Solutions Manual for

Elementary Particle Physics:

An Intuitive Introduction
Andrew Larkoski∗

Physics Department
Reed College

December 22, 2018

∗
[email protected]

1
Contents
1 Introduction 6
1.1 Energy of a Mosquito . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2 Yukawa’s Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 Mass of the Photon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 Planck Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.5 Expansion of the Universe . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.6 Decay Width of the Z boson . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.7 Decay of Strange Hadrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.8 PDG Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.9 InSpire and arXiv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2 Special Relativity 13
2.1 Properties of Lorentz Transformations . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Rapidity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3 Lorentz-Invariant Measure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.4 Properties of Klein-Gordon Equation . . . . . . . . . . . . . . . . . . . . . . 15
2.5 Maxwell’s Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.6 Properties of the Clifford Algebra . . . . . . . . . . . . . . . . . . . . . . . . 16
2.7 Relativity of Spin-1/2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.8 Dark Matter Searches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.9 Top Quark Decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3 A Little Group Theory 26

3.1 Representations of the Symmetric Group . . . . . . . . . . . . . . . . . . . . 26
3.2 Lorentz Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.3 Hermitian Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.4 Baker-Campbell-Hausdorff Formula . . . . . . . . . . . . . . . . . . . . . . . 30
3.5 Casimir Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.6 Helicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.7 Symplectic Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.8 π-p Scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4 Fermi’s Golden Rule & Feynman Diagrams 34

4.1 Galactic Collisions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.2 Integrating δ-functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.3 Three-Body Phase Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.4 e+ e− → e+ e− Scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.5 Non-Relativistic Limit of Feynman Diagrams . . . . . . . . . . . . . . . . . . 42
4.6 Proton-Proton Total Cross Section . . . . . . . . . . . . . . . . . . . . . . . 44
4.7 Proton Collision Beam at ATLAS . . . . . . . . . . . . . . . . . . . . . . . . 44
4.8 Upper Limits on LUX Bounds . . . . . . . . . . . . . . . . . . . . . . . . . . 46

2
5 Particle Collider Experiment 48
5.1 Synchrotron Losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.2 Limits of the Tracking System . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.3 Reconstructing Muons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.4 Data Quantity from the LHC . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.5 Properties of Poisson Statistics . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.6 Look-Elsewhere Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.7 Discovery of the Top Quark . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.8 Missing Energy and Neutrinos . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.9 Event Displays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

6 Quantum Electrodynamics in e+ e− Collisions 57

6.1 Lorentz Transformation of Spinors . . . . . . . . . . . . . . . . . . . . . . . 57
6.2 Helicity Spinors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
6.3 Spin Analysis of e+ e− → µ+ µ− . . . . . . . . . . . . . . . . . . . . . . . . . 62
6.4 Spin-0 Photon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6.5 e+ e− → scalars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6.6 Decays of the Z boson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
6.7 Inclusive vs. Exclusive Cross Sections . . . . . . . . . . . . . . . . . . . . . . 65
6.8 Finite Decay Width Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

7 Quarks & Gluons 68

7.1 Plus-Function Expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
7.2 Breit Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
7.3 Form Factor Evolution Equation . . . . . . . . . . . . . . . . . . . . . . . . . 72
7.4 Infrared and Collinear Safety . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
7.5 Properties of Helicity Spinors . . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.6 More Helicity Spinors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
7.7 The Drell-Yan Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
7.8 Thrust in Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

8 Quantum Chromodynamics 81
8.1 Masslessness of the Gluon . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
8.2 Bianchi Identity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
8.3 Instantons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
8.4 Wilson Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
8.5 su(2) Lie Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
8.6 Casimir Invariant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
8.7 Adjoint Representation of SU(3) . . . . . . . . . . . . . . . . . . . . . . . . . 91
8.8 Running Couplings of QED and QCD . . . . . . . . . . . . . . . . . . . . . . 92

3
9 Parton Evolution and Jets 95
9.1 Dilation Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
9.2 Expansion of Differential Cross Section for Thrust . . . . . . . . . . . . . . . 96
9.3 Jet Multiplicity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
9.4 Properties of the DGLAP Equation . . . . . . . . . . . . . . . . . . . . . . . 100
9.5 Resummation of Q2 with DGLAP . . . . . . . . . . . . . . . . . . . . . . . . 100
9.6 Jet Mass at the LHC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
9.7 Underlying Event at Hadron Colliders . . . . . . . . . . . . . . . . . . . . . . 104
9.8 Jet Event Display . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

10 Parity Violation 107

10.1 Time Reversal of Spinors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
10.2 Charge Conjugation of Spinors . . . . . . . . . . . . . . . . . . . . . . . . . . 108
10.3 CPT on Spinors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
10.4 C, P, T in Electromagnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
10.5 Electron Spin in Muon Decay . . . . . . . . . . . . . . . . . . . . . . . . . . 111
10.6 Endpoint of Electron Energy in Muon Decays . . . . . . . . . . . . . . . . . . 112
10.7 Kinematics of the IceCube Experiment . . . . . . . . . . . . . . . . . . . . . 114
10.8 High-Energy Neutrino Cross Sections . . . . . . . . . . . . . . . . . . . . . . 116

11 The Mass Scales of the Weak Force 122

11.1 Maxwell with a Massive Photon . . . . . . . . . . . . . . . . . . . . . . . . . 122
11.2 Scalar Higgs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
11.3 Charge 0 Higgs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
11.4 Forms of SU(2) Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
11.5 Unification of Couplings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
11.6 When V − A and When Electroweak? . . . . . . . . . . . . . . . . . . . . . . 126
11.7 Charged Current DIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
11.8 Left-Handed Coupling of the W Boson? . . . . . . . . . . . . . . . . . . . . . 131

12 Consequences of the Weak Interactions 134

12.1 Mass and Flavor Basis Commutator . . . . . . . . . . . . . . . . . . . . . . . 134
12.2 Unitarity of the CKM Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . 137
12.3 Jarlskog Invariant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
12.4 Extra Quark Generations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
12.5 Measuring the Cabibbo Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
12.6 Non-Relativistic Limit of Neutrino Oscillations . . . . . . . . . . . . . . . . . 141
12.7 Neutrinos for Nuclear Non-Proliferation . . . . . . . . . . . . . . . . . . . . 142
12.8 Neutrinos from SN 1987a . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
12.9 Solar Neutrino Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148

4
13 The Higgs Boson 148
13.1 W Boson Decays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
13.2 pp → W + W − Backgrounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
13.3 Searching for H → W + W − . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
13.4 H → γγ Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
13.5 Higgs Production Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
13.6 Landau-Yang Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
13.7 Combining Uncorrelated Measurements . . . . . . . . . . . . . . . . . . . . . 155
13.8 Testing the Spin-2 Higgs Boson Hypothesis . . . . . . . . . . . . . . . . . . . 156

14 Particle Physics at the Frontier 157

14.1 Neutrinoless Double-β Decay at EXO . . . . . . . . . . . . . . . . . . . . . . 157
14.2 Double-Higgs Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
14.3 Borel Summation of a Convergent Series . . . . . . . . . . . . . . . . . . . . 158
14.4 The International Linear Collider . . . . . . . . . . . . . . . . . . . . . . . . 158
14.5 The Largest Possible Collider . . . . . . . . . . . . . . . . . . . . . . . . . . 159

5
1 Introduction
1.1 Energy of a Mosquito
The mass of a mosquito is approximately m = 2.5 × 10−6 kg and it flies at approximately
v = 0.1 m/s, or so. Therefore, its kinetic energy is
1
K = mv 2 = 1.25 × 10−8 J . (1)
2
One electron volt is about 1.6 × 10−19 J, so the energy in eV of a flying mosquito is

1.25 × 10−8
K= eV ' 7.8 × 1010 eV . (2)
1.6 × 10−19
The energy per nucleon of the flying mosquito can be found by dividing the total energy
found above by the number of protons and neutrons in the mosquito. With a total mass of
2.5 × 10−6 kg and the mass of the proton/neutron is approximately mp = 1.67 × 10−27 kg,
the total number of nucleons in the mosquito are

2.5 × 10−6
Nn ' −27
' 1.5 × 1021 . (3)
1.67 × 10
Therefore the kinetic energy per nucleon of the mosquito is about

K 7.8 × 1010
' eV ' 5.2 × 10−11 eV . (4)
Nn 1.5 × 1021
This is about 23 orders of magnitude smaller than the energy of protons at the LHC!

1.2 Yukawa’s Theory

The radius of an atomic nucleus is on the order of a femtometer, 10−15 m. To turn this into
a mass or energy, we divide the product ~c by this distance. This is

~c (1.05 × 10−34 ) · (3 × 108 )

E= = J ' 3 × 10−11 J . (5)
x 10−15
To convert to eV, we divide by the ratio eV/J' 1.6 × 10−19 , so that

3 × 10−11
E= ' 2 × 108 eV = 200 MeV . (6)
1.6 × 10−19
That is, the pion has a mass of about 200 MeV.

6
1.3 Mass of the Photon
If Maxwell’s equations describe the magnetic field of the Milky Way galaxy, this sets an
upper bound on the mass of the photon. The diameter of the Milky Way is about 100,000
light-years, which in meters is approximately
100, 000 l-y = 105 · 3 × 108 · π × 107 ' 1021 m .



(7)
In this expression, we used the fact that, to better than 1% accuracy, the number of seconds
in a year is π × 107 . If electromagnetism as we understand it describes the galactic magnetic
field at this distance, the photon must be able to have a wavelength that is at least this size.
The corresponding upper bound on the minimum photon energy is
~c (1.05 × 10−34 ) · (3 × 108 )
E< ' 21
J ' 3 × 10−47 J . (8)
x 10
In electron volts, this corresponds to
3 × 10−47
E< eV ' 2 × 10−28 eV . (9)
1.6 × 10−19
So, the mass of the photon must be less than about 10−28 eV. Converting this to kg, we
divide the energy in Joules by c2 :
3 × 10−47 −64
m< 2 ' 3 × 10 kg . (10)
(3 × 10 )
8

The mass of the electron is about 1/2 MeV, so this is about 34 orders of magnitude smaller.
While this limit is extremely impressive, the assumptions necessary to describe the galac-
tic magnetic field and connect it to Maxwell’s equations in particular are a bit tenuous, so
this result is not used by the PDG to set a limit on the photon mass.

1.4 Planck Units

1.4 (a)
The Planck time, tP , can be expressed as a product of Newton’s constant GN , ~, and the
speed of light c raised to some powers:
tP = GαN ~β cγ , (11)
where α, β, and γ are some numerical powers. We can find the powers by matching units
on both sides of the expression. c is a velocity and so
[c] = LT −1 , (12)
where L is a length and T is a time unit. ~ has units of energy times time or that
[~] = M L2 T −1 , (13)

7
where M is a mass unit. Finally, the units of GN can be determined by Newton’s law of
gravitation and Newton’s second law:
GN m1 m2 d2~r
F~g = − r̂ = m1 , (14)
r2 dt2
from which it follows that
[GN ] = M −1 L3 T −2 . (15)
Plugging these into the expression for the Planck time tP we have

[tP ] = T = [GN ]α [~]β [c]γ = M −α L3α T −2α M β L2β T −β Lγ T −γ . (16)

Demanding that there is no mass unit requires that

− α + β = 0, (17)

or that α = β. Demanding that there be no length unit requires

3α + 2β + γ = 0 = 5α + γ = 0 , (18)

or that γ = −5α. Finally, demanding that there be one unit of time requires that

− 2α − β − γ = 1 = 2α , (19)

or that α = 1/2. It then follows that the Planck time is

r
GN ~
tP = . (20)
c5
The value of Newton’s constant in SI units is GN = 6.67 × 10−11 kg−1 m3 s−2 . It then
follows that the Planck time is
s
(6.67 × 10−11 ) · (1.05 × 10−34 )
tP = s ' 5.4 × 10−44 s , (21)
(3 × 108 )5

which is pretty small!

1.4 (b)
Now, we’re asked to find the Planck mass, mP . Because we already have the Planck time,
tP , we can find the Planck mass pretty easily. Note that the quantity
~
EP = , (22)
tP
is an energy. Then, the Planck mass is found by dividing by c2 :
r
~ ~c
mP = 2
= ' 2 × 10−8 kg . (23)
tP c GN

8
Expressed in eV, the Planck mass is

mP c2
−19
' 1.2 × 1028 eV . (24)
1.6 × 10
The proton mass is about a GeV, or 109 eV, so the Planck mass is about 19 orders of
magnitude larger!

1.4 (c)
For two particles of mass m1 and m2 and electric charges q1 and q2 , the ratio of the electric
force F~E to the gravitational force F~g between them is

|F~E | 1 |q1 ||q2 |

= , (25)
|F~g | 4π0 GN m1 m2

where 0 is the permittivity of free space. The proton and electron both have an electric
charge magnitude of the fundamental unit of charge e = 1.6 × 10−19 , so plugging in numbers,
the ratio of forces is
2
|F~E | 1 (1.6 × 10−19 )
' −12 ) (6.67 × 10−11 ) (1.67 × 10−27 ) (9.11 × 10−31 )
' 2.2 × 1039 . (26)
~
|Fg | 4π (8.85 × 10

1.5 Expansion of the Universe

1.5 (a)
The CMB has a temperature of 2.7 K. With Boltzmann’s constant kB , we can turn this into
a corresponding energy. We find

ECMB = kB T = 2.7 × 1.38 × 10−23 J ' 3.7 × 10−23 J . (27)

To determine the energy in eV, we divide by 1.6 × 10−19 :

3.7 × 10−23
ECMB ' −19
eV ' 2.3 × 10−4 eV . (28)
1.6 × 10

1.5 (b)
The ground state energy of hydrogen is Eg = −13.6 eV. So, when the temperature of the
universe was less than the energy of 13.6 eV, electrons and protons could become bound and
form hydrogen. To determine this temperature, we work backward from the steps of the first
part of this problem, multiplying by the factor 1.6 × 10−19 and then dividing by kB . We find

|Eg |
Trecomb. = 1.6 × 10−19 K ' 1.6 × 105 K . (29)
kB

9
1.5 (c)
We want to calculate the ratio of the wavelength of CMB photons observed today, λtoday , by
the wavelength at recombination, λrecomb. . This ratio is
λtoday frecomb. Erecomb. Trecomb.
= = = . (30)
λrecomb. ftoday Etoday Ttoday

In this chain of equalities, we used the fact that wavelength λ is inversely proportional to
frequency f and the frequency of light is proportional to its energy. From what was developed
in the previous parts, the energy of the photons is proportional to its temperature. So, the
redshift factor is just the ratio of the temperature at recombination to the temperature
today:
λtoday Trecomb. 1.6 × 105
= ' ' 5.9 × 104 . (31)
λrecomb. Ttoday 2.7
As mentioned in the problem, this is a factor of about 30 larger than the true result when
thermodynamics are properly taken into account.

1.6 Decay Width of the Z boson

1.6 (a)
From the plot of Fig. 1.5, the maximum value of the peak of the distribution is about 32
nb. Therefore, half of this is 16 nb. The lower point of the distribution at which it takes
a value of 16 nb is at approximately 90 GeV, while the higher point is at about 92.5 GeV.
Therefore, the width, or full-width at half-maximum is the difference of these two values, or
2.5 GeV.

1.6 (b)
To determine the lifetime in seconds of the Z boson from its width, we need to relate the
width to a time through the energy-time uncertainty relation:
~
∆t ' . (32)
∆E
To convert the width from natural units to SI, we need to multiply by the factor of 1.6×10−19
so that the lifetime in seconds is
~ 1.05 × 10−34
∆t ' ' ' 2.6 × 10−25 s . (33)
∆E (2.5 × 109 ) · (1.6 × 10−19 )

We also needed to include a factor of 109 to account for the fact that the width is 2.5 GeV =
2.5 × 109 eV.

10
1.6 (c)
Through the energy-time uncertainty principle, lifetime and decay width are inversely pro-
portional. Therefore, if the width approaches 0, then the lifetime diverges: the particle lives
forever. Correspondingly, if the width gets very large, then the lifetime approaches 0: the
particle decays instantly.

1.7 Decay of Strange Hadrons

The Ω− hadron travels about 3 cm = 0.03 m before decaying which corresponds to a lifetime
by multiplication by its velocity. While its velocity is not known, it will be an appreciable
fraction of the speed of light c, so we can just assume that it is c. Dividing by c, the lifetime
τ is
0.03
τ' s ' 10−10 s . (34)
3 × 108
To convert to the decay width ∆E, we use the energy-time uncertainty relationship, and the
factor 1.6 × 10−19 to convert to eV. We then find
~ 1.05 × 10−34
∆E = ' −10 eV ' 6.6 × 10−5 eV . (35)
τ 10 · (1.6 × 10−19 )

1.8 PDG Review

1.8 (a)
From the PDG, the lower bound on the lifetime of the proton is 2.1 × 1029 years. To have
a reasonable probability to observe one proton decay in a year, you would need at least
2.1 × 1029 protons, if the lifetime were exactly at the lower bound. Water, H2 O, consists of
10 protons (and 8 neutrons), so we would need to observe 2.1 × 1028 water molecules for a
year. The volume of this amount of water can be found by first identifying the number of
moles of water:
2.1 × 1028
mol ' 3.5 × 104 mol . (36)
6.02 × 1023
Because water consists of 18 total protons and neutrons, this amount of water has a mass of

3.5 × 104 · 18 g ' 630 kg .

(37)

Water has a density of 1000 kg/m3 , so the total volume of water needed is about
630 3
m = 0.63 m3 . (38)
1000
This is roughly the volume of a large bathtub.

11
1.8 (b)
From the “Particle Listings” section of the PDG, the masses of the W , Z, and Higgs bosons
and the top quark are:
mW = 80.379 GeV , (39)
mZ = 91.1876 GeV , (40)
mH = 125.18 GeV , (41)
mt = 173.0 GeV . (42)
The mass of the proton or neutron is approximately 1 GeV (actually slightly less), and so the
value of the mass in GeV can be approximately used to identify the atomic mass of elements
with about the same mass. For example, Krypton has an atomic mass of about 83, which is
close to the mass of the W boson. Zirconium has an atomic mass of about 91, close to the
mass of the Z boson. Tellurium has an atomic mass of about 127, close to the mass of the
Higgs boson. Finally, Ytterbium has an atomic mass of about 174, close to the mass of the
top quark.
Students may find a slightly different selection of elements from a more precise accounting
of the proton and neutron masses or identification of different isotopes.

1.8 (c)
We want to identify the masses of the particles involved in the bubble chamber trace of
Fig. 1.6. Again, we use the “Particle Listings” section of the PDG, and we have to do a bit
of sleuthing to identify all the particles by their symbols. Their masses are:
mK − = 493.677 MeV , (43)
mΩ− = 1672.43 MeV , (44)
mK 0 = 497.611 MeV , (45)
mπ − = 139.57061 MeV , (46)
mΞ0 = 1314.82 MeV , (47)
mK + = 493.677 MeV , (48)
mΛ0 = 1115.683 MeV , (49)
mp = 938.2720813 MeV . (50)
The width, or lifetime, from the PDG of the Ω− baryon is 0.821 × 10−10 s, which is very
close to our very simple estimate!

1.9 InSpire and arXiv

1.9 (a)
At InSpire, we can search for Noether’s papers and find her most highly-cited paper. In
fact, InSpire only has one of her papers listed (all others are pure mathematics), which is

12
“Invariant Variation Problems.”

1.9 (b)
We can also search by date. To find all papers from 1967, we use the command “find date =
1967”. Then, we can sort by decreasing order in citation count. The two most highly-cited
papers are:
S. Weinberg, “A Model of Leptons,” Phys. Rev. Lett. 19, 1264 (1967).
A. D. Sakharov, “Violation of CP Invariance, C asymmetry, and baryon asym-
metry of the universe,” Pisma Zh. Eksp. Teor. Fiz. 5, 32 (1967) [JETP Lett. 5,
24 (1967)] [Sov. Phys. Usp. 34, no. 5, 392 (1991)] [Usp. Fiz. Nauk 161, no. 5,
61 (1991)].

2 Special Relativity
2.1 Properties of Lorentz Transformations
We’re asked to verify that the matrix
 
γ 0 0 γβ
 0 1 0 0 
Λµν =
 0
 (51)
0 1 0 
γβ 0 0 γ
leaves the metric invariant:
Λ| ηΛ = η . (52)
|
First, note that the matrix Λ is symmetric: Λ = Λ. So, multiplying from the left, we have
    
γ 0 0 γβ 1 0 0 0 γ 0 0 −γβ
 0 1 0 0   0 −1 0 0  =  0 −1 0 0 

Λ| η =  
 0 0 1 0   0 0 −1 0   0
. (53)
0 −1 0 
γβ 0 0 γ 0 0 0 −1 γβ 0 0 −γ
Continuing, multiplying by Λ on the right produces
  
γ 0 0 −γβ γ 0 0 γβ
 0 −1 0 0  0 1 0
  0 
Λ| ηΛ =   (54)
 0 0 −1 0  0 0 1 0 
γβ 0 0 −γ γβ 0 0 γ
 2 
γ (1 − β 2 ) 0 0 0
 0 −1 0 0 
=  = η.
 0 0 −1 0 
0 0 0 −γ (1 − β 2 )
2

Recall that 1 − β 2 = γ −2 , from which the result follows.

13
2.2 Rapidity
The rapidity y is defined to be
1 E + pz
y= log . (55)
2 E − pz
We can perform a Lorentz boost along the ẑ axis by a velocity β by multiplication of the
momentum four-vector by a Lorentz matrix:
    
γ 0 0 γβ E γ(E + βpz )
 0 1 0 0  0   0 
 0 0 1 0  0  = 
    . (56)
0 
γβ 0 0 γ pz γ(pz + βE)

Therefore, under a Lorentz transformation, the rapidity transforms as

1 γ(E + βpz ) + γ(pz + βE) 1 (1 + β)(E + pz ) 1 1+β

y→ log = log = y + log . (57)
2 γ(E + βpz ) − γ(pz + βE) 2 (1 − β)(E − pz ) 2 1−β

That is, under a Lorentz boost along the ẑ axis, the rapidity transforms additively.

2.3 Lorentz-Invariant Measure

Under a Lorentz transformation, the coordinate four-vector xµ transforms to

x0µ = Λµν xν . (58)

Then, under a Lorentz transformation, the coordinate measure d4 x transforms as

d4 x0 = |J| d4 x , (59)

where J is the Jacobian formed from the determinant of the derivative matrix:
∂x0µ
J = det . (60)
∂xν
From the Lorentz transformation above in terms of the matrix Λ, this partial derivative is

∂x0µ
= Λµν . (61)
∂xν
Therefore, the Jacobian is just the determinant of the Lorentz-transformation matrix:

J = det Λ . (62)

By the definition of Λ, it satisfies

Λ| ηΛ = η , (63)

14
which enables us to calculate the determinant of Λ. First, note that transposition doesn’t
change the determinant:
det Λ = det Λ| . (64)
Also, the determinant of the metric η is just the product of its non-zero elements:

det η = −1 . (65)

Then, taking the determinant of the Lorentz transformation equation, we have

det(Λ| ηΛ) = − (det Λ)2 = −1 . (66)

Therefore, | det Λ| = 1 = J, and so after Lorentz transformation the measure d4 x is un-

changed.

2.4 Properties of Klein-Gordon Equation

The solution of the Klein-Gordon equation is an exponential phase function,

φ(x) = e−ip·x = e−i(Et−~p·~x) . (67)

We can find the frequency by adding the period T to the time t and demanding that the
field φ(x) is unchanged:
e−i(E(t+T )−~p·~x) = e−i(Et−~p·~x) , (68)
or that exp[−iET ] = 1. Then, the period T = 2π/E and so the frequency f = E/(2π). A
similar procedure can be used to determine the wavelength of the solution. The wavelength
λ is then
2π
λ= . (69)
|~p|
It then follows that the phase velocity is the just the product of the frequency and wavelength:
E
v = λf = . (70)
|~p|

2.5 Maxwell’s Equations

Gauss’s law has already been identified as the 0th component of the equations of motion of

∂µ F µν = J ν . (71)

Now, let’s take ν = i, a spatial coordinate. Then, Ji is the ith component of the current
vector and the left side of this equation is
 
µi ∂Ei ∂Bk ∂Bj
∂µ F = −∂0 F0i + ∂i Fii + ∂j Fji + ∂k Fki = − + − . (72)
∂t ∂j ∂k

15
This is just the ith component of the vector expression
~
∂E
− ~ = J~ .
+∇×B (73)
∂t
This is just Ampère’s law.
The remaining two Maxwell’s equations follow from components of the Bianchi identity,

∂µ Fνρ + ∂ρ Fµν + ∂ν Fρµ = 0 . (74)

First, taking µ = ν = 0 the requires that ρ = i, a spatial coordinate, because the field
strength tensor is anti-symmetric. However, with this choice, the Bianchi identity is trivial:

∂0 F0i + ∂0 Fi0 = ∂0 (F0i − F0i ) = 0 , (75)

so this produces no useful differential equations. If we instead choose µ = 0 and ν and ρ the
spatial coordinates i and j, we find
 
∂Bk ∂Ej ∂Ei
∂0 Fij + ∂j F0i + ∂i Fj0 = − − − = 0. (76)
∂t ∂i ∂j

This is just the k th component of the vector equation

~
∂B ~ = 0,
+∇×E (77)
∂t
which is Faraday’s law.
Finally, if all of the indices µ, ν, and ρ are different spatial indices i, j, and k, the Bianchi
identity is

∂i Fjk + ∂k Fij + ∂j Fki = −∇ · B = 0 , (78)

which is of course the final of Maxwell’s equations.

2.6 Properties of the Clifford Algebra

2.6 (a)
The Clifford algebra is the anti-commutation of two γ matrices:

{γµ , γν } = 2ηµν I4×4 , (79)

where I4×4 is the 4 × 4 identity matrix. While the question asks to verify the Clifford algebra
for all γ matrices, here we will just verify two anti-commuators. First, if µ = ν = 0, the
relevant γ matrix is  
0 I
γ0 = , (80)
I 0

16
where I is the 2 × 2 identity matrix. Note that
  
0 I 0 I
γ0 γ0 = = I4×4 . (81)
I 0 I 0

Therefore,
{γ0 , γ0 } = 2I4×4 = 2η00 I4×4 . (82)
Now, let’s consider the anti-commutator of γ0 and γ1 , where
 
0 σ1
γ1 = . (83)
−σ1 0

Performing the explicit multiplication, we find

     
0 I 0 σ1 0 σ1 0 I
{γ0 , γ1 } = γ0 γ1 + γ1 γ0 = + (84)
I 0 −σ1 0 −σ1 0 I 0
   
−σ1 0 σ1 0
= + = 0.
0 σ1 0 −σ1

This is as expected because η01 = 0.

2.6 (b)
We are asked to determine the unitary matrix S that enacts a similarity transformation
which relates the Weyl basis of γ matrices to this new basis. Because both sets of γ matrices
have a block form, we can generically express S as
 
A B
S= , (85)
C D

for 2 × 2 matrices A, B, C, and D.

Demanding that this matrix is unitary, SS † = I further restricts its entries. We have
 †
A C† AA† + BB † AC † + BD†
     
† A B I 0
SS = = = . (86)
C D B † D† CA† + DB † CC † + DD† 0 I

We then want the following relationships:

 †
A C† BA† + AB † BC † + AD†
      
Weyl † A B 0 I I 0
Sγ0 S = = = ,
C D I 0 B † D† DA† + CB † DC † + CD† 0 −I
       
Weyl Weyl A B 0 σi −Bσi Aσi 0 σi A B
Sγi = γi S = = =
C D −σi 0 −Dσi Cσi −σi 0 C D
 
σi C σi D
= (87)
−σi A −σi B

17
 The unitary condition and the similarity transformation for γ0 can be satisfied by setting
B = A and C = −D. Then, these conditions become
2AA†
   
0 I 0
= , (88)
0 2DD† 0 I
2AA†
   
0 I 0
= , (89)
0 −2DD† 0 −I
   
−Aσi Aσi −σi D σi D
= . (90)
−Dσi −Dσi −σi A −σi A
The final equation just enforces the relationships
Aσi = σi D , Dσi = σi A . (91)
We only need to demonstrate that one such unitary matrix exists. Therefore, a simple
solution is that A = D, which is proportional to the 2 × 2 identity matrix. By the unitary
condition, we find that
1
A = √ I, (92)
2
so that  
1 I I
S=√ . (93)
2 −I I

2.6 (c)
Now, we act with the similarity transformation that we found on the solutions to the Dirac
equation
   
ξ −imt ξ
ψ+ = e , ψ− = eimt , (94)
ξ −ξ
First, on ψ+ , we find
    
† 1 I −I ξ −imt √0
S ψ+ = √ e = e−imt . (95)
2 I I ξ 2ξ
On ψ− , we find
    √ 
† 1 I −I ξ imt 2ξ
S ψ− = √ e = eimt . (96)
2 I I −ξ 0

2.7 Relativity of Spin-1/2

2.7 (a)
Taking the dot product of a momentum four-vector p with the σ matrices, we find
 
p0 − p3 −p1 + ip2
p · σ = Ip0 − σ1 p1 − σ2 p2 − σ3 p3 = . (97)
−p1 − ip2 p0 + p3

18
The determinant of this matrix is

det(p · σ) = (p0 − p3 )(p0 + p3 ) − (−p1 + ip2 )(−p1 − ip2 ) = p20 − p21 − p22 − p23 = p2 , (98)

which is just the square of the four-vector p.

2.7 (b)
The trace of this matrix is then

tr(p · σ) = (p0 − p3 ) + (p0 + p3 ) = 2p0 . (99)

As a 2 × 2 matrix, it has two eigenvalues λ1 and λ2 . The determinant is the product of

eigenvalues and the trace is the sum:

λ1 λ2 = p2 , λ1 + λ2 = 2p0 . (100)

These can be combined into the quadratic equation

λ(2p0 − λ) = p2 , (101)

for an eigenvalue λ. The solutions to the quadratic formula are

q
λ1,2 = p0 ± p20 − p2 . (102)

2.7 (c)
A Lorentz transformation on the four-vector p expressed as the matrix p · σ is

p · σ → Ap · σB , (103)

for two 2 × 2 matrices A and B. The matrix p · σ is Hermitian, (p · σ)† = p · σ because each
of the σ matrices are Hermitian. A Lorentz transformation turns a four-vector into another,
valid, four-vector so we can express the Lorentz-transformed result as

Ap · σB = p0 · σ , (104)

for some new four-vector p0 . The matrix on the right is Hermitian, and therefore so too is
the matrix on the left. That is,

(Ap · σB)† = B† p · σA† = Ap · σB . (105)

For this to hold in general, we must have that B = A† .

19
2.7 (d)
Under a Lorentz transformation the square of a four-vector is unchanged. Therefore, the
determinant of the matrix p · σ is unchanged after Lorentz transformation. That is

det(p · σ) = p2 = det Ap · σA† = p2 det AA† .

(106)

Therefore, we must have that

det AA† = 1 .


(107)

2.7 (e)
Using this form of the four-vector and the expected form for Lorentz transformations, we
want to determine the matrices that perform some particular transformations. First, a
rotation by an angle φ in the x̂ŷ plane. After this rotation, the momentum four-vector is

p → (p0 , p1 cos φ − p2 sin φ, p2 cos φ + p1 sin φ, p3 ) . (108)

Therefore, after Lorentz transformation, the matrix becomes

 
p0 − p3 −(p1 cos φ − p2 sin φ) + i(p2 cos φ + p1 sin φ)
p·σ →
−(p1 cos φ − p2 sin φ) − i(p2 cos φ + p1 sin φ) p0 + p 3
(−p1 + ip2 )e−iφ
 
p0 − p 3
= iφ . (109)
(−p1 − ip2 )e p0 + p 3

Such a rotation doesn’t change the trace of the matrix, which results in an additional
constraint. The trace is cyclic:

tr(Ap · σA† ) = tr(A† Ap · σ) = tr(p · σ) , (110)

therefore AA† = I. That is, the matrix A is an element of SU(2), the set of 2 × 2 unitary
matrices with unit determinant. A generic SU(2) matrix can be written as
 iα 
e cos θ −eiβ sin θ
A= , (111)
e−iβ sin θ e−iα cos θ

for an angle θ and two phases α, β. For a rotation exclusively in the x̂ŷ plane, θ = 0 and so
 iα 
e 0
A= . (112)
0 e−iα

Multiplying this on the left and right of the matrix p · σ, we find

 iα    −iα 
† e 0 p0 − p3 −p1 + ip2 e 0
Ap · σA =
0 e−iα −p1 − ip2 p0 + p3 0 eiα
 
p0 − p 3 (−p1 + ip2 )e2iα
= . (113)
(−p1 − ip2 )e−2iα p0 + p3

20
For this to agree with the expression for the rotated matrix that we identified earlier, we
must have 2α = −φ so that  −iφ/2 
e 0
A= . (114)
0 eiφ/2
Now, let’s perform a boost by a velocity β along the ẑ axis. Under this transformation,
the energy and ẑ component of momentum transform as
p0 → γ(p0 + βp3 ) , (115)
p3 → γ(p3 + βp0 ) . (116)
Therefore, the four-momentum matrix transforms to
 
γ(1 − β)(p0 − p3 ) −p1 + ip2
p·σ → . (117)
−p1 − ip2 γ(1 + β)(p0 + p3 )
The form of the Lorentz transformation matrix A that implements this transformation is
 
a 0
A= = A† , (118)
0 a−1
for some real number a. Acting this on the left and right of the matrix p · σ we find
   
† a 0 p0 − p3 −p1 + ip2 a 0
Ap · σA =
0 a−1 −p1 − ip2 p0 + p3 0 a−1
 2 
a (p0 − p3 ) −p1 + ip2
= . (119)
−p1 − ip2 a−2 (p0 + p3 )
Therefore, it must be that
a2 = γ(1 − β) . (120)
Note that indeed
γ(1 − β) · γ(1 + β) = γ 2 (1 − β 2 ) = 1 , (121)
and so the coefficients of the diagonal entries are inverses of one another.

2.8 Dark Matter Searches

2.8 (a)
The radius of the Earth’s orbit about the Milky Way is about 25,000 light years, so the total
distance d over which the Earth travels in one orbit of the Milky Way is
d = 2π · 25, 000 ly ' 1.6 × 105 ly . (122)
The ratio of this distance to the orbital period of 230,000,000 years is the fraction of the
speed of light at which we orbit the galaxy. That is,
v 1.6 × 105
' ' 7 × 10−4 . (123)
c 2.3 × 108
Then, the velocity at which we orbit the galaxy, or move through the dark matter halo, is
v ' (7 × 10−4 ) · (3 × 108 ) m/s ' 2 × 105 m/s . (124)

21
2.8 (b)
Assuming that the initial xenon nucleus is at rest, its four-momentum is

pXe = (mXe , 0, 0, 0) . (125)

The initial WIMP momentum is aligned along the ẑ axis, so we can express its four-
momentum as q 
pχ = p2z + m2χ , 0, 0, pz . (126)
Note that this four-vector is on-shell.
For the final momenta, we restrict their three-momentum to still lie along the ẑ axis. For
the final-state xenon nucleus, we have
q 
0 Xe 2 2 Xe
pXe = (pz ) + mXe , 0, 0, pz . (127)

By momentum conservation, it then follows that the four-vector of the final state WIMP is
q 
0 2 Xe
pχ = (pz − pz ) + mχ , 0, 0, pz − pz
Xe 2 . (128)

All of these four-vectors are on-shell and three-momentum is conserved.

2.8 (c)
Enforcing conservation of energy allows us to solve for the recoiling xenon nucleus momen-
tum, pXe
z . We have the conservation equation
q q q
2 2 2
mXe + pz + mχ = (pz ) + mXe + (pz − pXe
2 2 Xe 2
z ) + mχ . (129)

Re-writing pXe
z = αpz , this becomes
q q q
mXe + pz + mχ = α pz + mXe + (1 − α)2 p2z + m2χ .
2 2 2 2 2
(130)

Squaring both sides of this equation, we then find

q
m2Xe + p2z + m2χ + 2mXe p2z + m2χ (131)
q q
= α2 p2z + m2Xe + (1 − α)2 p2z + m2χ + 2 α2 p2z + m2Xe (1 − α)2 p2z + m2χ .

Much can be simplified in this expression. Canceling and rearranging terms, we then find
q q q
mXe pz + mχ + α(1 − α)pz = α pz + mXe (1 − α)2 p2z + m2χ .
2 2 2 2 2 2
(132)

22
Now, we can square both sides again. We find
q
m2Xe p2z + m2χ + α2 (1 − α)2 p4z + 2mXe α(1 − α)p2z p2z + m2χ


(133)
= α2 p2z + m2Xe (1 − α)2 p2z + m2χ

= α2 (1 − α)2 p4z + α2 p2z m2χ + m2Xe (1 − α)2 p2z + m2Xe m2χ .

We can again cancel an rearrange terms. We then find

q  q 
2mXe p2z + m2χ + 2m2Xe = α m2χ + m2Xe + 2mXe p2z + m2χ . (134)

In doing this simplification, we have divided the whole expression by α. This implies that
α = 0 is a solution, but we ignore it because it is smaller than the other solution. Solving
for α, we find
2mXe p2z + m2χ + 2m2Xe
p
pXe
z
α= = 2 p . (135)
pz mχ + m2Xe + 2mXe p2z + m2χ
From this, we can solve for pXe
z by simply multiplying by pz .

2.8 (d)
The energy of the recoiling xenon nucleus is
Xe 2


p
q
2 2 z
z ) ' mXe +
E = mXe + (pXe , (136)
2mXe

where on the right we have Taylor expanded to first non-trivial order in pXe
z . The kinetic
energy K is then
2
pXe
z
K = E − mXe ' , (137)
2mXe
which is just what we would expect from the non-relativistic expression. To lowest order in
the WIMP momentum, the recoiling xenon nucleus momentum is

2mXe mχ + 2m2Xe 2mχ mXe

pXe
z ' pz 2 2
= v. (138)
mχ + mXe + 2mXe mχ (mχ + mXe )

The kinetic energy K is therefore

2
pXe
z 2m2χ mXe
K= = v2 . (139)
2mXe (mχ + mXe )2

23
2.8 (e)
The minimum kinetic energy that the LUX experiment can measure is about 1 keV, which
is 10−6 GeV. Using the expression derived in the previous part and the velocity v of the
WIMP halo from part (a), we then have
2m2χ mXe v 2 2m2χ mXe
Kmin = 10−6 GeV = GeV ' 5 × 10−7
GeV . (140)
(mχ + mXe )2 c2 (mχ + mXe )2
Rearranging and expressing the WIMP mass as a fraction of the xenon mass mχ = αmXe ,
we then have
α2 mXe
1 GeV = GeV . (141)
(1 + α)2
131
The mass of Xe is approximately 123 GeV, so this can further be arranged to solve for α:
(1 + α)2 ' 123α2 . (142)
Taking the square-root of both sides and solving for α, we find
mχ
α= ' 0.1 . (143)
mXe
That is, the minimal WIMP mass that LUX is sensitive to is approximately 1/10 the mass
of 131 Xe, or about 12 GeV. This is remarkably close to the lower mass bound in the plot.

2.9 Top Quark Decay

2.9 (a)
To evaluate the dot product of the W boson and bottom quark momentum, let’s square the
top quark’s momentum:
p2t = m2t = (pW + pb )2 = m2W + 2pW · pb , (144)
where we have used conservation of momentum in the top quark decay and assumed that
the bottom quark is massless. Therefore,
2pW · pb = m2t − m2W . (145)

2.9 (b)
In the W boson decay to a lepton and neutrino, momentum is conserved as
pW = pl + pν . (146)
The lepton and neutrino always have masses much, much less than the W boson, so we
assume they are massless. As such, the dot product of the lepton and neutrino momentum
is
p2W = m2W = (pl + pν )2 = 2pl · pν . (147)

24
2.9 (c)
Combining the top quark and W boson decays, we have the momentum conservation equa-
tion:
p2t = pb + pl + pν . (148)
Squaring both sides and using the relationships we derived above, we find

p2t = m2t = (pb + pl + pν )2 = 2pb · pl + 2pb · pν + 2pl · pν = 2pb · pl + 2pb · pν + m2W . (149)

Therefore, the dot product of the lepton and bottom quark momentum is

2pl · pb = m2t − m2W − 2pb · pν . (150)

2.9 (d)
Assuming that both the lepton and neutrino are massless, the smallest value that the dot
product 2pb · pν can take is 0. To see this, note that this dot product is

2pb · pν = 2Eb Eν − 2~pb · p~ν = 2Eb Eν (1 − cos θ) , (151)

where we used that |~p| = E for a massless particle and p~b · p~ν = |~pb ||~pν | cos θ, where θ is
the angle between the momenta. This angle can go to 0, at which point the four-vector dot
product vanishes.
With this minimum established, the maximum value that the lepton and bottom quark
four-momentum dot product can be is

2pl · pb |max = m2t − m2W . (152)

2.9 (e)
The maximum value of the lepton and bottom quark momentum dot product is therefore

m2lb = m2t − m2W , (153)

and so the top quark mass is q

mt = m2lb + m2W . (154)
From the plot of Fig. 2.4, the endpoint of the distribution is approximately 150 GeV. With
the W boson mass of 80 GeV, the top quark mass is approximately
q √
mt = m2lb + m2W ' 1502 + 802 GeV = 170 GeV . (155)

25
3 A Little Group Theory
3.1 Representations of the Symmetric Group
3.1 (a)
Let’s first study the three-dimensional representation of the symmetric group, S3 . We imag-
ine that the three elements are represented as a vector:
 
a
 b , (156)
c
We can determine the elements of S3 as represented by 3 × 3 as their action on this vector.
The identity element is of course just the identity matrix:
 
1 0 0
 0 1 0 . (157)
0 0 1
We are also told about the matrix that represents a 120◦ clockwise rotation:
 
0 0 1
 1 0 0 . (158)
0 1 0
The 6 matrices that implement the elements of S3 are defined by having one 1 in each
row and column, and zeroes otherwise. So, another matrix is
 
1 0 0
 0 0 1 , (159)
0 1 0
which is a reflection about vertex 1, or
 
0 1 0
 1 0 0 , (160)
0 0 1
which is a reflection about vertex 3, or
 
0 1 0
 0 0 1 , (161)
1 0 0
which is a 240◦ clockwise rotation, or finally
 
0 0 1
 0 1 0 , (162)
1 0 0
which is a reflection about vertex 2.

26
3.1 (b)
Now, let’s study the two-dimensional representation of S3 . The vector we consider now is
a two-dimensional vector whose entries represent the locations of the vertices in the (x, y)
plane. Let’s put the vertices of the equilateral triangle on the unit circle so that one vertex
is located at (0, 1), another vertex is located at
  √ !
7π 7π 3 1
cos , sin = − ,− , (163)
6 6 2 2

and the final vertex is at

  √ !
11π 11π 3 1
cos , sin = ,− , (164)
6 6 2 2

We of course have the identity element

 
1 0
, (165)
0 1

which leaves each of the vertices unchanged. Another element flips the second and third
vertices:  
−1 0
, (166)
0 1
and does nothing to the first vertex. Because we have put the vertices on the unit circle,
rotations are just implemented by the familiar elements of SO(2). A clockwise rotation by
120◦ is √ !
3
cos 2π 2π 1
 
3
− sin 3
−√2
− 2
2π 2π = 3
, (167)
sin 3 cos 3 −1 2 2
◦
while a clockwise rotation by 240 is
√ !
3
cos 4π − sin 4π −√21
 
3 3 = 2 . (168)
sin 4π
3
cos 4π
3 − 23 − 12

The remaining reflections about vertex 2 and 3 can be found by multiplying the reflection
about vertex 1 and the two rotations. First, the reflection and the 120◦ rotation:
√ ! √ !
1 3 1 3
 
−1 0 −√2
− 2 √2 2
3 1
= 3 1
. (169)
0 1 − −
2 2 2 2

And finally, the reflection and the 240◦ rotation:

√ ! √ !
3
−√12 1
− 23
 
−1 0 2 2√
= . (170)
0 1 − 23 − 12 − 2
3
− 12

27
3.2 Lorentz Group
A matrix that implements a Lorentz transformation Λ is defined to satisfy

Λ| ηΛ = η . (171)

Let’s prove that the set of all such matrices {Λ} forms a group. First, the identity matrix I
satisfies this requirement:
I| ηI = IηI = η . (172)
Also, matrix multiplication is associative, and so for three Lorentz transformation matrices
Λ1 , Λ2 , and Λ3 it is true that
(Λ1 Λ2 )Λ3 = Λ1 (Λ2 Λ3 ) . (173)
Next, the set of matrices are closed. For two matrices Λ1 and Λ2 which satisfy

Λ|1 ηΛ1 = η , Λ|2 ηΛ2 = η , (174)

their product also satisfies this constraint:

(Λ1 Λ2 )| η(Λ1 Λ2 ) = Λ|2 Λ|1 ηΛ1 Λ2 = Λ|2 ηΛ2 = η . (175)

Finally, we have to verify that the matrix Λ has an inverse Λ−1 which is also a Lorentz
transformation. To determine if the matrix Λ is invertible, we calculate its determinant. We
have that
det(Λ| ηΛ) = det(Λ)2 det(η) = det(η) = −1 . (176)
To write this, we used the fact that the determinant of a product of matrices is just the prod-
uct of determinants of each matrix and that the determinant of a matrix and its transpose
are identical. Further, we used the constraint on Λ such that it is a Lorentz transformation.
This then requires that
det(Λ)2 = 1 , (177)
which is non-zero. Therefore Λ has an inverse, Λ−1 . As Λ has an inverse, we can multiply
on the left and right of the constraint by Λ−1 :

|
|
Λ−1 Λ| ηΛΛ−1 = η = Λ−1 ηΛ−1 . (178)

Therefore, Λ−1 satisfies the same constraint equation to leave the metric invariant, so it is
also a Lorentz transformation. This proves that the set of matrices {Λ} that satisfy

Λ| ηΛ = η (179)

forms a group.

28
3.3 Hermitian Matrices
3.3 (a)
Let’s first prove that Hermitian matrices have exclusively real eigenvalues. The eigenvalue
equation is
M~v = λ~v , (180)
for an eigenvector ~v and an eigenvalue λ. Let’s act on the left of this equation by the
Hermitian conjugate vector ~v † :
~v † M~v = λ~v †~v . (181)
Because the eigenvalue equation is homogeneous, we are always allowed to normalize the
eigenvector to square to unity:
~v †~v = 1 . (182)
Therefore, the eigenvalue λ is just
~v † M~v = λ . (183)
Now, let’s Hermitian conjugate this expression:

†
~v † M~v = ~v † M†~v = λ∗ . (184)

Because M is Hermitian, M = M† , and so we have

~v † M†~v = λ∗ = ~v † M~v = λ . (185)

Therefore, λ = λ∗ , and so all eigenvalues must be real.

3.3 (b)
Now, consider two eigenvalues ~v1 and ~v2 such that

M~v1 = λ1~v1 , M~v2 = λ2~v2 , (186)

for λ1 6= λ2 . Note also that by Hermitian conjugating these expressions we also have

~v1† M = λ1~v1† , ~v2† M = λ2~v2† , (187)

where we have used the fact that the matrix M is Hermitian and its eigenvalues are real.
Now, consider the inner product:

~v1† M~v2 = λ2~v1†~v2 , (188)

where M acts on ~v2 . However, it could instead act on ~v1† first:

~v1† M~v2 = λ1~v1†~v2 , (189)

Because λ1 6= λ2 , the only way for these to be consistent is if ~v1 and ~v2 are orthogonal:
~v1†~v2 = 0.

29
3.4 Baker-Campbell-Hausdorff Formula
In this exercise, we are asked to explicitly determine the exponential matrix form of the
product eiT1 eiT2 = eiT3 . Taking the logarithm of both sides, we have

iT3 = log eiT1 eiT2 ,

(190)

where the exponential and logarithm are defined by their Taylor series. Through quadratic
order in the matrices T1 and T2 , the product of the exponentiated matrices is

T12 T22
  
iT1 iT2
e e = I + iT1 − + ··· I + iT2 − + ··· (191)
2 2
T2 T2
= I + i(T1 + T2 ) − 1 − 2 − T1 T2 + · · · .
2 2
Note the ordering: T1 T2 6= T2 T1 , in general.
Now, we can plug this into the Taylor expansion for the logarithm. Recall that

x2
log(1 + x) = x − + ··· . (192)
2


Therefore, the Taylor expansion of log eiT1 eiT2 is

T12 T22
 
iT1 iT2


log e e = log I + i(T1 + T2 ) − − − T1 T2 + · · · (193)
2 2
T2 T2 (T1 + T2 )2
= i(T1 + T2 ) − 1 − 2 − T1 T2 + + ···
2 2 2
T1 T2 T2 T1
= i(T1 + T2 ) − T1 T2 + + + ···
2 2
1
= i(T1 + T2 ) − [T1 , T2 ] + · · · .
2
Here, [T1 , T2 ] is the commutator:

[T1 , T2 ] = T1 T2 − T2 T1 . (194)

Therefore, the matrix T3 through quadratic order in T1 and T2 is

i
T3 = T1 + T2 + [T1 , T2 ] + · · · . (195)
2

3.5 Casimir Operator

The Casimir of the spin group Ŝ 2 is defined by its Lie algebra elements Ŝx , Ŝy , and Ŝz as

Ŝ 2 = Ŝx2 + Ŝy2 + Ŝz2 . (196)

30
The commutator of this object with Ŝx , for example, is

[Ŝ 2 , Ŝx ] = (Ŝx2 + Ŝy2 + Ŝz2 )Ŝx − Ŝx (Ŝx2 + Ŝy2 + Ŝz2 ) . (197)

Note that we can change the order of operator multiplication with commutators. For exam-
ple,
Ŝx Ŝy2 = Ŝy Ŝx Ŝy + [Ŝx , Ŝy ]Ŝy = Ŝy2 Ŝx + Ŝy [Ŝx , Ŝy ] + [Ŝx , Ŝy ]Ŝy . (198)
The commutation relations tell us that

[Ŝx , Ŝy ] = iŜz . (199)

Therefore, we have that

Ŝx Ŝy2 = Ŝy2 Ŝx + iŜy Ŝz + iŜz Ŝy . (200)
One can do a similar manipulation and show that

Ŝx Ŝz2 = Ŝz2 Ŝx − iŜy Ŝz − iŜz Ŝy . (201)

Putting this all together, the commutator of the Casimir and Ŝx is

[Ŝ 2 , Ŝx ] = (Ŝx2 + Ŝy2 + Ŝz2 )Ŝx − (Ŝx3 + Ŝy2 Ŝx + iŜy Ŝz + iŜz Ŝy + Ŝz2 Ŝx − iŜy Ŝz − iŜz Ŝy ) (202)
= 0.

Because the Casimir Ŝ 2 is symmetric in x, y, and z, this proves generally that

[Ŝ 2 , Ŝi ] = 0 , (203)

for i = x, y, z.

3.6 Helicity
The photon polarization vectors are
1 1
R (p) = √ (0, 1, i, 0) , L (p) = √ (0, 1, −i, 0) . (204)
2 2
A rotation in the x̂ŷ plane by an angle φ is just implemented by the familiar SO(2) matrix
 
cos φ − sin φ
M= . (205)
sin φ cos φ

Let’s first act this matrix on the vector formed from the x̂ and ŷ components of the right-
handed helicity polarization vector:

e−iφ 1
       
cos φ − sin φ 1 1 1 cos φ − i sin φ
√ =√ = √ . (206)
sin φ cos φ 2 i 2 i(cos φ − i sin φ) 2 i

31
For the left-handed polarization vector, we find

eiφ
       
cos φ − sin φ 1 1 1 cos φ + i sin φ 1
√ =√ =√ . (207)
sin φ cos φ 2 −i 2 −i(cos φ + i sin φ) 2 −i
The eigenvalues of these vectors under rotations are manifest. Note that the right-handed
polarization vector rotates with the angle φ. That is, if it initially has an angle θ, then its
new angle is θ − φ, which is a rotation in the direction of φ. The value of θ must be larger by
φ to correspond to the same point as before the rotation. Correspondingly, the left-handed
polarization vector rotates oppositely. Under a full rotation by φ = 2π, the polarization
vectors return to their original values.

3.7 Symplectic Group

3.7 (a)
We are asked to prove that the set of matrices {M} that satisfy
   
| 0 I 0 I
M M= . (208)
−I 0 −I 0

As usual, associativity of the group is satisfied because matrix multiplication is associative.

Also, the N × N identity matrix I satisfies this requirement, which we won’t write out
explicitly. If two matrices M and N satisfy the requirement,
       
| 0 I 0 I | 0 I 0 I
M M= , N N= , (209)
−I 0 −I 0 −I 0 −I 0

then so too does their product MN. Note that

       
| 0 I | | 0 I | 0 I 0 I
(MN) MN = N M MN = N N= . (210)
−I 0 −I 0 −I 0 −I 0

Finally, we have to show that the inverse of a matrix M ∈ Sp(N ) is also in Sp(N ). To
show that the inverse M−1 exists, we take the determinant of the requirement equation to
find
   
0 I
det M |
M = det(M)2 = 1 . (211)
−I 0

Here, we have used the facts that the determinant of a product of matrices is the product of
determinants, det M = det M| , and that the determinant of the off-diagonal matrix is unity:
 
0 I
det = 1. (212)
−I 0

32
Because det(M)2 = 1, the matrix M is non-singular, and therefore has an inverse, M−1 . Now,
we can just multiply by this inverse and its transpose on the left and right of the constraint
equation. This produces
   
−1 |

0 I −1 0 I
M M = . (213)
−I 0 −I 0

Therefore M−1 is also an element of Sp(N ). This completes the proof that this set of matrices
forms a group.

3.7 (b)
Restricting to Sp(2), we are looking for matrices M that satisfy
   
| 0 1 0 1
M M= . (214)
−1 0 −1 0

Let’s express the matrix M as  

a b
M= , (215)
c d
where a, b, c, d are in general complex numbers. Plugging this expression into the constraint
equation yields
       
a c 0 1 a b 0 ad − bc 0 1
= = . (216)
b d −1 0 c d −(ad − bc) 0 −1 0

That is, matrices that are elements of Sp(2) are those with unit determinant. The group
of all 2 × 2 matrices with complex elements with unit determinant is called SL(2, C). If we
further restrict to those unitary matrices M† M = I, then the resulting group is SU(2).

3.8 π-p Scattering

3.8 (a)
(i) We want to find the Clebsch-Gordan coefficient for π 0 n in the I = 3/2, I3 = −1/2
state. The isospin of the pion triplet is 1 and of the nucleon doublet is 1/2, so we look
in the 1 × 1/2 table. The π 0 has Ip3 = 0 while the neutron has I3 = −1/2. Therefore,
this Clebsch-Gordan coefficient is 2/3.

(ii) In that same table, we now look for the Clebsch-Gordan coefficient for π − p in the
I = 1/2, I3 = −1/2 state. π −p has I3 = −1 while the proton has I3 = 1/2, so this
Clebsch-Gordan coefficient is − 2/3.

(iii) Finally, we’re asked for the Clebsch-Gordan coefficient for π + p in the I = 3/2, I3 = 1/2
state. We immediately know that this Clebsch-Gordan is 0 because the sum of the π +
and proton third components of isospin is 3/2, and not 1/2.

33
3.8 (b)
Let’s now consider the production of a ∆ baryon from the scattering of either π + p or π − p.
Because the pion and nucleon doublets have isospin 1 and 1/2, respectively, we want to look
in the 1/2 table, as earlier. Now, because the ∆ baryon has total isospin of 3/2, we want
to find the Clebsch-Gordan coefficients for π + p and π − p that can produce total isospin 3/2.
For π + p, the only state that it can produce is I = 3/2, I3 = 3/2 because the sum of the
third components of isospin is 3/2. Therefore, the Clebsch-Gordan is 1. By contrast, for
π − p, the only combination of isospin I3 = −1 (π − ) and
√ I3 = 1/2 (p) that produces a total
isospin of 3/2 has a Clebsch-Gordan coefficient of 1/ 3. These values are proportional to
wavefunction coefficients, so probabilities are proportional to their squares. Thus, it is 3
times more likely for π + p to produce a ∆ baryon than π − p.

3.8 (c)
The minimum value of the mass of the pion-proton system is when both have zero momentum,
and so the system has a total energy of just the sum of pion and proton masses. The charged
pion has a mass of about 140 MeV (from the PDG), while the proton has a mass of about 938
MeV. Therefore, the minimum mass of the pion-proton system is the sum of their individual
masses, or about 1078 MeV. This indeed looks close to where the curve on the plot starts.

3.8 (d)
From the plot, the peak of the π + p scattering probability is located at a value of about
58, while the probability for π − p is a value of about 26. Their ratio is a factor of about
2.2, which is not unreasonably far from the predicted value of 3. However, our prediction
neglected several things, in particular that there is more than just the ∆ baryon that the
pion and proton can produce. These other particles contaminate the probability near the
peak, and shift the expected ratio a bit.

4 Fermi’s Golden Rule & Feynman Diagrams

4.1 Galactic Collisions
4.1 (a)
Recall that the definition of the luminosity is

NA NB |vA − vB |
L= , (217)
Vol
where NA and NB are the numbers of objects in each bunch, respectively, |vA − vB | is the
relative speed of the bunches, and Vol is the volumes of the bunches, assuming both bunches

34
are the same size. Because the Milky Way and Andromeda galaxies are about the same size,
this assumption holds. In particular, their volumes are approximately
 5 2
10
Vol ' π · 103 ly3 ' 1013 ly3 . (218)
2

One light-year in meters is approximately

1 ly = 3 × 108 · π × 107 m ' 1016 m . (219)

In this calculation, we used the fact that, to better than 1% accuracy, the number of seconds
in a year is π × 107 . Therefore, the volume of each galaxy is approximately

3
Vol ' 1013 · 1016 m3 = 1061 m3 . (220)

Further, the relative speed of the galaxies is about |vA − vB | = 105 m/s, and each galaxy has
about 100 billion stars. It then follows that the luminosity is

1011 · 1011 · 105 −2 −1

L' m s = 10−34 m−2 s−1 . (221)
1061
A barn is 10−28 m2 , so in terms of inverse barns, this luminosity is

L ' 10−62 b−1 s−1 , (222)

which is exceedingly small! This is nearly 75 orders of magnitude smaller than the luminosity
at the LHC.

4.1 (b)
From earlier, the total volume of one of these galaxies is

Vol ' 1061 m3 . (223)

Therefore, each star on average has an empty volume around them of approximately
Vol
11
' 1050 m3 . (224)
10
The cube-root of this per-star volume is then approximately the inter-stellar distance:
Distance
' 1050/3 m ' 1017 m . (225)
star
The ratio of this inter-stellar distance to the radius of a typical star, like our Sun is therefore

Distance 1017
' ' 108 . (226)
star · rS 109

35
 By contrast, for proton bunches at the LHC, the volume of a bunch is approximately

3 × 108
VolLHC = A` ' 3 × 10−10 · m3 ' 10−10 m3 . (227)
2 · 4 × 108

Therefore, each proton occupies a volume of approximately

VolLHC
' 10−21 m3 . (228)
1011
The inter-proton distance in bunches at the LHC is the cube-root of this:
Distance
' 10−21/3 m ' 10−7 m . (229)
proton
The ratio of this typical distance to the radius of the proton is therefore

Distance 10−7
' −15 ' 108 . (230)
proton · rp 10

This is remarkably close to the corresponding ratio for the galaxies!

4.1 (c)
The number of total collision events is the time integral of the luminosity times the cross
section: Z
events = dt Lσ ' T Lσ , (231)

where T is the total time of collision. Let’s first calculate this for the collision of galaxies.
We are given that the total collision time is approximately a billion years or

T ' 109 y ' π × 1016 s . (232)

The luminosity from earlier was

L ' 10−34 m−2 s−1 . (233)

The cross section of stars will be approximately just their actual cross sectional area. This
isn’t quite true because stars interact gravitationally, and so even if they are somewhat
widely separated, they could still be attracted to one another and collide. Nevertheless, we
will approximate σ as
2
σ ' 7 × 108 m2 ' 5 × 1017 m2 . (234)
Putting it together, the total approximate number of star collisions from the Milky Way and
Andromeda galaxies colliding is

events ' (π × 1016 ) · 10−34 · (5 × 1017 ) ' 2 . (235)

36
So, there will be about 2 individual star collisions, which is not something to worry about!
Performing the same exercise for the LHC, the time T that is takes the bunches to pass
through one another is approximately the time it takes light to travel the length of a bunch:
`
' 10−9 s .
T ' (236)
c
The luminosity as measured from CMS earlier in the chapter is approximately
L ' 1040 m−2 s−1 . (237)
The cross section σ of the proton is approximately just the square of its radius (with the
caveats discussed for stars), so we have
σ ' 10−30 m2 . (238)
Putting it all together, the number of individual proton collision events per bunch crossing
at the LHC is
events ' 10−9 · 1040 · 10−30 ' 10 . (239)
Again, this is very close to the expected number of stellar collision events. Remarkably, the
proton collisions at the LHC are an extremely good approximation of the fate of the Milky
Way galaxy!

4.2 Integrating δ-functions

4.2 (a)
Consider the integral Z
dx δ(y − f (x)) . (240)
Let’s make the change of variables
τ = f (x) . (241)
Note that we can solve for x as
x = f −1 (τ ) . (242)
Then, the Jacobian of this change of variables is
df 1
dτ = dx or that dx = df
dτ . (243)
dx dx

With this change of variables, the integral becomes

Z Z
1
dx δ(y − f (x)) = dτ  df  δ(y − τ ) . (244)
 
dx x=f −1 (τ )

Note that the Jacobian derivative is evaluated at x = f −1 (τ ) and the explicit δ-function
enforces τ = y. Therefore, this integral can be equivalently expressed as
Z Z
1
dx δ(y − f (x)) = dx  df  δ x − f −1 (y) .


(245)
 
dx

37
4.2 (b)
Now, we want to evaluate the integral
Z 2
dx δ(x2 − 1) . (246)
0

From the previous expression, we identify y = 1 and f (x) = x2 . Therefore,

df
= 2x , (247)
dx
so that the integral can be expressed as
Z 2 Z 2
2 1 1
dx δ(x − 1) = dx δ(x − 1) = . (248)
0 0 2x 2

4.3 Three-Body Phase Space

4.3 (a)
The xi variables are defined as
2Q · pi
xi = , (249)
Q2
and so the sum over i = 1, 2, 3 is

2Q · (p1 + p2 + p3 )
x1 + x2 + x3 = . (250)
Q2
Because Q is the total momentum four-vector, we necessarily have the relationship

Q = p1 + p2 + p 3 . (251)

Therefore,
2Q · (p1 + p2 + p3 ) 2Q2
x1 + x2 + x3 = = = 2. (252)
Q2 Q2

4.3 (b)
In the center-of-mass frame, we have that

Q = (Ecm , 0, 0, 0) , (253)

so that
2Q · pi 2Ecm Ei 2Ei
xi = = = , (254)
Q2 2
Ecm Ecm

38
where Ei is the energy of particle i. Therefore, the energy can be expressed as
xi
Ei = Ecm . (255)
2
The four-vector of particle i can be expressed as
x 
i
pi = (Ei , p~i ) = Ecm , p~i , (256)
2
and it squares to the mass of particle i, mi :

x2i 2
p2i = m2i = Ei2 − |~pi |2 = E − |~pi |2 . (257)
4 cm
Rearranging this, the magnitude of the three-momentum of particle i is
r
x2i 2
|~pi | = E − m2i . (258)
4 cm

4.3 (c)
The complete three-body phase space integral can be expressed as
Z 3
d p1 d3 p2 d3 p3 1 1 1
Z
dΠ3 = 3 3 3
(2π)4 δ (4) (Q − p1 − p2 − p3 ) . (259)
(2π) (2π) (2π) 2E1 2E2 2E3

In the center-of-mass frame, the total momentum Q is

Q = (Ecm , 0, 0, 0) , (260)

and so the momentum-conserving δ-function is

δ (4) (Q − p1 − p2 − p3 ) = δ(Ecm − E1 − E2 − E3 )δ (3) (~p1 + p~2 + p~3 ) . (261)

Therefore, we can use the three-momentum conservation δ-functions to integrate over all
three components of p~3 , enforcing that p~3 = −~p1 − p~2 . Correspondingly, the energy of
particle 3 can be expressed as
q q
E3 = |~p3 |2 + m23 = |~p1 + p~2 |2 + m23 . (262)

Therefore, the three-body phase space integral is

Z 3
d p1 d3 p2 1 1 1
Z
dΠ3 = 2π δ(Ecm − E1 − E2 − E3 ) . (263)
(2π)3 (2π)3 2E1 2E2 2E3

39
4.3 (d)
The remaining integrals can be expressed in spherical coordinates as

|~p1 |2 d|~p1 | d cos θ1 dφ1 |~p2 |2 d|~p2 | d cos θ2 dφ2 1 1 1

Z Z
dΠ3 = 2π δ(Ecm −E1 −E2 −E3 ) .
(2π)3 (2π)3 2E1 2E2 2E3
(264)
We are allowed to arbitrarily fixed the origin at which the angle θ2 , for example is 0. A
convenient choice is to set θ2 = 0 if θ2 = θ1 . That, is we just call θ2 the relative angle
between the momentum of particle 1 and 2, θ12 . We can correspondingly relabel θ1 = ψ so
that
|~p1 |2 d|~p1 | d cos ψ dφ1 |~p2 |2 d|~p2 | d cos θ12 dφ2 1 1 1
Z Z
dΠ3 = 2π δ(Ecm −E1 −E2 −E3 ) .
(2π)3 (2π)3 2E1 2E2 2E3
(265)
Now, we need to express the energy E3 in terms of cos θ12 . From earlier, we have
q q
E3 (cos θ12 ) = |~p1 + p~2 | + m3 = |~p1 |2 + |~p2 |2 + 2|~p1 ||~p2 | cos θ12 + m23 .
2 2
(266)

The δ-function that remains is a non-linear function of cos θ12 :

 q 
δ(Ecm − E1 − E2 − E3 ) = δ Ecm − E1 − E2 − |~p1 |2 + |~p2 |2 + 2|~p1 ||~p2 | cos θ12 + m23 ,
(267)
so we need to use the change of variables rule for δ-functions. First, we calculate the Jacobian
J:
d |~p1 ||~p2 |
q
−1
J = |~p1 |2 + |~p2 |2 + 2|~p1 ||~p2 | cos θ12 + m23 = . (268)
d cos θ12 E3
Then, we put this in the three-body phase space integrand, and can eliminate the integral
over cos θ12 using this δ-function accordingly. We therefore have

|~p1 |2 d|~p1 | d cos ψ dφ1 |~p2 |2 d|~p2 | dφ2 1 1 1

Z Z
E3
dΠ3 = 2π (269)
(2π) 3 (2π) 3 2E1 2E2 2E3 |~p1 ||~p2 |
|~p1 | |~p2 |
Z
1
= 5
d|~p1 | d cos ψ dφ1 d|~p2 | dφ2 . (270)
8(2π) E1 E2

4.3 (e)
The angular integrals that remain just rotate the center-of-mass, which isn’t interesting to
our applications here. So, we can just integrate over them. The integrals over both of the
azimuthal angles φ1 and φ2 are each 2π, and the integral over cos ψ is 2:
Z 1 Z π
d cos ψ = dψ sin ψ = 2 . (271)
−1 0

40
Therefore, these angular integrals contribute an overall factor of 8π 2 , for which the phase
space integral reduces to

|~p1 | |~p2 |
Z Z
1
dΠ3 = 3
d|~p1 | d|~p2 | . (272)
32π E1 E2

4.3 (f )
Finally, we want to change variables in the integrals that remain from |~p1 | and |~p2 | to x1 and
x2 . Recall that the magnitude of momentum |~pi | can be expressed as
r
x2i 2
|~pi | = E + m2i . (273)
4 cm
Perhaps more enlighteningly, this is

x2i 2
|~pi |2 = E + m2i , (274)
4 cm
and so
xi 2 Ecm
|~pi |d|~pi | = Ecm dxi = Ei dxi . (275)
4 2
It then follows that
|~pi | Ecm
d|~pi | = dxi . (276)
Ei 2
Plugging this into the expression for the phase space integral, we finally find
Z 2 Z
Ecm
dΠ3 = dx1 dx2 , (277)
128π 3
as promised.

4.4 e+ e− → e+ e− Scattering
The two Feynman diagrams for Bhabha scattering are:

e+ p2 e+
k2

p1 k1
e e
time

e e
p1 k1
41

p2 k2
e+ e+
 p1 k1
e e
time

e e
p1 k1

p2 k2
e+ e+
time
Note that the photon propagator connects different pairs of particles together in these
two diagrams; therefore, different momentum flows through it. There are only two diagrams
because a third possible diagram in which, for example, the initial electron was connected
to the final state positron through the photon does not conserve electric charge.

4.5 Non-Relativistic Limit of Feynman Diagrams

4.5 (a)
By momentum conservation at each vertex, the momentum p that flows into the upper vertex
is equal to the sum of the momentum p0 + q that flows out of it. Therefore,
q = p − p0 . (278)

4.5 (b)
Let’s first evaluate q 2 from the expression in the previous part. We find
q 2 = (p − p0 )2 = 2m2e − 2p · p0 . (279)
Note that the four-momenta p and p0 square to the electron mass. If q 2 = 0, this then means
that
p · p0 = m2e . (280)
To understand this physical configuration, we can express this dot product as
p p
p · p0 = EE 0 − |~p||~p0 | cos θ = |~p|2 + m2e |~p0 |2 + m2e − |~p||~p0 | cos θ . (281)
Here, we have written the energy of the initial electron as E and its three-momentum as p~,
and similar for the final electron. θ is the angle between the initial and final electron. This
expression makes it clear that the only way that p · p0 = m2e is if θ = 0 and p~ = p~0 . In that
case, then the four-vectors are equal: p = p0 .
If the four-vectors p and p0 are equal, then the electromagnetic interaction between the
electron and muon has no affect on the trajectory of the the particles. Therefore, the force
between the electron and muon is 0. The only way that the electric force between to charged
particles can be 0 is if the particles are infinitely far apart. Thus, as q 2 → 0, this corresponds
to the physical configuration in which the electron and muon get very far apart.

42
4.5 (c)
The momentum electric potential Ṽ (q) is

−e2 −e2 −e2

Ṽ (q) = = = (282)
q2 (p − p0 )2 (E − E 0 )2 − (~p − p~0 )2
On the right-most equation, we used the expression for the four-vector dot product p · p0
from the previous problem. With the non-relativistic limit me  |~p|, |~p0 |, the energies of the
initial electron is p
E = |~p|2 + m2e ' me . (283)
In the non-relativistic limit, this is the same energy as the final-state electron. Therefore,
the denominator simplifies as

(E − E 0 )2 − (~p − p~0 )2 ' −(~p − p~0 )2 = −|~q|2 . (284)

Therefore the non-relativistic potential is

e2
Ṽ (~q) = . (285)
|~q|2

4.5 (d)
Now, we want to Fourier transform this potential to position space. The Fourier transform
is defined as
d3 q e2 −i~q·~r
Z
V (~r) = e . (286)
(2π)3 |~q|2
In spherical coordinates, this integral can be expressed as
|~q|2 d|~q| d cos θ dφ e2 −i|~q||~r| cos θ
Z
V (~r) = e , (287)
(2π)3 |~q|2
where θ is the polar angle between the ~q and ~r vectors, and φ is the azimuthal angle. The
integral over φ just yields a factor of 2π, while the integral over cos θ is non-trivial because
of the dependence in the exponential:
Z ∞ Z 1
e2
V (~r) = 2 d|~q| d cos θ e−i|~q||~r| cos θ (288)
4π 0 −1
Z ∞
ie2 1 d|~q| −i|~q||~r|
− ei|~q||~r| .


= 2 e (289)
4π |~r| 0 |~q|
Note that the exponential factors are anti-symmetric in |~q| ↔ −|~q|, and so this integral can
be extended to range over |~q| ∈ (−∞, ∞):
Z ∞
ie2 1 d|~q| −i|~q||~r|
V (~r) = 2 e . (290)
4π |~r| −∞ |~q|

43
4.5 (e)
Using the definition of the δ-function, we can re-express the Fourier-transformed potential
as
Z ∞ Z |~r| Z ∞
ie2 1 d|~q| −i|~q||~r| e2 1 0 0
V (~r) = 2 e = 2 dr d|~q| e−i|~q|r (291)
4π |~r| −∞ |~q| 4π |~r| 0 −∞
Z |~r|
e2 1 e2 1
= dr0 δ(r0 ) = . (292)
4π |~r| 0 4π |~r|

This is exactly the electric potential we expect from Maxwell’s equations.

4.6 Proton-Proton Total Cross Section

The total cross sections of protons as measured by the TOTEM experiment is about 100
millibarns at 7 and 8 TeV center-of-mass collision energy. In square meters, a millibarn is

1 mb = 10−3 b = 10−3 · 10−28 m2 = 10−31 m2 . (293)

At 100 mb, the proton cross section in square meters is therefore

100 mb = 10−29 m2 . (294)

Assuming the proton is a sphere, its cross section σ is related to its radius rp through

σ = πrp2 . (295)

Therefore, the radius of the proton as measured at TOTEM is

r r
σ 10−29
rp = ' m ' 1.8 × 10−15 m . (296)
π π

4.7 Proton Collision Beam at ATLAS

4.7 (a)
The integrated luminosity at the end of data taking, as shown in Fig. 4.3, is 36 fb−1 . In
inverse square centimeters, this is

36 fb−1 = 36 · 1024 · 1015 cm−2 = 3.6 × 1040 cm−2 . (297)

The total time over which data was collected was from about May 16 through October 31.
Conveniently, each tick on the abscissa of Fig. 4.3 is one week; therefore there were about
24 weeks or 168 days of data collection. This is a total of

168 d = 168 · 3600 · 24 s ' 1.5 × 107 s . (298)

44
The average luminosity hLi is therefore the ratio of the integrated luminosity to the total
data collection time:
36
hLi ' fb−1 s−1 ' 2.4 × 10−6 fb−1 s−1 (299)
1.5 × 107

3.6 × 1040
' cm−2 s−1 ' 2.4 × 1033 cm−2 s−1 .
1.5 × 10 7

4.7 (b)
To determine the total number of collision events that occurred, we multiply the integrated
luminosity by the total proton cross section. From the TOTEM experiment result in Fig. 4.2,
we see that the total cross section is approximately

σ ' 100 mb = 1014 fb . (300)

Therefore, the total number of proton collisions during data taking in 2016 is

events = 3.6 × 1015 . (301)

As the time of total data collection was about 1.5 × 107 s, this corresponds to
3.6 × 1015 −1
events/sec ' s ' 2.4 × 108 s−1 . (302)
1.5 × 107

4.7 (c)
(i) The total cross section for pp → Z events observed in 13 TeV collisions is approximately
5 × 104 pb. Therefore, the total number of events recorded is

events(pp → Z) ' 5 × 104 · 36 × 103 = 1.8 × 109 . (303)

Note that the integrated luminosity of 36 fb−1 corresponds to 36 × 103 pb−1 .

(ii) The total cross section for pp → tt̄ events observed in 13 TeV collisions is approximately
8 × 102 pb. Therefore, the total number of events recorded is

events(pp → tt̄) ' 8 × 102 · 36 × 103 ' 2.9 × 107 . (304)

(iii) The total cross section for pp → H events observed in 13 TeV collisions is approximately
60 pb. Therefore, the total number of events recorded is

events(pp → H) ' 60 · 36 × 103 ' 2.2 × 106 . (305)

(iv) The total cross section for pp → tt̄Z events observed in 13 TeV collisions is approxi-
mately 1 pb. Therefore, the total number of events recorded is

events(pp → tt̄Z) ' 1 · 36 × 103 ' 3.6 × 104 . (306)

45
4.8 Upper Limits on LUX Bounds
4.8 (a)
To estimate the cross section of a WIMP with a xenon nucleus, we need an expression for
the total number of observed events. This is found from integrating the luminosity times
the cross section over the total exposure time of LUX:
Z
events = dt Lσ ' T Lσ , (307)

where T is the exposure time. This is 332 days or

332 days = 332 · 3600 · 24 s ' 2.9 × 107 s . (308)

The luminosity is
L = nχ nXe Vol|vχ − vXe | . (309)
nχ and nXe are the number densities of WIMPs and liquid 131 Xe. We are told that the mass
density of WIMPs must be 0.3 GeV/cm3 . If the WIMP has a mass of mχ GeV, then its
number density is
0.3 3 × 105 −3
nχ = cm−3 = m . (310)
mχ mχ
The mass density of liquid xenon is about 3000 kg/m3 . 1 mol of 131
Xe weighs 131 g, so the
number density of liquid xenon is
3 × 106 −3
nXe ' 6.02 × 1023 · m ' 1.4 × 1028 m−3 . (311)
131
In the expression for the luminosity, Vol is the volume of the LUX experiment as this is
the only volume in which WIMPs can interact with xenon nuclei and we observe it. At a
total mass of 250 kg, the LUX detector has a volume of
250 3
Vol = m ' 8.3 × 10−2 m3 . (312)
3000
Finally, the relative speed of the WIMPs with respect to the LUX experiment is

|vχ − vXe | = 240 km/s = 2.4 × 105 m/s . (313)

As a function of the WIMP mass mχ , the luminosity is therefore

3 × 105 8.7 × 1037 −2 −1
1.4 × 1028 8.3 × 10−2 2.4 × 105 m−2 s−1 '




L= m s . (314)
mχ mχ
Now, assuming that one event was observed, we can solve for the WIMP-xenon cross
section σ as
1 1 mχ
σχ−Xe ' ' m2 ' mχ · 4 × 10−46 m2 . (315)
TL 2.9 × 10 8.7 × 1037
7

46
1 barn is a cross section of 10−28 m2 , and therefore if one event had been observed, the
WIMP-xenon cross section would be

σχ−Xe ' mχ · 4 × 10−18 b = mχ · 4 × 103 zb . (316)

Again, mχ is measured in units of GeV.

4.8 (b)
From the WIMP-xenon cross section, the cross section per nucleon is a factor of 131 (the
number of protons and neutrons in 131 Xe) times smaller. That is,
σχ−Xe
σχ−N = ' mχ · 30 zb . (317)
131
Let’s compare this to the established LUX bounds shown in Fig. 4.5. Our calculation predicts
that the the cross section bound is linearly proportional to the WIMP mass. Indeed, above
masses of about mχ = 100 GeV, the cross section limit is observed to be linearly proportional
to mχ . However, in our assumptions, we find that the absolute scale of the limit is off by
a few orders of magnitude. At a WIMP mass of about mχ = 1 TeV, LUX finds a limit
on the cross section of about 1 zb, while we would predict that if one event were observed
the cross section would be about 3 × 103 zb. These statements are different however. LUX
observed no events in 332 days, while we assume that one event was observed, so they aren’t
necessarily badly in conflict. From the lack of observation of a signal, LUX established a
90% confidence limit on the cross section for WIMPs to scatter off of xenon. That is, with
90% confidence, LUX states that the cross section of WIMP and xenon is below the solid
black curve.

4.8 (c)
If only one WIMP particle passed through the LUX detector in 332 days, then the volume
of the WIMP halo that passed through the LUX detector must have contained at most one
WIMP particle. Given a total volume Vol of the WIMP halo that passed through the LUX
detector, the total number of WIMP particles that could passed through the LUX detector
is
3 × 105
Nχ = nχ Vol = Vol . (318)
mχ
The volume of the LUX detector is approximately 8.3 × 10−2 m3 , and so its cross sectional
area A is approximately

2/3 2
A ' 8.3 × 10−2 m ' 0.19 m2 . (319)

The total length of the WIMP halo that passed through the cross sectional area of the LUX
experiment ` is the relative speed times the total exposure time or

` = T |vχ − vXe | ' (2.9 × 107 )(2.4 × 105 ) m ' 7 × 1012 m . (320)

47
It then follows that the total number of WIMP particles that passed through the LUX
experiment is
3 × 105 4 × 1017
Nχ = nχ Vol = · 0.19 · 7 × 1012 ' . (321)
mχ mχ
If there was a single particle that passed through LUX during that time, it must have had
a mass of
mχ ' 4 × 1017 GeV , (322)
which is pretty heavy!

5 Particle Collider Experiment

5.1 Synchrotron Losses
5.1 (a)
In Sec. 5.1, we had calculated the power lost from synchrotron radiation per proton at the
LHC to be about 40 MeV/s. The total number of protons Np in the LHC ring is the number
of bunches times the number of protons per bunch:

Np = 2808 · 1.15 × 1011 ' 3.2 × 1014 . (323)

Therefore, the total power radiated away in synchrotron radiation is

Ptot ' 40 · 3.2 × 1014 MeV/s ' 1.3 × 1013 GeV/s . (324)

Converting this to J/s, we recall that 1 GeV is 1.6 × 10−10 J. Therefore, the power lost in
watts is
Ptot ' 1.6 × 10−10 · 1.3 × 1013 W ' 2 × 103 W . (325)
That is, the power in synchrotron radiation at the LHC is equivalent to about two microwave
ovens.

5.1 (b)
Recall the formula for power in synchrotron radiation:

e2
P = E 4v4 . (326)
6π0 m4 c11 R2
The only things that differ between electrons at LEP and protons at the LHC are the particle
masses and their energies. We can safely assume in both cases that the particle velocities
are essentially the speed of light. Then, the ratio of the synchrotron power at LEP to the
LHC is  4
PLEP mp ELEP
= , (327)
PLHC me ELHC

48
where mp and me are the masses of the proton and electron, respectively, while ELEP and
ELHC are the center-of-mass collision energies at LEP and the LHC. The ratio of the proton
mass to the electron mass is approximately
mp 938
' ' 1.8 × 103 . (328)
me 0.511
The ratio of the energies of LEP and LHC are
ELEP 0.206
' ' 1.6 × 10−2 . (329)
ELHC 13
Then, the mass–energy ratio of LEP to the LHC is
mp ELEP
' 1.8 × 103 · 1.6 × 10−2 ' 29 . (330)
me ELHC

Therefore, the synchrotron radiation power at LEP is about 294 ' 7 × 105 times larger than
at the LHC. So, if the LHC was approximately 40 MeV/s, at LEP the power in synchrotron
radiation was
PLEP ' 40 · 7 × 105 MeV/s ' 2.8 × 104 GeV/s . (331)

5.1 (c)
For the same power in synchrotron radiation, the proton energy at the LHC can be a factor
of the proton mass to the electron mass, mp /me ' 1800 times larger than the electron energy
at LEP. While the LHC is a much higher energy machine than LEP, it is not almost 2000
times more energetic. There are other factors, beyond synchrotron radiation, that limit the
possible energy of protons at the LHC given a fixed radius tunnel. In particular, the strength
of the bending magnets is a main limiting factor.

5.2 Limits of the Tracking System

5.2 (a)
A charged particle that does not reach the calorimetry has a trajectory that curls up in the
magnetic field with a diameter that is no larger than the distance from the collision point
to the edge of the tracking system. Therefore, the radius of the trajectory of a particle that
just reaches the calorimeter is just over 1.1/4 = 0.275 m. The corresponding p⊥ of such a
particle is therefore
~
R|B| 0.275 · 2
p⊥ ' ' GeV ' 0.18 GeV . (332)
3 3

49
5.2 (b)
On the other extreme, for a particle with extremely high p⊥ to just be observed to bend in
the magnetic field, by the time it reaches the edge of the semiconductor tracker, its trajectory
must have been displaced from straight by the size of one pixel. With this assumption, we
want to calculate the radius of curvature of the particle, R, from which we can determine the
corresponding p⊥ . We can construct a right triangle with one side the radius of curvature
R, one side the radius of the semiconductor tracker rt , and the hypotenuse of length R plus
the size of one pixel, ∆x. The Pythagorean theorem of this triangle is

R2 + rt2 = (R + ∆x)2 . (333)

∆x is much smaller than any other length, so we can expand to linear order in ∆x and solve
for R:
r2 0.52
R= t ' m ' 7.3 × 103 m . (334)
2∆x 2 · 1.7 × 10−5
The corresponding maximum p⊥ that bends in the magnetic field is
~
R|B| 7.3 × 103 · 2
p⊥ ' ' GeV ' 4.9 × 103 GeV . (335)
3 3

5.2 (c)
The uncertainty on the determination of momentum of such a high-p⊥ particle is large. From
Eq. 5.19, the uncertainty ∆p⊥ is
 
2 1
∆p⊥ ' p⊥ ∆ . (336)
R

From earlier, we had identified the curvature of the highest p⊥ track to be

1 2∆x
= 2 . (337)
R rt

This is also a representation of the uncertainty on such a measurement of the curvature. The
track could have hit anywhere in the pixel of size ∆x, and we would still estimate the same
value for the curvature. That is, the uncertainty on the p⊥ is
2∆x
∆p⊥ ' p2⊥ ' 3.2 × 103 GeV . (338)
rt2

With such a large uncertainty, the measurement of the p⊥ of such a high-p⊥ track is essentially
unknown.

50
5.3 Reconstructing Muons
Considering first the solenoidal magnetic field in the tracking system, the magnitude of
transverse momentum p⊥ of a muon in GeV is
Rsol Bsol
p⊥ ' GeV . (339)
3
We note that the radius Rsol is measured in meters and the magnetic field Bsol is measured in
Tesla. For the toroidal magnetic field, the components of momentum along the proton beam
~ = Btor φ̂).
axis (pz ẑ) and perpendicular to it (p⊥ ŝ) are perpendicular to the magnetic field (B
Therefore, the total magnitude of muon momentum is
Rtor Btor
q
p2⊥ + p2z ' GeV . (340)
3
Combining these results, it then follows that the magnitude of these different components
of momenta are:
Rsol Bsol
p⊥ ' GeV , (341)
q3
1 2 2 2 2
pz ' Rtor Btor − Rsol Bsol GeV . (342)
3

5.4 Data Quantity from the LHC

If ATLAS and CMS are only colliding protons for a total of a few months per year, this is
a fraction of 1/3 or so of the year. So, we’ll estimate it as data collection over 107 seconds
in a year. Further, let’s make the conservative estimate of 100 events recorded per second;
therefore, about a billion (109 ) events are collected at each of ATLAS and CMS in a year.
If the data of each of these events is a megabyte or so, then the total data collected is about
a petabyte, 1015 bytes.
This turns out to be a rather conservative estimate of the data collected at the LHC.
On average, the number of events actually recorded per second is closer to 1000, the data of
an individual event is a few megabytes, and data collection has even approached nearly six
complete months in the year. With these adjustments, the LHC has accumulated more that
several hundred petabytes of data since turning on in 2009.

5.5 Properties of Poisson Statistics

5.5 (a)
The mean value of the Poisson distribution hni is
∞ ∞ ∞
X λn e−λ X λn−1 e−λ X λm e−λ
hni = n=λ =λ = λ. (343)
n=0
n! n=1
(n − 1)! m=0
m!

51
5.5 (b)
For the variance, σ 2 = hn2 i − hni2 , it’s useful to just focus on the calculation of the second
moment, hn2 i, because we already know the mean hni = λ. We have
∞
X λn e−λ
hn2 i = n2 . (344)
n=0
n!
Let’s use the hint, and take the first derivative of the normalization of the Poisson distribution
wrt λ:
∞ ∞ ∞ ∞
d d X λn e−λ X λn−1 e−λ X λn e−λ X λn−1 e−λ
1=0= = n− = n − 1. (345)
dλ dλ n=0 n! n=0
n! n=0
n! n=0
n!
We could evaluate the explicit sum that remains on the right, but we don’t want to do that
yet in order to calculate hn2 i. Now, take another derivative wrt λ:
∞
! ∞ ∞
d d X λn−1 e−λ X λn−2 e−λ X λn−1 e−λ
0=0= n−1 = n(n − 1) − n. (346)
dλ dλ n=0 n! n=0
n! n=0
n!
Therefore, we have
∞ ∞
X λn−2 e−λ X λn−1 e−λ
n(n − 1) = n. (347)
n=0
n! n=0
n!
From the result in part (a), the right side of this equation is
∞
X λn−1 e−λ
n = 1. (348)
n=0
n!
Massaging the left side, we find
∞ ∞ ∞
X λn−2 e−λ X λn−2 e−λ 2
X λn−2 e−λ
n(n − 1) = n − n (349)
n=0
n! n=0
n! n=0
n!
∞ ∞
X λn e−λ X λn e−λ
= λ−2 n2 − λ−2 n = λ−2 hn2 i − hni


(350)
n=0
n! n=0
n!
hn2 i 1
= 2 − . (351)
λ λ
Therefore, we find that
hn2 i 1
− = 1, (352)
λ2 λ
or that
hn2 i = λ2 + λ . (353)
Finally, the variance is
σ 2 = hn2 i − hni2 = λ2 + λ − λ2 = λ , (354)
as promised.

52
5.6 Look-Elsewhere Effect
5.6 (a)
We’re asked to determine the probability that at least one of the Nbins has an excess at least
as large at Xσ. The probability for any one bin to have such an excess is pX . One approach
to this problem is to explicitly sum together the probability for exactly one bin with the
excess, exactly two such bins, exactly three, etc., but this is tedious. A much more direct
route is to note that the sum of the probabilities of any number of bins, including 0, with the
excess is unity. That is, the excess must exist in zero, or one, or two, etc., bins. Therefore,
the probability that at least one bin has the excess is 1 minus the probability that no bin
does:
P (at least 1 bin with excess) = 1 − P (0 bins with excess) . (355)
The probability that a bin does not have the excess is thus 1 − pX , and so the probability
that Nbins do not is
P (0 bins with excess) = (1 − pX )Nbins . (356)
Therefore, the probability for at least one bin to have the excess is

P (at least 1 bin with excess) = 1 − (1 − pX )Nbins . (357)

5.6 (b)
To determine the probability that there is an excess anywhere in the data for pX  1, we
just Taylor expand our result from the previous part. Note that

(1 − pX )Nbins = 1 − Nbins pX + O(p2X ) , (358)

and so the probability is approximately

pglobal
X = Nbins pX . (359)

5.6 (c)
A 3σ deviation of one bin of your data has a p-value of p3 ' 0.0013. If this excess could have
occurred anywhere in your data of Nbins = 100 bins, the p-value including the look-elsewhere
effect is significantly increased:

pglobal
3 ' 100 · 0.0013 ' 0.13 . (360)

This is correspondingly approximately the p-value for a 1σ deviation of the data. This
suggests that such a deviation is likely not interesting, and just a statistical fluctuation.

53
5.6 (d)
(i) In the three bins that span 720–780 GeV, there are approximately 19 excess events
above the estimated fitted background. This can be determined from the lower panel
in Fig. 5.10. From the main plot of Fig. 5.10, about 25 events were expected in those
three bins. Therefore, the standard deviation of the number of background events is
approximately the square-root of this, or about 5 events. With 19 excess events above
background, the significance is about 3.75σ, which is very intriguing. This explains the
excitement of the community.

(ii) On the other hand, we didn’t need to be inspired by CMS, and should instead consider
the look-elsewhere effect. To do this, we need to count the number of three consecutive
bins the data. There’s one slight subtlety with this procedure. Note that the expected
background passes below one event per bin at about 1200 GeV, which is also where
the data bins become much more sporadically filled. So, to estimate the look-elsewhere
effect, we’ll ignore those bins above 1200 GeV. Then, up to that invariant mass, there
are about 48 bins. The number of consecutive triples of bins is then 46.
An excess of 3.75σ has a p-value of
 
3.75
1 erf √
2
p3.75 = − ' 8.8 × 10−5 . (361)
2 2
Including the look-elsewhere effect, the global p-value is

pglobal
3.75 ' 46 · 8.8 × 10
−5
' 4.1 × 10−3 . (362)

This p-value corresponds to a deviation of only about 2.6σ, which is significantly less
interesting.

(iii) Assuming that this excess is just a statistical fluctuation, to reduce it to 1σ, we need
to record enough events in those three bins for the 19 excess events to be the standard
deviation of the number of events in the bin. Assuming Poisson statistics, we would
need a total number of events of

Ntot = 192 = 361 . (363)

From the 25 expected events plus the 19 excess events, the plot of Fig. 5.10 has about
44 total events in data in the bins of 720–780 GeV. Therefore, we would need to collect
about
361 − 44 = 317 (364)
more events or increase the dataset by about a factor of 7.

54
5.7 Discovery of the Top Quark
5.7 (a)
From Fig. 5.11, the total number of events observed in this search for the top quark is just
the sum of all events within the solid black histogram. This is 19 events. If 6.9 events were
expected from the null hypothesis, then the standard deviation of the null hypothesis would
be about √
6.9 ' 2.6 (365)
events. Therefore, the data differs from the null hypothesis by about
19 − 6.9
' 5.0σ! (366)
2.6

5.7 (b)
The p-value of a 5σ deviation is p5 ' 2.7 × 10−7 . This is extremely unlikely, indeed justifying
the claim of discovery.

5.8 Missing Energy and Neutrinos

5.8 (a)
Assuming that the W boson has no momentum transverse to the beam axis, the charged
lepton and neutrino from its decay must have equal and opposite transverse momentum. The
magnitude of transverse momentum is maximized if the decay products are emitted trans-
verse to the beam in the W boson rest frame. In that configuration, the three-momentum of
the charged lepton is just its transverse momentum which is just half of the W mass mW /2,
because the charged lepton is massless.

5.8 (b)
For massless particles, like the charged lepton and neutrino we consider here, their transverse
mass mT is just equal to their transverse momentum, pT . Therefore, we can express the W
boson transverse mass as
W
√ q √ q q p
mT = 2 p⊥ p⊥ − p~⊥l · p~⊥ν = 2 p⊥ p⊥ − p⊥ p⊥ cos θlν = 2pl⊥ pν⊥ 1 − cos θlν . (367)
l ν l ν l ν

In this expression, we have introduced θlν as the angle between the transverse momen-
tum vectors of the charged lepton and neutrino. However, because the charged lepton and
neutrino transverse momenta are equal and opposite, this angle is just θlν = π. Further,
the maximum magnitude of transverse momenta of the charge lepton or neutrino is mW /2.
Therefore, the maximum value of the W boson transverse mass is

mW mW √
r

mW 
T max = 2 1 − cos π = mW . (368)
2 2

55
5.8 (c)
Admittedly, it is a bit tricky to estimate the maximum value of the electron’s transverse
momentum from the plot of Fig. 5.12. Nevertheless, there is a clear peak at about 38 GeV,
and above that, the data quickly drops off. So, if the “maximum” value of the electron
transverse momentum is pl⊥ . 40 GeV, the mass of the W boson is about twice this value,
or mW ' 80 GeV. This extremely naı̈ve estimate is actually very close to the actual value.

5.9 Event Displays

5.9 (a)
First, note that the particles denoted by the light mark have a track, and so must be
electrically charged. Further, they deposit their energy in the electromagnetic calorimeter.
This suggests that these particles are electrons.
The dark particles, by contrast, still have a track, but do not deposit energy in the
calorimetry. Instead, they are detected by the muon system at the outer edge of ATLAS, so
these particles are muons.

5.9 (b)
From the Lego plot, we’ll find the four particles’ four-momentum vectors. To do this, we
identify the p⊥ , η, and φ of each particle and then convert that into the four-momentum
with the expression
p = (p⊥ cosh η, p⊥ cos φ, p⊥ sin φ, p⊥ sinh η) . (369)
Working from left to right in the Lego plot, the p⊥ , η, and φ of the four particles are:

particle 1: p⊥ ' 29 GeV , η ' −1 , φ ' 68◦ , (370)

particle 2: p⊥ ' 23 GeV , η ' −0.3 , φ ' 80◦ , (371)
particle 3: p⊥ ' 19 GeV , η ' 0.5 , φ ' 203◦ , (372)
particle 4: p⊥ ' 31 GeV , η ' 0 , φ ' 255◦ . (373)

Plugging these values into the expression for the four-vector, we find

p1 ' (45, 11, 27, −34) GeV , (374)

p2 ' (24, 4, 23, −7) GeV , (375)
p3 ' (21, −17, −7.4, 9.9) GeV , (376)
p4 ' (31, −8, −30, 0) GeV . (377)

(i) The total transverse momentum of these particles is just the vector sum of their x̂ and
ŷ components. The x̂ momentum is then

px ' 11 + 4 − 17 − 8 GeV = −10 GeV . (378)

56
 The ŷ component is, correspondingly,

py ' 27 + 23 − 7.4 − 30 GeV ' 12 GeV . (379)

Therefore, the transverse momentum vector is

p~⊥ ' (−10, 12) GeV . (380)

(ii) Summing the four-momenta of these four particles, we find the total momentum vector:

p1 + p2 + p3 + p4 ' (121, −10, 12, −31) GeV . (381)

Dotting this vector with itself, we find

(p1 + p2 + p3 + p4 )2 ' 1212 − 102 − 122 − 312 ' 1.34 × 104 . (382)

Now, taking the square-root and interpreting it as a mass m, we find

p
m = p2 ' 116 GeV . (383)

This is quite close to the mass of the Higgs boson for which mH ' 125 GeV. Indeed, this
event display corresponds to a candidate event in which the Higgs boson was produced
and decayed to electrons and muons.
(iii) Assuming that the velocity transverse to the beam is 0, the ẑ component of momentum
of this particle can be expressed as
mvz
pz = γmvz = p , (384)
1 − vz2
where vz is the velocity as a fraction of the speed of light of the particle along the ẑ
axis. Solving for vz , we find s
p2z
|vz | = . (385)
p2z + m2
Plugging in the values for pz and m, we find
r
312
|vz | ' ' 0.26 , (386)
312 + 1162
or about a quarter of the speed of light.

6 Quantum Electrodynamics in e+e− Collisions

6.1 Lorentz Transformation of Spinors
We want to explicitly construct the 2 × 2 matrices that implement particular Lorentz trans-
formations on spinors.

57
6.1 (a)
First, we want to perform an azimuthal rotation by an angle χ. With this rotation, the
right-handed spinor transforms to
√ √
 −i(φ−χ)/2  iχ/2 −iφ/2
cos 2θ cos 2θ
 
e e e
uR (p) → 2E = 2E . (387)
ei(φ−χ)/2 sin 2θ e−iχ/2 eiφ/2 sin 2θ

The matrix M that implements this transformation as uR → MuR is therefore

 iχ/2 
e 0
M= . (388)
0 e−iχ/2

Note that this matrix is indeed unitary M† M = I and has determinant 1.

6.1 (b)
Now, let’s consider a polar rotation by an angle ω. The right-handed spinor then becomes
√ √
 −iφ/2  −iφ/2
cos θ−ω θ ω θ ω


e e cos cos + sin sin
uR (p) → 2E 2 = 2E 2 2 2 2
. (389)
eiφ/2 sin θ−ω
2
eiφ/2 sin 2θ cos ω2 − cos 2θ sin ω2

If we denote the elements of the matrix M that implements this rotation as

 
a b
M= , (390)
c d

then its action on the right-handed spinor is

√
 −iφ/2
cos 2θ + beiφ/2 sin 2θ

ae
MuR (p) = 2E . (391)
ce−iφ/2 cos 2θ + deiφ/2 sin 2θ

We can then match terms from the transformed spinor and the action of the matrix to
determine the entries of M:
cos ω2 e−iφ sin ω2
 
M= . (392)
−eiφ sin ω2 cos ω2

This matrix is unitary and has determinant 1.

6.1 (c)
Finally, we want to determine the matrix that implements a Lorentz boost along the ẑ axis
by velocity β. Under such a transformation, the energy becomes

E → γ(E + βpz ) , (393)

58
while the ẑ component of momentum transforms as
pz → γ(pz + βE) . (394)
As always, the boost factor γ is related to the velocity of the boost by
1
γ=p . (395)
1 − β2
In the spherical coordinates used in this problem, these can be expressed as
E → γE(1 + β cos θ) , (396)
pz → γE(cos θ + β) . (397)
The ratio of pz to E is the cosine of the polar angle. The new polar angle, θ0 , is therefore
cos θ + β
cos θ0 = . (398)
1 + β cos θ
Note that under this boost, the azimuthal angle φ is unchanged.
In the expression for the spinors, we actually need the cosine or sine of half of the polar
angle. This can be found from half-angle trigonometric identities:
r s
θ0 1 + cos θ0 (1 + β)(1 + cos θ)
cos = = , (399)
2 2 2(1 + β cos θ)
r s
0 0
θ 1 − cos θ (1 − β)(1 − cos θ)
sin = = . (400)
2 2 2(1 + β cos θ)

We have written these formulae appropriate for the first quadrant, but the final result will
hold for any angles. Further, because the energy has changed, the normalization of the
spinors is affected. The new energy, E 0 , multiplies these angles correspondingly:
s
√ θ 0 p (1 + β)(1 + cos θ) p √ θ
2E 0 cos = 2γE(1 + β cos θ) = γ(1 + β) 2E cos , (401)
2 2(1 + β cos θ) 2
s
√ θ 0 p (1 − β)(1 − cos θ) p √ θ
2E 0 sin = 2γE(1 + β cos θ) = γ(1 − β) 2E sin . (402)
2 2(1 + β cos θ) 2

So, when the dust settles, this Lorentz boost is just a simple rescaling of the components
of the spinor. It therefore follows that the matrix M that implements this boost on the
right-handed spinor uR (p) is
 p 
γ(1 + β) p 0
M= . (403)
0 γ(1 − β)
For the left-handed spinor, the diagonal entries are switched. Unlike the rotation matrices,
this matrix is not unitary, but has determinant 1.

59
6.2 Helicity Spinors
6.2 (a)
From the expressions for the spinors, the outer product of the right-handed spinor and its
conjugate is
√ cos 2θ √
 −iφ/2   
† e iφ/2 θ −iφ/2 θ
uR uR = 2E 2E e cos e sin (404)
eiφ/2 sin 2θ 2 2
2 cos2 2θ 2e−iφ cos 2θ sin 2θ
 
=E
2eiφ cos 2θ sin 2θ 2 sin2 2θ
1 + cos θ e−iφ sin θ
 
=E .
eiφ sin θ 1 − cos θ

Recall that the matrix p · σ̄ is

p · σ̄ = EI + E sin θ cos φσ1 + E sin θ sin φσ2 + E cos θσ3 (405)

1 + cos θ e−iφ sin θ
 
=E . (406)
eiφ sin θ 1 − cos θ

This is identical to the spinor outer product uR u†R .

We can similarly compute uL u†L and find

1 − cos θ −e−iφ sin θ

 
†
uL uL = E = p·σ. (407)
−eiφ sin θ 1 + cos θ

6.2 (b)
The trace of either of the outer product matrices is identical, and just the sum of the diagonal
elements. We find
tr p · σ̄ = tr p · σ = 2E . (408)
We can express the trace in a more illuminating way. The trace is

tr p · σ̄ = (p · σ̄)11 + (p · σ̄)22 = δij (p · σ̄)ij (409)

= δij uRi u†Rj = u†R uR . (410)

In this manipulation, we used the Kronecker δ and Einstein notation to sum over the diagonal
entries of the matrix, but then we identified this with the inner product of the spinors.

6.2 (c)
In spherical coordinates, the helicity operator ĥ can be represented as a matrix:

cos θ e−iφ sin θ

 
1 1
ĥ = p̂ · ~σ = . (411)
2 2 eiφ sin θ − cos θ

60
Acting on the right-handed spinor, we find

cos θ e−iφ sin θ √

 −iφ/2
cos 2θ
  
1 e
ĥuR = 2E (412)
2 eiφ sin θ − cos θ eiφ/2 sin 2θ
1√
 −iφ/2
cos θ cos 2θ + sin θ sin 2θ



e
= 2E
2 eiφ/2 sin θ cos 2θ − cos θ sin 2θ
1√
 −iφ/2
cos 2θ

e 1
= 2E iφ/2 θ = uR .
2 e sin 2 2

To get the last line, we used angle sum and difference formulae. Doing the same exercise for
the left-handed spinor, we find
1
ĥuL = − uL . (413)
2

6.2 (d)
To evaluate the spinor product for this problem, we can choose a nice frame. Let’s align the
electron and positron momenta along the ẑ axis with equal energy and opposite direction so
that
√ √
 
1
uR (p1 ) = 2E , vL† (p2 ) = 2E (0 − i) . (414)
0
The muon and anti-muon also have equal and opposite momentum, but at a relative angle
of θ from the ẑ axis. We can therefore write their spinors as
√ √ −i sin 2θ
   
† θ θ
uR (k1 ) = 2E cos sin , vL (k2 ) = 2E . (415)
2 2 i cos 2θ

In writing these spinors, we have set the momentum of the muons in the x̂ẑ plane, which is
why there is the factor of i.
Doing the spinor products, we have
√ √ −i sin 2θ
 
† θ
vL (p2 )vL (k2 ) = 2E (0 − i) 2E θ = 2E cos , (416)
i cos 2 2
√ θ √
   
† θ 1 θ
uR (k1 )uR (p1 ) = 2E cos sin 2E = 2E cos .
2 2 0 2

Note also that Ecm = 2E, so multiplying all the spinors together, we find

2 1 + cos θ
  
vL† (p2 )vL (k2 ) †
uR (k1 )uR (p1 ) = Ecm . (417)
2
So, up to a factor of 2 and a minus sign, this spinor inner product is identical to the product
with Pauli spin matrices.

61
6.3 Spin Analysis of e+ e− → µ+ µ−
In this exercise, we consider the scattering of helicity eigenstates e− + − +
R eL → µR µL . This
scattering occurs in the center-of-mass frame in which the electron has momentum aligned
along the +ẑ axis and the positron has momentum aligned along the −ẑ axis. With these
momentum assignments, note that the net spin of the initial state is one unit in magnitude,
pointing along the +ẑ axis. If the scattering angle θ = π, then this means that the muon
momentum is aligned along the −ẑ axis, and the anti-muon’s momentum along the +ẑ axis.
However, in this configuration, the final state has one unit of spin aligned along the −ẑ axis.
Therefore, the angular momentum of the initial and final states in this configuration have no
overlap. Because angular momentum must be conserved, this means that this configuration,
when the scattering angle θ = π, has zero probability to occur.

6.4 Spin-0 Photon

If the photon was a spin-0 particle, then the spins of the initial state electron and positron
must be anti-aligned. Similarly, the muon and anti-muon spins must too be anti-aligned. So,
the total spin of both the electron-positron and muon–anti-muon pairs is zero. A zero total
spin projected onto any axis is zero; therefore if the photon spin were zero, the product of
electron and positron wavefunctions must be rotationally invariant. The product of muon
and anti-muon wavefunctions is also rotationally invariant. Therefore, there is no correlation
between the alignment of spins of initial and final states. This then means that in the
scattering matrix element for e+ e− → µ+ µ− with a spin-0 photon, there is no dependence
on the scattering angle. So, if we observe any non-trivial scattering angle dependence in the
differential cross section, then the photon must have non-zero spin.

6.5 e+ e− → scalars
6.5 (a)
The fact that there are spin-0 particles in the final state doesn’t affect the spin configurations
of the initial electron and positron. Their spins still have to sum to ±1 to be able to produce
the intermediate photon. In the center-of-mass frame, then, the helicities of the electron
and positron must be opposite: either a right-handed electron and left-handed positron, or
vice-versa.

6.5 (b)
To evaluate the matrix element for e+ e− → φφ∗ , we can recycle many results from the
e+ e− → µ+ µ− analysis in the chapter. In particular, the left side of the diagram that consists
of the product of electron and positron wavefunctions and the photon propagator, is identical
to the e+ e− → µ+ µ− case. The only difference is the coupling of the scalars to the photon,
but we are given that Feynman rule. Further, there are only two helicity configurations of

62
the electron and positron that produce a non-zero matrix element. Therefore, the matrix
elements to compute are

† e2
M(e− + ∗ µ
R eL → φφ ) = vL (p2 )σ uR (p1 ) (k1 − k2 )µ , (418)
(p1 + p2 )2
† e2
M(e− e
L R
+
→ φφ∗
) = v (p
R 2 )σ̄ µ
u (p
L 1 ) (k1 − k2 )µ . (419)
(p1 + p2 )2

We had already calculated the spinor products in the center-of-mass frame, and we can
express the four-momenta k1 and k2 in spherical coordinates:
Ecm
k1 = (1, sin θ cos φ, sin θ sin φ, cos θ) , (420)
2
Ecm
k2 = (1, − sin θ cos φ, − sin θ sin φ, − cos θ) , (421)
2
where θ is the scattering angle and the total three-momentum of the final state is 0. Putting
this all together, the first matrix element becomes

† e2
M(e− + ∗ µ
R eL → φφ ) = vL (p2 )σ uR (p1 ) (k1 − k2 )µ (422)
(p1 + p2 )2
2
µ e
= Ecm (0, −i, 1, 0) 2 Ecm (1, sin θ cos φ, sin θ sin φ, cos θ)µ (423)
Ecm
2
= e (i cos φ − sin φ) sin θ . (424)

We are free to orient our axes however we want, so we can set φ = 0 for which

M(e− + ∗ 2
R eL → φφ ) = ie sin θ , (425)

as advertised. Performing a nearly identical analysis, the other matrix element is

† e2
M(e− + ∗ µ
L eR → φφ ) = vR (p2 )σ̄ uL (p1 ) (k1 − k2 )µ (426)
(p1 + p2 )2
e2
= Ecm (0, −i, −1, 0)µ 2 Ecm (1, sin θ, 0, cos θ)µ (427)
Ecm
2
= ie sin θ , (428)

where we have set φ = 0 from the beginning. These matrix elements with different electron
and positron helicities are identical.

6.5 (c)
To calculate the differential cross section for this process, we need to square the matrix
elements, sum them, and average over initial electron and positron helicities. The averaging

63
factor is just a factor of 1/4 and because the matrix elements are identical this calculation
is particularly easy. We have
|M(e− + ∗ 2 − + ∗ 2 4 2
R eL → φφ )| = |M(eL eR → φφ )| = e sin θ , (429)
and so their sum and spin average is
1
e4
|M(e− + ∗ 2 − +
R eL → φφ )| + |M(eL eR → φφ )|
∗ 2
= sin2 θ . (430)
4 2
We can then plug this into the expression for two body phase space to determine the differ-
ential cross section:
 2 2
dσ π e 2 α2 π
= 2
sin θ = 2
(1 − cos2 θ) . (431)
d cos θ 4Ecm 4π 4Ecm
Note that this differential cross section vanishes if θ = 0, π. In that limit, the initial and
final state momentum are all collinear. In this configuration, the scalars have no orbital
angular momentum with respect to the initial collision axis, so angular momentum cannot
be conserved.

6.5 (d)
To determine the total cross section, we just integrate over the scattering angle cos θ:
Z 1
+ − ∗ α2 π 2 α2 π
σ(e e → φφ ) = 2
d cos θ (1 − cos θ) = 2
. (432)
4Ecm −1 3Ecm
This is a factor of 4 smaller than the total cross section for e+ e− → µ+ µ− . The reason
for this difference is due to the fact that scalars have fewer accessible states than spin-1/2
particles.

6.6 Decays of the Z boson

6.6 (a)
The Z boson is hundreds of times more massive than either electrons or up quarks, so,
effectively, we can imagine that they are massless. Neither the Z boson nor the electron
carries color, while the up quark can be one of three colors. Both the electron and up quark
are spin-1/2 particles, so the only thing that differs between them is the color of the up quark.
There are thus three times as many possible final states with up quarks than electrons, so
we would expect that the Z boson decays to up quarks three times as often as electrons.

6.6 (b)
The Z boson can decay to any quarks whose mass is smaller than the mass of itself. Those
quarks are the up, down, strange, charm, and bottom quarks. All three charged leptons, the
electron, muon, and tau, are less massive than the Z boson, so it can decay to any charged
lepton.

64
6.6 (c)
Each quark can carry one of three colors, so with 5 quarks, there are a total of 5 × 3 = 15
possible final states involving quarks. By contrast, leptons are color neutral, so there are
only three possible final states involving leptons. This counting predicts that, all else being
equal, Z bosons decay to quarks, and by extension hadrons, five times more often than to
charged leptons.

6.6 (d)
From the PDG, the Z boson decays to each charged lepton about 3.36% of the time, or a
total of about 10% to any charged lepton. The rate of Z boson decay to hadrons is about
70%, or about 7 times as often as to charged leptons. This isn’t too far off from our estimate
of a factor of 5 larger and the difference is described by the theory of the electroweak force.

6.7 Inclusive vs. Exclusive Cross Sections

6.7 (a)
The requirement that two jets i and j must satisfy is

2 min[Ei2 , Ej2 ](1 − cos θij ) > ycut Ecm

2
. (433)

To find the smallest energy scale Emin to which this corresponds to, we want to maximize
the angle between the jets, so θij = π. In that case, the minimum energy is
2 ycut 2
Emin = E . (434)
4 cm
Plugging in values of Ecm = 206 GeV and the minimum ycut = 10−5 , we find an energy of

Emin ' 323 MeV . (435)

This is a factor of a few less than the proton mass, but also a couple times larger than the
pion mass.

6.7 (b)
From Fig. 6.6, the dijet cross section is a poor description of the e+ e− → hadrons process
below about ycut . 10−2 . By this point, the three-jet process is becoming increasingly
important and the dijet cross section is only 60% of the total cross section.

6.7 (c)
With the relationship that the three-jet cross section is related to the dijet cross section of

σ(e+ e− → 3 jets) ' αs log2 ycut σ(e+ e− → 2 jets) , (436)

65
to solve for the value of the strong coupling, we consider the value of ycut at which the
two- and three-jet cross sections are equal. From Fig. 6.6, this occurs at approximately
ycut ' 10−2.5 , and so the strong coupling value is

αs ' log−2 10−2.5 ' 0.03 . (437)

This is only a factor of a few away from the accepted value of the strong coupling, αs .

6.7 (d)
As ycut → 0, the probability that there are only two jets in the final state vanishes, according
to Fig. 6.6. Another way to phrase this is that as ycut → 0, jets can be increasingly close
in angle to one another and have very low energy. For any finite collection of particles in
the final state, there will be a maximum value of ycut below which every particle will be
identified as an individual jet. So, at sufficiently low values of ycut , the number of resolved
jets is just the total number of particles produced.

6.8 Finite Decay Width Effects

6.8 (a)
The propagator P (q 2 ) for a massive, unstable particle like the Z boson takes the form
1
P (q 2 ) = , (438)
q2 − m2Z + imZ ΓZ
where q is the momentum flowing through the propagator, mZ is the mass of the Z boson,
and ΓZ is the decay width of the Z boson. The absolute square of this propagator is then
1 1 1
|P (q 2 )|2 = = 2 . (439)
q2 − m2Z 2
+ imZ ΓZ q − mZ − imZ ΓZ
2 (q − mZ ) + m2Z Γ2Z
2 2

We want to find the full width at half the maximum value of this squared propagator.
Because the denominator is a sum of non-negative terms, it is maximized for q 2 = m2Z . At
that point, it takes the value
1
|P (m2Z )|2 = 2 2 . (440)
mZ ΓZ
Therefore, the squared propagator is at half of this maximum value when

(q 2 − m2Z )2 = m2Z Γ2Z , (441)

or for
q 2 = m2Z ± mZ ΓZ . (442)
With q 2 = Ecm
2
, we can express this relationship at which the distribution as
ΓZ
q
Ecm = m2Z ± mZ ΓZ ' mZ ± . (443)
2
66
On the right, we have Taylor expanded the square-root, assuming that ΓZ  mZ . Therefore,
the full width at half maximum (FWHM) is the difference of

ΓZ ΓZ
FWHM ' (mZ + ) − (mZ − ) = ΓZ , (444)
2 2
justifying the term “width.”

6.8 (b)
In the center-of-mass frame, the photon propagator for the process e+ e− → hadrons is just
2
the inverse square of the center of mass energy, 1/Ecm . This propagator is squared in the
cross section, so with the propagator identified, we can also express the cross section with
an intermediate photon as

11 4πα2 2 1
σ(e+ e− → γ ∗ → hadrons) = Ecm · 4 . (445)
3 3 Ecm
4
We just need to replace the 1/Ecm with the absolute squared Z boson propagator discussed
earlier to find the cross section with an intermediate Z boson instead:

+ − ∗ 11 4πα2 4
Ecm
σ(e e → Z → hadrons) = 2 (E 2 − m2 )2 + m2 Γ2
. (446)
3 3Ecm cm Z Z Z

6.8 (c)
To find the total cross section with either an intermediate photon or a Z boson in the process
e+ e− → hadrons, to this approximation, we just sum up their individual cross sections. We
then find
11 4πα2 4
 
+ − Ecm
σ(e e → hadrons) = 1+ 2 . (447)
2
3 3Ecm (Ecm − m2Z )2 + m2Z Γ2Z

6.8 (d)
Fig. 6.7 enables us to determine the mass and the width √ of the Z boson. As labeled on the
plot, the Z boson corresponds to the peak at around s = 90 GeV. As argued earlier, the
maximum value of the propagator and by extension the cross section occurs when q 2 = s =
2
Ecm = m2Z . Therefore, we estimate the mass of the Z boson to be the location of the peak of
the cross section in Fig. 6.7; at about 90 GeV. The PDG says that the mass of the Z boson
is mZ = 91.1876 GeV, so we are quite close.
To estimate the width of the Z boson, we can use the information about the height of the
peak. The peak at 90 GeV has a height of about 5 × 10−5 mb = 50 nb. From the analysis
in Sec. 6.1.3, we can evaluate the pre-factor of the cross section at 90 GeV in nb. We have

4πα2 80
2
' 2 nb ' 10−3 nb . (448)
3(90 GeV) 90

67
Then, at s = m2Z , the expression for the cross section becomes

m2Z
 
+ −
 −3 11
σ(e e → hadrons) Ecm =m ' 10 nb ·
 1+ 2 . (449)
Z 3 ΓZ

Taking the ratio of this expression with the observed peak height from Fig. 6.7, we find

m2Z
1.4 × 104 ' 1 + . (450)
Γ2Z

Because this factor is so large, the “1” term on the right is essentially irrelevant, so we can
ignore it. It then follows that the Z boson width is
mZ
ΓZ ' ' 750 MeV . (451)
1.2 × 102
This is only a factor of a few off of the measured result from the PDG of ΓZ = 2.4952 GeV.
So, our very rough estimates aren’t too bad!

7 Quarks & Gluons

7.1 Plus-Function Expansion
7.1 (a)
Our first step in understanding the +-function expansion is to demonstrate that the naı̈ve
Taylor expansion is unjustified. A naı̈ve application of the Taylor expansion would give us
∞
  X ( log x)n
= . (452)
x1− x n=0 n!

Integrating the left side over x, we find

Z 1

dx = 1. (453)
0 x1−
By contrast, the integral of the terms on the right side are infinite, order-by-order in . For
instance, for n = 0, we have Z 1

dx → ∞ . (454)
0 x
This is clearly not helpful so the proper Taylor expansion isn’t the way to proceed.

68
7.1 (b)
If x > 0, then note that x−1+ is finite for any value of . Therefore, for x > 0, the limit as
 → 0 is

lim 1− = 0 . (455)
→0 x
From the previous part, we had calculated the integral of the function f (x), which was defined
to be 1. So, in the limit that  → 0, f (x) is a function that only has non-zero support at
x = 0 and integrates to unity. These properties define the δ-function with argument x:

lim 1− = δ(x) . (456)
→0 x

7.1 (c)
The observations that the Taylor expansion in  is not justified and the δ-function is the
 → 0 limit of the function f (x) motivates the +-function expansion:
∞
n+1 logn x
 
 X
= δ(x) + . (457)
x1− n=0
n! x +

If this is an equality, these two expressions must integrate to the same value, which is just 1.
On the right, the δ-function already integrates to 1, so the infinite sum must integrate to 0:
∞
n+1 logn x
Z 1 X  
0= dx . (458)
0 n=0
n! x +

Every term in this expression has a coefficient at a different order in , and so for this to
hold for arbitrary , each term must separately integrate to 0. That is
Z 1  n 
log x
0= dx , (459)
0 x +

for any n.

7.1 (d)
We now want to evaluate the integral
logn x
Z 1  
dx g(x) , (460)
0 x +

where g(x) is an analytic function. Because g(x) is analytic, it admits a Taylor expansion
and we can integrate order-by-order in the expansion. We then have
Z 1  n  Z 1  n  X ∞ ∞ Z 1  n 
log x log x i
X log x
dx g(x) = dx ci x = ci dx xi (461)
0 x + 0 x + i=0 i=0 0 x +
∞ Z 1  n 
X log x
= ci dx xi . (462)
i=1 0 x +

69
Note that the i = 0 term explicitly integrates to 0 from the previous part of this problem.
Also, note that the coefficient c0 = g(0), the value of the function at x = 0. Further, because
xi → 0 for x → 0 with i > 0, the +-function distribution is unnecessary; all issues with what
happens at x = 0 are irrelevant. Therefore, we can remove it:
∞  n  ∞ ∞
logn x i logn x X i
Z 1 Z 1 Z 1
X log x i
X
ci dx x = ci dx x = dx ci x (463)
i=1 0 x + i=1 0 x 0 x i=1
∞
! Z
logn x X i logn x
Z 1 1
= dx ci x − c0 = dx (g(x) − g(0)) ,
0 x i=0 0 x

as promised.

7.2 Breit Frame

7.2 (a)
The momentum of the intermediate photon in DIS in the Breit frame is

q = (0, ~q) , (464)

for some three-vector ~q. For initial and final electron momenta of p1 and p2 , respectively,
the momentum of the intermediate photon is

q = p1 − p2 . (465)

Because the Breit frame photon has 0 energy, the initial and final electrons must have the
same energy: E1 = E2 ≡ E. Further, conservation of three-momentum means that

p~1 − p~2 = ~q . (466)

Therefore, we can express the four-momenta of the initial and final state electrons as

p1 = (E, p~) , p2 = (E, p~ − ~q) . (467)

Additionally, both of these electrons are on-shell, and in the high-energy limit, are effectively
massless. This enforces that

E 2 − |~p|2 = 0 , E 2 − |~p − ~q|2 = 0 . (468)

Expanding out the relationship on the right and using the relationship on the left, we find

E 2 − |~p − ~q|2 = E 2 − |~p|2 − |~q|2 + 2~p · ~q = −|~q|2 + 2~p · ~q = 0 , (469)

or that
~q · (2~p − ~q) = 0 . (470)

70
The unique solution to this expression is that the momentum of the electron p~ must be
~q
p~ = . (471)
2
Therefore, the four-vectors of the initial and final-state electrons are
1 1
p1 = (|~q|, ~q) , p2 = (|~q|, −~q) . (472)
2 2
So, in this frame, it looks like the electron bounces off the intermediate photon with the
same kinetic energy and opposite momentum, just like bouncing a ball off of a brick wall.

7.2 (b)
In the frame in which the proton is at rest, the momentum of the intermediate photon is

q = pe − pe0 = (Ee − Ee0 , Ee0 sin θ, 0, Ee − Ee0 cos θ) . (473)

The Breit frame is the one in which the energy of the photon Eq is 0. The Lorentz boost
velocity β that accomplishes this can be found by Lorentz boosting the energy along the ẑ
axis and setting it to 0. The boosted energy is

Eq → γ(Eq − qz β) = 0 , (474)

or that
Eq Ee − Ee0
β= = . (475)
qz Ee − Ee0 cos θ
Note that this corresponds to a velocity less than the speed of light because cos θ < 1, in
general.
With this boost established, we can simply Lorentz transform the proton’s momentum
from at rest to the Breit frame. The proton energy becomes:
mp
Ep = γmp = r  2 , (476)
Ee −Ee0
1− Ee −Ee0 cos θ

while the ẑ component of momentum is then

mp Ee − Ee0 mp
pz = γβmp = r  2 Ee − Ee0 cos θ = r 2 . (477)
Ee −Ee0 Ee −E 0 cos θ
1− Ee −Ee0 cos θ
e
Ee −Ee0
−1

The four-momentum of the proton in the Breit frame is then

pp = (Ep , 0, 0, pz ) . (478)

71
7.3 Form Factor Evolution Equation
7.3 (a)
The form factor F2 (x, Q2 ) from earlier in the chapter assuming scale invariance is
  π 
−/2
F2 (x, Q2 ) = cos Γ()F2 (x) r02 Q2 . (479)
1− 2
Taking the derivative with respect to Q2 is straightforward, because only the final factor
depends on Q2 . Note that

∂ 2    π 
− 2 −/2−1

 F2 (x, Q2 )
F 2 (x, Q ) = − cos Γ()F 2 (x)r0 Q = − . (480)
∂Q2 21− 2 2 Q2
Rearranging, we then find the differential equation
∂ 
Q2 2
= − F2 (x, Q2 ) . (481)
∂Q 2

7.3 (b)
This differential equation can be solved in the usual separation of variables manner. Note
that
∂F2 (x, Q2 )  ∂Q2
= − . (482)
F2 (x, Q2 ) 2 Q2
Integrating both sides, we find

log F2 (x, Q2 ) = − log Q2 + c , (483)
2
where c is some integration constant. Q2 has units of squared energy, so taking its logarithm
isn’t desirable. Without loss of generality, the integration constant c can be itself chosen to
eliminate the dimensions of the argument of the logarithm:

c= log Q20 + c0 , (484)
2
where Q20 is a reference energy scale and c0 is another integration constant. It then follows
that the solution of the differential equation can be written as

 Q2
log F2 (x, Q2 ) = − log 2 + c0 . (485)
2 Q0

If Q2 = Q20 , the logarithm on the right vanishes, which enables us to solve for c0 :

c0 = log F2 (x, Q20 ) . (486)

72
Now, exponentiating both sides, we find the expression for the form factor at an arbitrary
energy scale Q2 is
Q2
2 −  log
F2 (x, Q ) = F2 (x, Q20 )e 2 Q2
0 . (487)
Note that the exponential factor in this expression is just
2  2 −/2
− 2 log Q2 Q
Q0
e = . (488)
Q20

Comparing this to first expression for the form factor, if we identify r0 = 1/Q0 , then the
expression for the form factor at Q2 = Q20 is
  π 
F2 (x, Q20 ) = cos Γ()F2 (x) . (489)
1− 2

7.3 (c)
As Q2 → ∞, the logarithmic factor in the exponent also diverges:
Q2
lim log 2 = ∞ . (490)
Q2 →∞ Q0
Because we assume that  > 0, this divergence correspondingly suppresses the form factor
and we find
lim
2
F2 (x, Q2 ) = 0 . (491)
Q →∞
2
The other limit, when Q → 0 leads to a different limit of the form factor. Now, the
logarithmic factor diverges to −∞:
Q2
lim log = −∞ . (492)
2
Q →0 Q20
Because  > 0, this increases the size of the form factor to be arbitrarily large:

lim F2 (x, Q2 ) = ∞ . (493)

Q2 →0

7.4 Infrared and Collinear Safety

7.4 (a)
The C-parameter on three-body phase space is defined to be
(1 − xq )(1 − xq̄ )(xq + xq̄ − 1)
C=6 . (494)
xq xq̄ (2 − xq − xq̄ )
The divergences of the cross section doubly differential in xq and xq̄ lie along the lines of
xq = 1 and xq̄ = 1. To consider where the C-parameter maps these divergences, let’s first

73
let xq → 1, for xq̄ 6= 1. In this limit, all the factors present in the C-parameter are finite
with the factor 1 − xq forcing the value of C to 0 in this limit:

lim C = 0 . (495)
xq →1

Because the expression for the C-parameter is symmetric in xq and xq̄ , this is the same limit
as xq̄ → 1:
lim C = 0 . (496)
xq̄ →1

Both of these limits map to the same point, which is essential for IRC safety.
There is another limit we can take that is not covered by these two limits. In particular,
the difference between xq and xq̄ from 1 could scale similarly. That is, let’s take

x q = 1 − cq  , xq̄ = 1 − cq̄  , (497)

for two constants cq and cq̄ , for  → 0. To first-order in , the expression for the C-parameter
is
(cq )(cq̄ )
lim C = lim = 0. (498)
→0 →0 (cq + cq̄ )

These three limits are all the possible divergent regions of the differential cross section.
Because they all map to the same point, we say that the C-parameter is IRC safe.

7.4 (b)
Now, let’s consider the relative squared energy fractions:

x22 + x23
 
3
E= −1 , (499)
2 x21 + x22 + x23

where the center-of-mass energy fractions are ordered x1 > x2 > x3 . It suffices to only one
limits to show that this is not IRC safe. For concreteness, we can imagine that x1 = xq ,
x2 = xq̄ , and x3 = xg = 2 − xq − xq̄ . The largest that xq could be is 1, for arbitrary xq̄ ,
at which the double differential cross section diverges. In this limit, the energy fraction
observable becomes
x2q̄ + (1 − xq̄ )2
 
3
lim E = −1 . (500)
xq →1 2 1 + x2q̄ + (1 − xq̄ )2
This is a non-trivial function of xq̄ , which means that there is (at least) a line on the phase
space defined by a range of values of E over which the matrix element diverges. Therefore,
this fails the requirement that the divergences are isolated to a single point on phase space,
and hence this observable is not IRC safe.

74
7.4 (c)
The broadening B is defined as
r
x2
B= (1 − x1 )1/2 (1 − x2 )1/2 , (501)
x1
for x1 > x2 > x3 . As the divergences in the differential cross section occur when xq or xq̄ go
to 1, this expression for the broadening makes it clear that in those limits, it always takes
the value of 0. In particular, note that it is impossible for x3 → 1, because the sum of the
three phase space variables is fixed to be 2. Therefore, broadening is IRC safe, because every
limit of xq or xq̄ to 1 is mapped to B = 0.

7.4 (d)
The ratio of 1 − thrust to the broadening takes the value
s
1−τ x1 (1 − x1 )
= , (502)
B x2 (1 − x2 )

for x1 > x2 > x3 . It suffices to demonstrate that one limit is ill-defined to show that this is
not IRC safe. Consider the assignments that x1 = xq and x2 = xq̄ and we take

x q = 1 − cq  , xq̄ = 1 − cq̄  , (503)

for  > 0. For the ordering xq > xq̄ to hold, we enforce cq < cq̄ . With these expressions, the
value of the ratio becomes
s s s
1−τ xq (1 − xq ) (1 − cq )cq  (1 − cq )cq
= = = . (504)
B xq̄ (1 − xq̄ ) (1 − cq̄ )cq̄  (1 − cq̄ )cq̄

The limit as  → 0 is where the differential cross section diverges, in which we find
1−τ cq
r
lim = . (505)
→0 B cq̄
Other than the requirement that cq < cq̄ , this ratio is completely arbitrary, so the divergence
is not isolated to a single point. Therefore, the ratio of 1 − thrust to broadening is not IRC
safe.

7.5 Properties of Helicity Spinors

7.5 (a)
We will first prove that spinor inner products are anti-symmetric. The angle bracket spinor
product is
hpki = u†L (p)uR (k) . (506)

75
The left- and right-handed spinors can be expressed as

e k cos θ2k
   −iφ /2 
†
p iφp /2 θp −iφp /2 θp p
uL (p) = 2Ep e sin −e cos , uR (k) = 2Ek ,
2 2 eiφk /2 sin θ2k
(507)

where the p and k subscripts denote the corresponding momentum. Taking their inner
product, we have
 
†
p i(φp −φk )/2 θp θk −i(φp −φk )/2 θp θk
uL (p)uR (k) = 2 Ep Ek e sin cos − e cos sin . (508)
2 2 2 2

This expression as written is manifestly anti-symmetric in p ↔ k, therefore it follows that

hpki = u†L (p)uR (k) = −u†L (k)uR (p) = −hkpi . (509)

7.5 (b)
Our proof of the Schouten identity will greatly simplify the components of the spinors. In
particular, note that the spinor pi, for example, has two components which we can label as
 
p1
pi = . (510)
p2

Correspondingly, because the inner product hpki is anti-symmetric, it necessarily is formed

from the combination of spinor entries as:

hpki = p1 k2 − p2 k1 . (511)

With this identification, we can simply multiply all terms together and collect them. The
right-hand side of the Schouten identity is then

hpqihkri + hprihqki = (p1 q2 − p2 q1 )(k1 r2 − k2 r1 ) + (p1 r2 − p2 r1 )(q1 k2 − q2 k1 ) (512)

= −p1 k2 q2 r1 − p2 k1 q1 r2 + p1 k2 q1 r2 + p2 k1 q2 r1 (513)
= (p1 k2 − p2 k1 )(q1 r2 − q2 r1 ) (514)
= hpkihqri . (515)

7.6 More Helicity Spinors

7.6 (a)
If momentum p~ is aligned along the +ẑ axis, its spinor is
 
p 1
uR (p) = vL (p) = 2Ep . (516)
0

76
The spinor for momentum k is
sin 2θ
p  
uL (k) = vR (k) = 2Ek , (517)
− cos 2θ
where we can choose, without loss of generality, the azimuthal angle of the momentum to
be φ = 0. Then, the first spinor products are
 
† µ µ 1
uR (p)σ vL (p) = 2Ep (1 0)σ = 2Ep (1, 0, 0, 1)µ , (518)
0
sin 2θ
   
† θ θ
vR (k)σ̄µ uL (k) = 2Ek sin − cos σ̄µ = 2Ek (1, sin θ, 0, cos θ)µ . (519)
2 2 − cos 2θ
Note that the resulting four-vectors are simply the momenta p and k, up to a factor of 2.
Therefore, we indeed have established that
1 † µ

†

2p · k = u (p)σ vL (p) vR (k)σ̄µ uL (k) . (520)
2 R
Next we need to evaluate the spinor product
" # 
sin 2θ
   

†

†
 θ θ 1
uR (p)uL (k) vR (k)vL (p) = 4Ep Ek (1 0) θ
sin − cos (521)
− cos 2 2 2 0
θ
= 4Ep Ek sin2 = 2Ep Ek (1 − cos θ) = 2p · k . (522)
2

7.6 (b)
Aligning the momentum p~ along the +ẑ direction, the spinors are
   
p 1 p 0
uR (p) = 2Ep , uL (p) = 2Ep . (523)
0 −1
Acting the rotation matrix M(φ) on these spinors rotates them about the ẑ axis by an angle
φ. It is useful to note that these spinors are eigenvectors of the σ3 matrix:
   
1 0 p 1
σ3 uR (p) = 2Ep = uR (p) , (524)
0 −1 0
   
1 0 p 0
σ3 uL (p) = 2Ep = −uL (p) . (525)
0 −1 −1
Therefore, the action of the rotation matrix on these spinors is:
    
φ 1 0 p 1 φ
M(φ)uR (p) = exp i 2Ep = ei 2 uR (p) , (526)
2 0 −1 0
    
φ 1 0 0 φ
= e−i 2 uL (p) .
p
M(φ)uL (p) = exp i 2Ep (527)
2 0 −1 −1

77
Then, using this rule for the rotation of the spinors, the rotation of the polarization vector
is
φ
1 u† (r)σ µ (M(φ)uR (p)) 1 u† (r)σ µ ei 2 uR (p)
M(φ)µR (p) = − √ R† = − √ R† φ (528)
2 uR (r) (M(φ)uL (p)) 2 uR (r)e−i 2 uL (p)
= eiφ µR (p) , (529)

which indeed corresponds to the property of a right-handed spin-1 polarization vector under
a rotation.

7.6 (c)
We can perform the exact same rotation analysis as earlier, but this time rotate about the
vector r. Now, we find
φ
1 (M(φ)uR (r))† σ µ uR (p) 1 e−i 2 u†R (r)σ µ uR (p)
M(φ)µR (p) = −√ = −√ (530)
2 (M(φ)uR (r))† uL (p)
φ
2 e−i 2 u†R (r)uL (p)
= µR (p) , (531)

indicating indeed that the vector r does not affect the spin of the polarization vector.

7.7 The Drell-Yan Process

7.7 (a)
The squared center-of-mass collision energy of the protons is s, and the momenta of the
colliding partons is scaled by the appropriate factors of x1 and x2 . Therefore, the squared
partonic center-of-mass collision energy ŝ = Q2 is

ŝ = x1 x2 s = Q2 = m2Z . (532)

Therefore, the momentum fraction of parton 2, x2 , is

m2Z
x2 = . (533)
x1 s

7.7 (b)
Recall that the rapidity y for any particle is
1 E + pz
y= log , (534)
2 E − pz
where E is the energy of the particle and pz is its momentum along the ẑ axis. Assuming
that the Z boson is produced on-shell in the parton collision, its energy is just the sum of

78
the energies of the colliding partons. If the energy of a proton is Ep , the energy E of the Z
boson is
E = (x1 + x2 )Ep . (535)
The z-component of momentum pz is sum of the momenta of the initial partons. Assuming
that the collision energy is high enough that the protons are effectively massless, the net z
momentum is proportional to the difference of the partonic fractions:

pz = (x1 − x2 )Ep . (536)

Therefore, the rapidity of the Z boson is

1 (x1 + x2 )Ep + (x1 − x2 )Ep 1 x1
y= log = log . (537)
2 (x1 + x2 )Ep − (x1 − x2 )Ep 2 x2

Plugging in the value for x2 that we found in the previous part, we then find that
√
1 x1 1 x21 s x1 s
y = log = log 2 = log , (538)
2 x2 2 mZ mZ

as promised.

7.7 (c)
From the previous part, the rapidity is maximized if the√ momentum fraction x1 is maximal;
that is, if x1 = 1. The center-of-mass collision energy is s = 1.96 TeV and the mass of the
Z boson (from the PDG) is mZ = 91.1876. Therefore, the maximum rapidity is
√
s 1960
ymax = log = log ' 3.07 . (539)
mZ 91.1876
The maximal value of the rapidity display in Fig. 7.2 appears to be about y ' 3, which is
very close to this maximal possible value.

7.7 (d)
Because there is a direct map between rapidity and momentum fraction provided in part (b)
of this problem, we can directly relate the probability distribution for rapidity p(y) to the
parton distribution function f (x). The probability for the rapidity to be within dy of y is
equal to the probability for the momentum fraction to be within dx of x:

p(y) dy = f (x) dx . (540)

We can then relate the probability distributions by dividing by dy on both sides of the
equation:
dx
p(y) = f (x) . (541)
dy

79
From part (b), the rapidity y should be evaluated at
√
x s
y = log , (542)
mZ
and the derivative of y with respect to x is then
dy 1
= . (543)
dx x
We can insert this into the relationship between the probability distributions and finally find
p(y)|y=log x√s = xf (x) . (544)
mZ

7.7 (e)
From Fig. 7.2, a collection of 10 points of the Z boson rapidity distribution are approximately:
(0.05, 0.27) , (0.35, 0.27) , (0.65, 0.26) , (0.95, 0.24) , (1.25, 0.23) , (545)
(1.55, 0.19) , (1.85, 0.14) , (2.15, 0.09) , (2.45, 0.04) , (2.75, 0.01) .
The first entry is the rapidity y and the second is the value of the rapidity probability distri-
bution p(y). Recall that Fig. 7.2 only plots positive rapidity; the full probability distribution
is symmetric in y ↔ −y. To convert these values to (x, xf (x)), we just need to use the map
to momentum fraction x:
mZ
x = √ ey . (546)
s
The smallest value of x accessible in this process corresponds to the collision when one parton
takes all of the momentum of its proton (say, x1 = 1) which means that
mZ
e−ymax = √ . (547)
s
Therefore, the smallest value of x accessible in this process is
mZ m2
xmin = √ e−ymin = Z ' 0.0022 . (548)
s s
Now, remembering that the full rapidity distribution is symmetric about y = 0, the
corresponding x and xf (x) values are:
(0.003, 0.01) , (0.004, 0.04) , (0.0054, 0.09) , (0.0073, 0.14) , (0.0099, 0.19) , (549)
(0.013, 0.23) , (0.018, 0.24) , (0.024, 0.26) , (0.033, 0.27) , (0.044, 0.27) ,
(0.049, 0.27) , (0.067, 0.27) , (0.089, 0.26) , (0.12, 0.24) , (0.16, 0.23) ,
(0.22, 0.19) , (0.3, 0.14) , (0.4, 0.09) , (0.54, 0.04) , (0.73, 0.01) .
Note that this peaks in the range of x ∈ [0.033, 0.067] and the value at the peak is xf (x) =
0.27.

80
7.7 (f )
Even though our pdf extraction procedure is very simplistic, we find qualitative agreement
with the plots in Fig. 7.6. The peaks of the up and down quark pdfs in those plots is at
about x = 0.1, which is only a factor of 2 larger than where we estimate the peak. Note the
logarithmic scale in x in the figure, as well. A factor of 2 difference is relatively small when
the value of x ranges over four decades! Further, the height of the peak from our extraction
was about xf (x) = 0.27, which lies between the peak heights for the up and down quark
pdfs from the figure.
Nevertheless, our simplistic extraction does miss some features like the tail of the distri-
bution extending to very small values of x. Within our framework, however, there’s nothing
we can do about this unless we have another mechanism for probing
√ smaller values of x, like
making a larger collider that can collide protons at a larger s.

7.8 Thrust in Data

In this exercise, we are asked to compare data on the thrust observable to our lowest-order
prediction in equation 7.141. First, the plot of data from HEPDATA is shown in Fig. 1a. The
plot of the distribution of Eq. 7.141 rescaled by an appropriate factor is shown in Fig. 1b.
Broadly, our predicted distribution looks very similar to the data, but there are a few key
differences. First, the data extends out to almost 1−τ = 1/2, while our predicted distribution
terminates at 1 − τ = 1/3. If there are more particles in the final state than just q q̄g, the
value of 1 − τ can extend beyond 1/3, and has an absolute maximum of 1/2. Further, the
data exhibits a peak at about 1 − τ = 0.03, and decreases as 1 − τ decreases. This is in
contrast to our prediction which has no maximum and in fact diverges as 1 − τ → 0. This
divergence is unphysical and other considerations must be made to correctly describe this
behavior.

8 Quantum Chromodynamics
8.1 Masslessness of the Gluon
A possible mass term for the gluon in the QCD Lagrangian would be

LQCD ⊃ −m2g tr[Aaµ T a Aµ b T b ] . (550)

From Eq. 8.31, the Lie algebra-valued gluon field Aaµ T a transforms under a gauge transfor-
mation as  
1
Aµ T → U Aµ T + ∂µ α (x)T U† .
a a a a a a
(551)
g

81
 ��

��

� (� - τ)
�σ
��

�
��� ��� ��� ��� ��� ���
�-τ

(a) (b)

Figure 1: Plot of thrust distributions as measured in various e+ e− colliders (left) and the
distribution from Eq. 7.141 (right).

Then, under a gauge transformation, the mass term in the Lagrangian becomes
     
1 1 µ b
−mg tr[Aµ T A T ] → −mg tr U Aµ T + ∂µ α (x)T U U A T + ∂ α (x)T U†
2 a a µb b 2 a a a a † µb b b
g g
  
2 a a 1 a a µb b 1 µ b b
= −mg tr Aµ T + ∂µ α (x)T A T + ∂ α (x)T (552)
g g
2m2g  a a µ b
= −m2g tr[Aaµ T a Aµ b T b ] − tr Aµ T ∂ α (x)T b

g
m2g 
− 2 tr (∂µ αa (x)) T a ∂ µ αb (x) T b .


g
In evaluation of this expression, we used the cyclicity of the trace and that the gauge trans-
formation matrix U is unitary. In general, unless the basis of color factors is identical at
every point in space-time, so that ∂µ αa (x) = 0, such a mass term is not invariant to gauge
transformations.

8.2 Bianchi Identity

8.2 (a)
a
Recall that the field-strength tensor Fµν is defined to be
a
− igFµν T a = [Dµ , Dν ] , (553)

82
where the covariant derivative Dµ is
Dµ = ∂µ − igAaµ T a . (554)
Therefore, the Bianchi identity can be equivalently expressed as
[Dµ , [Dν , Dρ ]] + [Dρ , [Dµ , Dν ]] + [Dν , [Dρ , Dµ ]] = 0 . (555)
Let’s evaluate these nested commutators. First, the commutator of covariant derivatives is
[Dν , Dρ ] = (∂ν − igAbν T b )(∂ρ − igAcρ T c ) − (∂ρ − igAcρ T c )(∂ν − igAbν T b ) (556)
= −ig∂ν Acρ T c + ig∂ρ Abν T b − g 2 Abν Acρ [T b , T c ]
= −ig ∂ν Abρ − ∂ρ Abν T b − ig 2 f bcd T d Abν Acρ .
 

Then, we can add another covariant derivative to the commutator:

[Dµ , [Dν , Dρ ]] = (∂µ − igAaµ T a )(−ig ∂ν Abρ − ∂ρ Abν T b − ig 2 f bcd T d Abν Acρ )
 
(557)
b b
  b 2 abc a b c a a
− (−ig ∂ν Aρ − ∂ρ Aν T − ig f T Aν Aρ )(∂µ − igAµ T )
= −ig ∂µ ∂ν Abρ − ∂µ ∂ρ Aaν T a − ig 2 f abc T a ∂µ (Abν Acρ ) − g 2 Aaµ ∂ν Abρ − ∂ρ Abν [T a , T b ]
 

− g 3 f bcd Aaµ Abν Acρ [T a , T d ]

= −ig ∂µ ∂ν Abρ − ∂µ ∂ρ Aaν T a − ig 2 f abc T a ∂µ (Abν Acρ ) − ig 2 Aaµ ∂ν Abρ − ∂ρ Abν f abd T d
 

− ig 3 f bcd f ade Aaµ Abν Acρ T e .

In the evaluation of these expressions, we have extensively used the Lie algebra relationship
[T a , T b ] = if abc T c , (558)
where f abc are the structure constants.
This is just one nested commutator; we must sum it with two others in which the in-
dices µ, ν, ρ are permuted. All terms involving derivatives explicitly cancel in the sum. For
example, the terms with two derivatives in the sum are of the form:
−ig ∂µ ∂ν Abρ − ∂µ ∂ρ Aaν T a − ig ∂ρ ∂µ Abν − ∂ν ∂µ Aaρ T a − ig ∂ν ∂ρ Abµ − ∂ρ ∂ν Aaµ T a = 0 ,
     

(559)
because all terms cancel pairwise and ∂µ ∂ν = ∂ν ∂µ . The remaining terms cancel in the sum
because they arrange themselves into terms that vanish by the product rule of differentiation,
using anti-symmetry of the structure constants. Therefore, the only terms that do not
explicitly cancel are those with no derivatives. In the sum over the three nested commutators,
this is
− ig 3 f bcd f ade Aaµ Abν Acρ T e − ig 3 f bcd f ade Aaρ Abµ Acν T e − ig 3 f bcd f ade Aaν Abρ Acµ T e = 0 . (560)
We can freely relabel indices that are summed over, so this can equivalently be expressed as
− ig 3 Aaµ Abν Acρ T e (f bcd f ade + f abd f cde + f cad f bde ) = 0 . (561)

83
For this to hold for any gluon field Aaµ we then must require
f bcd f ade + f abd f cde + f cad f bde = 0 , (562)
which is indeed the Jacobi identity. Note the slight re-ordering of indices on the structure
constants from what is stated in the problem; these can reconciled by noting that
f abc = −f bac . (563)

8.2 (b)
The Jacobi identity expressed as nested commutators of a Lie algebra is
[T a , [T b , T c ]] + [T b , [T c , T a ]] + [T c , [T a , T b ]] = 0 . (564)
Expanding out the first nested commutator, we find
[T a , [T b , T c ]] = [T a , T b T c − T c T b ] = T a T b T c − T a T c T b − T b T c T a + T c T b T a . (565)
Then, the sum over the three nested commutators is
[T a , [T b , T c ]] + [T b , [T c , T a ]] + [T c , [T a , T b ]] = 0 (566)
= T aT bT c − T aT cT b − T bT cT a + T cT bT a + T bT cT a − T bT aT c − T cT aT b + T aT cT b
+ T cT aT b − T cT bT a − T aT bT c + T bT aT c .
Each product of matrices cancels pairwise, summing to 0.

8.3 Instantons
8.3 (a)
To determine the relationship between the electric and magnetic fields imposed by self-
duality, we will just focus on one entry of the field strength tensor, and then use Lorentz
covariance to determine the vectorial relationship. Let’s just consider µ = 0, ν = 1 for which
the field strength tensor entry is
F01 = Ex , (567)
which can be found from Eq. 2.122. The dual of this component is then
i i i
01ρσ F ρσ = 0123 F 23 + 0132 F 32 . (568)
2 2 2
Note that there are only two non-zero terms in the sum because all four indices of the
totally anti-symmetric  symbol must be distinct. Further, the field strength tensor is anti-
symmetric, so
F 23 = −F 32 = −Bx . (569)
Combining these terms, we then find that
Ex = −iBx . (570)
It then follows that the vectorial constraint for a self-dual electromagnetic field is
~ = −iB
E ~. (571)

84
8.3 (b)
This is a significantly challenging exercise. It also assumes significant comfort with index
notation and anti-symmetry.
To massage the self-dual field strength Lagrangian, we first consider the product of two
field strengths:
a a
= ∂µ Aaν − ∂ν Aaµ + gf abc Abµ Acν ∂ρ Aaσ − ∂σ Aaρ + gf ade Adρ Aeσ



Fµν Fρσ (572)
= (∂µ Aaν )(∂ρ Aaσ ) − (∂µ Aaν )(∂σ Aaρ ) − (∂ν Aaµ )(∂ρ Aaσ ) + (∂ν Aaµ )(∂σ Aaρ )
+ gf ade (∂µ Aaν )(Adρ Aeσ ) − gf ade (∂ν Aaµ )(Adρ Aeσ ) + gf abc (∂ρ Aaσ )(Abµ Acν )
− gf abc (∂σ Aaρ )(Abµ Acν ) + g 2 f abc f ade Abµ Acν Adρ Aeσ .

We will focus on terms ordered by their number of derivatives. First, let’s study the final
term, proportional to g 2 . In the Lagrangian, this term becomes
i
L ⊃ − g 2 f abc f ade µνρσ Abµ Acν Adρ Aeσ . (573)
8
Because the  symbol is totally anti-symmetric, we can perform a sub-analysis to show that
this term vanishes. What we will do is fix µ and cyclically rotate through ν, ρ, σ. Every
other term is related by a sign to the sum of these three terms. Keeping track of the signs,
the Lagrangian contains the terms:
i i i
L ⊃ − g 2 f abc f ade µνρσ Abµ Acν Adρ Aeσ − g 2 f abc f ade µνρσ Abµ Acσ Adν Aeρ − g 2 f abc f ade µνρσ Abµ Acρ Adσ Aeν .
8 8 8
(574)
We can isolate the dependence on the structure constants by relabeling the Lorentz indices
µ, ν, ρ, σ, as they are also summed over. Doing this rearrangement, we find
i
L ⊃ − g 2 µνρσ Abµ Acν Adρ Aeσ f abc f ade + f abe f acd + f abd f aec .


(575)
8
The factor in parentheses is called the Jacobi identity, and is required to vanish in an
associative Lie algebra. Therefore, the terms in the self-dual action with four gauge fields
vanishes.
Next, let’s consider those terms with two derivatives, proportional to g 0 . While there are
four terms here, when contracted with the  symbol it simplifies to:
i
L ⊃ − µνρσ (∂µ Aaν )(∂ρ Aaσ ) − (∂µ Aaν )(∂σ Aaρ ) − (∂ν Aaµ )(∂ρ Aaσ ) + (∂ν Aaµ )(∂σ Aaρ )
 
(576)
8
i
= − µνρσ (∂µ Aaν )(∂ρ Aaσ ) − (∂µ Aaν )(∂σ Aaρ )
 
4
i
= − µνρσ ∂µ Aaν ∂ρ Aaσ − ∂σ Aaρ .


4
In writing this, we note that partial derivatives commute, which simplifies the expression on
the final line.

85
 Finally, let’s study the O(g) terms in the expansion. When contracting with the  symbol,
we have
i
L ⊃ − µνρσ gf ade (∂µ Aaν )(Adρ Aeσ ) − gf ade (∂ν Aaµ )(Adρ Aeσ ) + gf abc (∂ρ Aaσ )(Abµ Acν ) − gf abc (∂σ Aaρ )(Abµ Acν )
 
8
i
= − µνρσ gf abc (∂µ Aaν )(Abρ Acσ ) − (∂ν Aaµ )(Abρ Acσ ) + (∂ρ Aaσ )(Abµ Acν ) − (∂σ Aaρ )(Abµ Acν ) .
 
8
We now want to write this so the Lorentz index ν is always associated with the group index
a, ρ with b, and σ with c. Being mindful of the signs, we have
i
L ⊃ − µνρσ gf abc (∂µ Aaν )(Abρ Acσ ) + (∂µ Aaν )(Abρ Acσ ) + (∂µ Aaν )(Abρ Acσ ) + (∂µ Aaν )(Abρ Acσ )
 
8
i
= − µνρσ gf abc (∂µ Aaν )(Abρ Acσ ) (577)
2
i
= − µνρσ ∂µ gf abc Aaν Abρ Acσ


6
i i
= − µνρσ ∂µ gf abc Aaν Abρ Acσ + µνρσ ∂µ gf abc Aaν Abρ Acσ .



4 12
On the final line, we used the derivative product rule and total anti-symmetry of both the 
symbol and the structure constants f abc .
Putting all of these terms together, we find that the self-dual Lagrangian is indeed a total
derivative:
i
L = − µνρσ Fµν
a a
Fρσ (578)
8
i µνρσ
h
a a a

abc a b c g abc a b c i
= − ∂µ  Aν ∂ρ Aσ − ∂σ Aρ + gf Aν Aρ Aσ − f Aν Aρ Aσ (579)
4 3
i µνρσ

a a g abc a b c 
= − ∂µ  Aν Fρσ − f Aν Aρ Aσ , (580)
4 3
as expected.

8.4 Wilson Lines

8.4 (a)
We will show that the Wilson line defined by
 Z x 
µ
W (x, y) = exp ie ds Aµ (s) , (581)
y

satisfies the differential equation

Dµ W (x, y) = (∂µ − ieAµ )W (x, y) = 0 . (582)

86
Taking the partial derivative of the Wilson line with respect to xµ and using the Fundamental
Theorem of Calculus, we find
 Z x   Z x  Z x
µ µ ∂
∂µ exp ie ds Aµ (s) = exp ie ds Aµ (s) µ
ie dsµ Aµ (s) (583)
y y ∂x y
= ieAµ (x)W (x, y) .
Inserting this into the parallel transport differential equation we indeed see that it vanishes,
as expected.

8.4 (b)
Under a gauge transformation, the electromagnetic vector potential transforms as
Aµ → Aµ + ∂µ λ , (584)
where λ is any function of spacetime coordinate x. Therefore, the Wilson line transforms as
 Z x 
µ
W (x, y) → exp ie ds (Aµ (s) + ∂µ λ(s)) (585)
y
 Z x 
= exp[ieλ(x)] exp ie ds Aµ (s) exp[−ieλ(y)] = eieλ(x) W (x, y)e−ieλ(y) ,
µ
y

where again, we used the Fundamental Theorem of Calculus. λ is an arbitrary function and x
and y are in general distinct points, so λ(x) and λ(y) are just some real numbers. Therefore,
the Wilson line transforms with an exponential phase under gauge transformation.

8.4 (c)
The time-independent Wilson loop along a closed path P is
 I 
WP = exp −ie d~s · A . ~ (586)
P

Stokes’s theorem is the statement that this line integral is equivalent to an integral of the
curl of the integrand over any surface bounded by path P :
I Z
Stokes
~ = ~ · (∇
~ × A)
~ .
d~s · A dΩ (587)
P Ω

Here, we have denoted the surface as Ω and dΩ ~ is the differential area element on the surface.
The curl of the vector potential is just the magnetic field:
~ ×A
∇ ~=B
~, (588)
so the Wilson loop is just the exponential of the magnetic flux through any surface bounded
by the path P :  I   Z 
WP = exp −ie d~s · A ~ = exp −ie dΩ ~ ·B
~ . (589)
P Ω

87
8.5 su(2) Lie Algebra
8.5 (a)
The matrices T 1 , T 2 , and T 3 are
 
0 0 0
T 1 = i1jk =  0 0 i , (590)
0 −i 0
 
0 0 −i
T 2 = i2jk = 0 0 0 , (591)
i 0 0
 
0 i 0
T 3 = i3jk =  −i 0 0 . (592)
0 0 0

These matrices are indeed Hermitian and traceless.

8.5 (b)
Now, we are asked to evaluate the commutation relations of these matrices. Here, we will
just calculate one explicitly. Let’s calculate

[T 1 , T 2 ] = T 1 T 2 − T 2 T 1 (593)
     
0 0 0 0 0 −i 0 0 −i 0 0 0
=  0 0 i   0 0 0  −  0 0 0   0 0 i 
0 −i 0 i 0 0 i 0 0 0 −i 0
     
0 0 0 0 −1 0 0 1 0
=  −1 0 0  −  0 0 0  =  −1 0 0 
0 0 0 0 0 0 0 0 0
= iT 3 .

Therefore, the structure constant f 123 = 1. By the total anti-symmetry of the structure
constants, we necessarily have that
f abc = abc . (594)
This is identically the structure constants for the su(2) Lie algebra, so these matrices form
a three-dimensional representation of su(2).

8.5 (c)
To calculate the Killing form, we’ll do a trick. Note that the product of two matrices can be
expressed as
T a T b = ajk blk , (595)

88
where a and b denote the matrix, the second index is the row and the third index is the
column of the matrix. The trace of this matrix product is then

tr[T a T b ] = ajk bjk . (596)

Note that because  is totally anti-symmetric, for this to be non-zero we must have a = b.
So, this trace is proportional to the Kronecker symbol δ ab . To determine the constant of
proportionality, let’s set a = b = 1 so that

tr[T 1 T 1 ] = 1jk 1jk = 123 123 + 132 132 = 2 . (597)

Therefore, the Killing form of this matrix vector space is

tr[T a T b ] = 2δ ab . (598)

8.6 Casimir Invariant

8.6 (a)
We are asked to show that the sum of squared Lie algebra elements commutes with every
member of the Lie algebra. So, let’s calculate the commutation relation
" m #
X
a 2 b a 2 b
[(T ) , T ] = (T ) , T , (599)
a=1

for any element of the Lie algebra T b , assuming that the dimension of the Lie algebra is m.
The commutation relation of the Casimir in the Lie algebra with another matrix in the Lie
algebra is

[(T a )2 , T b ] = T a T a T b − T b T a T a = T a T b T a − T b T a T a + T a [T a , T b ] (600)
b a a b a a a a b a b a
= T T T − T T T + T [T , T ] + [T , T ]T
= if abc T a T c + if abc T c T a = if abc (T a T c + T c T a ) = 0 . (601)

Recall that repeated indices are summed over. Because the structure constants f abc are
totally anti-symmetric, the final expression is manifestly 0.

8.6 (b)
The Casimir for the fundamental representation of a Lie algebra CF is defined as the sum of
the squares of the elements of the Lie algebra:
N 2 −1
X
CF IN ×N = (T a )2 . (602)
a=1

89
In this expression, we have explicitly written the Casimir proportional to the N × N identity
matrix. Taking the trace of both sides of this expression, we use the Killing form and that
the trace of the identity matrix is N to find:

N2 − 1
N CF = . (603)
2
Solving for CF , we find the quoted result.

8.6 (c)
If N = 2, the SU(2) Casimir is
22 − 1 3
CF = = . (604)
2×2 4

8.6 (d)
If N = 3, the SU(3) Casimir is
32 − 1 4
CF = = . (605)
2×3 3

8.6 (e)
Finally, we are asked to calculate the fundamental Casimir from explicit calculation with
the Gell-Mann matrices. Because there are a lot of matrix products to perform, we won’t
do that here. We will just calculate one matrix product and show that the Casimir follows.
The first Gell-Mann matrix is  
0 1 0
1
T1 =  1 0 0  . (606)
2
0 0 0
The square of this matrix is  
1 0 0
1
(T 1 )2 =  0 1 0  . (607)
4
0 0 0
One can show that the square of all other Gell-Mann matrices takes a similar form. They
all exhibit an overall 1/4, with two unit entries on the diagonal and one 0 entry. Therefore,
the fundamental Casimir is just the factor of 1/4 times the eight Gell-Mann matrices, and
then multiplied by a factor of 2/3 to account for the 0 on the diagonal. This is
1 2 4
CF = ×8× = , (608)
4 3 3
as expected from the previous part.

90
8.7 Adjoint Representation of SU(3)
8.7 (a)
The adjoint Casimir of SU(N ), CA , is defined to be
N 2 −1
X
a 2
CA I(N 2 −1)×(N 2 −1) = (Tadj ) , (609)
a=1

where we have expressed the Casimir proportional to the (N 2 −1)×(N 2 −1) identity matrix.
Taking the trace of both sides of this equation, we find

CA (N 2 − 1) = N (N 2 − 1) . (610)

Simplifying this expression, we indeed find CA = N .

8.7 (b)
The commutation relations of a Lie algebra are

[T a , T b ] = if abc T c . (611)

Assuming that the matrices T a are in the fundamental representation, their Killing form is
1
tr[T a T b ] = δ ab . (612)
2
So, to isolate the structure constants, we can multiply the commutation relations by another
basis matrix T d and then take the trace. That is
i i
tr [T a , T b ]T d = if abc tr[T c T d ] = f abc δ cd = f abd .
 
(613)
2 2
Now, we can solve for the structure constants f abd and relabel d = c to find

f abc = −2i tr [T a , T b ]T c ,
 
(614)

as promised.

8.7 (c)
Now, we’re asked to calculate the explicit structure constants from the matrices of the fun-
damental representation of SU(3), the Gell-Mann matrices. Because there are many possible
combinations of matrices, we will just present the calculation of one structure constant. Let’s
consider
f 458 = −2i tr [T 4 , T 5 ]T 8 = −2i tr T 4 T 5 T 8 − T 5 T 4 T 8 .
   
(615)

91
The relevant Gell-Mann matrices are
    
0 0 1 0 0 −i 1 0 0
1 1 1
T4 =  0 0 0  , T5 =  0 0 0  , T8 = √  0 1 0  . (616)
2 2 2 3
1 0 0 i 0 0 0 0 −2

The relevant matrix products are:

     
0 0 1 0 0 −i 1 0 0 i 0 0
4 5 8 1  1 
T T T = √ 0 0 0   0 0 0  0 1 0  = √ 0 0 0  , (617)
8 3 1 0 0 i 0 0 0 0 −2 8 3 0 0 2i
     
0 0 −i 0 0 1 1 0 0 −i 0 0
5 4 8 1  1 
T T T = √ 0 0 0   0 0 0  0 1 0  = √ 0 0 0 .
8 3 i 0 0 1 0 0 0 0 −2 8 3 0 0 −2i

The structure constant then follows:

√
458 4 5 8 5 4 8 3
f = −2i tr[T T T − T T T ] = . (618)
2
From this one structure constant, we actually get many more. By the cyclicity of the trace,
cyclic permutations of the indices 4, 5, 8 all produce the same value. Transpositions of two
indices flip the sign of the structure constant because of the relative − sign in the trace.
Therefore, we have that
√
458 845 584 548 485 854 3
f = f = f = −f = −f = −f = . (619)
2

8.8 Running Couplings of QED and QCD

8.8 (a)
The expression for the running coupling in QED, the fine structure constant, is
α(Q0 )
α(Q) = . (620)
1− 2
3π
α(Q0 ) log QQ0

With me = 511 keV and α(me ) = 1/137, the Landau pole occurs at the energy Q when the
denominator vanishes:
2 Q 3π
1− α(me ) log = 0 , or when Q = me e 2α(me ) . (621)
3π me
Plugging in numbers, the location of the Landau pole is

Q ' 10277 GeV . (622)

This is about 200 orders of magnitude more than the total energy of the universe.

92
8.8 (b)
With nf = 6, the numerical factor in the running coupling for the strong coupling is
2
11 − · 6 = 7. (623)
3
So, setting the value of the fine structure constant equal to the strong coupling produces:
α(me ) αs (mZ )
= . (624)
1− 2
3π
α(me ) log Qmeqe 1+ 7
α (mZ ) log Q
2π s mZ
eq

Massaging this expression, the energy at which the couplings are equal is
Q2eq 6π αs (mZ ) − α(me )
log = . (625)
me mZ 25 α(me )αs (mZ )
It then follows that the energy at which the couplings are equal is

Qeq ' 2.4 × 1020 GeV . (626)

While still a large energy, this is more understandable. This energy corresponds to approxi-
mately the mass energy of 10−7 kg.

8.8 (c)
Including the effects of all of the charged fermions of the Standard Model, the β-function for
the fine structure constant is
2 α2 X
βSM (α) = Q2i . (627)
3 π i fermions

There are three quarks with charge 1/3 and three quarks with charge 2/3, so the sum over
the square of the electric charges of all quarks is
 
X
2 1 4
Qi = 3 3 × + 3 × = 5. (628)
i quarks
9 9

Note that we have multiplied by the factor of 3 to account for color. There are only three
charged leptons of the Standard Model, each with 1 unit of charge. Leptons carry no color,
so the sum of their charges is X
Q2i = 3 . (629)
i leptons

It then follows that the β-function of the fine structure constant is

∂α 16 α2
Q = βSM (α) = . (630)
∂Q 3 π

93
The expression for the running fine structure constant is then
α(Q0 )
α(Q) = . (631)
1− 16
3π
α(Q0 ) log QQ0

We can find the Landau pole in the exact same way as earlier. Again, setting the fiducial
energy scale Q0 = me , the Landau pole now occurs at
3π
Q = me e 16α(me ) ' 6 × 1031 GeV . (632)

This energy corresponds to a mass of about 104 kg.

8.8 (d)
The only charged fermion of the Standard Model with mass greater than the bottom quark
is the top quark. So, to determine the β-function of the fine structure constant above the
bottom quark threshold but below the top quark, we need to re-calculate the sum over
fermion charges. The top quark has electric charge of 2/3, so the relevant sum over the
quark charges is now
 
X
2 1 4 11
Qi = 3 3 × + 2 × = . (633)
9 9 3
i quarks\top

The sum over lepton charges is the same as earlier so the β-function is now
∂α 40 α2
Q = βb (α) = . (634)
∂Q 9 π
To compare the slope to the plot, we need to take the derivative with respect to Q2 . Note
that
∂α ∂α 1 20 α2
Q2 2 = = β b (α) = . (635)
∂Q ∂ log Q2 2 9 π
Over the range in the plot, the value of the coupling doesn’t change more than a few percent,
and is approximately α ' 0.0077. The slope, as determined by the β-function is then
approximately
20 α2
' 4.2 × 10−5 . (636)
9 π
From the plot the value of α changed from about 0.00755 at Q2 = 100 GeV2 to about 0.00775
at Q2 = 10000 GeV2 . The change in the logarithms of Q2 over this range is then
104
log = 2 log 10 ' 4.6 . (637)
102
From the plot, the slope of the value of the fine structure constant is then about
0.00775 − 0.00775
slope ' ' 4.3 × 10−5 , (638)
4.6
94
close enough to our prediction!
The data point on the plot of Fig. 8.5 corresponds to Q ' 55 GeV, which is why the final
question asks about this value. The running coupling assuming only the electron contributes
to the running of α is (from above)
α(me )
α(Q) = . (639)
1− 2
3π
α(me ) log mQe
Plugging in the value Q = 55 GeV and α(me ) = 1/137, the value of the running coupling
would be:
α(55 GeV) ' 0.0074 . (640)
This is significantly lower than the measured value and explicitly demonstrates that all the
relevant Standard Model charged fermions contribute to the running of α.

9 Parton Evolution and Jets

9.1 Dilation Operator
9.1 (a)
The commutator of the dilation and momentum operators is
[D̂, P̂µ ] = [−ixν ∂ν , −i∂µ ] = −xν ∂ν ∂µ + ∂µ (xν ∂ν ) (641)
= −xν ∂ν ∂µ + (∂µ xν )∂ν + xν ∂µ ∂ν = ∂µ .
In the final equation, we used the derivative rule
∂µ xν = δµν , (642)
the spacetime Kronecker δ. Up to a factor of i, the partial derivative is the momentum
operator P̂µ , so the commutation relation of the dilation and momentum operators is
[D̂, P̂µ ] = iPµ . (643)

9.1 (b)
Let’s act the dilation operator on the function
f (x) = (x · x)∆/2 . (644)
We then have
∆ µ ∆
D̂f (x) = −ixµ ∂µ (ηρσ xρ xσ )∆/2 = −i x (x · x) 2 −1 ηρσ ∂µ (xρ xσ ) (645)
2
∆ µ ∆ ∆
= −i x (x · x) 2 −1 ηρσ (δµρ xσ + δµσ xρ ) = −i∆xµ (x · x) 2 −1 xµ
2
= −i∆(x · x)∆/2 = −i∆f (x) .
Therefore, the eigenvalue of the dilation operator is −i∆. ∆ is a measure of the mass
dimension of the eigenfunction f (x).

95
9.2 Expansion of Differential Cross Section for Thrust
9.2 (a)
From example 7.2, the differential cross section for thrust at O(αs ) is
 2
6τ − 6τ + 4 2τ − 1 3(3τ − 2)(2 − τ )

1 dσ αs
= CF log − . (646)
σ0 dτ 2π τ (1 − τ ) 1−τ 1−τ
In this expression, we ignore the phase space constraints. Let’s expand this result in the
limit in which τ → 1. To do this, we will just set τ = 1 wherever the result does not diverge.
We then find:
1 dσ τ →1 6−6+4 2 − 1 3(3 − 2)(2 − 1)
 
αs
= CF log − (647)
σ0 dτ 2π 1−τ 1−τ 1−τ
 
αs 4 1 3
= CF log − .
2π 1−τ 1−τ 1−τ
The remaining terms indeed diverge as τ → 1, and the logarithmic term diverges most
rapidly. Keeping only this term, we then have
1 dσ τ →1 2αs log(1 − τ )
=− CF . (648)
σ0 dτ π 1−τ

9.2 (b)
Now, the result we found from calculating the Sudakov form factor is
αs CF log(4(1 − τ )) h α
s 2
i
P (τ ) = −2 exp − CF log (4(1 − τ )) . (649)
π 1−τ π
Expanding this to lowest order in αs just means we ignore the Sudakov form factor:
αs CF log(4(1 − τ ))
P (τ )αs →0 = −2 . (650)
π 1−τ

9.2 (c)
Comparing the τ → 1 and αs → 0 limits of the fixed-order and resummed cross sections, we
see that they are almost the same. However, our resummed cross section consists of a term
that is less divergent as τ → 1. Note that
αs CF log(4(1 − τ )) αs CF log(1 − τ ) + log 4
P (τ )αs →0 = −2 = −2 , (651)
π 1−τ π 1−τ
and so the most divergent term in the lowest-order expression of the resummed cross section
is
αs CF log(1 − τ )
P (τ )αs →0 = −2 . (652)
π 1−τ
This is indeed identical to the expansion of the fixed-order cross section.

96
9.3 Jet Multiplicity
9.3 (a)
The phase space restriction that ALEPH imposed to identify jet multiplicity is
2
2 min[Eq2 , Eg2 ](1 − cos θ) > ycut Ecm , (653)
where we assume that this tests the relationship between a quark q and a gluon g. If the
quark has energy fraction z = xq , then its energy is
xq z
Eq = Ecm = Ecm . (654)
2 2
Correspondingly, the energy of the gluon in the collinear limit θ → 0 is
1−z
Eg = Ecm . (655)
2
Then, in the soft gluon z → 1 and collinear limit, the phase space restriction becomes:
(1 − z)2 Ecm
2
θ2 2
> ycut Ecm , (656)
2 2
where we note that 1 − z < z as z → 1. That is,
(1 − z)2 θ2
> ycut . (657)
4

9.3 (b)
The region defined by the relationship
(1 − z)2 θ2
> ycut , (658)
4
is a triangle in the Lund plane. This triangle has a right angle at the origin, where z = 0 and
θ2 = 1. The length of the base of the triangle can be determined by turning the inequality
into an equality and setting z = 0. That is,
θ2
= ycut , (659)
4
or that
1 1
log 2
= log . (660)
θ 4ycut
The height of the triangle can be found through a similar procedure, setting θ2 = 1:
(1 − z)2
= ycut , (661)
4
or that
1 1 1
log = log . (662)
1−z 2 4ycut

97
9.3 (c)
From this identified triangle, its area is the product of the base times the height, divided by
2:
1 1 1 1
area = log 2 log = log2 (4ycut ) . (663)
2 θ 1−z 4
To find the probability that the final state has two jets in it, we must forbid emissions in the
area of this triangle, which is accomplished by exponentiating the area times the factor
2αs CF
− . (664)
π
Therefore, the probability that there are no emissions in the triangle, or equivalently, the
probability that there are only two jets in the final state is
 
αs CF 2
P2 (ycut ) = exp − log (4ycut ) , (665)
π 2

as expected.

9.3 (d)
A plot of the distribution of P2 (ycut ) is shown in Fig. 2. To make this plot, we set αs = 0.1,
and we see rather remarkable agreement with the data plot of Fig. 9.6, down to about ycut &
10−3 . Below that point, the data distribution vanishes more quickly than the prediction.
With our prediction in hand, we can immediately identify the value of ycut below which the
cross section for two final state jets is not a good approximation for the cross section of
the process e+ e− → hadrons. When the exponent of P2 (ycut ) is order-1, there is significant
suppression of the dijet cross section. That is, when
αs CF
1' log2 (4ycut ) , (666)
π 2
or correspondingly  r 
1 2π
ycut ' exp − ' 2.6 × 10−4 , (667)
4 αs CF
the dijet cross section is definitely not a good approximation of the total cross section.

9.3 (e)
To calculate the probability for three, four, or more jets in the final state we can continue
the process that we started. Note that in the Lund plane, all emissions are independent and
identically distributed, so filling the Lund plane is a Poisson process. That is, the probability
distribution, to the approximation that we are working here, of the number of jets in the

98
 ���

���

�� (���� ) ���

���

���

��� -�
�� ��-� ����� ����� ����� �
����

Figure 2: Distribution of the probability for two jets in the final state, as a function of ycut .

final state is just given by a Poisson distribution. The exponent factor of the probability for
two jets in the final state is the mean of the Poisson distribution,
αs CF
λ= log2 (4ycut ) . (668)
π 2
Then, the probability for the final state to have 2 + n jets, where n ≥ 0 is
αs CF 2

n  
π 2
log (4y cut ) αs CF 2
P2+n (ycut ) = exp − log (4ycut ) . (669)
n! π 2

Note that the sum over all n is the total probability which is indeed unity:
∞ ∞
n
αs CF
log2 (4ycut )
 
X X
π 2 αs CF 2
1= P2+n (ycut ) = exp − log (4ycut ) . (670)
n=0 n=0
n! π 2

99
9.4 Properties of the DGLAP Equation
The integral of the difference of up and anti-up quark parton distribution functions is 2:
Z 1
dx(fu (x, Q2 ) − fū (x, Q2 )) = 2 . (671)
0
This is supposed to be independent of Q2 , so we can take a derivative of both sides:
Z 1 Z 1  2 2

2 d 2 2 2 dfu (x, Q ) 2 dfū (x, Q )
0=Q dx(fu (x, Q ) − fū (x, Q )) = dx Q −Q . (672)
dQ2 0 0 dQ2 dQ2
Now, in the expression on the right, we can replace the derivatives of the parton distribution
functions by the results of the DGLAP equation:
Z 1  2 2

2 dfu (x, Q ) 2 dfū (x, Q )
dx Q −Q (673)
0 dQ2 dQ2
Z 1
αs 1 dz
Z x 
fu (z, Q2 ) − fū (z, Q2 ) .

= dx Pqg←q
0 2π x z z
Now, let’s make the change of variables from x to
x
y≡ , (674)
z
so that the integrals become
Z 1
αs 1 dz
Z x 
fu (z, Q2 ) − fū (z, Q2 )

dx Pqg←q (675)
0 2π x z z
Z 1
αs 1  αs 1
Z Z
2 2

= dy dz Pqg←q (y) fu (z, Q ) − fū (z, Q ) = dy Pqg←q (y) .
0 2π 0 π 0
In the final equality, we have used that the integral over the difference of parton distributions
is 2. Therefore, for this normalization to be independent of Q2 , we must have that
Z 1
dx Pqg←q (x) = 0 . (676)
0

9.5 Resummation of Q2 with DGLAP

9.5 (a)
The derivative of the explicit sum form of the parton distribution function is
2 ∞
2 dfq (x, Q ) dfq (x, Q2 ) d X
2 n Q
2
Q = 2 = Q2
c n (x, Q0 ) log (677)
dQ2 d log Q d log Q Q20
Q2 0
2 n=0
0
∞ ∞
X d nQ2 X n−1 Q
2
= cn (x, Q20 ) 2 log = n cn (x, Q2
0 ) log
d log Q Q20 Q20
n=0 Q2
0
n=0
∞
X Q2
= (n + 1)cn+1 (x, Q20 ) logn .
n=0
Q20

100
On the other side, the integral of the parton distribution in this form is
∞
αs 1 dz αs 1 dz Q2
Z x Z x X
Pqg←q 2
fq (z, Q0 ) = Pqg←q cn (z, Q20 ) logn 2 (678)
2π x z z 2π x z z n=0 Q0
∞ 2 Z 1
n Q αs dz
X x
= log 2
Pqg←q cn (z, Q20 ) .
n=0
Q0 2π x z z

Now, we can match these expressions order-by-order in n to derive the recursion relation:
1 αs 1 dz
Z x
2
cn+1 (x, Q0 ) = Pqg←q cn (z, Q20 ) . (679)
n + 1 2π x z z

9.5 (b)
With the identification that
c0 (x, Q20 ) = fq (x, Q20 ) , (680)
the n = 1 coefficient is
Z 1
αs dz x
c1 (x, Q20 ) = Pqg←q fq (z, Q20 ) . (681)
2π x z z
Then, the n = 2 coefficient is
1  αs 2 1 dz1
Z  Z 1  
2 x dz2 z1
c2 (x, Q0 ) = Pqg←q Pqg←q fq (z2 , Q20 ) . (682)
2 2π x z1 z1 z1 z2 z2
We can continue this process, iteratively determining the coefficients. At the nth level, this
coefficient is
1  αs n 1 dz1
Z   Z 1  
2 x dzn zn−1
cn (x, Q0 ) = Pqg←q ··· Pqg←q fq (zn , Q20 ) . (683)
n! 2π x z1 z1 zn−1 zn zn

9.5 (c)
Now, let’s Mellin transform the DGLAP equation. The derivative side is trivial as the Mellin
integral has no dependence on the energy Q2 . The integral side of the DGLAP equations
becomes
Z 1 Z 1 Z 1 Z 1
N αs dz x
2 N αs
dx x Pqg←q fq (z, Q ) = dy y dz z N Pqg←q (y) fq (z, Q2 ) .
0 2π x z z 0 2π 0
(684)
On the right, we have made the change of variables from x to y = x/z. In this form, we can
integrate over z to get the Mellin-transform of the parton distribution function. Then, the
DGLAP equations can be written in Mellin space as
˜ 2
 Z 1 
2 dfq (N, Q ) αs
Q 2
= dx x Pqg←q (x) f˜q (N, Q2 ) .
N
(685)
dQ 2π 0

101
9.5 (d)
Finally, in Mellin space, the solution to the DGLAP equations is very simple. The claim is
that the solution is
 Z 1  2

αs Q
f˜q (N, Q ) = f˜q (N, Q0 ) exp
2 2 N
dx x Pqg←q (x) log 2 . (686)
2π 0 Q0

Let’s take a derivative of this expression. Note that

d d
Q2 = 2 , (687)
dQ2 d log Q
Q2
0

and so

df˜q (N, Q2 ) df˜q (N, Q2 ) αs 1 Q2

Z  
d
Q2 = 2 = f˜q (N, Q20 )Q2
exp N
dx x Pqg←q (x) log 2
dQ2 d log Q 2 d log 2
2π 0 Q0
Q0 Q0
 Z 1  Z 1
Q2
  
αs N ˜ 2 αs N
= dx x Pqg←q (x) fq (N, Q0 ) exp dx x Pqg←q (x) log 2
2π 0 2π 0 Q0
 Z 1 
αs
= dx xN Pqg←q (x) f˜q (N, Q2 ) . (688)
2π 0

This is exactly the DGLAP equation in Mellin space.

9.6 Jet Mass at the LHC

9.6 (a)
In the (p⊥ , η, φ) coordinates, a massless four-momentum can be expressed as

p = p⊥ (cosh η, cos φ, sin φ, sinh η) . (689)

Now, taking the dot product of two vectors p1 and p2 in this representation, we have

2p1 · p2 = 2p⊥1 p⊥2 (cosh η1 cosh η2 − cos φ1 cos φ2 − sin φ1 sin φ2 − sinh η1 sinh η2 ) (690)
= 2p⊥1 p⊥2 (cosh(η1 − η2 ) − cos(φ1 − φ2 )) .

To write the second line, we have used (hyperbolic) trigonometric angle difference formulas.

9.6 (b)
Now, let’s Taylor expand this result in the small-angle limit. To lowest order, the hyperbolic
function expands as
(η1 − η2 )2
cosh(η1 − η2 ) = 1 + + ··· . (691)
2

102
Cosine, on the other hand, expands as

(φ1 − φ2 )2
cos(φ1 − φ2 ) = 1 − + ··· . (692)
2
Then, to lowest non-trivial order in the expression of the dot product, the pseudorapidity
and azimuthal angle dependence appears as

m2J = 2p1 · p2 = 2p⊥1 p⊥2 (cosh(η1 − η2 ) − cos(φ1 − φ2 )) (693)

(η1 − η2 )2 (φ1 − φ2 )2
  
= p⊥1 p⊥2 (η1 − η2 )2 + (φ1 − φ2 )2
 
' 2p⊥1 p⊥2 1 + − 1−
2 2
2
= p⊥1 p⊥2 R .

On the final line we note that

R2 = (η1 − η2 )2 + (φ1 − φ2 )2 . (694)

9.6 (c)
As a further simplification, it is useful in the small-angle limit to express the transverse
momentum of each particle as a fraction of the total jet transverse momentum p⊥J . Assuming
that p⊥1 = zp⊥J and so p⊥2 = (1 − z)p⊥J , the expression for the jet mass becomes

m2J = z(1 − z)R2 p2⊥J . (695)

9.6 (d)
With this result, we can then predict the mean square jet mass hm2J i, using the quark collinear
splitting function. Inserting the various pieces into the expression from the problem, the
mean square jet mass is
R02 1
d2 σ
Z Z
hm2J i = dθ 2
m2 (z, θ)
dz (696)
0 0 dz dθ2 J
Z R02 2 Z 1
αs dθ 1 + z2
= CF dz z(1 − z)θ2 p2⊥J
2π 0 θ 2
0 1 − z
Z R02 Z 1
αs 2 2
= CF p⊥J dθ dz (1 + z 2 )z
2π 0 0
3 αs 2 2
= CF p⊥J R0 .
4 2π

9.6 (e)
Now, let’s plug in some numbers and compare this prediction to data. To compare to the
ATLAS result in Fig. 9.7, we set the jet radius R0 = 1.2 and 300 GeV < p⊥J < 400 GeV.

103
From our expression above, there’s no range of p⊥J that can be used, so we’ll just pick the
middle value, and say that p⊥J = 350 GeV. For approximate comparison to data, this is
perfectly sufficient. Further, we assume that the value of the strong coupling αs ' 0.1.
Plugging in these numbers, the mean square jet mass is approximately
3 αs
hm2J i = CF p2⊥J R02 ' 2.8 × 103 GeV2 . (697)
4 2π
The root-mean square mass is the square-root of this:
q
hm2J i ' 53 GeV. (698)

A mass of 53 GeV is near the peak of the distribution, so it’s not a bad approximation.
However, the root-mean square of any probability distribution is never less than the mean,
so we expect the root-mean square of the data to be closer to 100 or 120 GeV. This is
explained by several factors that we have ignored in our simple analysis. The largest of
which is that a significant fraction of the jets in the data sample will have been initiated by
a gluon, rather than a quark. Because gluons carry more color than quarks, they exhibit a
greater probability for emitting particles and correspondingly gluon jets are more massive
than quark jets. Including the effect of gluon jets in our analysis would increase our predicted
root-mean square mass to be closer to the value from the data.

9.7 Underlying Event at Hadron Colliders

9.7 (a)
Our first step in analyzing underlying event is to change variables in the splitting probability
Pq (z, θ2 ) to coordinates in a hadron collider. The splitting probability in the soft gluon and
collinear limit is
2 αs dz dθ2
Pq (z, θ ) = CF . (699)
π 1 − z θ2
The initial quark pulled out of the proton has energy
x
E= Ecm , (700)
2
and so the energy of the quark after the splitting Eq is
zx
Eq = zE = Ecm . (701)
2
In terms of the energy Eq and the angle θq of the quark off of the ẑ axis, the transverse
momentum of the quark p⊥q is
zx
p⊥q = Eq sin θq ' Ecm θq , (702)
2

104
in the small-angle limit. By conservation of transverse momentum, the magnitude of the
transverse momentum of the gluon must be equal to that of the quark:
(1 − z)x
p⊥g ' Ecm θg = p⊥q . (703)
2
In this expression, we used the gluon energy fraction 1 − z and the angle of the gluon from
the ẑ axis, θg . The angle θ in the splitting probability is the angle between the quark and
the gluon after splitting which is necessarily the sum of θq and θg :

θ = θq + θg . (704)

For the equality p⊥q = p⊥g to hold, this the requires that

θg = zθ . (705)

Therefore, the transverse momentum of the gluon in the small-angle limit is

x
p⊥g ' z(1 − z)θ Ecm . (706)
2
Further simplifying to the soft limit, in which z → 1, the gluon transverse momentum is
x
p⊥g ' (1 − z)θ Ecm . (707)
2
The pseudorapidity of the gluon, ηg , is then
θg zθ
ηg = − log tan ' − log , (708)
2 2
in the small angle limit. Additionally, in the soft limit, z → 1, the pseudorapidity is exclu-
sively a function of θ:
θ
ηg ' − log . (709)
2
Solving for θ2 , we find
θ2 = 4e−2ηg . (710)
Note that the Jacobian of this change of variables (taking absolute values) is

dθ2 = 8e−2ηg dηg , (711)

or that
dθ2
= 2 dηg . (712)
θ2
Continuing, plugging in the expression for rapidity into the expression for the gluon
transverse momentum, we have
x
p⊥g = 2(1 − z)e−ηg Ecm . (713)
2
105
The Jacobian relating p⊥g to energy fraction z is then
x
dp⊥g = dz 2e−ηg Ecm . (714)
2
Plugging this and the expression for 1 − z into the splitting probability we have
dz dp⊥g
= . (715)
1−z p⊥g
Then, putting all these pieces together, the splitting probability Pq (p⊥ , η) is
αs dp⊥
Pq (p⊥ , η) = 2 CF dη . (716)
π p⊥
Here, we have dropped the subscripts denoting that the transverse momentum p⊥ and pseu-
dorapidity η are for the gluon.

9.7 (b)
This probability distribution is independent of pseudorapidity. This means that underlying
event radiation in your detector will have a very flat, uniform distribution in η. This also
demonstrates why η is a useful coordinate at a hadron collider. Underlying event is always
there in your proton collisions. More interesting things like jet production will exhibit a
distribution in η that is sharply peaked at particular values of η. Therefore, by naturally using
pseudorapidity to express the data, any deviations from uniformity is, to first approximation,
interesting physics.

9.7 (c)
From earlier, the full expression for the pseudorapidity is
θ
η = − log tan , (717)
2
so the angle θ is
θ = 2 tan−1 e−η . (718)
Plugging in η = 4.5, we find an angle of

θ = 2 tan−1 e−4.5 ' 0.022 rad ' 1.3◦ . (719)

9.7 (d)
The data presented in Fig. 9.8 of the pseudorapidity distribution of the underlying event
are approximately consistent with our prediction that the underlying event is uniform in
pseudorapidity. Over the entire plotted range, the data only deviate by about 25%, and over
the restricted range |η| ∈ [0, 3] (down to about 5.7◦ above the beam), the deviation from
uniformity is much smaller, at most about 10%.

106
9.8 Jet Event Display
This event display shows three prominent jets, one of which has significantly lower transverse
momentum than the other two. The lowest p⊥ jet is located at

(η, φ) ' (−0.5, 75◦ ) , (720)

and has a transverse momentum of approximately 30 GeV, when the various calorimeter
cells are summed over. The next jet is located at

(η, φ) ' (−1.5, 250◦ ) , (721)

and has a transverse momentum of approximately 200 GeV, when the various calorimeter
cells are summed over. Finally, the last prominent jet is located at

(η, φ) ' (2, 60◦ ) , (722)

and has a transverse momentum of nearly 250 GeV.

10 Parity Violation
10.1 Time Reversal of Spinors
The time reversal operator T flips the direction of time t → −t and complex conjugates
i → −i. For the derivative operator iσ · ∂, this time reversal transforms as

T[iσ · ∂] = T[iI∂0 − iσ1 ∂1 − iσ2 ∂2 − iσ3 ∂3 ] (723)

= iI∂0 + iσ1 ∂1 − iσ2 ∂2 + iσ3 ∂3 .

In writing the second line, we note that

∂
∂0 = , (724)
∂t
and that  
0 −i
σ2 = , (725)
i 0
while the other σ matrices have only real number elements. Then, the time reversal trans-
formation of the Dirac equation is

T[iσ · ∂ψR ] = (iI∂0 + iσ1 ∂1 − iσ2 ∂2 + iσ3 ∂3 )T[ψR ] = 0 . (726)

Now, writing the solution to the time-reversed Dirac equation as

T[ψR ] = ũT[e−ip·x ] = ũe−iEt−i~p·~x , (727)

107
for some spinor ũ and a massless four-vector p. In this expression, we time-reversed the
exponential phase factor. We can then take derivatives and re-express the Dirac equation
as:

T[iσ · ∂ψR ] = (IE + σ1 E sin θ cos φ − σ2 E sin θ sin φ + σ3 E cos θ) ũ = 0 (728)

 
1 + cos θ eiφ sin θ
=E ũ
e−iφ sin θ 1 − cos θ
cos2 2θ eiφ sin 2θ cos 2θ
 
= 2E ũ .
e−iφ sin 2θ cos 2θ sin2 2θ

This then enforces the following relationship between the two entries of the spinor ũ:
θ θ
cos ũ1 + eiφ sin ũ2 = 0 . (729)
2 2
Properly normalizing the spinor, this can be satisfied by
√ eiφ/2 sin 2θ
 
ũ = 2E . (730)
−e−iφ/2 cos 2θ

Note that this spinor is neither right-handed nor left-handed, but is simply related to the
right-handed spinor as:

0 1 √
 −iφ/2
cos 2θ
  
e
ũ = 2E = uR , (731)
−1 0 eiφ/2 sin 2θ

as expected.

10.2 Charge Conjugation of Spinors

10.2 (a)
Under charge conjugation, the electric and magnetic fields transformed as:
~ = −E
C[E] ~, ~ = −B
C[B] ~, (732)

because the action of charge conjugation negates electric charges. In terms of the vector
potential Aµ , the electric and magnetic fields are (from Chap. 2):

~
E ~ − ∂A ,
~ = −∇V ~ =∇
B ~ ×A
~. (733)
∂t
~ and ∂/∂t are unchanged by charge conjugation, if the electric and
Because the derivatives ∇
magnetic fields are negated under charge conjugation, then so too must the vector potential:

C[Aµ ] = −Aµ . (734)

108
10.2 (b)
Now, let’s act the charge conjugation operator on the Dirac equation coupled to electromag-
netism:
(iσ · ∂ − eσ · A)ψR = 0 . (735)
As discussed earlier, the charge conjugation operator passes through derivatives, Pauli ma-
trices, and numbers (the fundamental unit of charge e), and so only acts on the fields:
C[(iσ · ∂ − eσ · A)ψR ] = (iσ · ∂ − eσ µ C[Aµ ])C[ψR ] = 0 . (736)
We had just argued that charge conjugation negated the vector potential, so the Dirac
equation becomes:
(iσ · ∂ − eσ µ C[Aµ ])C[ψR ] = (iσ · ∂ + eσ · A)C[ψR ] = 0 . (737)
The only thing that had effectively changed in this Dirac equation is the sign of the electric
charge: −e → +e. That is, the solution to this new Dirac equation C[ψR ] does indeed
describe an anti-particle with electric charge +e, with all other properties of the spinor
unchanged.

10.3 CPT on Spinors

Let’s determine the complete action of charge conjugation, parity, and time reversal on the
spinor uR . First, from Tab. 10.1, the action of time reversal on the spinor is
T[uR ] = uR , (738)
where  is the two-index anti-symmetric object. Next, the action of parity transforms uR →
uL , and it doesn’t act on . Combining T and P, we have
PT[uR ] = P[uR ] = P[uR ] = uL . (739)
Finally, as long as there are no charges around, the charge conjugation operator is just
effectively the identity on a two-component spinor. That is, adding charge conjugation
transforms uL → vR , but these two spinors are identical with no electric charge. Putting it
all together, the combined CPT transformation of a spinor is
CPT[uR ] = C[uL ] = C[uL ] = vR . (740)
To connect this directly to the expression given in the exercise, we need to evaluate this
on the explicit spinor solutions. Recall that the spinor vR in spherical coordinates is
√
 −iφ/2
sin 2θ

e
vR (p) = 2E . (741)
−eiφ/2 cos 2θ
Acting with  on this spinor, we have
0 1 √ √
 −iφ/2
sin 2θ −eiφ/2 cos 2θ
    
e
vR = 2E = 2E = −u∗R , (742)
−1 0 −eiφ/2 cos 2θ −e−iφ/2 sin 2θ
as claimed.

109
10.4 C, P, T in Electromagnetism
10.4 (a)
In this exercise, we’re not asking for a proof that the scattering amplitude M(γγ → γγγ)
vanishes, just a plausibility argument. In the calculation of the matrix element, we need
wavefunctions for each of the external photons. These wavefunctions are just free solutions
to Maxwell’s equations for the vector potential Aµ with given momenta. Then, under charge
conjugation, because the vector potential is negated, these external wavefunctions are simi-
larly negated. Because the scattering process we are considering consists of an odd number
of external photons, the matrix element is correspondingly negated:
C[M(γγ → γγγ)] = −M(γγ → γγγ) . (743)
From another perspective, however, the scattering process γγ → γγγ has no external charged
particles so there’s in some sense nothing to charge conjugate. This implies that the charge
conjugate matrix element must be identical to the original matrix element:
C[M(γγ → γγγ)] = M(γγ → γγγ) . (744)
However, if the matrix element is equal to its opposite, the only possibility is if it simply
vanishes:
M(γγ → γγγ) = 0 . (745)

10.4 (b)
Parity was defined to negate the electric field (as a vector) and leave the magnetic field
unchanged (as a pseudovector). Using the expressions for the electric and magnetic fields in
terms of the vector potential, this is
" #
∂ ~
A ~
∂P[A]
~ = P −∇A
P[E] ~ 0− ~
= ∇P[A 0] − = −E~, (746)
∂t ∂t
h i
~ ~ ~
P[B] = P ∇ × A = −∇ ~ × P[A]
~ =B ~.

For these transformations to be consistent, we must require that

P[A0 ] = A0 , ~ = −A
P[A] ~. (747)

10.4 (c)
Earlier, we had shown that under time reversal, the electric field is unchanged, while the
magnetic field is negated. So, using this requirement, we do the same exercise as in the
previous part to determine the time reversal transformations of the vector potential:
" #
∂ A~ ∂T[A]~
~ = T −∇A
T[E] ~ 0− ~
= −∇T[A 0] + =E ~, (748)
∂t ∂t
h i
~ =T ∇
T[B] ~ ×A~ =∇ ~ × T[A]
~ = −B ~.

110
 spin spin
=) =)
⌫µ µ ⌫¯e

spin spin
=)
=)
e µ ⌫¯e
⌫µ
spin
)=

Figure 3: Illustration of the decay of the muon at the kinematic endpoint at which the energy
of the electron is half the mass of the muon. The direction of momentum of the particles is
identified as well as their direction of spin.

For these transformations to be consistent, we must require that

T[A0 ] = A0 , ~ = −A
T[A] ~. (749)

10.4 (d)
Putting this all together, we first note that under the combined action of PT, the vector
potential is unchanged:
PT[Aµ ] = Aµ . (750)
Therefore, the combined action of CPT on the vector potential is equivalent to just the action
of charge conjugation:
CPT[Aµ ] = C[Aµ ] = −Aµ , (751)
as stated.

10.5 Electron Spin in Muon Decay

In the decay of the muon, the energy of the electron from decay is maximized if the momenta
of the neutrino and anti-neutrino are collinear and opposite in direction to the electron. In
that case, then, by momentum conservation, the energy of the electron is half of the mass
of the muon (assuming the electron is massless). The neutrino must be left-handed and
the anti-neutrino must be right-handed, by the structure of the weak decay of the muon.
Because their momenta are in exactly the same direction, their spins are perfectly opposite,
so the muon neutrino and the electron anti-neutrino carry a net angular momentum of 0.
Therefore, the total angular momentum of the final state is carried by the spin of the electron
exclusively. By angular momentum conservation, the spin of the initial muon must therefore
be identical to the spin of the final electron. Fig. 3 shows the configuration of final state
particles in the muon decay, with a left-handed electron, as required by the weak interaction.

111
10.6 Endpoint of Electron Energy in Muon Decays
10.6 (a)
The propagator for the electron in the emission of a photon will have a denominator that
consists of a four-vector dot product of electron and photon momentum, pe · pγ . Without
loss of generality, we can align the momentum of the electron along the +ẑ axis and so it
can be expressed as  
p
pe = Ee , 0, 0, Ee − me .
2 2 (752)

Note that this momentum indeed describes a massive, on-shell electron: p2e = m2e . The
photon is massless, and its momentum is some angle θ with respect to the +ẑ axis:

pγ = Eγ (1, 0, sin θ, cos θ) . (753)

Then, the dot product of these momenta is

s !
p m2
pe · pγ = Ee Eγ − Ee2 − m2e Eγ cos θ = Ee Eγ 1− 1 − 2e cos θ . (754)
Ee

10.6 (b)
Assuming that the mass of the electron is small compared to its energy and the angle of
photon emission θ is also small, we can Taylor expand this expression. Note that
s
m2 m2
1 − 2e = 1 − e2 + · · · , (755)
Ee 2Ee

and
θ2
cos θ = 1 − + ··· . (756)
2
Then, the lowest-order Taylor expansion in these limits of the four-momentum dot product
is
s !
m2e θ2
   
m2e
pe · pγ = Ee Eγ 1 − 1 − 2 cos θ ' Ee Eγ 1 − 1 − 1− (757)
Ee 2Ee2 2
 2
θ2

me
' Ee Eγ + .
2Ee2 2
Therefore, if both the photon splitting angle and the mass of the electron are relevant in this
expression on the second line, then we have
me
θc = , (758)
Ee
the angle at which the non-zero electron mass is important.

112
10.6 (c)
Let’s now calculate the distribution for the emitted photon energy in the soft and collinear
limit, p(Eγ ). We have
Z 1 Z 1
p(Eγ ) = dz dθ2 p(z, θ2 ) δ(Eγ − zEe ) (759)
0 θc2
1 1 1
dθ2
Z Z Z
α dz α 1 dz
= 2
δ(Eγ − zEe ) = log 2 δ(Eγ − zEe )
π 0 z θc2 θ π θc 0 z
1 Ee2
α log θc2 α log m2e
= = .
π Eγ π Eγ

In getting to the third line, we had to manipulate the δ-function into a form that could be
integrated explicitly. This required a Jacobian:
Z 1 Z 1  
dz 1 dz Eγ 1
δ(Eγ − zEe ) = δ −z = . (760)
0 z Ee 0 z Ee Eγ

10.6 (d)
Now, we can integrate over this distribution to determine the mean emitted photon energy
hEγ i. To do the integral, we need to determine the bounds of integration. First, the minimal
photon energy is 0; the photon can possibly take away no energy at all. The largest natural
energy that the photon can take is just the total energy of the electron. This may seem a
bit unnatural because we are working in the soft and collinear photon emission limit, but by
dimensional analysis, it’s the only possible energy scale. We assume that the electron mass
is much smaller than its energy me  Ee , so me does not provide a relevant energy scale for
the photon. The only energy scale around is the energy of the electron and by conservation
of energy, the photon cannot have energy larger than Ee . Then, the mean photon energy is
Ee2
Ee Ee
α log m2e
Z Z
hEγ i = dEγ p(Eγ ) Eγ = dEγ Eγ (761)
0 0 π Eγ
α E2
= Ee log e2 .
π me

10.6 (e)
Now, plugging in numbers, we can estimate the amount of energy carried away by emitted
photons. We have
1
α Ee 30
hEγ i = 2 Ee log ' 2 137 30 log MeV ' 0.6 MeV . (762)
π me π 0.511

113
In Fig. 10.2, the data are plotted at about every MeV in electron energy, and so the difference
between the data and our prediction in the range 20 < Ee . 40 MeV is no more than about
one MeV. If the electron energy that we predicted in the chapter was reduced by about 0.6
MeV in this energy range, we would find much better agreement between our prediction and
data. Indeed, this difference seems to largely be accounted for once we include the effect of
photon radiation from the electron.

10.7 Kinematics of the IceCube Experiment

10.7 (a)
With the proton at rest and the initial muon anti-neutrino momentum aligned along the +ẑ
axis, their momenta can be expressed as

pp = (mN , 0, 0, 0) , (763)
pν̄ = Eν̄ (1, 0, 0, 1) . (764)

The four-momentum of the anti-muon, assuming that it is massless, is then

pµ = Eµ (1, 0, sin θ, cos θ) , (765)

where θ is the scattering angle and Eµ is the energy of the anti-muon. By energy and
momentum conservation, the four-momentum of the neutron is then

pn = (mN + Eν̄ − Eµ , 0, −Eµ sin θ, Eν̄ − Eµ cos θ) . (766)

At this stage, however, the momentum of the neutron is not on-shell. Enforcing it to square
to p2n = m2N then enables us to solve for the anti-muon energy, Eµ . Doing the explicit square,
we find

p2n = m2N = (mN + Eν̄ − Eµ )2 − Eµ2 sin2 θ − (Eν̄ − Eµ cos θ)2 (767)
= m2N + 2mN Eν̄ − 2mN Eµ − 2Eν̄ Eµ (1 − cos θ) .

This can then be solved for Eµ and we find

mN Eν̄
Eµ = . (768)
mN + Eν̄ (1 − cos θ)

Finally, the four-momenta of the anti-muon and the neutron are

mN Eν̄
pµ = (1, 0, sin θ, cos θ) , (769)
mN + Eν̄ (1 − cos θ)
 
mN Eν̄ mN Eν̄ sin θ mN Eν̄ cos θ
pn = mN + Eν̄ − , 0, − , Eν̄ − .
mN + Eν̄ (1 − cos θ) mN + Eν̄ (1 − cos θ) mN + Eν̄ (1 − cos θ)

114
10.7 (b)
We can re-arrange the expression for the anti-muon energy and solve for the energy of the
initial anti-neutrino. From
mN Eν̄
Eµ = , (770)
mN + Eν̄ (1 − cos θ)
the anti-neutrino energy is
mN Eµ
Eν̄ = . (771)
mN − Eµ (1 − cos θ)
For a given anti-muon energy Eµ , the smallest that the anti-neutrino energy could have been
is if the scattering angle θ = 0 for which

Eν̄,min = Eµ . (772)

So, if Eµ is observed to be 2 PeV, this is the minimum energy that the anti-neutrino could
have had. There is in fact no upper bound on the anti-neutrino energy; it could have been
infinite. The denominator of the expression for the anti-neutrino vanishes if the scattering
angle is
mN
1 − cos θ = , (773)
Eµ
which is some finite angle.

10.7 (c)
This result then tells us what the minimum and maximum angles between the anti-neutrino
and anti-muon momenta could be. Of course, their momentum could be collinear, with
θmin = 0, which corresponds to the case when their energies are also equal. Even in the case
in which the anti-neutrino has an infinite energy, the maximum scattering angle is

mN θ2
1 − cos θmax = ' max . (774)
Eµ 2

On the right, we have Taylor expanded in the small angle limit because mN  Eµ . With
mN = 1 GeV and Eµ = 2 PeV = 2 × 106 GeV, this angle is
r
mN √
θmax = 2 ' 10−6 rad = 10−3 rad ' 0.057◦ . (775)
Eµ

Thus, even if the anti-neutrino had infinite energy, the anti-muon momentum would still be
within about four arcminutes of the anti-neutrino momentum.

115
10.8 High-Energy Neutrino Cross Sections
10.8 (a)
From the expression for the matrix element given in the exercise, we can use the σ ma-
trix identities derived in the calculation of the muon decay rate calculation to simplify the
expression. We have that
µ
σ̄ab σ̄µ cd = 2ac bd . (776)
Then, the matrix element can be written as
4GF  †  
M(ν̄µ p → µ+ n) = √ vR (pν )σ̄ µ vR (pµ ) u†L (pn )σ̄µ uL (pp ) (777)
2
8GF  †  
=− √ vR (pν )uR (pn ) u†R (pp )vR (pµ ) .
2
Taking the absolute square of this matrix element, we find

|M(ν̄µ p → µ+ n)|2 = 32G2F |vR† (pν )uR (pn )|2 |u†R (pp )vR (pµ )|2 . (778)

In our approximations, we assume that both the anti-neutrino and the anti-muon are
massless, so we can evaluate these spinor products using the techniques derived in the chap-
ter. For both of the spinor products that exist, we contract a massless with a massive spinor.
So, we have the following identities:

|vR† (pν )uR (pn )|2 = pν · pn , |u†R (pp )vR (pµ )|2 = pp · pµ . (779)

Therefore, the absolute squared matrix element is

|M(ν̄µ p → µ+ n)|2 = 32G2F (pν · pn )(pp · pµ ) . (780)

10.8 (b)
Now, if the proton is at rest, then the dot product of the proton and the anti-neutrino
four-momenta is
pp · pµ = mN Eµ . (781)
Next, with momentum conservation, we can re-express the dot product of the anti-neutrino
and neutron momenta as

pν · pn = pν · (pν + pp − pµ ) = pν · pp − pν · pµ (782)
= mN Eν − Eν Eµ (1 − cos θ) . (783)

In the evaluation of this expression, we note that the anti-neutrino is massless, and the scat-
tering angle is the angle between the direction of the anti-neutrino and anti-muon momenta.
Putting this together, the squared matrix element is

|M(ν̄µ p → µ+ n)|2 = 32G2F mN Eµ Eν [mN − Eµ (1 − cos θ)] . (784)

116
10.8 (c)
Two-body phase space relevant for this scattering process is
Z 3
d pµ d3 pn
Z
1
dΠ2 = 3 3
(2π)4 δ (4) (pν + pp − pµ − pn ) . (785)
(2π) (2π) 4Eµ En
Let’s first cancel off factors of 2 and π:
Z Z
1 1
dΠ2 = 2
d3 pµ d3 pn δ (4) (pν + pp − pµ − pn ) . (786)
16π Eµ En
We can eliminate the integrals over the neutron momentum with the three-momentum δ-
functions: Z Z
1 1
dΠ2 = 2
d3 pµ δ(Eν + mN − Eµ − En ) . (787)
16π Eµ En
The only δ-function that remains is that which imposes energy conservation. Using this
δ-function, we can replace the neutron energy in the integrand:
Z Z
1 3 1
dΠ2 = d p µ δ(Eν + mN − Eµ − En ) . (788)
16π 2 Eµ (Eν + mN − Eµ )
Now, let’s expand out the remaining integration measure of the muon. Because we assume
that the muon is massless, note that this enforces Eµ = |~pµ |, so the integration measure is
d3 pµ = Eµ2 dEµ d cos θ dφ , (789)
where θ is the scattering angle and φ is the azimuthal angle. The integrand in spherical
coordinates becomes
Z Z
1 Eµ
dΠ2 = dE µ d cos θ dφ δ(Eν + mN − Eµ − En ) . (790)
16π 2 Eν + mN − Eµ
We can freely integrate over φ ∈ [0, 2π] which yields
Z Z
1 Eµ
dΠ2 = dEµ d cos θ δ(Eν + mN − Eµ − En ) . (791)
8π Eν + mN − Eµ

10.8 (d)
Of the remaining phase space integrals, we do the integral over the scattering angle because
IceCube does not measure this angle. From part (a) of exercise 10.7, the neutron energy is
Eν mN
En = Eν + mN − . (792)
mN + Eν (1 − cos θ)
Plugging this into the energy conservation δ-function, we find
 
Eν mN
δ(Eν + mN − Eµ − En ) = δ − Eµ . (793)
mN + Eν (1 − cos θ)

117
We will rearrange this to solve for cos θ, which then enables us to do the integral over the
scattering angle. First, we solve for cos θ from:
Eν mN
= Eµ , (794)
mN + Eν (1 − cos θ)
which yields
(Eν − Eµ )mN
cos θ = 1 − . (795)
Eµ Eν
Next, we need to find the Jacobian from rearranging the δ-function. Taking the derivative
with respect to cos θ, we find
d Eν mN mN Eν2
J −1 = = (796)
d cos θ mN + Eν (1 − cos θ) (mN + Eν (1 − cos θ))2
Eµ2
= . (797)
mN
On the second line we inserted the expression from Eq. 795 for cos θ which dramatically
simplifies the Jacobian. Putting this all together, the energy conservation δ-function can be
expressed in terms of the scattering angle as
(Eν − Eµ )mN
 
mN
δ(Eν + mN − Eµ − En ) = 2 δ cos θ − 1 + . (798)
Eµ Eµ Eν

10.8 (e)
Now, using this expression for the scattering angle, we can further simplify the phase space
integral to be
Z Z
1 Eµ
dΠ2 = dEµ d cos θ δ(Eν + mN − Eµ − En ) (799)
8π Eν + mN − Eµ
(Eν − Eµ )mN
Z  
1 mN
= dEµ d cos θ δ cos θ − 1 + .
8π Eµ (Eν + mN − Eµ ) Eµ Eν
To connect the phase space integral to the cross section, we use Fermi’s Golden Rule:
Z
+ 1
σ(ν̄µ p → µ n) = dΠ2 |M(ν̄µ p → µ+ n)|2 (800)
4Eν mN
(Eν − Eµ )mN
Z  
1 1 mN
= dEµ d cos θ δ cos θ − 1 +
4Eν mN 8π Eµ (Eν + mN − Eµ ) Eµ Eν
2
× 32GF mN Eµ Eν [mN − Eµ (1 − cos θ)]
G2F mN (Eν − Eµ )mN
Z  
1
= dEµ d cos θ δ cos θ − 1 +
π Eν + mN − Eµ Eµ Eν
× [mN − Eµ (1 − cos θ)] .

118
Note that the velocity difference between the particles in the initial state is 1 (= c), because
the initial proton is at rest. Now, we can insert the expression for cos θ using the δ-function.
Note that
Eµ
mN − Eµ (1 − cos θ) = mN . (801)
Eν
The expression for the cross section finally simplifies to

G2F m2N (Eν − Eµ )mN

Z  
+ Eµ
σ(ν̄µ p → µ n) = dEµ d cos θ δ cos θ − 1 + .
πEν Eν + mN − Eµ Eµ Eν
(802)
The scattering angle can be anything on cos θ ∈ [−1, 1], so we integrate over this domain:
Z 1
(Eν − Eµ )mN (Eν − Eµ )mN (Eν − Eµ )mN
     
d cos θ δ cos θ − 1 + =Θ Θ 2− .
−1 Eµ Eν Eµ Eν Eµ Eν
(803)

These two Θ-functions follow from enforcing that the value of cos θ from the δ-function is
compatible with the domain of integration. These can be simplified to
Z 1
(Eν − Eµ )mN
   
mN Eν
d cos θ δ cos θ − 1 + = Θ(Eν − Eµ )Θ Eµ − . (804)
−1 Eµ Eν 2Eν + mN

Therefore, the cross section is

G2F m2N
Z  
+ Eµ mN Eν
σ(ν̄µ p → µ n) = dEµ Θ(Eν − Eµ )Θ Eµ − . (805)
πEν Eν + mN − Eµ 2Eν + mN

10.8 (f )
With the cross section in this simple form, we can now integrate over the muon energy. Let’s
massage the integral into a form that we’ve seen before. We have

G2F m2N
Z  
+ Eµ mN Eν
σ(ν̄µ p → µ n) = dEµ Θ(Eν − Eµ )Θ Eµ − (806)
πEν Eν + mN − Eµ 2Eν + mN
Eµ
G2F m2N
Z  
Eν +mN mN Eν
= dEµ Θ(Eν − Eµ )Θ Eµ −
πEν 1 − Eν E+m
µ
N
2Eν + mN
G2F m2N
Z    
x Eν mN Eν
= (Eν + mN ) dx Θ −x Θ x− .
πEν 1−x Eν + mN (2Eν + mN )(Eν + mN )

In going from the second to the third line, we made the change of variables
Eµ
x≡ . (807)
Eν + mN

119
In example 7.2, we had shown that
Z
x
dx = − log(1 − x) − x . (808)
1−x
Applied to the problem at hand and enforcing the Θ-functions for the integration bounds
we find
G2F m2N
  
+ Eν Eν
σ(ν̄µ p → µ n) = − (Eν + mN ) log 1 − + (809)
πEν Eν + mN E + mN
  ν 
mN Eν mN Eν
− log 1 − −
(2Eν + mN )(Eν + mN ) (2Eν + mN )(Eν + mN )
2 2 2
   
GF mN Eν + mN 2Eν 2Eν
= log 1 + − ,
π Eν mN (2Eν + mN ) 2Eν + mN

as claimed.

10.8 (g)
To start, we will make the following observation. The number of events per unit time that
are observed is a product of the luminosity and the cross section:

events/time = nν nN Vol cσ , (810)

where nν and nN is the number density of neutrinos and nucleons in IceCube, respectively,
and Vol is the total volume of ice of IceCube. The number density of nucleons times the
volume of IceCube is just the number of nucleons in IceCube, NN , so the event rate is also

events/time = nν NN cσ . (811)

To calculate the total number of events detected, we just multiply both sides of this expression
by the total exposure time texp :

events = nν NN ctexp σ . (812)

Now, the total number of neutrinos Nν is the number density times the total volume of
neutrinos that passes through IceCube during the exposure time. Because neutrinos travel
at the speed of light, the length of this volume ` = ctexp and the cross sectional area is just
the cross sectional area of IceCube, A. Using these results, we can express the total number
of observed events in the simple form of
Nν NN
events = σ. (813)
A
Now, we just need to calculate each of these components.

120
 To estimate the cross section for PeV neutrinos, we will just set Eν = 1 PeV = 106 GeV
and the nucleon mass to mN = 1 GeV. Because the nucleon mass is so much smaller than
the energy of the neutrino, the cross section simplifies to

G2F m2N
 
+ Eν
σ(ν̄µ p → µ n) ' log −1 , (814)
π mN

to leading-order in mN /Eν . The Fermi constant GF is

GF ' 1.17 × 10−5 GeV−2 , (815)

so the value of the cross section for a PeV neutrino is approximately

σ(ν̄µ p → µ+ n) ' 5.6 × 10−10 GeV−2 ' 2.2 × 10−42 m2 . (816)

In this expression, we have used the relationship that 1 GeV corresponds to a distance of
about 0.2 femtometers.
The volume of IceCube is
Vol = 1 km3 = 109 m3 . (817)
The density of ice is about 1000 kg/m3 , and so the total mass of the ice in IceCube is

mice = 103 · 109 kg = 1012 kg = 1015 g . (818)

That is, there are 1015 mol of nucleons in the IceCube detector which is a total of

NN ' 6.02 × 1023 · 1015 ' 6 × 1038 . (819)

The cross sectional area of this volume of neutrinos A is the cross sectional area of
IceCube, which is approximately

A = 1 km2 = 106 m2 . (820)

Plugging in the numbers into the expression for the number of events
Nν NN
events = σ, (821)
A
we have
2.2 × 10−42
3 ' 6 × 1038 · Nν . (822)
106
Solving for the neutrino number Nν , we find

Nν ' 2.3 × 109 , (823)

or a few billion PeV neutrinos.

121
11 The Mass Scales of the Weak Force
11.1 Maxwell with a Massive Photon
11.1 (a)
The Lagrangian for electromagnetism with a massive photon is

1 m2
L = − Fµν F µν − J µ Aµ + Aµ Aµ . (824)
4 2
The Maxwell’s equations arising from the Bianchi identity are unchanged because that is
simply a consequence of the antisymmetry of the field strength tensor and its construction
from partial derivatives of the vector potential. From Chap. 2, the massless equations of
motion of electromagnetism are
∂µ F µν − J ν = 0 . (825)
In the case of the massive photon, all we need to do to include the photon mass is to add
the variation of the mass term to this equation of motion. This can be found by simply
differentiating the Lagrangian with respect to the field Aµ :

δL δ m2
⊃ Aµ Aµ = m2 Aµ . (826)
δAµ δAµ 2

We just add this to the massless equations of motion to find

∂µ F µν − J ν + m2 Aν = 0 . (827)

11.1 (b)
Now, we perform a gauge transformation of the action that follows from this Lagrangian.
We can ignore the term in the action proportional to the square of the field strength tensor,
as that is gauge invariant regardless of the mass of the photon. A gauge transformation of
the two other terms in the action yields:

m2
Z  
4 µ µ
S[Aµ ] ⊃ d x −J Aµ + Aµ A (828)
2
m2
Z  
4 µ µ µ
→ d x −J (Aµ + ∂µ λ) + (Aµ + ∂µ λ)(A + ∂ λ)
2
m2
Z  
4 µ 2 µ µ
= S[Aµ ] + d x λ∂µ J − λm ∂µ A + (∂µ λ)(∂ λ) .
2

In going from the second to third line, we have used integration by parts and eliminated
terms evaluated at the boundary of spacetime. In general, what remains is an absolute

122
mess, and further depends on the gauge transformation parameter λ. Further performing
integration by parts on the term with two λs, this reduces to

m2 2
Z  
4 µ 2 µ
S[Aµ ] → S[Aµ ] + d x λ∂µ J − λm ∂µ A + λ∂ λ . (829)
2

Now, the differential equation that J µ must satisfy for the action to be “gauge invariant” is

m2 2
∂µ J µ − m2 ∂µ Aµ + ∂ λ = 0. (830)
2
This is very much so not a conservation law and further it’s not gauge invariant itself. So,
when the photon is massive, electric charge is not conserved.

11.2 Scalar Higgs

If the Higgs boson were a field with non-zero spin, then to define its vev would require both
a magnitude as well as a direction of its spin. The Higgs field permeates space, and so the
vev direction of its spin would create a preferred direction. If there is a preferred direction
to space, then the universe would not be Lorentz invariant, as the effect of a rotation could
be detected. Therefore, in a universe that is Lorentz invariant, the Higgs boson cannot have
a non-zero spin.

11.3 Charge 0 Higgs

If the vev were to be in the electrically charged component of the scalar doublet φ, ~ then
the vev would couple to the photon. Because the vev is some non-zero value, this would
correspondingly result in a mass for the photon in the broken theory, like we observed in the
case of BCS superconductivity. To the best of our knowledge, the photon is truly massless
(electric charge is conserved), and so the vev must be electrically neutral. Fluctuations about
the vev is the Higgs boson, so if the vev is electrically neutral, then so too must the Higgs
be.

11.4 Forms of SU(2) Matrices

Let’s first Taylor expand the exponential matrix form for a 2 × 2 unitary matrix to linear
order. To first order, we don’t need to worry about commutators and the Baker-Campbell-
Hausdorff formula, which will make our job easier. We have
~
σ ~σ
ei~α· 2 = I + i~α · + ··· (831)
2
1 + i α23 i α21 + α22
 
' .
i α21 − α22 1 − i α23

123
In the second line, we just used the explicit form of the Pauli matrices. Now, Taylor expand-
ing the other form of the unitary matrix we have
 iχ   
e 1 cos ψ −eiχ2 sin ψ 1 + iχ1 −(1 + iχ2 )ψ
' . (832)
e−iχ2 sin ψ e−χ1 cos ψ (1 − iχ2 )ψ 1 − iχ1
From this, we first immediately read off that
α3
χ1 = . (833)
2
Matching the real parts in the off diagonal entries of the matrices, we also see that
α2
ψ=− . (834)
2
Then, matching the imaginary parts on the off-diagonal entries, we see that
α2 α1
− χ2 ψ = χ2 = , (835)
2 2
or that
α1
χ2 = . (836)
α2
This final result demonstrates that these relationships only hold to lowest order in the
parameters. Clearly if α2 = 0, this doesn’t mean that χ2 = ∞, but rather that the value of
ψ is not close to 0 and so the function sin ψ is important.

11.5 Unification of Couplings

11.5 (a)
The general form of these β-function differential equations is
dα α2
Q =β , (837)
dQ 2π
where β is some numerical value. We can solve this general form by separation of variables:
dα β dQ
= , (838)
α2 2π Q
and integrate both sides to find
1 1 β Q
− + = log . (839)
α(Q) α(mZ ) 2π mZ
Here, mZ is introduced as a reference energy. Solving for α(Q), we find
α(mZ )
α(Q) = β
. (840)
1− 2π
α(mZ ) log mQZ

124
Using the explicit forms for the β-functions for the three couplings of the Standard Model,
we have
α1 (mZ ) α2 (mZ ) α3 (mZ )
α1 (Q) = 41 α1 (mZ )
, α2 (Q) = 19 α2 (mZ )
, α3 (Q) = α3 (mZ )
.
1− 10 2π
log mQZ 1+ 6 2π
log mQZ 1+ 7 2π log mQZ
(841)

11.5 (b)
Now we are asked to find the energies at which pairs of these couplings take the same value.
We can do this analysis in general, and then we just plug in the particular values of the
couplings. Let’s set α1 (Q) = α2 (Q):

α1 (mZ ) α2 (mZ )
α1 (Q) = = α2 (Q) = , (842)
1− β1 α1 (m
2π
Z)
log Q
mZ
1 − β2 α2 (m
2π
Z)
log mQZ

where β1 and β2 are the respective numerical factors in the β-function. Rearranging, we
have
α1 (mZ )α2 (mZ ) Q
α1 (mZ ) − α2 (mZ ) = (β2 − β1 ) log . (843)
2π mZ
Solving for the energy Q at which the couplings are equal we find

2π α1 (mZ ) − α2 (mZ )
 
Q = mZ exp . (844)
β2 − β1 α1 (mZ )α2 (mZ )

With this general expression, we then plug in the values of the couplings and their β-
function coefficients. The differences in β-function coefficients is
19 41
β2 − β1 = − − ' −7.23 , (845)
6 10
41
β3 − β1 = −7 − = −11.1 , (846)
10
19
β3 − β2 = −7 + ' −3.83 . (847)
6
The ratio of couplings evaluated at the Z boson mass are:

α1 (mZ ) − α2 (mZ )
' −73.7 , (848)
α1 (mZ )α2 (mZ )
α1 (mZ ) − α3 (mZ )
' −96.9 , (849)
α1 (mZ )α3 (mZ )
α2 (mZ ) − α3 (mZ )
' −23.2 . (850)
α2 (mZ )α3 (mZ )

125
Finally, the ratio of these coupling factors to the β-function differences is
1 α1 (mZ ) − α2 (mZ )
' 10.2 , (851)
β2 − β1 α1 (mZ )α2 (mZ )
1 α1 (mZ ) − α3 (mZ )
' 8.73 , (852)
β3 − β1 α1 (mZ )α3 (mZ )
1 α2 (mZ ) − α3 (mZ )
' 6.06 . (853)
β3 − β2 α2 (mZ )α3 (mZ )
Perhaps remarkably, these factors which appear in the exponent differ by no more than a
factor of 2, even though the actual values of the couplings differ by up to a factor of 10.
However, when exponentiated these small differences are magnified and the energies at which
the couplings are equal are
Q12 ' 6.2 × 1029 GeV , Q13 ' 6 × 1025 GeV , Q23 ' 3.1 × 1018 GeV . (854)
In terms of energy scales typically involved with identifying Landau poles, the spread of these
energies at which the couplings are equal is quite small. It can be further shrunk by several
orders of magnitude by a more careful accounting of the running of the coupling from the Z
boson mass through the threshold for producing the top quark. The observation that these
couplings become equal at close to the same energy scale is motivation for grand unification,
which is the theory that the forces of nature unify into a single force at very high energies,
much like the weak and electromagnetic forces unify at energies comparable to the Higgs
vev.

11.6 When V − A and When Electroweak?

11.6 (a)
If the neutrino and electron are massless, then the value of the Mandelstam variable t is
|t| = |(pν − pe )2 | = 2pν · pe . (855)
Two massless four-vectors can have a dot product that is zero for non-zero energy if they are
collinear. So, if the neutrino and electron momenta are collinear and there is no momentum
transfer through the W boson, |t| = 0, which is its minimum value.

11.6 (b)
The dot product of the four-momentum of the electron and the neutrino can be evaluated
in terms of their energies and scattering angle θ as
|t| = 2pν · pe = 2Eν Ee (1 − cos θ) . (856)
Using the expression for the electron energy given in the problem, we can re-write this as
2Eν2 mN (1 − cos θ)
|t| = . (857)
mN + Eν (1 − cos θ)

126
Solving this for cos θ, we find

mN |t|
cos θ = 1 − . (858)
2Eν mN − Eν |t|
2

Now, as requested in the problem, we set |t| = m2W and find the scattering angle at which
the full electroweak theory is important:

mN m2W
cos θ = 1 − . (859)
2Eν2 mN − Eν m2W

11.6 (c)
Note that the cosine of the scattering angle is unphysical at sufficiently low neutrino en-
ergies. This means that the electroweak theory is unnecessary, and the V − A theory is
perfectly adequate to describe the low-energy processes. The minimum energy at which the
electroweak theory is required is when cos θ = −1 or that

m2W mN
= 2. (860)
2mN Eν2 − m2W Eν

Rearranging this expression, we find a quadratic equation for Eν :

m2W m2
Eν2 − Eν − W = 0 . (861)
2mN 4
The physical solution to this quadratic equation is
s !
2 2
m 4m
Eν = W 1 + 1 + 2N . (862)
4mN mW

The mass of a nucleon mN is about 80 times smaller than the mass of the W boson mW , so to
good approximation, we can Taylor expand this result to lowest order in the ratio m2N /m2W .
We then find the minimal neutrino energy of

m2W
Eν ' ' 3.2 × 103 GeV . (863)
2mN
To evaluate this energy, we used mW ' 80 GeV and the nucleon mass of mN ' 1 GeV. So,
the lowest neutrino energy at which you have to worry about the full electroweak theory is
large (several TeV), but significantly smaller than the energies that IceCube has measured.
So, IceCube needs to account for the full electroweak theory in their analysis.

127
11.7 Charged Current DIS
11.7 (a)
To calculate the differential cross section for charged-current DIS, we first need to evaluate
the matrix element M(e− L uL → νeL dL ) in the electroweak theory. From the expression for
the covariant derivative in the broken theory,
√ Eq. 11.86, the strength with which left-handed
fermions couple to the W boson is gW / 2, so this matrix element is
2
gW 1
M(e−
L uL → νeL dL ) = u†L (pν )σ̄ µ uL (pe ) 2
u†L (pd )σ̄ µ uL (pu ) , (864)
2 (pe − pν ) − mW
2

where pν , pe , pd , and pu are the momenta of the neutrino, electron, down quark, and up
quark, respectively. We can make the relabeling

Q2 ≡ −t = −(pe − pν )2 , (865)

and simplify the spinor products. Using the Fierz identity developed in the calculation of
the decay rate of the muon, we have
     
u†L (pν )σ̄ µ uL (pe ) u†L (pd )σ̄ µ uL (pu ) = −2 u†L (pν )uR (pd ) u†R (pu )uL (pe ) . (866)

With these simplifications, the matrix element is

  
u†L (pν )uR (pd ) u†R (pu )uL (pe )
M(e− 2
L uL → νeL dL ) = gW (867)
Q2 + m2W

Now, we can take the absolute square of this expression and find

|u†L (pν )uR (pd )|2 |u†R (pu )uL (pe )|2
|M(e−
L uL → νeL dL )| =2 4
gW . (868)
(Q2 + m2W )2

The spinor products simplify to four-vector dot products, as we have derived earlier:

|u†L (pν )uR (pd )|2 = 2pν · pd = |u†R (pu )uL (pe )|2 = 2pe · pu . (869)

Note that this string of equalities follows from momentum conservation which is

pe + pu = pν + pd . (870)

However, note that

(pe + pu )2 = 2pe · pu = ŝ , (871)
so the matrix element simplifies further:

ŝ2
|M(e− 2 4
L uL → νeL dL )| = 4gW . (872)
(Q2 + m2W )2

128
In terms of the electron-proton center-of-mass collision energy s, we have

ŝ = xs , (873)

where x is the momentum fraction of the up quark in the proton. That is, the matrix element
becomes
x2 s 2
|M(e− L uL → ν eL dL )|2
= 4g 4
W . (874)
(Q2 + m2W )2
Now, we can plug this matrix element into Fermi’s Golden Rule. In the center-of-mass
frame for the electron-up quark scattering, the cross section is then
4 Z
− gW x2 s 2
σ(eL uL → νeL dL ) = d cos θ 2 . (875)
8πŝ (Q + m2W )2

In this expression, we used the expression for two-body phase space in terms of the scattering
angle θ. We want the cross section differential in Q2 , so we need to change variables. Recall
that
ŝ
Q2 = −t = 2Ee Eν (1 − cos θ) = (1 − cos θ) , (876)
2
because the scattering occurs in the center-of-mass frame in which
√
ŝ
Ee = Eν = . (877)
2
Then,
2
d cos θ = dQ2 , (878)
ŝ
so the cross section can be written as
4
x2 s 2
Z
gW
σ(e−
L uL → νeL dL ) = dQ2 . (879)
4πx2 s2 (Q2 + m2W )2

The cross section differential in the momentum transfer Q2 is then

dσ(e−
L uL → νeL dL )
4
gW 1
= . (880)
dQ2 4π (Q2 + m2W )2

Finally, to get the cross section for the electron-proton scattering process, we multiply by the
parton distribution function for the up quark, which then makes the expression additionally
differential in x:
d2 σ(e− p → νe + X) 4
gW 1
= fu (x) . (881)
dx dQ2 4π (Q2 + m2W )2

129
11.7 (b)
To determine if the data in Fig. 11.5 exhibits Bjorken scaling, we should extract the predicted
form of the parton distribution function from these data and Eq. 11.151. Our predicted cross
section has some Q2 dependence beyond what may exist in the parton distribution, so we
need to extract this. However, at relatively low values of Q2 for which Q2  m2W , the
differential cross section is approximately independent of Q2 , assuming Bjorken scaling:

d2 σ(e− p → νe + X) Q2 m2W
4
gW
→ fu (x) . (882)
dx dQ2 4πm2W

Thus, Bjorken scaling and the low-energy limit predicts that the differential cross sections
are simply independent of Q2 . This is quite well-represented in the data at Q2 = 1000 and
2000 GeV2 , and then deviations are observed at higher values of Q2 . This suggests that
Bjorken scaling is well-exhibited, but at high Q2 , the low-energy assumption is beginning to
fail.

11.7 (c)
In particular, assuming Bjorken scaling, we can predict the mass of the W boson from
these data. We’ll present a very simple way of doing this here; one can do something more
sophisticated even from the data that’s presented. Assuming Bjorken scaling, note that the
ratio of differential cross sections at two different values of Q2 , say Q21 and Q22 is very simple
and independent of x:
d2 σ(e− p→νe +X)
dx dQ21 (Q22 + m2W )2
d2 σ(e− p→νe +X)
= . (883)
(Q21 + m2W )2
dx dQ22

So, let’s just pick two values of Q2 and compare the differential cross sections at a single
value of x. For instance, let’s pick Q21 = 1000 GeV2 and Q22 = 5000 GeV2 and study the bin
near x = 0.08 in both distributions. From the plots, the ratio of the cross sections at this
point is approximately

d2 σ(e− p→νe +X) 
dx dQ21  0.06

d2 σ(e− p→νe +X) 
' ' 2.72 . (884)
2
0.022
dx dQ2 
x=0.08

The square-root of this is approximately

√
2.72 ' 1.65 . (885)

Then, from the ratio of cross sections, we have

5000 + m2W
1.65 ' , (886)
1000 + m2W

130
or, solving for m2W , we find
m2W ' 5153 GeV2 . (887)
Taking the square-root, we find
mW ' 72 GeV . (888)
From this very naive procedure, we find a result that is remarkably close to the accepted
value of about 80 GeV.

11.8 Left-Handed Coupling of the W Boson?

11.8 (a)
The first step to calculating the differential cross section for top quark decay in this left-right
symmetric model is to contract some Pauli matrices. We would like to prove that
µ
σab σ̄µ cd = 2δad δbc . (889)
To do this we will proceed in a similar fashion to what was done in Sec. 10.4.1. There, we
expressed the Pauli matrices with Kronecker-δs as
Iab = δa1 δb1 + δa2 δb2 , (890)
σ1 ab = δa1 δb2 + δa2 δb1 , (891)
σ2 ab = −iδa1 δb2 + iδa2 δb1 , (892)
σ3 ab = δa1 δb1 − δa2 δb2 . (893)
Now, taking the matrix product, we have
µ
σab σ̄µ cd = (I, σ1 , σ2 , σ3 )µab (I, −σ1 , −σ2 , −σ3 )µ cd (894)
= (δa1 δb1 + δa2 δb2 )(δc1 δd1 + δc2 δd2 ) + (δa1 δb2 + δa2 δb1 )(δc1 δd2 + δc2 δd1 )
+ (−iδa1 δb2 + iδa2 δb1 )(−iδc1 δd2 + iδc2 δd1 ) + (δa1 δb1 − δa2 δb2 )(δc1 δd1 − δc2 δd2 )
= 2(δa1 δb1 δc1 δd1 + δa1 δb2 δc2 δd1 + δa2 δb1 δc1 δd2 + δa2 δb2 δc2 δd2 )
= 2δad δbc ,
as expected.

11.8 (b)
In the narrow-width approximation, the right-handed top quark decay matrix element be-
comes   
† µ †
g 2 uR (pb )σ uR (pt ) uL (pν )σ̄µ vR (pl )
M(tR → bR lR +
νL ) → −i W . (895)
2 mW ΓW
Using the Fierz identity derived in the previous part, the spinor products can be simplified
to   
u†R (pb )vR (pl ) u†L (pν )uR (pt )
+ 2
M(tR → bR lR νL ) = −igW . (896)
mW ΓW

131
Now, the absolute square of this matrix element is
4
gW
+
|M(tR → bR lR νL )|2 = 2 2
|u†R (pb )vR (pl )|2 |u†L (pν )uR (pt )|2 (897)
mW ΓW
2g 4
= 2 W2 (pb · pl )(pν · pt ) .
mW ΓW
In the second line, we used results derived in the calculation of the muon decay rate to
evaluate the spinor products, where we only assume that the top quark has non-zero mass.

11.8 (c)
To calculate the differential decay rate in this left-right symmetric model, we need the left-
handed top quark decay matrix element from example 11.2. It is
4
+ 2gW
|M(tL → bL lR νL )|2 = (pb · pν )(pt · pl ) . (898)
m2W Γ2W
The top quark spin-averages matrix element is then
1 g4
+
(|M(tL → bL lR +
νL )|2 + |M(tR → bR lR νL )|2 ) = 2 W 2 [(pb · pν )(pt · pl ) + (pb · pl )(pν · pt )] .
2 mW ΓW
(899)
In the frame in which the top quark is at rest, this matrix element can be expressed in terms
of the three-body phase space variables xi , where
2pt · pi
xi = . (900)
m2t
Using this, we have
xl m2t xν m2t
pt · pl = , p ν · pt = . (901)
2 2
Note also that
p2t m2 (1 − xl ) m2t (1 − xν )
p b · pν = − pt · pl = t , p b · pl = . (902)
2 2 2
Then, the matrix element in terms of the three-body phase space variables is
1 g 4 m4
+
(|M(tL → bL lR +
νL )|2 + |M(tR → bR lR νL )|2 ) = W2 t2 [xl (1 − xl ) + xν (1 − xν )] . (903)
2 4mW ΓW
With this result, we can then plug it into Fermi’s Golden Rule, as done in example 11.2.
The result from doing this is
4 Z 1 Z 1
m5t m2W
  
gW
Γt = dxν dxl Θ(xν + xl − 1) [xl (1 − xl ) + xν (1 − xν )] δ xb − 1 − 2 .
1028π 3 m2W Γ2W 0 0 mt
(904)

132
The δ-function that constrains the energy fraction of the bottom quark isn’t so useful in this
expression because xb isn’t explicit anywhere. However, note that xb = 2 − xν − xl , so this
δ-function can be equivalently expressed as

m2W m2W
     
δ xb − 1 − 2 = δ xl − 1 − xν + 2 . (905)
mt mt

Using this δ-function to integrate over xl , we find that the decay rate is
4 Z 1
m5t m2W m2W
   
gW
Γt = dxν 1 − xν + 2 xν − 2 + xν (1 − xν ) . (906)
1028π 3 m2W Γ2W m2W2 mt mt
mt

Now, to evaluate the decay rate differential in the lepton-bottom quark invariant mass,
we need to change integration variables from xν to mlb . The relationship between these two
parameters is √
mlb = mt 1 − xν , (907)
or that
m2lb
xν = 1 − . (908)
m2t
Then, the Jacobian of the change of variables is
2mlb
dxν = dmlb . (909)
m2t

With this change, the decay rate becomes

4 Z √m2t −m2W
m3t
 2
mlb + m2W m2lb + m2W m2lb m2lb
   
gW
Γt = dmlb mlb 1− + 2 1− 2 .
512π 3 m2W Γ2W 0 m2t m2t mt mt
(910)

Removing the integral, we find the differential decay rate to be

4
m2lb + m2W m2lb
     q 
dΓt gW mt mlb 2 2 2 2 2
= (mlb + mW ) 1 − + mlb 1 − 2 Θ mt − mW − mlb .
dmlb 512π 3 m2W Γ2W m2t mt
(911)

11.8 (d)
Unlike the purely left-handed electroweak
p prediction, this cross section does not vanish at
the upper endpoint. Indeed, when mt − m2W = mlb , we find
2

4
(m2t − m2W )3/2

dΓt  gW
= 6= 0 . (912)
dmlb √m2 −m2 =mlb 512π 3 mt Γ2W
t W

133
This is not surprising from conservation of angular momentum. The lepton–bottom quark
invariant mass is maximized when their momenta are back-to-back, as illustrated in example
11.2. While we have still assumed that the neutrino is left-handed, the bottom quark in the
decay can be right-handed, and as such can render the final state to have a total angular
momentum of 1/2. Thus, with a right-handed coupling in its decay, there is a non-zero
probability for the decay products to assume their maximal invariant mass of the bottom
quark and lepton because that configuration is allowed by angular momentum conservation.

11.8 (e)
To find the peak of the distribution of this new prediction we take the derivative of the decay
rate with respect to mlb and set it equal to 0:

m2lb + m2W m2lb

    
d 2 2 2
mlb (mlb + mW ) 1 − + mlb 1 − 2 =0 (913)
dmlb m2t mt
−10m4lb + (6m2lb + m2W )(m2t − m2W )
= .
m2t

The value of mlb that satisfies this is

q p
2 2
peak
3(m t − m W ) + (m2t − m2W )(9m2t + m2W )
mlb = √ ' 123 GeV . (914)
10
To evaluate the location of the peak, we used mt = 173 GeV and mW = 80 GeV. This is
nearly 10 GeV higher than the peak predicted in the purely left-handed electroweak decay,
which is even further from the peak in data.

11.8 (f )
From this analysis of the upper kinematic endpoint and the location of the peak in the dis-
tribution, the purely left-handed coupled electroweak force gives a more accurate prediction
to data than does the left-right symmetric model.

12 Consequences of the Weak Interactions

12.1 Mass and Flavor Basis Commutator
12.1 (a)
In the Wolfenstein parametrization, the CKM matrix is represented as
 
1 − λ2 /2 λ Aλ3 (ρ − iη)
VCKM '  −λ 1 − λ2 /2 Aλ2 . (915)
3 2
Aλ (1 − ρ − iη) −Aλ 1

134
All explicit parameters λ, A, ρ, and η are real numbers. The Hermitian conjugate of this
form of the CKM matrix is then
 
1 − λ2 /2 −λ Aλ3 (1 − ρ + iη)
†
VCKM ' λ 1 − λ2 /2 −Aλ2 . (916)
3 2
Aλ (ρ + iη) Aλ 1

Then, the product of the CKM matrix and its Hermitian conjugate is
 
1 − λ2 /2 λ Aλ3 (ρ − iη)
†
VCKM VCKM = −λ 1 − λ2 /2 Aλ2  (917)
3 2
Aλ (1 − ρ − iη) −Aλ 1
 
1 − λ2 /2 −λ Aλ3 (1 − ρ + iη)
× λ 1 − λ2 /2 −Aλ2 
3 2
Aλ (ρ + iη) Aλ 1
 1 2 2 2 2 2 1

4
+ A λ (ρ + η ) A λ(ρ − iη) − 2
Aλ(1 − ρ + iη)
= I + λ4  A2 λ(ρ + iη) 1
4
+ A2 − 12 A(1 − 2ρ + 2iη) .
1 1 2 2 2 2 2
− 2 Aλ(1 − ρ − iη) − 2 A(1 − 2ρ − 2iη) A + A λ ((1 − ρ) + η )

So, as claimed, the Wolfenstein parametrization of the CKM matrix is unitary, up to correc-
tions of order λ4 .

12.1 (b)
The matrix X is just the CKM matrix minus the identity matrix:
 
−λ2 /2 λ Aλ3 (ρ − iη)
X= −λ −λ2 /2 Aλ2 . (918)
Aλ3 (1 − ρ − iη) −Aλ2 0

To determine the exponentiated Hermitian matrix Fu , we use the Taylor expansion trick
described in the problem. Do do this, we need the square and the cube of the matrix X,
only keeping those terms up through λ3 . Ignoring terms higher than cubic in λ, the square
of X is  
−λ2 −λ3 Aλ3
X2 =  λ3 −λ2 0  . (919)
3
Aλ 0 0
The cube of X, ignoring terms beyond λ3 , is simply
 
0 −λ3 0
X3 =  λ3 0 0 . (920)
0 0 0

135
Then, with these three matrices, we can construct the exponentiated Hermitian matrix,
through λ3 order as

X2 X3
Fu = iX − i +i + ··· (921)
2 3
3
λ + λ6 − 12 Aλ3 (1 − 2ρ + 2iη)
 
0
3
= i −λ − λ6 0 Aλ2 .
1
2
Aλ3 (1 − 2ρ − 2iη) −Aλ2 0

Note that this matrix is indeed Hermitian.

12.1 (c)
The up-type quark Yukawa mass matrix is
 
yu
v
Mu = √  yc , (922)
2 yt

where the Yukawa couplings are in general non-zero and all different. With the flavor matrix
identified in the previous part, the product of the mass and flavor matrix is
  2
 
0 yu λ 1 + λ6 − 12 Ayu λ3 (1 − 2ρ + 2iη)
v    
M u Fu = i √  −yc λ 1 + 6 λ2
0 Ayc λ 2  . (923)
2 
1 3 2
2
Ayt λ (1 − 2ρ − 2iη) −Ayt λ 0

The opposite order of the matrix product yields

   
λ2
0 yc λ 1 + 6 − 12 Ayt λ3 (1 − 2ρ + 2iη)
v    
Fu Mu = i √  −yu λ 1 + λ6
2
0 Ayt λ2
 . (924)
2  
1
2
Ayu λ3 (1 − 2ρ − 2iη) −Ayc λ2 0

The commutator of the mass and flavor matrices is then

[Mu , Fu ] = Mu Fu − Fu Mu (925)
  2
 
0 (yu − yc )λ 1 + λ6 − 12 A(yu − yt )λ3 (1 − 2ρ + 2iη)
v    
= i√  −(yc − yu )λ 1 + λ6
2
0 A(yc − yt )λ2
.
2 
1
2
A(yt − yu )λ3 (1 − 2ρ − 2iη) −A(yt − yc )λ2 0

This matrix is indeed non-zero, and has contributions at O(λ).

136
12.1 (d)
This commutator of the mass and flavor matrices is a measure of the affect that a mass
measurement has on a subsequent flavor measurement. To measure this commutator, we
could, for example, perform a mass measurement, then a flavor measurement, then a mass
measurement in succession. To measure the mass of an electrically charged particle like
the quarks, we could have them pass through a region with non-zero magnetic field. The
radius of curvature of the charged particle is a measure of its momentum and if we know
the energy, we can determine the mass. Then, to measure the flavor of the quark, we can
have it interact with matter (= atomic nuclei), and identify the hadrons produced from the
interaction. Then, one could measure the mass of the resulting system by a total energy and
momentum measurement.
That said, such a measurement is nearly impossible to ever accomplish because quarks
never exist in isolation due to confinement. Nevertheless, elements of the CKM matrix can
be determined by measuring hadron decays rates and quark masses can be determined by
matching hadron masses to predictions from Lattice QCD, so this exercise is relevant.

12.2 Unitarity of the CKM Matrix

The form of the CKM matrix in Eq. 12.72 is more general than that in 12.73, so here, we
will just explicitly demonstrate its unitarity. The CKM matrix written there is
  
1 0 0 cos θ13 0 sin θ13
VCKM =  0 eiδ cos θ23 ei(δ+ζ) sin θ23   0 1 0  (926)
i i(+ζ)
0 −e sin θ23 e cos θ23 − sin θ13 0 cos θ13
 iα 
e cos θC −ei(α+γ) sin θC 0
×  eiβ sin θC ei(β+γ) cos θC 0  .
0 0 1

The Hermitian conjugate of this form of the CKM matrix is

e−iα cos θC e−iβ sin θC

  
0 cos θ13 0 − sin θ13
†
VCKM =  −e−i(α+γ) sin θC e−i(β+γ) cos θC 0   0 1 0  (927)
0 0 1 sin θ13 0 cos θ13
 
1 0 0
× 0 e−iδ cos θ23 −e−i sin θ23  .
−i(δ+ζ)
0 e sin θ23 e−i(+ζ) cos θ23

We now want to multiply these matrices together and verify that their product is the identity
matrix. Because of this nice factorized form, we can consider simpler products of matrices.

137
 †
Multiplying the matrix on the right of VCKM and on the left of VCKM we have

e−iα cos θC e−iβ sin θC

 iα  
e cos θC −ei(α+γ) sin θC 0 0
 eiβ sin θC ei(β+γ) cos θC 0   −e−i(α+γ) sin θC e−i(β+γ) cos θC 0  (928)
0 0 1 0 0 1
 
1 0 0
=  0 1 0 .
0 0 1
The product of the middle matrices is
    
cos θ13 0 sin θ13 cos θ13 0 − sin θ13 1 0 0
 0 1 0  0 1 0 = 0 1 0 . (929)
− sin θ13 0 cos θ13 sin θ13 0 cos θ13 0 0 1
†
Finally, the product of the left-most matrix within VCKM and the right-most matrix in VCKM
yields
    
1 0 0 1 0 0 1 0 0
 0 eiδ cos θ23 ei(δ+ζ) sin θ23   0 e−iδ cos θ23 −e−i sin θ23  =  0 1 0  .
−i(δ+ζ)
i
0 −e sin θ23 e i(+ζ)
cos θ23 0 e sin θ23 e−i(+ζ) cos θ23 0 0 1
(930)
That is, this form of the CKM matrix is expressed as the product of three unitary matrices
and so ensures that
†
VCKM VCKM = I. (931)

12.3 Jarlskog Invariant

12.3 (a)
We are first asked to calculate the Jarlskog invariant from the general unitarity triangle
relationship
Vud Vub∗ + Vcd Vcb∗ + Vtd Vtb∗ = 0 . (932)
Each term in this expression is a complex number that is represented by an ordered pair of
coordinates that represent a two-dimensional vector. These three ordered pairs compose the
sides of a triangle. Given the length of the three sides of a triangle, one can calculate its
area with Heron’s formula. We’ll provide a derivation of that result here.
Call the length of the three sides of a triangle A, B, and C. The area of the triangle can
be expressed as
1
Area = AB sin θ , (933)
2
where θ is the angle between sides A and B. The length of side C can be determined by the
Law of Cosines:
C 2 = A2 + B 2 − 2AB cos θ . (934)

138
One can then use the Law of Cosines to re-express the area of the triangle exclusively in
terms of the lengths of the three sides A, B, and C. Note that
A2 + B 2 − C 2
cos θ = , (935)
2AB
and the area is
1 √ 1√ 2 2
Area = AB 1 − cos2 θ = 2A B + 2A2 C 2 + 2B 2 C 2 − A4 − B 4 − C 4 . (936)
2 4
The squared length of each side of this triangle is just the absolute square of each of the
terms in the unitarity expression from above. We have, for example,

A2 = |Vud |2 |Vub |2 , B 2 = |Vcd |2 |Vcb |2 , C 2 = |Vtd |2 |Vtb |2 . (937)

Further, using the unitarity relationship, we can eliminate dependence on Vtd and Vtb in place
of the other elements. Note that

C 2 = |Vtd |2 |Vtb |2 = |Vud |2 |Vub |2 + |Vcd |2 |Vcb |2 + 2 Re(Vud Vub∗ Vcd∗ Vcb ) (938)
= A2 + B 2 + 2 Re(Vud Vub∗ Vcd∗ Vcb ) .

where Re denotes the real part of the expression in parentheses. Plugging this expression
into Heron’s formula, we find
1
q
Area = |Vud |2 |Vub |2 |Vcd |2 |Vcb |2 − Re(Vud Vub∗ Vcd∗ Vcb )2 . (939)
2
We go go a bit further by noting that

|Vud Vub∗ Vcd∗ Vcb |2 = |Vud |2 |Vub |2 |Vcd |2 |Vcb |2 = Re(Vud Vub∗ Vcd∗ Vcb )2 + Im(Vud Vub∗ Vcd∗ Vcb )2 , (940)

where Im is the imaginary part. It then follows that the Jarlskog invariant J is simply
1
J = Im(Vud Vub∗ Vcd∗ Vcb ) . (941)
2

12.3 (b)
To determine the maximum value of the Jarlskog invariant, we use another constraint from
unitarity of the CKM matrix. The probability for any given quark to mix into another quark
is of course 1, which leads to constraints such as

|Vud |2 + |Vus |2 + |Vub |2 = 1 . (942)

The area of the unitarity triangle is therefore maximized if each element has the same
absolute value:
1
|V | = √ , (943)
3

139
where V is any element of the CKM matrix. Then, to calculate the maximal Jarlskog
invariant, we consider an equilateral triangle with each side of length 1/3, because each
side is a product of two elements of the CKM matrix. As an equilateral triangle, the angle
between two sides is π/3, and so its area is
1 1 1 π 1
Jmax = · · sin = √ ' 4.8 × 10−2 . (944)
2 3 3 3 12 3
This is about 150 times larger than the measured value, justifying the claim that CP violation
in the quark sector is indeed relatively small.

12.4 Extra Quark Generations

If our universe had N generations of quarks they could in general mix with an N ×N unitary
matrix. There are in general N 2 free parameters that define a general N × N unitary matrix.
These parameters can be separated into rotation angles and complex phases. The rotation
angles compose an N × N orthogonal matrix. The number of rotation angles is equal to the
number of independent, orthogonal planes about which we can rotate. In N dimensions, the
number of such planes is the number of unique pairs of axes or

N (N − 1)
 
N
= . (945)
2 2

Then, the number of complex phases in a general unitary matrix is

N (N − 1) N (N + 1)
N2 − = . (946)
2 2
Many of these phases can be eliminated, however. With N generations of quarks, there are a
total of 2N quarks, and we are able to redefine 2N − 1 of their phases to remove the complex
phases in the mixing matrix. That is, there are actually only

N (N + 1) (N − 1)(N − 2)
− (2N − 1) = (947)
2 2
independent, irreducible complex phases. Note that the number of complex phases if N = 2
is
(2 − 1)(2 − 2)
= 0, (948)
2
as expected. That is, two quark generations do not exhibit CP violation. If N = 3, the
number of complex phases is
(3 − 1)(3 − 2)
= 1, (949)
2
again as expected.

140
12.5 Measuring the Cabibbo Angle
To determine the Cabibbo angle θC , we can predict the decay rates of D mesons to different
final states. The matrix element for these decays is proportional to the relevant element
of the two-generation mixing matrix, presented in Sec. 12.2.3. The decay of the D+ meson
to the K̄ 0 meson mixes a charm quark into a strange quark, proportional to cos θC . By
contrast, the decay of the D+ meson to the π 0 meson mixes a charm quark into a down
quark, proportional to sin θC . That is, the matrix elements for these decays are

M(D+ → K̄ 0 e+ νe ) ∝ cos θ , M(D+ → π 0 e+ νe ) ∝ sin θ . (950)

In a decay rate or the calculation of the branching fractions, the matrix elements are squared.
Further, the final state for both decays are three-body and have essentially the same relevant
masses, therefore to good approximation the ratio of the branching fractions is equal to the
ratio of the squared matrix elements:
Br(D+ → K̄ 0 e+ νe ) |M(D+ → K̄ 0 e+ νe )|2 cos2 θC
' = 2 = cot2 θC . (951)
Br(D → π e νe )
+ 0 + |M(D → π e νe )|
+ 0 + 2 sin θC
The measured ratio of the branching fractions is
Br(D+ → K̄ 0 e+ νe ) 8.6 × 10−2
' ' 19.5 . (952)
Br(D+ → π 0 e+ νe ) 4.4 × 10−3
Then, we can solve for the Cabibbo angle θC :
√
θC ' cot−1 19.5 ' 0.22 . (953)

The sin of this angle is therefore

sin θC ' sin 0.22 ' 0.22 , (954)

which is very close to the accepted value in the PDG!

12.6 Non-Relativistic Limit of Neutrino Oscillations

The relativistic neutrino oscillation probabilities are
 
2 2 L 2 2
P (ν̄e → ν̄e ) = 1 − sin (2θ) sin (m − m2 ) , (955)
4E 1
 
2 2 L 2 2
P (ν̄e → ν̄µ ) = sin (2θ) sin (m − m2 ) . (956)
4E 1
We will first restore factors of c and ~. As written, the argument of the second sine function
has dimensions of  
L
(m1 − m2 ) = [mass][length]−1 [time]2 .
2 2
(957)
4E

141
So, we need to multiply by appropriate factors of c and ~ to render this argument dimen-
sionless. Recall that the dimensions of these constants are

[c] = [length][time]−1 , [~] = [mass][length]2 [time]−1 . (958)

Therefore, the combination ~c−3 has the same dimensions as the argument of the sine. That
is, the combination
L
(m2 c4 − m22 c4 ) (959)
4~cE 1
is dimensionless. This is still completely relativistic, so we can replace the baseline L for the
travel time T with
L = cT . (960)
The argument of sine is then
L T
(m21 c4 − m22 c4 ) = (m2 c4 − m22 c4 ) . (961)
4~cE 4~E 1
Now we take the non-relativistic limit. In the non-relativistic limit, the sum of the
neutrino energies is essentially just the sum of their rest energies:

E → m1 c2 + m2 c2 . (962)

Using this expression, the argument of sine becomes

T T
(m21 c4 − m22 c4 ) → (m1 c2 − m2 c2 ) . (963)
4~E 4~
Therefore, the neutrino oscillation probabilities in the non-relativistic limit are
 
2 2 T 2 2
P (ν̄e → ν̄e ) = 1 − sin (2θ) sin (m1 c − m2 c ) , (964)
4~
 
2 2 T 2 2
P (ν̄e → ν̄µ ) = sin (2θ) sin (m1 c − m2 c ) . (965)
4~

12.7 Neutrinos for Nuclear Non-Proliferation

12.7 (a)
The total volume of the WATCHMAN detector is

Vol = 10.8 · π5.42 m3 ' 9.9 × 102 m3 . (966)

The density of water is 1000 kg/m3 , and so the total mass of the water in WATCHMAN is
about
mH2 O ' 9.9 × 105 kg . (967)

142
Because gadolinium is only 0.1% of the contents of the detector by weight, we can ignore its
contribution to estimate the total mass. Then, the mass of gadolinium would be

mGd ' 9.9 × 102 kg = 9.9 × 105 g . (968)

158
The number of Gd atoms in the detector would then be
9.9 × 105
NGd ' mol ' 3.8 × 1027 . (969)
158

12.7 (b)
In one second, a 10 MW reactor would produce 10 MJ of energy. In eV, 10 MJ is

10 MJ = 107 J ' 6.3 × 1025 eV . (970)

Then, the number of uranium atoms that fission per second NU at such a reactor must be

6.3 × 1025
NU ' ' 3.1 × 1017 . (971)
2 × 10 8

The mass of a uranium atom is 235 times that of a proton, and the mass of the proton is
about 1.67 × 10−27 kg. Therefore, the total fissioning uranium per second in mass is

mU ' 235 · (1.67 × 10−27 ) · (3.1 × 1017 ) kg ' 1.2 × 10−7 kg . (972)

12.7 (c)
With two electron anti-neutrinos per 235 U fission and 3.1 × 1017 fissioning uranium atoms per
second, there are a total of about 6.2 × 1017 neutrinos emitted per second from the reactor.
The density of these neutrinos nν that pass through a surface with radius of 25 km is then

6.2 × 1017 −2
nν ' m ' 2.5 × 1013 m−2 . (973)
2.5 × 104
As a cylinder, the cross section of WATCHMAN is approximately a rectangle with cross
sectional area
A ' 10.8 · 10.8 m2 ' 117 m2 . (974)
The rate of neutrinos Rν that pass through WATCHMAN per second is then the product of
the neutrino density times the cross sectional area of WATCHMAN:

Rν = nν A ' 2.9 × 1015 s−1 . (975)

143
12.7 (d)
Recall that the expression for the event rate is
Nν NN
event/time = cσβ −1 . (976)
Volν
Here, Nν and NN are the number of neutrinos that pass through WATCHMAN and NN is
the number of nucleons in the detector, σβ −1 is the cross section for inverse-β decay, and
Volν is the total volume of neutrinos that pass through WATCHMAN during the exposure
time. If there is a total exposure time of texp to see one event, multiplying by this exposure
time we have
Nν NN
1= ctexp σβ −1 . (977)
Volν
The total volume of neutrinos that passes through WATCHMAN during that time is the
cross sectional area A of WATCHMAN times ctexp , which is the length of the region traveling
at the speed of light that passes through the detector in texp . Then, to observe one event,
we have
Nν NN
1= σβ −1 . (978)
A
Finally, the total number of neutrinos that pass through WATCHMAN is the rate times the
exposure time:
Rν NN
1= texp σβ −1 . (979)
A
Solving for the exposure time necessary to observe one neutrino interaction, we find
A
texp = . (980)
Rν NN σβ −1

To evaluate this expression, we need to determine the number of nucleons in the detector.
The total mass of water is

mH2 O ' 9.9 × 105 kg = 9.9 × 108 g , (981)

or nearly a billion mol of nucleons. That is, the total number of nucleons in the water is

NN ' 6 × 1032 . (982)

Then, the exposure time to observe one neutrino event is approximately

A
texp = ' 6.7 s . (983)
Rν NN σβ −1

To evaluate this, we also used that 1 b = 10−28 m2 .

144
12.7 (e)
We now want to determine the amount of water necessary to observe neutrinos at 200 km
from the reactor at the same rate as WATCHMAN at 25 km. Note that the rate of neutrinos
passing through the detector Rν is inversely proportional to the square of the distance to
the reactor. So, the rate of neutrinos passing through the new detector Rν0 is a factor of

252
Rν0 = Rν ' 1.6 × 10−2 Rν . (984)
2002
That is, for the same rate of neutrino observation, we need to increase the number of nucleons
NN by this factor:
2002
NN0 = NN = 64NN . (985)
252
That is, we need to increase the mass of water in the detector by a factor of 64. The mass of
water in WATCHMAN is about 9.9 × 105 kg so the mass of water in the new detector must
be
mH2 O ' 64 · 9.9 × 105 kg ' 6.3 × 107 kg . (986)

12.8 Neutrinos from SN 1987a

12.8 (a)
At 168,000 light years from SN 1987a, the total area through which the 1058 neutrinos pass
through at this distance is

area = 4π(1.68 × 105 )2 ly2 ' 3.2 × 1043 m2 . (987)

To convert from light years to meters, one light-year is the speed of light times the number of
seconds in a year. By contrast, the cross sectional area of the Earth through which neutrinos
from SN 1987a might have passed through is only

cross-sec of Earth = π(6 × 106 )2 m2 ' 1.1 × 1014 m2 . (988)

This is a factor of
cross-sec of Earth
' 3.6 × 10−30 (989)
area
of the total area at the distance to Earth through which neutrinos could have passed through.
Therefore, the total number of neutrinos that passed through Earth out of the total 1058
neutrinos is
Earth neutrinos ' 1058 · 3.6 × 10−30 ' 3.6 × 1028 , (990)
which is still a lot of neutrinos!

145
12.8 (b)
The cross sectional area of Kamiokande-II is approximately the product of its height and
diameter, as a cylinder. This is

cross-sec of K-II ' 256 m2 . (991)

This is a further factor of

cross-sec of K-II
' 2.3 × 10−12 (992)
cross-sec of Earth
the cross sectional area of Earth. Therefore, out of the approximately 3.6 × 1028 neutrinos
that passed through Earth only

Kamiokande-II neutrinos ' 3.6 × 1028 · 2.3 × 10−12 ' 8.3 × 1016 (993)

passed through Kamiokande-II.

12.8 (c)
To estimate the cross section of neutrinos to interact with water molecules, we start from
the expression for the event rate from Chap. 4:

events/time = nν nH2 0 Vol cσ . (994)

Here, nν and nH2 0 are the number densities of neutrinos and water molecules, σ is the
neutrino-water cross section, and Vol is the volume in which they could interact; that is, the
volume of Kamiokande-II. Note, however, that the number density of the water times the
volume of Kamiokande-II is just the number of water molecules, NH2 0 . So, the event rate
can also be written as
events/time = nν NH2 0 cσ . (995)
To find the total number of events, we multiply by the exposure time texp over which the
neutrinos were observed:
events = nν NH2 0 ctexp σ . (996)
Now, the total volume of neutrinos that interact with Kamiokande-II can be found in the
following way. The length of the neutrino volume ` is just the speed of light (their velocity)
times the exposure time:
` = ctexp . (997)
The cross sectional area of the volume of neutrinos that could possibly interact with Kamiokande-
II is just the cross sectional area A of the detector itself. Therefore, the number of neutrinos
Nν that pass through Kamiokande-II is

Nν = Actexp nν . (998)

146
Using this expression, we can re-write the number of events as
Nν NH2 0
events = σ. (999)
A
The neutrino-water cross section is then
A · events
σ= . (1000)
Nν NH2 0
We know almost all of the quantities on the right, except for the number of water
molecules in the detector. The volume of Kamiokande-II is
Vol = 16 · π82 m3 ' 3.2 × 103 m3 . (1001)
The density of water is about 1000 kg/m3 and so there is a total mass of water of
mH2 O ' 3.2 × 106 kg = 3.2 × 109 g . (1002)
The molar mass of water is 18, and so the number of water molecules in Kamiokande-II are
3.2 × 109
NH2 0 ' mol ' 1032 . (1003)
18
Putting all these pieces together, the cross section for neutrino-water scattering is
A · events 256 · 11
σ= ' ' 3.4 × 10−46 m2 = 3.4 × 10−18 b = 3.4 ab . (1004)
Nν NH2 0 8.3 × 1016 · 1032

12.8 (d)
If neutrinos were massless, then if SN 1987a is 168,000 light-years away, it would take the
neutrinos 168,000 years to reach Earth. If instead the neutrinos had an energy of 1 MeV
and a mass of 16 eV, the boost factor γ would be
E 106
γ= = ' 6.3 × 104 . (1005)
m 16
The velocity β as a fraction of the speed of light is then
r
1
β = 1 − 2 ' 1 − 1.3 × 10−10 . (1006)
γ
That is, the neutrinos travel slower than the speed of light by about 1 part in 10 billion. Thus,
their travel time to Earth will increase by about one part in 10 billion. One ten-billionth of
168,000 years is about
1.68 × 105
yr ' 1.3 × 10−5 yr ' 410 s ' 6.8 min . (1007)
1.3 × 1010
This is less than the 2 hour time difference between neutrinos and light from SN 1987a by
a factor of about 20. So, definitely, if the mass of the neutrinos was much larger, the arrival
time between the neutrinos and photons couldn’t be as large as two hours (with assumptions
on supernova dynamics), hence the upper bound on neutrino masses.

147
12.9 Solar Neutrino Problem
The key to this exercise is the statement that neutrinos are produced incoherently in the
Sun. This means they are produced at random times, random locations, and with random
phases in the Sun. That is, the factor in the survival probability that oscillates with baseline
distance:  
2 L 2 2
sin (m − m2 ) , (1008)
4E 1
is smeared out, and averaged over by the incoherent processes of the Sun. The average value
of sin2 over its period is 1/2, and so we just replaced the oscillating factor by 1/2. With the
original, coherent, survival probability given by
 
2 2 L 2 2
Pcoher (ν̄e → ν̄e ) = 1 − sin (2θ12 ) sin (m − m2 ) , (1009)
4E 1
the incoherent survival probability is
sin2 (2θ12 )
Pincoher (ν̄e → ν̄e ) = 1 − . (1010)
2
While we can just use the mixing angle given in the problem to evaluate this, it’s nice to use
some double angle formulas to simplify it as well. We have
sin2 (2θ12 )
Pincoher (ν̄e → ν̄e ) = 1 − = 1 − 2 sin2 (θ12 ) cos2 (θ12 ) (1011)
2
= 1 − 2 sin2 (θ12 ) 1 − sin2 (θ12 ) = 1 − 2 sin2 (θ12 ) + 2 sin4 (θ12 ) .

Now, plugging in the value for sin2 θ12 , we find

Pincoher (ν̄e → ν̄e ) ' 0.574 . (1012)

13 The Higgs Boson

13.1 W Boson Decays
If each W boson decays to hadrons 70% of the time, then the probability that both decay
to hadrons is the square of this:
p(2W → hadrons) = 0.72 = 0.49 . (1013)
Similarly, the probability for both W bosons to decay leptonically is the square of 30%:
p(2W → leptons) = 0.32 = 0.09 . (1014)
Then, the probability for one W boson to decay hadronically and the other leptonically (or
vice-versa) is the product of these two rates, times a factor of two because there are two
possible configurations of decays:
p(2W → semi-leptonic) = 2 · 0.3 · 0.7 = 0.42 . (1015)
Note that the sum of these three rates is indeed 1.

148
13.2 pp → W + W − Backgrounds
From Fig. 4.4, the measured cross section at 13 TeV for pp → W + W − is about 100 pb. If
each W boson decays to leptons about 30% of the time, then they both decay to leptons
with a probability that is the square of that, or about 9% of the time. Therefore, the cross
section for pp → W + W − → l+ νl l0− ν̄l0 is approximately

σ(pp → W + W − → l+ νl l0− ν̄l0 ) ' 0.09 · 100 pb ' 9 pb . (1016)

13.3 Searching for H → W + W −

13.3 (a)
By momentum conservation, we have the following equalities:

p2H = m2H = (pW + + pW − )2 = 2m2W + 2pW + · pW − = (pe + pνe + pµ + pνµ )2 (1017)

= 2pe · pνe + 2pµ · pνµ + 2pe · pµ + 2pνe · pνµ = 2m2W + 2pe · pµ + 2pνe · pνµ .

In this expression, we have assumed that the W bosons are on-shell and the invariant mass
of the lepton and its respective neutrino is the W boson mass. To determine the maximum
value of the positron-muon invariant mass m2eµ = 2pe · pµ , we set the neutrino invariant to
as small as possible, which is simply 0, if the momenta are collinear. Then, rearranging the
relationships above, we find that the maximum invariant mass of the electron and muon is

m2eµ ≥ m2H − 2m2W . (1018)

13.3 (b)
Assuming that the leptons are massless, the minimum value of the positron muon invariant
mass is 0, which occurs when their momenta are collinear.

13.3 (c)
To determine the maximum value of the missing transverse momentum, we just note a few
things. First, the total momentum is zero because the Higgs decays at rest and so the total
transverse momentum is also 0. The total energy is just the Higgs mass which is somehow
shared among the four massless leptons. Note that if the MET was larger than mH /2 then it
would be impossible for the electron and muon to have a total momentum that could cancel
this, and render the total momentum 0. It is possible for the MET to be smaller than mH /2
and still have total momentum 0; this just means that the neutrinos have some components
of momentum in opposite directions. Therefore, the maximum MET is just mH /2.

149
13.4 H → γγ Rate
13.4 (a)
We are first asked to estimate the matrix element for Higgs decays to photons using NDA.
There’s actually nothing we need to do here; we can simply re-use the result for the process
gg → H. The matrix element for gluon-gluon fusion and decay to photons is essentially
identical: both gluons and photons are massless, and the processes proceed via a top loop.
The only differences are the couplings of the gluon and photon to the top quark and the
direction of time, but that doesn’t affect the result of the matrix element because these
processes are time-reversal invariant (there are no electroweak vector bosons). Instead of
gluons coupling to top quarks with a factor of αs , we replace it with the fine structure
constant α and include factors of the top’s electric charge Qt . That is, the matrix element
for H → γγ is
αQ2t yt m2H
M(H → γγ) ' . (1019)
4π mt
The electric charge of the top quark is Qt = 2/3, but we will leave it implicit for now.

13.4 (b)
Fermi’s Golden Rule for two body decays with massless final state particles such as what we
consider here, yields a decay rate of
Z 1
1 1
ΓH→γγ = d cos θ |M(H → γγ)|2 . (1020)
2mH 16π −1

The integral over the scattering angle is just 2 because the Higgs decays isotropically. Then,
the decay rate to photons is
α2 Q4t yt2 m3H
ΓH→γγ ' . (1021)
256π 3 m2t
One can plug in the appropriate numbers, but we won’t do that until comparing directly to
the decay rate to bottom quarks.

13.4 (c)
The mass of the bottom quark is related to its Yukawa coupling via
yb v
mb = √ , (1022)
2
so that the Yukawa coupling is
√
2mb
yb = ' 2.4 × 10−2 . (1023)
v

150
13.4 (d)
The decay matrix element for H → bb̄ is simpler than the decay to photons; the Higgs boson
couples directly to the bottom quarks. While it’s simple enough to just explicitly calculate
the Feynman diagram with spinor products, we’ll apply NDA again to estimate the matrix
element. Assuming that the only mass scale is the Higgs boson mass because mH  mb , all
dimensions of the matrix element must be made up from factors of mH . Further, the direct
coupling of the Higgs to the bottom quarks means that the only coupling that the matrix
element is sensitive to is the bottom quark Yukawa, yb . Therefore, all the matrix element
can be is
M(H → bb̄) ' yb mH . (1024)
The ratio of decay rates to photons and bottom quarks is equal to the ratio of the square
of matrix elements, assuming that the mass of the bottom quark is much smaller than the
Higgs boson. That is,
ΓH→bb̄ 16π 2 m2 16π 2 y 2 m2
= yb2 m2H 2 4 2 t 4 = 2 4 b2 2t . (1025)
ΓH→γγ α Qt yt mH α Qt yt mH
Now, plugging in numbers with yt ' 1, α = 1/137, mt = 173 GeV, and mH = 125 GeV, we
find a decay rate ratio of
ΓH→bb̄
' 1.7 × 104 . (1026)
ΓH→γγ
That is, the Higgs boson decays to bottom quarks about 10000 times more often than to
photons. The actual result is closer to a factor of about 1000, when properly accounting for
the masses of the top and bottom quarks.

13.5 Higgs Production Rate

13.5 (a)
The cross section for Higgs boson production from gluon-gluon fusion is
Z 1 Z 1
α2 y 2 m2
σ(pp → H) ' dx1 dx2 s t H2 fg (x1 )fg (x2 )δ(x1 x2 s − m2H ) . (1027)
0 0 16π mt
To integrate over x2 , we need the Jacobian factor J from the δ-function. This is
d
J −1 = (x1 x2 s − m2H ) = x1 s . (1028)
dx2
Further, the momentum fraction x2 as enforced by the δ-function is
m2H
x2 = . (1029)
x1 s
Then, using the δ-function, the cross section is
αs2 yt2 m2H 1 dx1
Z  2 
mH
σ(pp → H) ' 2
fg (x1 )fg Θ(x1 s − m2H ) . (1030)
16πs mt 0 x1 x1 s

151
13.5 (b)
The expression for the rapidity of the Higgs boson y is
1 E + pz
y= log . (1031)
2 E − pz
The energy of the Higgs boson is the sum of the colliding gluon energies:
√
s
E = (x1 + x2 ) . (1032)
2
The ẑ momentum of the Higgs boson is the sum of the ẑ momenta of the gluons, which are
oriented oppositely: √
s
pz = (x1 − x2 ) . (1033)
2
Then, the rapidity is
√ √
s s
(x1 + x2 ) 2
+ (x1 − x2 ) 2 1 x1
y= √ √ = log . (1034)
(x1 + x2 ) 2s − (x1 − x2 ) 2s 2 x2

Now, putting in the expression for the momentum fraction x2 , we have

√
x1 s
y = log . (1035)
mH
Inverting this, solving for x1 , we find
mH
x1 = √ ey . (1036)
s

Then, the derivative of this expression to determine the integration measure is

mH
dx1 = √ ey dy = x1 dy . (1037)
s

With this identification, the cross section can be expressed as an integral over rapidity y:
√
s
log
α 2 y 2 m2
Z    
mH mH y mH −y
σ(pp → H) ' s t H2 dy fg √ e fg √ e . (1038)
16πs mt log
mH
√ s s
s

13.5 (c)
From this analysis, it then follows that the maximum rapidity of the Higgs boson at the 13
TeV LHC is √
s 13000
ymax = log ' log ' 4.6 . (1039)
mH 125

152
13.5 (d)
The cross section for Higgs production differential in the rapidity of the Higgs boson is
especially sensitive to the parton distribution functions. From the data for the rapidity of
the Higgs boson, we can map that probability distribution onto the differential cross section
we derived in part (b) of this problem. This then enables us to determine the product of the
gluon parton distribution at different values of x.

13.6 Landau-Yang Theorem

13.6 (a)
In the rest frame of the Higgs boson, we can align the momentum of the two photons along
the ẑ axis. Additionally, the momenta are equal and opposite: p~2 = −~p1 . Polarization
vectors for the photon are always transverse to their momentum, as there are no oscillating
electric or magnetic fields in the direction of propagation. That is,
p~1 · ~1 = p~2 · ~2 = 0 . (1040)
Because the momenta are equal and opposite, transverse to p~1 means transverse to p~2 , and
vice-versa. So, we additionally have
p~2 · ~1 = p~1 · ~2 = 0 . (1041)

13.6 (b)
Because of this result, there can be no dot products of momenta and polarization vectors
present in the matrix element. It is possible to have dot products of polarization vectors or
momenta separately, however. Dot products of momenta greatly simplify. The only mass
scale in the process is the Higgs mass mH , and so all dot products must be proportional to
some power of mH . For instance,
m2H m2 m2
− H =− H,
p~1 · p~2 = E1 E2 − p1 · p2 = (1042)
4 2 2
where we have used the fact that the two photons have equal energy of mH /2 and their
invariant mass is the Higgs mass:
(p1 + p2 )2 = m2H = 2p1 · p2 . (1043)
Therefore, the matrix element can only depend on polarization vectors and the Higgs mass,
up to numerical factors. Then, we can write
M(H → γγ) = mH κijk iH (j1 k2 + k1 j2 ) . (1044)
Note that a three-point matrix element has dimensions of mass, which is accounted for by
the Higgs mass. The photon polarization vectors appear symmetrized because photons are
identical bosons. Finally, κijk is some dimensionless three-index object that contracts the
polarization vector indices.

153
13.6 (c)
As a matrix element, this object must be Lorentz invariant and subsequently rotationally
invariant. A requirement of this is that, for some rotation matrix M ∈ SO(3), the three-index
object κijk is invariant:
κijk Mil Mjm Mkn = κlmn . (1045)
Let’s contract indices l and m. We accomplish this by multiplying by the Kronecker-δ δlm
on both sides of this expression:

κijk Mil Mjm Mkn δlm = κlmn δlm . (1046)

Because of the orthogonality of the rotation matrix, we have

Mil Mjm δlm = δij , (1047)

so our expression becomes

κijk Mkn δij = κlmn δlm . (1048)
Because l and m are summed over, we can freely call them whatever we want, so

κijk Mkn δij = κijn δij , (1049)

or that
(κijk Mkn − κijn )δij = 0 . (1050)
The rotation matrix M was completely arbitrary, so this must hold for any rotation matrix
M. Elements of a rotation matrix are bounded in absolute value by 1, so the factor in
parentheses is in general non-zero. However, for the expression to hold, when contracted
with the δij this must vanish, and so in general must vanish if i = j. One can use a similar
argument for contraction of other pairs of indices. Therefore, for κijk to be non-zero, all
indices must be distinct.

13.6 (d)
Let’s start from the expression with the anti-symmetric symbol and massage it into a nice
form. We have

ijk M1i M2j M3k = M11 M22 M33 − M11 M23 M32 − M12 M21 M33 + M12 M23 M31 (1051)
+ M13 M21 M32 − M13 M23 M31
     
 M22 M23   M21 M23 
 + M13  M21 M23
 
= M11   − M12  
M32 M33   M31 M33   M31 M32 
 
 M11 M12 M13 
 
=  M21 M22 M23  = det M .
 M31 M32 M33 

154
 As an element of SO(3), the matrix M has determinant 1, so under any rotation, the
determinant is always 1. Note that 123 = 1 and under a rotation it transforms to
123 → ijk M1i M2j M3k = det M = 1 = 123 . (1052)
That is, the totally anti-symmetric symbol ijk is invariant to rotations. Therefore, the
three-index object κijk = κijk , to ensure that the matrix element is rotationally invariant.

13.6 (e)
Putting the anti-symmetric symbol into the matrix element, we have
M(H → γγ) = κmH ijk iH (j1 k2 + k1 j2 ) = κmH ikj iH (j1 k2 + k1 j2 ) (1053)
= −M(H → γγ) .
We are free to reorder the indices j and k by the Bose symmetry of the photons, but this
introduces a minus sign from the anti-symmetric object. This then proves that the decay of
a spin-1 Higgs boson to photons vanishes.

13.7 Combining Uncorrelated Measurements

In Fig. 13.3, we had argued that the excess around the Higgs boson mass was about 4.5σ,
as the total number of observed events was about 8400, and the expected number was 8000.
In Fig. 13.8, in the two bins near 125 GeV, we see that there are 13 observed events, while
only about 5 were expected (summing the values of the top of the red band in those bins).
Assuming
√ Poisson statistics, the standard deviation of the number of background events is
about 5 ' 2.2, and so the 13 observed events are a deviation of
13 − 5
√ ' 3.6σ . (1054)
5
With these results, we can then use the expression in example 13.3 to find the joint
deviation of these results, assuming that they are uncorrelated. We first evaluate the product
of error functions for these deviations:
   
4.5 3.6
1 − erf 2 √ 1 − erf √ 2
· ' 5.4 × 10−10 . (1055)
2 2
This is the p-value associate to observing these two excesses. To determine the combined
deviation σtot , we solve the implicit equation
 
1 − erf σ√tot2
= 5.4 × 10−10 . (1056)
2
This can be solved numerically and one finds
σtot ' 6.1 . (1057)
This is remarkably close to the official combined value from ATLAS of almost exactly 6σ.

155
13.8 Testing the Spin-2 Higgs Boson Hypothesis
13.8 (a)
Gluon 1 is traveling along the +ẑ axis and has left-handed helicity, and it is an initial state
particle so its polarization vector. As an initial state-particle, we must complex conjugate
its polarization vector, just like we identified with crossing external spinors from the final
state to the initial state. Its left-handed polarization vector is therefore:
1
∗g1 = √ (0, 1, i, 0) . (1058)
2
Gluon 2 is traveling in the opposite direction as gluon 1 and has opposite helicity. Therefore,
its spin is in the same direction as gluon 1, so it has the same polarization vector:
1
∗g2 = √ (0, 1, i, 0) . (1059)
2

13.8 (b)
Now, we are asked to determine the polarization vectors of the photons, γ1 and γ2 . Note
that the two photons also have back-to-back momenta and opposite helicity. This means
that their spin is in the same direction and that their polarization vectors are the same:

γ1 = γ2 . (1060)

So, to determine the polarization vectors, we just need to find one of them. The three-
momentum of photon 1 is rotated with respect to the +ẑ axis. In particular, note that the
direction of momentum is
    
0 1 0 0 0
∗  ∗ ∗ 
p̂γ1 =  sin θ =  0 cos θ sin θ 0 . (1061)
∗ ∗ ∗
cos θ 0 − sin θ cos θ 1

The matrix that implements the rotation away from the +ẑ axis is an element of SO(3), so
we can just use it to rotate the corresponding left-handed polarization vector. When first
aligned along the +ẑ axis, the polarization vector is just that which we identified in the
previous part. The rotated polarization vector is
     
1 0 0 1 1
1 1
~γ1 =  0 cos θ∗ sin θ∗  √  −i  = √  −i cos θ∗  . (1062)
0 − sin θ∗ cos θ∗ 2 0 2 i sin θ∗

Note that this polarization vector is indeed orthogonal to the photon momentum:

p̂γ1 · ~γ1 = 0 . (1063)

156
13.8 (c)
With these polarization vectors, we would like to evaluate their dependence in the matrix
element. By the spin configurations established in this problem, we have
M(gg → Hspin-2 → γγ) ∝ (∗g1 · γ1 )(∗g2 · γ2 ) + (∗g1 · γ2 )(∗g2 · γ1 ) (1064)
∝ (∗g · γ )2 .
In the second line, we note that the polarization vectors for the two gluons are equal to
one another and the polarization vectors for the two photons are equal to one another. As
shorthand, we denote these two relevant polarization vectors as ∗g and γ , respectively. Now,
taking the dot product of the polarizations from earlier, we have
1 1 + cos θ∗
∗g · γ = (0, 1, i, 0) · (0, 1, −i cos θ∗ , i sin θ∗ ) = − . (1065)
2 2
Then, the angular dependence in the matrix element is
M(gg → Hspin-2 → γγ) ∝ (∗g · γ )2 ∝ (1 + cos θ∗ )2 .
This vanishes if θ∗ = π, which corresponds to the left-handed photon traveling in the opposite
direction as the left-handed gluon (and similarly for right-handed particles). This is exactly
as our general discussion anticipated.

14 Particle Physics at the Frontier

14.1 Neutrinoless Double-β Decay at EXO
136
The EXO detector consists of 200 kg of Xe. This is 2 × 105 g and corresponds to
2 × 105
mol ' 8.9 × 1026 (1066)
136
xenon atoms. If the half-life of the neutrinoless double β-decay is 1025 years, this means that
if we watched 1025 xenon atoms for one year, there is 50% probability that such a decay will
be observed. Watching 2 × 1025 for one year would have probability 75% to have one decay,
etc., increasing to probability 1 as the number of xenon atoms increases. So, with a few
times 1025 xenon atoms, we will observe a decay in one year. EXO has about 9 × 1026 xenon
atoms, so we only need to watch it about one-tenth of the time to see a decay, or about one
month.

14.2 Double-Higgs Production

To use the NDA method to estimate the matrix element of the double Higgs process, we
can re-use results from the calculation of gluon-gluon fusion from Chap. 13. There, we had
found
αs yt m2H
M(gg → H) ' . (1067)
4π mt

157
This gluon-gluon fusion diagram is the left-side of diagram for gg → HH, so this gets us
most of the way to calculating the double Higgs diagram. To get the whole thing, we need to
multiply by the coupling of three Higgs bosons, which is λv. Further, because the four-point
matrix element is dimensionless, we need to multiply by appropriate factors of a mass to
make the matrix element dimensionless. The only relevant mass scale that we have to use
is mH , so we multiply by enough factors of mH to make the matrix element dimensionless.
That is,
αs yt m2H 1 αs yt λv
M(gg → HH) ' · λv · 2 = . (1068)
4π mt mH 4π mt
This is the complete result, but we can massage it into a nice form. The mass of the Higgs
boson is
m2H = 2λv 2 , (1069)
so we can eliminate the quartic coupling λ:

αs yt m2H
M(gg → HH) ' . (1070)
8π vmt

14.3 Borel Summation of a Convergent Series

To perform the Borel summation, we multiply and divide by a factorial:
∞
X xn ∞ Z ∞
1 X
= n
x = dt tn e−t . (1071)
1 − x n=0 n=0
n! 0

Now, we exchange the sum and integral:

∞ ∞ ∞
(xt)n
Z Z
−t
X 1
dt e = dt e−(1−x)t = . (1072)
0 n=0
n! 0 1−x

So, for a convergent series, the Borel summation technique gives us the same result as the
explicit sum. The final integral in this case converges for all x < 1.

14.4 The International Linear Collider

One major advantage to a linear collider is that there is no synchrotron radiation because
there is no acceleration due to change of motion. However, some drawbacks of a linear
collider are the fact that particles can’t be reused after the collision point. Further, you
can’t have multiple detectors located at different locations around a ring, like what was done
at Fermilab, LEP, and LHC.

158
14.5 The Largest Possible Collider
14.5 (a)
The power emitted in synchrotron radiation is

e2
P = E 4v4 . (1073)
6π0 m4 c11 R2
The relevant parts of this expression is E, the energy of the protons, and R, the radius of
the collider. Between the LHC and the Earth-collider, every other factor in this expression
is identical. That is, because the ratio of synchrotron power is 1 between these two colliders
we have the relationship
4 2
ELHC REarth
4 2
= 1. (1074)
EEarth RLHC
That is, the accessible energy of an Earth collider is
1/2
r
REarth 6 × 106
EEarth = 1/2 ELHC ' 13 TeV ' 485 TeV . (1075)
RLHC 4.3 × 103

14.5 (b)
Now, if we use all of Earth’s resources to counter synchrotron radiation, the energy of an
Earth collider would be
1/4
6π0 m4 c7 REarth
2

PEarth
EEarth = ' 44 J ' 2.8 × 1020 eV . (1076)
e2

14.5 (c)
Now, if we use all of Earth’s resources to counter synchrotron radiation, the energy of an
Earth orbital collider would be
1/4
6π0 m4 c7 REarth
2

orbit PEarth
EEarth = ' 7.1 × 103 J ' 4.4 × 1022 eV . (1077)
e2

For context, these energies are comparable to the highest energy cosmic rays that have ever
been observed, right at the GZK cutoff limit.

159