Transcript Signals And Backgrounds for the LHC -or- What They Were Thinking When

Signals And Backgrounds for
the LHC
-orWhat They Were Thinking When
They Designed ATLAS and CMS
Thomas J. LeCompte
Argonne National Laboratory
Lecture Series Outline
 Original outline
– A description of the ATLAS and CMS detectors
– A laundry list of potentially interesting signals, followed by their
backgrounds, stretching over two hours
 This bored me to tears
– I’d hate to think what it would do to you
Zzzzzz
 Revised outline
– I’ll try and explain why the detectors look like they do, in the context of
“simple” physics measurements
– I’ll discuss some early and interesting measurements
Think of this as one long talk, split over two days.
STOP ME if I go too fast or you have questions!!
2
Outline
 Four facts about detectors
 Representative signals and how they influence experimental design
– High pT muons
– Higgs via H → gg
– Jets
– Top Quarks
If you come away from this with the
– High pT electrons
perspective that one experiment is better
– Missing ET and Exotica
than the other, I haven’t done my job. I
hope to outline what choices were made
 Some early physics &
and why – not which choices were “right”
future directions
I’m not kidding…
STOP ME if I go too fast or you have questions!!
3
The Most Important Slide I Will Show
jets
From Claudio Campagnari/CMS
Measured cross-sections (except
for Higgs) at the Tevatron
How to extrapolate to the LHC
4
“Nobody won any money betting against Michael Jordan”
 I’m going to assume here that the
Tevatron finds no new physics
– Personally, I wouldn’t take this bet
• The Tevatron is running very well
• The experiments are experienced,
and also running very well
– Nonetheless, one has to assume
something
 If the Tevatron does find something, this
talk becomes very simple:
– Take whatever the Tevatron found
– Make a zillion of them
– Study the heck out of it
5
Fact One: The basic design of experiments is the same:
Tracking
Calorimeters
Muon detectors
 Driven by the physics of
the interaction of high
energy particles with
matter.
 Because the physics is
the same, successful
experimental designs are
similar
6
The Compact Muon Solenoid
7
What CMS Looks Like Today
8
ATLAS = A Toroidal LHC ApparatuS
9
What ATLAS Looks Like Today
10
Fact Two: Tracking measures 1/p
Charged
 particles in a uniform magnetic
field B  B0 zˆ move in helices:
It’s convenient to work in
the transverse plane (i.e.
the plane normal to the Z
direction)
In this plane, the helices
project to circles.
11
Fact Two: Tracking measures 1/p (II)
Radial line from origin to
point where the particle
exits the tracker:
The sagitta (“arrow”) s is
the distance of maximum
deflection from a straight
2
line track:
qBL
s
8 pT
or
qBL2
pT 
8s
Which leads to the expression
p
p

s
s
p
As momentum increases, tracking
becomes more difficult.
12
Fact Three: Sampling Calorimeters
Absorber layers
 A (constant) fraction of the
incoming particle’s energy
gets converted to something
that we can count (photons,
electrons, etc…)
E
E

N
N

1
1

N
E
 In the approximation that
each layer absorbs a
constant fraction of the
energy, the calorimeter
depth grows logarithmically
with energy.
Sensitive layers
As energy increases, calorimetry
resolution improves.
13
Fact Three: Sampling Calorimeters (II)
 EM showers all
look the same
 Hadronic
showers are like
snowflakes
– Every one is
unique
A schematic of an
electromagnetic shower
A GEANT simulation of an
electromagnetic shower
14
Fact Four: Compromise is a fact of life
 Like it or not, experiments are constrained by resources
– Every dollar that goes into one subsystem is a dollar
that doesn’t go into some other subsystem
– Every inch that goes into one subsystem is an inch
that doesn’t go into some other subsystem
– A collaboration with N members can’t design an
experiment that takes 2N members to build or operate.
– Industrial production capacity is finite
• Evidence: crystals, silicon wafers, liquid noble gasses
 Most experimenters have their own ideas on the best optimization
– Individual interests and experience varies
– The goal of a collaboration is to design a detector that everyone can
live with – even if no single person thinks it’s ideal.
15
The Large Hadron Collider
 The LHC:
– Collides protons on protons
– Crossing time of 25 (75) ns
– Design Luminosity 1033 cm-2/s,
increasing to 1034 after 1-2 years,
and hopefully to 1035 (Super LHC)
after that
 The Tevatron
– Collides protons on antiprotons
– Crossing time of 396 ns
– Design Luminosity 2 x 1032 cm-2/s – exceeded routinely.
16
Why Is The LHC More Luminous?
 It has to be
– No valence antiquarks in the proton. If you need an antiquark, you
need to get it from the sea, or a gluon induced higher order process
2
 It can be
b
p
– It uses protons instead of antiprotons
*
• Protons cost $3/oz.
n
• Antiprotons cost $500,000,000,000,000,000/oz.
– Energy is 7x higher
– More beam bunches
• LHC’s bunch intensity is very conservative, compared to what the
Tevatron achieves routinely
L
fE n N


17
LHC Stored Energy in Perspective
Luminosity
Equation:
L
2
fE nb N p
n
*
 Luminosity goes as the
square of the stored
energy.
 LHC stored energy at
design ~700 MJ
– Power if that
energy is
deposited in a
single orbit: ~10
TW (world energy
production is ~13
TW)
– Battleship gun
kinetic energy
~300 MJ
 It’s best to increase the
luminosity with care
USS New Jersey (BB-62)
16”/50 guns firing
18
Muon Detection:
The Iron Ball
19
What Is The Iron Ball?
 Suppose you wanted a detector to look for a very heavy (800 GeV) Higgs
via the decay H → ZZ followed by Z → mm (both Z’s)
– This is a very rare process → increase the luminosity
– Handle this increased luminosity by only looking at muons:
B-field region
20
CMS Muon Detectors
 CMS uses the return
field of their central
solenoid to measure
muon momenta
 Four planes of detector
stations inside the steel
measure the muon’s
tracks.
 Low pT muons range
out in the steel,
providing an additional
measurement.
21
The ATLAS Muon Spectrometer
How muon trajectories bend in the magnetic
field of the toroids.
Energy stored in the magnetic field is ~1.2 GJ.
Beam’s eye view
Energy stored in a lightning bolt is ~1.5 GJ.
Pictures from Jim Shank, Boston University
22
Comparing Design Philosophies
Emphasizes
 CMS uses as their magnetic field the return
field through the iron
– Allows one to go to large fields…
– …which means small radii… (Figure of
merit is BL2)
– …which means that their calorimeter can
be small (and expensive per unit volume).
 ATLAS uses air core toroids
– Very good resolution at high momentum
• Up to ~TeV scale muons
– Requires a lot of space
B
L2
Neither experiment is an iron ball, but elements of the iron ball design influenced
both detectors.
23
Anticipated Muon Backgrounds
 The main background
to muons is other
muons
 What I mean is the
main background to
muons are muons
from a source we are
not particularly
interested in.
– Example: pion or
kaon decays
 Improving purity
beyond ATLAS or
CMS hardly helps.
ATLAS
24
Unanticipated Muon Backgrounds
 There is no such thing as a muon
detector:
– They are charged particle
detectors, behind lots of steel
– If there is any path around the
steel, particles will find it
– Since there are a million hadrons
per muon, even if this is unlikely,
it can be an important background
The “CMX Ricochet”
 At the LHC, the cavern backgrounds
(secondary particles) might be large.
– The experiments have to worry
about particles “raining down” into
their detectors.
– This is VERY difficult to predict
from first principles. It has to be
measured.
25
Dimuon Mass
ATLAS
Requiring two muons
is enough by itself to
give a clean Z signal.
The background is
real muons – just
from other sources.
Note that this is a logarithmic plot
26
Dimuon Mass – Post Cuts
It’s possible to
remove virtually all of
the background in the
previous slide.
(Exactly how is not
really the subject of
this talk)
27
Photon Detection
28
Higgs decays to diphotons
From ATLAS Physics TDR
 The background under the peak
are real gg events: just not from
Higgs decay
– This is the irreducible
background
– Note the suppressed zero
 To see the peak, there are two
things you need to do:
– Get the mass resolution as
good as you can
• make the peak narrow
– Get the reducible
background as small as you
can
• Keep the background from
getting any worse than it
already is
29
Higgs Event Displays
CMS H → gg event
ATLAS H → gg event
30
Why Pointing is Helpful (Part 1)
g
m  2E1E2 (1  cos q )
2
q
Beam axis
Can improve resolution
g
31
Why Pointing is Helpful (Part 2)
g
q
Beam axis
Can reduce background
g
32
Identifying Photons – Basics of Calorimeter Design
Not too much or too
little energy here.
You want exactly one
photon – not 0 (a likely
hadron) or 2 (likely p0)
Not too wide here.
One photon and not
two nearby ones
(again, a likely p0)
Not too much energy
here.
A schematic of an
electromagnetic shower
A GEANT simulationof an
electromagnetic shower
Indicative of a hadronic
shower: probably a
neutron or KL.
33
ATLAS Electromagnetic Calorimeter
Design resolution:
E
E

10%
0.2 GeV
 0.7% 
E
E
Technology: uses lead as an absorber and
liquid argon as an ionization medium.
Energy deposited in the calorimeter is
converted to an electrical signal.
34
ATLAS Liquid Argon Calorimeter Module
 Highly segmented
– Allows measurement of
shower development
• Rejects background
– Has some pointing ability
 Very good (but not as good as
CMS) energy resolution
 “Accordion” faster than
other LAr calorimeters
– Still slower than crystals
35
ATLAS Calorimeter in Real Life
Before installation
– it’s now in a
cryostat and
impossible to see.
36
CMS Calorimeter Crystals
16X0
22X0
26X0
Design resolution:
E
E

2.7%
0.16 GeV
 0.55% 
E
E
Photo: Ren-yuan Zhu, Caltech
 CMS uses Lead Tungstate
crystals
– Scintillator: energy is
converted to light
– Exceptional energy
resolution, because there
are no inert absorbers
 The focus is to get the best
possible energy resolution, no
matter what it takes
– Energy resolution is ~2x
better than ATLAS’ in the
region where Higgs decay
is important
Another nice feature – low noise
37
CMS EM Calorimeter
Figure: Ren-yuan Zhu, Caltech
38
Comparing Design Philosophies
 CMS emphasizes energy resolution
– Use PWO crystals
• Expensive – means go to small radius to keep the detector within budget
• Only handful of vendors worldwide
 ATLAS emphasizes background rejection
– Able to go to larger radius: separates showers better
– Highly segmented calorimeter allows measurement of shower development
• One photon? Two? A hadron masquerading as a photon?
 Both calorimeters are quite thick
– Improves resolution (showers are contained)
– Degrades electron-hadron separation
• ATLAS measurement of shower development is intended to compensate
39
Jets
When you’re a jet, you’re a jet all the way…
S. Sondheim
40
A Two Slide Review of Jets
A “blast” of particles, all going in roughly the same direction.
2 jets
2 jets
2
2
3 jets
5 jets
3
5
Calorimeter View
Same Events, Tracking View
41
More on Jets
 Where do they come from?
 The force between two colored objects
(e.g. quarks) is independent of distance
– Therefore the potential energy grows
linearly with distance
– When it gets big enough, it pops a
quark-antiquark pair out of the vacuum
– These quarks and antiquarks
ultimately end up as a collection of
hadrons
• Process is called “fragmentation” or
“hadronization”
g
g
g
g
One (of several)
processes that produce
jets in collisions.
42
Measuring Jets
 Lego plot of a top quark event
has EM calorimeter energy in
green and hadronic calorimeter
energy in red.
 The prevalence of green comes
from two facts:
– Half of the particles (and
~40% of the energy) in a jet
are photons from neutral
meson decay
– The LHC EM calorimeters
are thick, and many hadrons
begin their showers inside
the electromagnetic
calorimeters.
ATLAS
43
ATLAS “Tile” Hadronic Calorimeter
Uses steel as an absorber, and scintillating tiles as the
active medium. Energy is converted to light.
44
CMS HCAL (Hadronic Calorimeter)
 Also a scintillating tilebased sampling
calorimeter
 Technology is similar to
ATLAS:
– Absorber is brass
instead of steel
– Tile orientation is
different (more
conventional in CMS)
 Calorimeter is relatively
thin
45
Comparing Design Philosophies
 CMS’ calorimeter is inside their magnet
– Additional depth is very expensive (since it requires a larger radius magnet)
– To increase the effective thickness, it’s made of a denser material (brass)
– ATLAS is quite thick – the outer muon detector resembles the “iron ball”
design.
 ATLAS would “naturally” use liquid argon for both the hadronic and
electromagnetic calorimeter
– Both D0 and H1 designed their detectors this was
– This is also very expensive
 Both experiments have chosen to economize here
– Slightly better performance would cost a lot more money
46
Jets & Hadron Calorimetry
Many things besides
hadronic calorimeter
response affect the jet
energy determination: the
EM calorimeter response,
out of cone corrections,
dead material and dead
material corrections,
hadronic decays or
interactions upstream of
the calorimeter…
Event with the best calorimeters, you are taking a very clear picture of a very fuzzy
object. Economizing on the hadron calorimeter is often the “least bad” option.
47
Tops
48
Top Quark Pair Production
BF (t  Wb )  99%
Top pair events are characterized by the
decay of the two W’s in the event.
49
Early Top Quarks at the LHC
 Start with a lepton (e or m)
plus four jets sample.
 Make all three jet mass
combinations, requiring
m(jj) = m(W)
 Identifying one jet as
containing a b quark (“btagging”) is not required.
Why is the top signal so clear without b-tagging?
Especially since the Tevatron needed b-tagging to
discover the top quark?
50
Top Quark Production
 At the Tevatron, the left column diagram
dominates.
– The W+jets background is also
produced by quark-antiquark
annihilation.
 At the LHC, the right column diagrams
dominate.
51
The Most Important Slide I Will Show (Again!)
jets
From Claudio Campagnari/CMS
Measured cross-sections (except
for Higgs) at the Tevatron
How to extrapolate to the LHC
52
Why Bother To B-Tag at the LHC?
 A signal to background of
1:2 is fine for a discovery,
but one would want to do
better for precision
physics.
 B-tagging keeps 80-90%
of the signal, but cuts the
blue background by an
order of magnitude.
 B-tagging also helps with the combinatoric background:
– Without b-tagging, there are 12 combinations in every event
– Tagging one b-jet reduces this to 6
– Tagging both b-jets reduces this to 2 (keeps ~40% of the signal)
53
Fake B Tags
Missing one hit
can turn this
into this:
 Solution: redundancy, redundancy, redundancy
– More hits reduces your sensitivity to lost/misassigned hits
– More hits increases the standalone tracking capability of your silicon
• It takes 5 numbers to parameterize a helix
– More hits improves the resolution of each track
54
LHC Silicon Vertex Detectors
ATLAS Pixels
ATLAS
CMS
55
Electrons & Other Tracks
CMS Silicon Detector
 The figure of merit for momentum
measurement is BL2. A half dozen
layers of silicon close in has great
impact parameter resolution – but L
is tiny.
 To improve the momentum
resolution, the experiments need to
build out from their silicon detectors.
 CMS chose to keep going with
silicon, and add another half dozen
layers
– By far, the largest silicon HEP
detector ever built.
The key idea: BL2 – relative to what?
Reminder: to tell an electron from a
photon, recognize that an electron has a
track (of the right momentum) in front of
the cluster, and a photon has no track.
56
ATLAS Tracking Philosophy
 Instead of adding 6 precise points
from silicon, ATLAS adds 36
“pretty good” points from a
tracker based on wires inside
straws.
 The detectors are designed to be
sensitive to transition radiation,
so can be used to aid in electron
identification.
 This detector design will not work
at the very highest luminosity –
when that happens, ATLAS may
replace with silicon.
57
Missing ET
58
Missing Transverse Energy
 We know momentum is
conserved.
 An apparent imbalance of
momentum can be due to an
escaping neutrino
– Calculated by adding up all
the other momenta and
reversing the sign
W  e
Momentum of the
“underlying event”
Missing ET
Electron momentum
 We work only in the xy
(transverse) plane
– Many particles escape
unmeasured down the
beampipe (this will be
important later)
59
Pink is the New Black
…and neutralinos are the new neutrinos.
 In Supersymmetry, every fermion has a boson that’s its partner and vice
versa
– The spin-1 photon’s partner is the spin-½ photino
 The lightest supersymmetric particle is stable
g~ 
(violates conservation of angular momentum)
 
 
g~ 
 e e  (violates conservation of lepton number)
g~ 
 n nn (violates conservation of baryon number)
 This is called “R parity conservation” and it keeps your supersymmetric
theory from violating experimental limits, such as that for proton decay
– It leads to a common signature in SUSY
models: particles that exit your detector
Footnotes:
1. The g, Z and H all have partners, and these
without interacting, leading to missing
partners have the same quantum numbers, so
momentum
they mix.
2. There are ways to contrive an R-parity violating
theory that evades experimental bounds
60
Variations on a Theme
 More exotic: theories with extra dimensions can also
have missing ET signatures
– The entire standard model is replicated
(so-called Kaluza-Klein modes)
• The “KK graviton” is one candidate for a particle that gives large
missing ET.
• Depending on the model, you might get more.
 Even more exotic: theories with names like “Hidden Valleys” and “Shadow
Matter”.
The point is not whether these particular theories are right
or wrong. The point is that Missing ET is a common
signature present in multiple models, so it should be
looked for.
61
Improving Missing ET
 To keep particles from escaping, one can:
– Make the holes smaller
• There’s a limit to this
– Make the detector longer
• The hole is the same size, but
subtends a smaller angle
62
Hermiticity
 A fancy way of saying “holes are bad”
 Particles escape down holes and cracks,
and generate missing ET
– “Real missing ET, because it truly is
missing
– “Fake missing ET” in the sense that it
wasn’t what you were looking for
• Difference between an undetected
and an undetectable particle
 Holes, gaps and cracks are necessary
– Minimum of two holes (for the beam)
– Cables need to come out somewhere
– Cooling and cryogens need to go in
• If they go in, they have to come out
63
Why Are Detectors as Long As They Are?
 Why not make detectors longer and longer and longer?
– Resource limitations
• Making it twice as big costs twice as much money
• …and takes twice as many people
• …or takes twice as long
 Relativistic kinematics affects the design of a detector
– Heavy objects are produced almost at rest
I wish
• Their decay products populate all 4p of solid angle
someone had
told me this
• What matters is solid angle
sooner.
– Light objects are produced uniformly across rapidity
• In principle, argues for a long detector
• But if the mass is low, the cross-section is high, and you’re making
a lot of them – one unit of rapidity is as good as any other
– In either case, there’s a natural point where it’s no longer cost
effective to keep going forward
64
Why are LHC experiments a little longer than Tevatron
experiments?
D0 detector
 One reason is for Missing ET performance (long is good)
 Another reason is from kinematics
– Quark-antiquark collisions at the LHC are asymmetric.
– Even for light objects, extra coverage doesn’t hurt you – it just costs $
65
Mismeasured Jets
 One way to generate fake missing ET is to
mismeasure an object in the event.
 Jets are the usual suspects:
– There are a lot of them
– There are several things that can go
wrong:
• Plain old mismeasurement
• Particles down cracks
• Particles in dead regions of the
detector
– Undercorrection and
overcorrection are both possible
• Particles in the jet decaying with a
leading neutrino
• And so on…
With apologies to Spinal Tap
These all sound unlikely.
That’s because they are
unlikely.
The reason that this is
important is…
66
The Most Important Slide I Will Show (Yet Again)
jets
From Claudio Campagnari/CMS
Measured cross-sections (except
for Higgs) at the Tevatron
How to extrapolate to the LHC
67
Triggering – the Oft Overlooked Component
 At the LHC, there are 40,000,000 beam crossings per second.
 Of these, perhaps 200 can be written to tape for analysis
 It’s the job of the trigger to select which 200
– There are no do-overs in baseball
– There are no do-overs in triggering
 Triggers are usually designed in tiers
– Low level triggers tend to be hardware-based, fast, and select events for
higher trigger levels to look at
– Higher level triggers are software based, and can take much more time to
decide whether to keep this event, or some other event.
 The Three Laws of Triggering
– 1. You cannot analyze an event you didn’t trigger on
– 2. If you aren’t going to analyze an event, it doesn’t help to trigger on it
– 3. If you are going to cut an event, cut it as early in the chain as you can.
68
Early and/or Interesting Measurements
69
Quark Contact Interactions
 Contact interactions
look different than
QCD.
– QCD is
predominantly
t-channel gluon
exchange.
Quark Compositeness New Interactions
q
q
M~L
M~L
q
q
Dijet Mass << L
Quark Contact Interaction
q
q
L
q
q
QCD
t - channel
Diagrams: R. Harris, CMS
 New physics at a scale L above the
observed dijet mass is modeled as an
effective contact interaction.
– Quark compositeness.
– New interactions from massive
particles exchanged among
partons.
70
“Week One” Jet Measurements
Jet Transverse Energy
5
pb-1
of (simulated) data:
corresponds to 1 week running at
1031 cm-2/s (1% of design)
 Expected limit on contact interaction:
L(qqqq) > ~6 TeV
– Rule of thumb: 4x the ET of the
most energetic jet you see
– Present PDG limit is 2.4-2.7 TeV
– Ultimate limit: ~20 TeV
– The ATLAS measurement is at
lower x than the Tevatron: PDF
uncertainties are less problematic
Note that after a very short time,
LHC will be seeing jets beyond
the Tevatron kinematic limit.
71
Making the Measurement
 There are only two hard things in making
this plot:
– The x-axis
– The y-axis
 The y-axis has two pieces: counting the
events, and measuring the luminosity
– The first is easy
– The second is hard, and I won’t talk about it
 The key to the x-axis is correctly measuring the jet energy
72
Balancing Jets
 The problem of setting the jet energy
scale can be split into two parts:
– 1. Establish that all jets share
the same scale
– 2. Establish that all jets share
the right scale.
 A good start to #1 is to look at dijet
events and show there is no bias to
the jet energy as a function of jet
position, jet composition, energy
deposition, pile-up, etc.
 A good start to #2 is to use known particles
(electrons and Z’s) to set the overall scale.
Getting the jet energy scale right
to 20% is easy. Getting it right to
2% is hard – and will take time.
20% in JES = a factor of 2 in data
73
Jet Energy Scale Job List
 See that the Z decay to electrons ends up in the right spot
– Demonstrates that the EM calorimeter is calibrated
 Balance jets with high and low EM fractions
– Demonstrates that the EM and hadronic calorimeters have the same
calibration
 Balance one jet against two jets
– Demonstrates that the calorimeter is linear
 Balance jets against Z’s and photons
– Verifies that the above processes work in an independent sample
– Demonstrates that we have the same scale for quark and gluon jets
 Use top quark decays as a final check that we have the energy scale right
– Is m(t) = 175 and m(W) = 80? If not, fix it!
Note that most of the work isn’t in getting the jet energy scale right. It’s in
convincing ourselves that we got the jet energy scale right – and that we have
assigned an appropriate and defensible systematic uncertainty to it.
74
Angular Distribution of a Contact Interaction
 Contact interaction is often more
isotropic than QCD
– QCD is dominated by tchannel gluon exchange.
– c.f. Eichten, Lane and Peskin
(Phys. Rev. Lett. 50, 811-814
(1983)) for distributions from a
contact interaction
Center of Momentum
Frame
Jet
q*
Parton
Parton
Jet
QCD Background
 CMS (and D0) compress this
distribution into a single ratio of
central-to-forward jets
Signal
0
1
cos q*
Diagrams: R. Harris, CMS
 It’s harder to grossly mismeasure
a jet’s position than its energy.
75
Angular Distribution of a Contact Interaction II
 The D0 (hep-ex/980714) dijet
ratio:
N(|h| < 0.5)/N(0.5 < | h | < 1)
– This is essentially a
measurement of the
position of the leading jet.
 CMS plans to do the same
thing (see plot)
 ATLAS is leaning more
towards a combined fit of
energy and angle.
– Same idea, different
mathematics
New physics changes the shape of this plot. You
aren’t counting on having a precise prediction of
the QCD value.
76
Changing Gears
 Why did we build the LHC?
– Wrong Answer: To find the Higgs Boson
D– Right Answer: To study the electroweak
sector at high energies / small distances.
• There might be a Higgs
A+ • There may not be a Higgs
• There may not be only one Higgs
• Finding something other than a single scalar Higgs is not failure.
It may even be better than finding a single scalar Higgs.
77
What is the Standard Model?
g
The (Electroweak)
Standard Model is the
theory that has
interactions like:
W+
W+
Z0
but not
W+
W+
Z0
g
Z0
Z0
&
W-
Z0
g
W-
Z0
Z0
Z0
g
Z0
g
&
but not:
Z0
g
Only three parameters - GF, a and
sin2(qw) - determine all couplings.
78
The Semiclassical W
 Semiclassically, the interaction between the W and the electromagnetic
field can be completely determined by three numbers:
– The W’s electric charge
• Effect on the E-field goes like 1/r2
– The W’s magnetic dipole moment
• Effect on the H-field goes like 1/r3
– The W’s electric quadrupole moment
• Effect on the E-field goes like 1/r4
 Measuring the Triple Gauge Couplings is equivalent to measuring the 2nd
and 3rd numbers
– Because of the higher powers of 1/r, these effects are largest at small
distances
– Small distance = short wavelength = high energy
79
Triple Gauge Couplings
 There are 14 possible WWg and WWZ couplings
 To simplify, one usually talks about 5 independent, CP conserving, EM
gauge invariance preserving couplings: g1Z, kg, kZ, lg, lZ
– In the SM, g1Z = kg = kZ = 1 and lg = lZ = 0
• Often useful to talk about Dg, Dk and l instead.
• Convention on quoting sensitivity is to hold the other 4 couplings at
their SM values.
– Magnetic dipole moment of the W = e(1 + kg + lg)/2MW
– Electric quadrupole moment = -e(kg - lg)/2MW2
– Dimension 4 operators alter Dg1Z,Dkg and DkZ: grow as s½
– Dimension 6 operators alter lg and lZ and grow as s
Do we live in a Standard Model universe? Or some other
universe?
80
Why Center-Of-Mass Energy Is Good For You
Approximate
Run II Tevatron
Reach
Tevatron
kinematic limit
 The open histogram is the
expectation for lg = 0.01
– This is ½ a standard
deviation away from
today’s world average fit
 If one does just a counting
experiment above the Tevatron
kinematic limit (red line), one
sees a significance of 5.5s
– Of course, a full fit is more
sensitive; it’s clear that the
events above 1.5 TeV have
the most distinguishing
power
From ATLAS Physics TDR:
30 fb-1
81
Not An Isolated Incident
 Qualitatively, the same thing
happens with other couplings
and processes
 These are from WZ events with
Dg1Z = 0.05
– While not excluded by data
today, this is not nearly as
conservative as the prior
plot
• A disadvantage of having
an old TDR
Plot is from ATLAS Physics TDR: 30 fb-1
Insert is from CMS Physics TDR: 1 fb-1
82
Not All W’s Are Created Equal
 The reason the inclusive W and
Z cross-sections are 10x higher
at the LHC is that the
corresponding partonic
luminosities are 10x higher
– No surprise there
 Where you want sensitivity to
anomalous couplings, the
partonic luminosities can be
hundreds of times larger.
 The strength of the LHC is not
just that it makes millions of
W’s. It’s that it makes them in
the right kinematic region to
explore the boson sector
couplings.
Here’s Claudio’s plot again…
83
TGC’s – the bottom line
Coupling
Present Value
LHC Sensitivity
(95% CL, 30 fb-1 one experiment)
Dg1Z
0.005-0.011
 0.01600..022
019
Dkg
0.03-0.076
 0.027 00..044
045
DkZ
0.06-0.12
 0.07600..061
064
lg
0.0023-0.0035
 0.02800..020
021
lZ
0.0055-0.0073
 0.08800..063
061
 Not surprisingly, the LHC does best with the Dimension-6 parameters
 Sensitivities are ranges of predictions given for either experiment
84
Early Running
 Reconstructing W’s and Z’s quickly will not be hard
 Reconstructing photons is harder
– Convincing you and each other that we understand the efficiencies and jet
fake rates is probably the toughest part of this
 We have a built in check in the events we
are interested in
– The Tevatron tells us what is happening
over here.
– We need to measure out here.
 At high ET, the problem of jets faking
photons goes down.
– Not because the fake rate is
necessarily going down – because the
number of jets is going down.
85
Things I Left Out and Really Shouldn’t Have
 Angular distributions have additional resolving power
– Remember, the W decays are self-analyzing
– Different couplings yield different angular distributions
• Easiest to think about in terms of multipole moments
 Neutral Gauge Couplings
– In the SM, there are no vertices containing only g’s and Z’s
– At loop level, there are ~10-4 corrections to this
– It is vital that these be explored
86
Putting it all together
 Complex signatures break down into simpler ones
– Suppose you were looking for stop squarks:
~
~
t t  (tg~)(t g~)  (Wbg~)(Wbg~)
• Signal would be a lepton + 4 jets (2 b-tagged) + lots and lots of
missing ET.
• One background is real top events + a mismeasured jet leading
into large missing ET.
 Many backgrounds are similar – additional jets from QCD radiation, which
may or may not be reconstructed correctly
– Could be misreconstructed as a photon, a b-jet, missing ET, etc…
– These jets are often correlated with some other object in the event
• You have a radiator, and a radiatee.
• Signal usually does not have this correlation – allows
discrimination.
But there’s a fly in the ointment…
87
Double Parton Scattering
 Two independent partons in the proton scatter:
s As B
s AB 
s Effective
might be better
characterized by
s AB
s As B
 Asˆ 
s Inelastic
 Searches for complex signatures in the presence of QCD background
often rely on the fact that decays of heavy particles are “spherical”, but
QCD background is “correlated”
– This breaks down in the case where part of the signature comes from
a second scattering.
– The jet cross-section is very high at the LHC, so this is proportionally
a larger background than at lower energies
 We’re thinking about bbjj as a good signature to measure this
– Large rate/large kinematic range
– Relatively unambiguous which jets go with
which other jets.
88
Comments on Double Parton Scattering
 The naïve parton model assumes independence
 We don’t expect partons to be completely independent of each other
– Quarks are confined, after all.
 This is very difficult to calculate
– We need to measure this
 DPS looks a lot like pileup
– Cuts that kill pileup also kill DPS
– This may be necessary at high luminosity, so the issue of DPS may
be moot.
89
Summary
 I hope to have given you some insight on why the LHC detectors look like
they do:
– Why the design choices are what they are
– What signals they are intended to accept
– What backgrounds they are intended to reject
– Of course, this is incomplete doing this right would take all week
 I hope to have given you some idea
on which strings we will be tugging at to
unravel the Standard Model:
90
Let’s Do It One More Once…
jets
From Claudio Campagnari/CMS
Measured cross-sections (except
for Higgs) at the Tevatron
How to extrapolate to the LHC
91