2-jet
204
52
0.340.07
14216
266
0.350.18
185
38
(At 126.0 GeV, =1.30.5 GeV)

CUHK
February 25, 2015
52
Observation paper
On July 31st, 2012 both
ATLAS and CMS
submitted paper drafts
to PLB
ATLAS incorporated
the WW analysis into
the combination
CUHK
February 25, 2015
53
Observation paper
Combined significance
Local p
Adding 2012 HWW data to the ATLAS Higgs combination:

ATLAS
2011 - 2012
Obs.
Exp.
1
s = 7 TeV: Ldt = 4.6-4.8 fb-1

s = 8 TeV: Ldt = 5.8-5.9 fb-1
1
10-1
10-2
10-3
10-4
10-5
10-6
10-7
10-8
10-9
10-10
10-11
110
ATLAS
2011 - 2012
mH = 126.0 GeV
W,Z H bb
s = 7 TeV: Ldt = 4.7 fb-1
0
1
2
H WW l l
s = 7 TeV: Ldt = 4.6-4.7 fb
-1
(*)
s = 7 TeV: Ldt = 4.7 fb-1

s = 8 TeV: Ldt = 5.8 fb-1
s = 7 TeV: Ldt = 4.8 fb-1
-1
s = 8 TeV: Ldt = 5.9 fb
(*)
6
115
120
125
130
135
140
145 150
mH [GeV]
H ZZ 4l
s = 7 TeV: Ldt = 4.8 fb-1
s = 8 TeV: Ldt = 5.8 fb-1
Combined
s = 7 TeV: Ldt = 4.6 - 4.8 fb
-1
s = 8 TeV: Ldt = 5.8 - 5.9 fb
-1
= 1.4 0.3
-1
For 6, the chances that the events observed

were due to random fluctuations are less than
one in 1000 million.
CUHK
February 25, 2015
0
1
Signal strength ()
54
Conclusion
55
Three most sensitive channels

Strength at
mH=126 GeV
1.8 0.5
Local p
10 categories
102
10
Obs. combined
Exp. combined
Obs. 2011
Exp. 2011
Obs. 2012
Exp. 2012
ATLAS
(*)
H ZZ 4l
1
10-1
10-2
4 channels
1.2 0.6
10-3
10-4
10-5
110 120 130 140 150 160 170 180
mH [GeV]
l+
H
W-
Local p
W+
l_
v
3 channels
(based on jet multiplicity)
CUHK
ATLAS
103
Observed
102
Expected
10
1
10-1
10-2
10-3
10-4
10-5
10-6
10-7
110 120 130 140
(*)
HWW ll
s = 7 TeV: Ldt = 4.7 fb-1
s = 8 TeV: Ldt = 5.8 fb-1
0
1
2
3
4
1.4 0.5
5
150
160
170
February 25, 2015
180 190
mH [GeV]
56
Combination summary
4+4+2+10+18+6+4+7+1+3+4+4+4+10+6 = 87 !!!
Common parameters: mH, , lumi uncertainty,
CUHK
February 25, 2015
57
Correlated systematic uncertainties

Integrated luminosity (3.9% for 2011, 3.6% for 2012)
Electron and photon trigger and identification efficiencies
Electron and photon energy scales: five parameters
(calibration method, presampler ES in B and EC,
material)
Muon reconstruction, separate for ID and MS
Jet energy scale and missing transverse energy
(dependent on pT, , jet flavor, specific treatment for bjets)
Sources affecting 7 & 8 TeV data fully correlated
Uncertainties on background estimates based on control
samples considered uncorrelated between 7 and 8 TeV
CUHK
February 25, 2015
58
ATLAS combination model
Each channel has its own data streams, triggers, control

regions, main backgrounds, systematic uncertainties,
Each team develops its own code for the analysis
All are put it into a common file format to allow the combination
Non-negligible amount of work on just naming conventions!
CUHK
February 25, 2015
59
Three views of the combination
As a limit: fluctuations around

expected except ~125 GeV
Probability that the excess comes

from background only: below 2
everywhere except ~125 GeV
Signal strength (SM=1): compatible

with 0 except ~125 GeV
CUHK
February 25, 2015
60
Conclusion
at the time
A new resonance has been observed in the search for the SM

Higgs boson
Mass ~ 126 GeV

No other Higgs-like particle between 111 and 559 GeV
It decays to 2 photons and to 4-leptons: it is a boson
Its couplings to gg, ZZ and WW are ~ compatible with SM Higgs
Observed
significance
Expected
significance
Signal strength ()
[ at mH=126.0 GeV]
4.5
2.5
1.8 0.5
H4-leptons
3.6
2.7
1.2 0.6
HWW
2.8
2.3
1.3 0.5
Combined with
other channels
5.9
4.9
1.4 0.3
CUHK
February 25, 2015
61
Since then
Measurement of the properties of the new particle
Spin measurement
Confirmed to be consistent with 0 on March 2013
All elementary particles we know so far are either or 1
This is the first elementary scalar
Couplings to fermions and cross section

@discovery, only boson couplings: ZZ, WW,
Strong evidence for fermion couplings came later (Jan 2014)
Currently, preparation work for the new data-taking

period, at 13 TeV (instead of 8), starting this May.
CUHK
February 25, 2015
62
Conclusion
Possibilities
If the large signal strength () for H (1.90.5 for ATLAS, 1.60.4 for
CMS) is confirmed with more data, new physics may surface:
In the SM, the most important Higgs production process in pp interactions is
gluon fusion through a top triangle.
production
decay
t, b, W, etc. (charge)
W, t
W, t
W, t
t, b, etc. (color)
If the SM is not the whole story, additional heavy particles could contribute to
these triangles:
New particles with color would contribute to the production triangle
New particles with charge would contribute to the decay loop
The presence of such new particles may enhance Higgs production and its
decay into two photons
CUHK
February 25, 2015
63
CUHK
February 25, 2015
64
~ 70%
~ 25%
~ 5%
CUHK
February 25, 2015
65
~ 70%
~ 25%
~ 5%
CUHK
February 25, 2015
66
Backup
67
Multivariate analysis tools
CUHK
February 25, 2015
68
Multivariate analysis tools
CUHK
February 25, 2015
69
CUHK
February 25, 2015
70
HZZ(*)4l
Event selection
Updated analysis for 2011 and 2012 data

Improved expected sensitivity for low mH
Estimate backgrounds using data (sidebands, control regions)
Development based only on 2011 data and 2012 control regions
Selection
At least two pairs of opposite-charge, same-flavor leptons (e,)
pT thresholds: 20, 15, 10, 7 GeV (6 GeV for muons)
50 < m12 < 106 GeV, m4l-dependent cut on m34, m34 < 115 GeV
All same-flavor, opposite-sign pairs mll>5 GeV (J/ veto)
Tracking and calorimeter -based isolation
Impact parameter significance
CUHK
February 25, 2015
71
HZZ(*)4l
Mass reconstruction, efficiency
Mass resolution in simulated events at mH=130 GeV

(with Z mass constraint for m12)
4: =1.8 GeV
2e2: =2.0 GeV
Combined reconstruction / selection

efficiency for mH=130 GeV
Significant gains from increased
kinematic acceptance and e-ID
CUHK
4e: =2.5 GeV
Efficiency (%)
2e2
4e
2011 data (old)
27
18
14
2011 data
43
23
17
2012 data
41
27
23
February 25, 2015
72
HZZ(*)4l
Background estimation
Irreducible (ZZ(*)): MC simulation normalized to theory cross section

Reducible (ll+jets and tt):
Comparable to ZZ in the low mass region
Estimated using data-driven methods
Background composition depends on flavor of sub-leading lepton pair
(i.e., the flavor of the pair of leptons that make the off-shell Z)
different approaches for ll+ and ll+ee:
ll+ (4, 2e2):
ttbar and Zbb from a fit to m12
ll+ee (22e, 4e):
Z+XX control samples
General strategy: Loosen or revert selection, obtain composition,
extrapolate to signal region
Two methods for each reducible background (nominal, cross-check)
CUHK
February 25, 2015
73
HZZ(*)4l
Background estimation
Fit to m12
Change selection on subleading leptons to enhance
ttbar and reduce ZZ:
No isolation cuts
At least one lepton should fail
the impact parameter cut
tt and Z+jets estimated via a fit
Chebychev +
BreitWignerCrystalBall
Extrapolate to signal region via

MC-based transfer factor
(validated in Z+ control region)
CUHK
At least two estimates for each

background for each subchannel;
good agreement in all cases.
February 25, 2015
74

Higgs Discovery

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Higgs Discovery

Hochgeladen von

Copyright:

Verfügbare Formate

Topics in Contemporary Physics

Chinese University of Hong Kong

Hong Kong SAR

The Higgs discovery

February 25, 2015

February 25, 2015

February 25, 2015

Reminder: Relativistic kinematics

c for all observers

February 25, 2015

Symmetry, conservation laws, groups

A symmetry is an operation on a system that leaves it invariant.

February 25, 2015

Orbital angular momentum (l)

integer or half integer

any (integer) value

For each particle type

Spin angular momentum (s)

A particle with spin 1 :

Spin-statistics connection: half-integer spin: fermions; integer spin: bosons.

February 25, 2015

Addition of Angular Momenta

Cmj jm1 1j2m2 j m , with m = m1 + m2

(Particle Physics Booklet,

Linear combination of angular

February 25, 2015

Operators for the three

and complex, and

Construct matrix representing observable A

Nucleon carries isospin

Strong interactions are invariant under rotations in isospin space.

February 25, 2015

Discrete Symmetries. Parity

time, m, E, charge density

helicity, magnetic flux

vector (polar vector)

angular momentum, mag. field,

February 25, 2015

From Proposal to Publications

Detector and capability design

February 25, 2015

Rate in Hz Dead time ms. Live time % Trigger rate Hz

To reduce dead time: multi-level systems

Low granularity / precision

Processor farms (1 ev/proc)

3D reco, fine-grain calo, tracking,

Proc. farms (1 ev/proc)

Event building (data from all detectors)

February 25, 2015

February 25, 2015

Some pdfs, The MC Method

It can be used to produce rv with other pdfs

February 25, 2015

The Monte Carlo Method

electromagnetic, hadronic showers

particle decays (generate lifetime)

ionization energy loss (generate E)

electronics response , ...

February 25, 2015

We frame it so that a very small value of p allows us to reject H.

February 25, 2015

The Discovery of the Higgs boson

CERNs Large Hadron Collider