You are on page 1of 261

Phenomenology of Particle Physics

NIU Spring 2011 PHYS 686 Lecture Notes
Stephen P. Martin
Physics Department
Northern Illinois University
DeKalb IL 60115
spmartin@niu.edu
For corrections, see: http://zippy.physics.niu.edu/ppp/
January 5, 2011

Contents
1 Introduction

4

1.1

Fundamental forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4

1.2

Resonances, widths, and lifetimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5

1.3

Leptons and quarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5

1.4

Hadrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

7

1.5

Decays and branching ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

11

2 Special Relativity and Lorentz Transformations

13

3 Relativistic quantum mechanics of single particles

23

3.1

Klein-Gordon and Dirac equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

23

3.2

Solutions of the Dirac equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

29

3.3

The Weyl equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

37

3.4

Majorana fermions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

38

4 Maxwell’s equations and electromagnetism

40

5 Field Theory and Lagrangians

42

5.1

The field concept and Lagrangian dynamics . . . . . . . . . . . . . . . . . . . . . . . . .

42

5.2

Quantization of free scalar field theory . . . . . . . . . . . . . . . . . . . . . . . . . . . .

48

5.3

Quantization of free Dirac fermion field theory . . . . . . . . . . . . . . . . . . . . . . .

53

1

6 Interacting scalar field theories
φ4

57

6.1

Scalar field with

coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

57

6.2

Scattering processes and cross-sections . . . . . . . . . . . . . . . . . . . . . . . . . . . .

61

φ3

6.3

Scalar field with

coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

69

6.4

Feynman rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

76

7 Quantum Electro-Dynamics (QED)

83

7.1

QED Lagrangian and Feynman rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

83

7.2

e− e+

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

91

7.3

e− e+ → f f. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

96

7.4
7.5

µ− µ+

Helicities in

e− e+

µ− µ+

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

Bhabha scattering (e− e+ → e− e+ ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

7.6

Crossing symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

7.7

e− µ+ → e− µ+ and e− µ− → e− µ− . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

7.8
7.9

Møller scattering (e− e− → e− e− ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

Gauge invariance in Feynman diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

7.10 Compton scattering (γe− → γe− ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

7.11 e+ e− → γγ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
8 Decay processes

128

8.1

Decay rates and partial widths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128

8.2

Two-body decays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

8.3

Scalar decays to fermion-antifermion pairs; Higgs decay . . . . . . . . . . . . . . . . . . 131

8.4

Three-body decays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

9 Fermi theory of weak interactions

137

9.1

Weak nuclear decays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

9.2

Muon decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139

9.3

Corrections to muon decay
(e− ν

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

→ νe µ− ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

9.4

Inverse muon decay

9.5

e− ν e → µ− ν µ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

9.6
9.7

µ

Charged currents and π ± decay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Unitarity, renormalizability, and the W boson . . . . . . . . . . . . . . . . . . . . . . . . 160

10 Gauge theories

165

10.1 Groups and representations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
10.2 The Yang-Mills Lagrangian and Feynman rules . . . . . . . . . . . . . . . . . . . . . . . 175

2

11 Quantum Chromo-Dynamics (QCD)

182

11.1 QCD Lagrangian and Feynman rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
11.2 Quark-quark scattering (qq → qq) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
11.3 Renormalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186

11.4 Gluon-gluon scattering (gg → gg) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196

11.5 The parton model, parton distribution functions, and hadron-hadron scattering . . . . . 197

11.6 Top-antitop production in pp and pp collisions . . . . . . . . . . . . . . . . . . . . . . . . 205
11.7 Kinematics in hadron-hadron scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
11.8 Drell-Yan scattering (ℓ+ ℓ− production in hadron collisions) . . . . . . . . . . . . . . . . 211
12 Spontaneous symmetry breaking

215

12.1 Global symmetry breaking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
12.2 Local symmetry breaking and the Higgs mechanism . . . . . . . . . . . . . . . . . . . . 218
12.3 Goldstone’s Theorem and the Higgs mechanism in general . . . . . . . . . . . . . . . . . 222
13 The standard electroweak model

224

13.1 SU (2)L × U (1)Y representations and Lagrangian . . . . . . . . . . . . . . . . . . . . . . 224

13.2 The Standard Model Higgs mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
13.3 Fermion masses and Cabibbo-Kobayashi-Maskawa mixing . . . . . . . . . . . . . . . . . 233
13.4 Neutrino masses and the seesaw mechanism . . . . . . . . . . . . . . . . . . . . . . . . . 241
13.5 Search for the Standard Model Higgs boson . . . . . . . . . . . . . . . . . . . . . . . . . 243
13.5.1 Higgs decays revisited . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
13.5.2 Limits from the LEP2 e− e+ collider . . . . . . . . . . . . . . . . . . . . . . . . . 246
13.5.3 Searches at the Tevatron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
13.5.4 Searches at the LHC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
Appendix A: Natural units and conversions

250

Appendix B: Dirac Spinor Formulas

251

Appendix C: Further Reading

253

Index

253

3

1

Introduction

In this course, we will explore some of the tools necessary for attacking the fundamental questions of
elementary particle physics. These questions include:
• What fundamental particles is everything made out of?
• How do the particles interact with each other?
• What principles underlie the answers to these questions?
• How can we use this information to predict and interpret the results of experiments?
The Standard Model of particle physics proposes some answers to these questions. Although it is highly
doubtful that the Standard Model will survive intact beyond this decade, it is the benchmark against
which future theories will be compared. Furthermore, it is highly likely that the new physics to be
uncovered at the CERN Large Hadron Collider (LHC) can be described using the same set of tools.
This Introduction contains a brief outline of the known fundamental particle content of the Standard
Model, for purposes of orientation. These and many other experimental results about elementary
particles can be found in the Review of Particle Properties, hereafter known as the RPP, authored
by the Particle Data Group, and in its pocket-sized abridged form, the Particle Physics Booklet. The
current edition of the RPP was published as K. Nakamura et al. (Particle Data Group), J. Phys. G
37, 075021 (2010), and is about the size of a telephone book directory for a city of 50,000 people. It
is updated and republished every even-numbered year. An always-up-to-date version is available online, at http://pdg.lbl.gov, and it is usually considered the definitive source for elementary particle
physics data and analysis. Each result is referenced according to the experiments that provided it.
Background theoretical material needed for interpreting the results is also included. Copies can be
ordered for free from their website.

1.1

Fundamental forces

The known interaction forces in the Standard Model (besides the universal attraction of gravity) are the
electromagnetic force, the weak nuclear force, and the strong nuclear force. These forces are mediated
by the spin-1 (vector) bosons listed in Table 1:

photon

Width (GeV/c2 )

Lifetime (sec)

Force

0

EM

Charge

γ

0

±1

80.399 ± 0.023

2.085 ± 0.042

0

91.1876 ± 0.0021

2.4952 ± 0.0023

Z0
gluon

Mass (GeV/c2 )

Boson

g

0

0

3.14 × 10−25

2.64 × 10−25

“0”

weak
weak
strong

Table 1: The fundamental vector bosons of the Standard Model.

4

The photon is the mediator of the electromagnetic force, while the W ± and Z 0 bosons mediate the
weak nuclear force, which is seen primarily in decays and in neutrino interactions. The W + and W −
bosons are antiparticles of each other, so they have exactly the some mass and lifetime. The gluon has
an exact 8-fold degeneracy due to a degree of freedom known as “color”. Color is the charge associated
with the strong nuclear force. Particles that carry net color charges are always confined by the strong
nuclear force, meaning that they can only exist in bound states, Therefore, no value is listed for the
gluon lifetime, and the entry “0” for its mass is meant to indicate only that the classical wave equation
for it has the same character as that of the photon. Although the fundamental spin-1 bosons are often
called force carriers, that is not their only role, since they are particles in their own right.

1.2

Resonances, widths, and lifetimes

Table 1 also includes information about the width Γ of the W and Z particle resonances, measured in
units of mass, GeV/c2 . In general, resonances can be described by a relativistic Breit-Wigner lineshape,
which gives the probability for the kinematic mass reconstructed from the production and decay of the
particle to have a particular value M , in the idealized limit of perfect detector resolution and an isolated
state. For a particle of mass m, the probability is:
P (M ) =

f (M )
,
(M 2 − m2 )2 + m2 Γ(M )2

(1.1)

where f (M ) and Γ(M ) are functions that usually vary slowly over the resonance region M ≈ m, and
thus can be treated as constants. The resonance width Γ ≡ Γ(m) is equivalent to the mean lifetime,
which appears in the next column of Table 1; they are related by

τ (in seconds) = (6.58212 × 10−25 )/[Γ (in GeV/c2 )].

(1.2)

The RPP lists the mean lifetime τ for some particles, and the width Γ for others. Actually, the
Standard Model of particle physics predicts the width of the W boson far more accurately than the
experimentally measured width indicated in Table 1. The predicted width, with uncertainties from
input parameters, is ΓW = 2.091 ± 0.002 GeV/c2 .

1.3

Leptons and quarks

The remaining known indivisible constituents of matter are spin-1/2 fermions, known as leptons (those
without strong nuclear interactions) and quarks (those with strong nuclear interactions). All experimental tests are consistent with the proposition that these particles have no substructure. The leptons
are listed in Table 2. They consist of negatively charged electrons, muons, and taus, and weakly interacting neutrinos. There is now good evidence (from experiments that measure oscillations of neutrinos
produced by the sun and in cosmic rays) that the neutrinos have non-zero masses, but their absolute
values are not known except for upper bounds as shown. The quarks come in 6 types, known as
“flavors”, listed in Table 3.
5

due to the confining nature of the strong force. The lifetimes of the d.08 to 0. or generations. All baryons are fermions with half-integer spin. s. or anti-green). the second is µ− .13 charm c 2/3 1.9 × 10−4 0 < 0. ν µ . A colorless bound state can be formed either from three quarks (a baryon). µ+ .3 Table 3: The quarks of the Standard Model. d. and the same mass and spin. c. ν τ . νe . t). c. c. u.197 × 10−6 −1 1. b quarks are also fuzzy. but other definitions give quite different values.018 2. and even the definition of the mass of a quark is subject to technical difficulties and ambiguities. The masses of the five lightest quarks (d.3 × 10−3 4. or from a quark with a given color and an antiquark with the corresponding anti-color (a meson).47 to 1. u. with the opposite charge and color. The masses of the fermions of a given charge increases with the family. c. b. This is related to the fact that quarks exist only in colorless bound states.electron Lepton Charge Mass (GeV/c2 ) Mean Lifetime (sec) e− −1 5. νµ . Antiquarks carry anticolor (anti-red. b. anti-blue. In contrast. ντ .83 bottom b −1/3 top t 2/3 1. or from three antiquarks (an anti-baryon).1 × 10−3 to 5.105658367(4) 2. The fermions listed in Tables 2 and 3 are often considered as divided into families. but it is an experimental fact that these family-changing reactions are highly suppressed. there is an antiquark (d. s.8 × 10−3 up u 2/3 strange s −1/3 0.0 173. For each quark. and all mesons are bosons with integer spin.91 × 10−13 Table 2: The leptons of the Standard Model. The antileptons are positively charged e+ . d. and are best described in terms of the hadrons in which they live. s. u. called hadrons. u.1 ± 1.10998910(13) × 10−4 ∞ −1 0. s.77682(16) νe muon 0 µ− νµ tau 0 τ− ντ <2× 10−9 < 1. b) are somewhat uncertain. s. Quark Charge Mass (GeV/c2 ) down d −1/3 4. t. τ + and antineutrinos ν e .7 × 10−3 to 3. The first family is e− . b. and pole masses for c. The quark mass values shown in Table 3 correspond to particular technical definitions of quark mass used by the RPP (I’ve chosen to quote the MS masses for u. All of the fermions listed above also have corresponding antiparticles. t) with the same mass but the opposite charge.7 to 5. The weak interactions mediated by W ± bosons can change quarks of one family into those of another. and the third is τ − . the top quark 6 .

6 × 1036 n (udd) 0 0.939565 885. Also. with spin 0 and charge 0.9 × 10−10 1.6 × 10−25 seconds) is so short that it decays before it can form hadronic bound states (which take roughly 3 × 10−24 seconds to form). some extensions of the Standard Model (for example technicolor) do not have any fundamental Higgs boson.31486 Ξ− (dss) −1 1. 1.4 GeV/c2 lower mass bound from LEP2.938272 > 6.4 Hadrons As remarked above. and experimental constraints from the LEP2 experiment require its mass to be larger than 114.63 × 10−10 (uus) +1 1. it is possible to evade the 114.mass is relatively well-known.19745 Ξ0 −1 (uss) 0 1. but these are just arbitrary labels. In the simplest version of the Standard Model.4 × 10−20 1. Other extensions of the Standard Model (notably supersymmetry) imply the existence of more than one fundamental Higgs boson.11568 Σ+ 2. and so its mass and width can be defined in a way that is not subject to large ambiguities. associated with the color that is the source charge for the strong force. They and other similar baryons with total angular momentum (including both constituent spins and orbital angular momentum) J = 1/2 are listed in Table 4. However. resulting in only 7 . nor even a well-defined resonance that plays the role of a Higgs boson. Each of these quarks has an exact 3-fold degeneracy.3217 8. there is no experiment that could tell a red quark from a green quark. The proton may be absolutely stable. The most common are the nucleons (the proton and the neutron).75%.48 × 10−10 2.19264 Σ− (dds) 1.64 × 10−10 Table 4: Baryons with J = 1/2 made from light (u. the baryons that make up most of the directly visible mass in the universe. with an uncertainty at this writing of order only about 0.18937 Σ0 (uds) 0 1. d. and blue. experiments to try to observe its decays have not found any. in extensions of the minimal Standard Model.02 × 10−11 7. green. J = 1/2 baryon Charge Mass (GeV/c2 ) Lifetime (sec) p (uud) +1 0. but there are also virtual (or “sea”) quark-antiquark pairs and virtual gluons in each of these and other hadrons. The colors are often represented by the labels red. The quarks listed in parentheses are the valence quarks of the bound state. at 95% confidence. even in principle. the Higgs boson is is required to exist as part of the model specification. There may also be a fundamental Higgs boson. quarks and antiquarks are always found as part of colorless bound states.4 GeV/c2 in the context of that model.7 Λ (uds) 0 1. Therefore it behaves like a free particle during its short life. This is because the top-quark mean lifetime (about 4. s) quarks.

except for the Σ0 baryon.672 8. The mass of the baryons in Table 4 increases with the number of valence strange quarks contained.6 × 10−23 8. and in particular it is definitely not the explanation for most of the mass that is directly observed in the universe.21 × 10−11 Table 5: J = 3/2 baryons.8 × 10−23 (sud) 0 1.532 0. Very narrow resonances correspond to very long-lived states. electron.118 ∆+ 5. J = 3/2 baryon Charge Mass (GeV/c2 ) Γ (GeV/c2 ) Lifetime (seconds) ∆++ (uuu) +2 1.7 × 10−23 7. 8 . but it decays into a proton.036 Σ∗− (sdd) 1.0091 Ξ∗− (ssd) 1.384 0. which has the same valence quark content: Σ0 → Λγ. the Ω− is by far the narrowest and most stable of the ten J = 3/2 baryon ground states listed. bottom.232 0.039 Ξ∗0 −1 (ssu) 0 1. and antineutrino (n → pe− ν¯e ).383 0. so a little off from the present best values) in parentheses. The neutron lifetime is also relatively long. In that sense. with some of the more common ones listed in Table 5. so Σ(1385) and Ξ(1530). Most of the masses of the W ± and Z bosons and the top. The other J = 1/2 baryons decay in times of order 10−10 seconds by weak interactions. the Higgs mechanism is by no means necessary to understand the origin of all mass. which decays extremely quickly by an electromagnetic interaction into the Λ. the RPP chooses to denote them by their approximate mass in MeV (as determined by older experiments. Note that the masses of the proton and the neutron (and all other hadrons) are much larger than the sums of the masses of the valence quarks that make them up. However.535 0. Instead of the ∗ notation to differentiate these states from the corresponding J = 1/2 baryons with the same quantum numbers. There are also J = 3/2 baryons. not listed here.0 × 10−15 1. Therefore.2 × 10−23 6. and strange quarks and the leptons are indeed believed to come from the Higgs mechanism.8 × 10−23 1. one can think of the Σ0 as being an excited state of the Λ. it is quite wrong (although fashionable) to say that the search for the Higgs boson is needed to understand the “origin of mass”.a very high lower bound on the mean lifetime.6 × 10−24 (uud) +1 "" "" "" ∆0 (udd) 0 "" "" "" ∆− (ddd) "" "" "" Σ∗+ −1 (suu) +1 1. Nucleons dominate the visible mass of particles in the universe. There are other excited states of these baryons.387 0.0099 Ω− −1 (sss) −1 1. The RPP uses a slightly different notation for the Σ∗ and Ξ∗ J = 3/2 baryons. to be discussed below. These nucleon masses come about from the strong interactions by a mechanism known as chiral symmetry breaking. charm.036 Σ∗0 1.

Σb . you can find them yourself in the RPP. the mean lifetimes are respectively 5. and Σb . generally with heavier masses. dd. Ωb . One of the interaction eigenstates (KL0 ) is actually much longer-lived than the other (KS0 ).62 GeV/c2 to 5.139570 (u¯ s). (sd) ±1 0.82 GeV/c2 . c) Mass (GeV/c2 ) Table 7: J = 0 mesons containing one heavy and one light quark and antiquark. However.7 GeV/c2 . with masses ranging from 5. down or strange). dd) ¯ (d¯ (ud). The lowest lying states with a 0 0 + 0 + ++ charm quark include the Λ+ c .493677 0 0.366 D± Ds± B± Bs0 (c¯ u). that can be thought of as excited states of the more common ones listed above. (s¯ c) (u¯b). (b¯ u) ¯ (d¯b). K 0 η η′ Charge ¯ (u¯ u.12 × 10−8 and 8. s) ¯ s¯ (u¯ u. d. so they are exactly degenerate mass pairs with the same lifetime. The lifetimes (and the widths) of the other J = 0 mesons are not listed here.There are also baryons containing heavy (c or b) quarks. The only ones that have been definitively observed so far have J = 1/2 and contain exactly one heavy quark. Ξc .497614 ¯ s¯ (u¯ u. there are other baryons.279 B 0. and Ωc resonances with masses ranging from 2. s) quarks and antiquarks. The most common ones are listed in Table 7. s) 0 0.95778 Table 6: J = 0 mesons containing light (u.280 Bs0 . More information about them can be found in the RPP. (b¯ s) 0 5. (u¯ c) ¯ (d¯ (cd).9685 ±1 5. (bd) 0 5.54785 0 0. Σc . u) Mass (GeV/c2 ) 0 0. the K 0 and K 0 mesons are mixed and not quite exactly degenerate in mass. Ξc .29 − + − GeV/c2 to 2. They always carry integer total angular momentum J. (s¯ u) ¯ (d¯ s). 9 . dd. Besides the J = 0 mesons listed above. The charged pions π ± are antiparticles of each other. Ξ0b . Σc . D0 Charge 0 1. there are counterparts containing a single heavy (charm or bottom) quark or antiquark. (s¯b).134977 ±1 0.95 × 10−11 seconds. The most common J = 0 mesons are listed in Table 6.8696 ±1 1. B 0 (c¯ s). Bound states of a valence quark and antiquark are called mesons. Ξ− b . J = 0 meson π0 π± K± K 0. J = 0 meson D0 . Again. with the other antiquark or quark light (up. as are the charged kaons K ± . Here the bar over a quark name denotes the corresponding antiquark. Σc . and those with a bottom quark include the Λ0b .8648 ±1 1.

302 5. dd) ω0 ¯ (u¯ u.7755 0 "" 0 0. (b¯ s) ±1 0 5. s) valence quarks and antiquarks: J = 1 meson ρ0 ¯ (d¯ (ud). J = 1 meson D∗0 . Table 9 lists the most common ones that contain only light (u. s) 0 1. (s¯ u) ¯ (d¯ s).7827 ±1 0. dd.348 (s¯b). (sd) 0 0. d. (b¯ u) ±1 2. K ∗0 (u¯ s). (b¯ c) (b¯b) ηb “bottomonium” Mass (GeV/c2 ) 0 2. Note that these have the same charges and slightly larger masses than the corresponding J = 0 mesons in Table 7.01946 Table 9: J = 1 mesons containing light quarks and antiquarks. The ones with the lowest-lying masses are listed in Table 8: J = 0 meson Charge ηc “charmonium” (c¯ c) Bc± (c¯b). Bs∗0 Charge (c¯ u).There are also J = 0 mesons containing only charm and bottom quarks and antiquarks. Vector (J = 1) mesons are also very important. (u¯ c) ¯ (d¯ (cd).007 ±1 2. (s¯ c) ¯ (d¯b). c) Mass (GeV/c2 ) 0 2.277 0 9. (bd) (u¯b). u) ¯ (u¯ u.980 ±1 6. The most common J = 1 mesons containing one heavy (c or b) quark or antiquark are likewise shown in Table 10.112 0 5.8917 K ∗0 .391 Table 8: J = 0 mesons containing a heavy quark and a heavy antiquark.8959 φ ¯ s¯ (u¯ u.415 Table 10: J = 1 mesons containing one heavy and one light quark and antiquark.010 (c¯ s). D ∗0 D ∗± Ds∗± B ∗0 . B ∗0 B ∗± Bs∗0 . J = 1 mesons with both quark and antiquark heavy are shown in Table 11: 10 . dd) K ∗± ρ± Charge Mass (GeV/c2 ) ±1 0.

Υ from Tables 8 and 11. but with excited radial bound-state wavefunctions. The spectroscopy of these states provides a striking confirmation of the quark model for hadrons and of the strong force. The heavy quarkonium (cc and bb) systems have other states besides the ηc . bound states of gluons. including the decay widths and the decay products. 1. However. J/ψ and ηb . these states are expected to mix with excited quark-antiquark bound states.8) → Λγ (1.13) τ (1.3) 0 − → π π (1. there are J = 0 mesons χc0.12) K + → π+ π0 B + Ω − → D 0 µ+ ν − → K Λ. (1.6) ω → π π π There are also decays that are mediated by electromagnetic interactions.9) + − → π π γ.1.5 Decays and branching ratios In some cases. ψ(2S).4603 Table 11: J = 1 mesons containing a heavy quark and a heavy antiquark. Much more detailed information on all of these hadronic bound states (and many others not listed above). there could also be J = 1 Bc∗± mesons.11) − → µ ν¯µ (1.J = 1 meson Charge J/ψ “charmonium” (c¯ c) (b¯b) Υ “bottomonium” Mass (GeV/c2 ) 0 3. Υ(4S). and they will be extremely difficult to identify experimentally. hadrons can decay through the strong interactions.7) ∆+ → pγ Σ 0 ρ 0 (1. there are excited bottomonium states Υ(2S). Some examples include: ∆++ → pπ + ρ − (1. Υ(10860). Theoretically. In principle. ψ(3770).4) + − 0 (1. that is. in the bb system. ψ(3872) that are similar to the ηc and J/ψ.2 that have the quark and antiquark in P -wave orbital angular momentum states.1. and Υ(11020) with J = 1. for example: π n → pe− ν¯e − (1. for example: π 0 → γγ (1. but (unlike their J = 0 counterparts in Table 8) their existence has not been established experimentally.14) (1. There are also states ηc (2S). can be found in the RPP.10) The smallest decay widths for hadrons are those mediated by the weak interactions.5) φ → K + K −. 11 (1. Υ(3S). For the cc system.1. Similarly.096916 0 9. with widths of order tens or hundreds of MeV. one also expects exotic mesons that are mostly “gluonium” or glueballs. χb0.15) .2 (2P ). and P -wave orbital angular momentum states with total J = 0.2 (1P ) and χb0. (1.

In other applications.7 ± 0. The theoretical road aims to match these results onto predictions of quantum field theories specified in terms of a small number of parameters. of the decays: BR(ω → π + π − π 0 ) = (89. production rates. and the sum of the partial widths is equal to the total decay width. which is highly successful as indicated by the impressive volume and detail in the RPP. providing amazingly accurate predictions for observable quantities such as magnetic moments and interaction rates. and eventually the description of the Standard Model in terms of it.21) (1. or do not converge even in principle.20) (1. decay rates. There are two roads to enlightenment regarding the Standard Model and its future replacement. In most cases.1 ± 0. As a randomly chosen example.7)% BR(ω → π 0 γ) = (8. In the following notes. we will try to understand some of the basic calculation methods of quantum field theory as a general framework.16) τ − → ντ µ− ν¯µ (1. If the total decay width for a parent particle X is Γ(X).3)% (EM) (1. but not all.22) with other final states totaling less than 1%. (1.19) Experimentally.9 ± 0. and provides understanding that is often elusive in the raw data. (1. etc. For some calculations. The fraction that each final state contributes to the total decay width is known as the branching ratio (or branching fraction). quantum field theory is only partly successful. “3-prong” (if exactly three charged hadrons). branching ratios. then the partial decay width of X into a particular final state Y is Γ(X → Y ) = BR(X → Y )Γ(X). However.3)% (strong) (EM) BR(ω → π + π − ) = (1. finds the answers to masses.17) → ντ + hadrons. In such cases. only rough or even qualitative results are possible. In the case of electromagnetic interactions.18) τ τ − − − → ντ e ν¯e (1. The experimental road. a variety of different decay modes contribute to each total decay width. quantum field theory is systematic and elegant. the hadronic τ decays are classified by the number of charged hadrons present in the final state.23) The sum of all of the branching ratios is equal to 1. perturbation theory and other known methods are too difficult to carry out. as either “1-prong” (if exactly one charged hadron). It is also common to present this information in terms of the partial widths into various final states. and even more detailed information like kinematic and angular distributions directly from data in high-energy collisions. 12 . quantum field theory is extremely successful.The weak interactions are also entirely responsible for the decays of the charged leptons: µ− → νµ e− ν¯e (1. in the case of the ω meson the strong interaction accounts for most. usually abbreviated as BR or B.

1) The Greek indices µ.5) where β = v/c. The laws of physics should not depend on what coordinate system we use. It is often useful to change our coordinate system (or “inertial reference frame”) according to xµ → x′µ = Lµ ν xν . Spacetime has four dimensions. suppose we rotate our coordinate system about the z-axis by an angle α.2) Such a change of coordinates is called a Lorentz transformation.4) Alternatively. x. xµ is an example of a contravariant four-vector. z ′ ) (2. 2. For any given event (for example. x′ . . or a particle decays to two other particles) one can assign a four-vector position: (ct. 1. (2. x2 . 3. y. and c is the speed of light in vacuum. As a matter of terminology. however. x3 ) = xµ (2.3) where ct′ = ct x′ = x cos α + y sin α y ′ = −x sin α + y cos α z ′ = z. As a simple example of a Lorentz transformation. Then: ct′ = γ(ct − βz) x′ = x y′ = y z ′ = γ(z − βct). run over the values 0. with the origins of the two frames coinciding at time t = t′ = 0.6) . It is not arbitrary. Then in the new coordinate system: x′µ = (ct′ . (2. as we will soon see. ρ. ν. this is a guiding principle in making a sensible theory. a firecracker explodes. . z) = (x0 . y ′ . x1 . we could go to a frame moving with respect to the original frame with velocity v along the z direction. Let us begin by reviewing some important features of special relativity. q γ = 1/ 1 − β 2 13 (2. .2 Special Relativity and Lorentz Transformations A successful description of elementary particles must be consistent with the two pillars of modern physics: special relativity and quantum mechanics. (2. Here Lµ ν is a 4 × 4 real matrix that parameterizes the Lorentz transformation.

All contravariant four-vectors transform the same way under a Lorentz transformation: a′µ = Lµ ν aν . (2. 1 0  0 γ (L−1 )µ ν  γ  0 =  0 βγ 0 1 0 0 (2. which is independent of the Lorentz frame. So.5). (2.9) In particular.5).12) where gµν   1 0 0 0  0 −1  0 0  = 0 0 −1 0  0 0 0 −1 14 (2. (2. so that γ = cosh ρ and βγ = sinh ρ.11) A key property of special relativity is that for any two events one can define a proper interval.Another way of rewriting this is to define the rapidity ρ by β = tanh ρ. 0. x′0 = x0 cosh ρ − x3 sinh ρ x′1 = x1 x′2 = x2 x′3 = −x0 sinh ρ + x3 cosh ρ. where dµ is some four-vector displacement. 0  γ (2. In the example of eq. (2. Then we can rewrite eq. 0.8) In the rest frame of a particle of mass m. The proper interval between the events is (∆τ )2 = (d0 )2 − (d1 )2 − (d2 )2 − (d3 )2 = gµν dµ dν (2. Another example of a contravariant four-vector is given by the 4-momentum formed from the energy E and spatial momentum p~ of a particle: pµ = (E/c. the 4-momentum of a particle is related to its mass by the Lorentz transformation that relates the frame of reference in which it is measured and the rest frame. its 4-momentum is given by pµ = (mc. 0). and which tells us how far apart the two events are in a coordinate-independent sense.13) . consider two events occurring at xµ and xµ + dµ . one has: Lµ ν  γ  0 =  0 −βγ 0 1 0 0 and the inverse Lorentz transformation is aµ = (L−1 )µ ν a′ν . (2.10) 0 0 1 0  βγ 0  . p~ ).  0 −βγ 0 0  .7) This change of coordinates is called a boost (with rapidity ρ and in the zˆ direction).

and from now on. and otherwise = 0. 15 . . in which repeated indices µ. −px . if pµ and q µ are the four-momenta of any two particles. If aµ and bµ are any four-vectors. plus the on-shell condition (2. −x. −py .19) Lµ ν = gµρ gνσ Lρ σ .22) The Lorentz invariance of dot products of pairs of 4-momenta. then p · q is a Lorentz-invariant. (2. people often use a harmlessly sloppy terminology and neglect the distinction.20) where (note the positions of the indices!) Because one can always use the metric to go between contravariant and covariant four-vectors.17) (2. (2. a particle with mass m satisfies the on-shell condition p2 = pµ pµ = E 2 /c2 − p~ 2 = m2 c2 .21) is a scalar quantity. (2. It follows that gµν Then one has. (2. −z). −y. The existence of the metric tensor allows us to define covariant four-vectors by lowering an index: xµ = gµν xν pµ = gµν pν = (ct. Here. it does not depend on which inertial reference frame it is measured in. are taken to be summed over.  1 0 = 0 0  0 0 0 −1 0 0  . aµ = gµν aν . then aµ bν gµν = aµ bν gµν = aµ bµ = aµ bµ ≡ a · b (2.18) It follows that covariant four-vectors transform as a′µ = Lµ ν aν (2.16) where δνµ = 1 if µ = ν. simply referring to them as four-vectors. we adopt the Einstein summation convention. (2. for any vector aµ . 0 −1 0  0 0 −1 aµ = gµν aν . . one can define an inverse metric gµν so that gµν gνρ = δρµ . It is an assumption of special relativity that gµν is the same in every inertial reference frame. For example.is known as the metric tensor. (2.22). is enough to solve most problems in relativistic kinematics.15) Furthermore. ν. −pz ). plus the conservation of total fourmomentum.14) = (E/c. In particular. .

0.28) . Let us find the threshold energy and momentum 4-vectors for this process in the COM (center-of-momentum) frame and in the frame in which one of the initial-state particles is at rest. − E 2 − m2 ).22) and the fact that the masses are assumed equal. 0. in the COM frame. (2.24) The spatial momenta are required to be opposite by the definition of the COM frame. colliding. The total 4-momentum of the initial state is pµ = (2E. by a choice of units. each of mass m.27) q q The angle θ parametrizes the arbitrary direction of the scattering. In order for the spatial momentum components to be real.25) is a Lorentz invariant. 0. cos θ Ef2 − M 2 ) (2. we will take c = 1. we can take the colliding initial-state particles to be moving along the z-axis. p E 2 − m2 ) (2. which in turn requires the energies to be the same.Let us pause to illustrate this with an example. − cos θ Ef2 − M 2 ). we have taken the scattering to occur within the yz plane. Then their 4-momenta are: pµ1 = (E. Now. 0. 0. so let us consider that case first. so Ef = E.26) k2µ = (Ef .23) p pµ2 = (E. requiring conservation of total 4-momentum gives k1µ + k2µ = pµ1 + pµ2 . Furthermore. using eq. Without loss of generality. 16 (2. Similarly. the same in any inertial frame. and thus the energies to be equal to a common value Ef because of the assumed equal masses M . 0). sin θ Ef2 − M 2 . Throughout most of the following. 0. (2. we therefore find the energy threshold condition in the COM frame E > Ethresh = M. and so this must be equal to the total 4-momentum of the final state in the COM frame as well. Without loss of generality. the final state 4-momenta can be written as: q q k1µ = (Ef . − sin θ Ef2 − M 2 . 0. 0. Suppose the result of the collision is two final-state particles each of mass M . (2. Relativistic kinematics problems are often more easily analyzed in the COM frame. Consider the situation of two particles. p2 = 4E 2 (2. as shown: ~k1 θ p~1 p~2 ~k2 The fact that we are in the COM frame again requires the spatial momenta to be opposite.

the Lab frame threshold energy condition for the scattering event to be possible is m(E ′ + m) > 2M 2 . 3 components of the µ ν equation p′µ 2 = Λ ν p2 :  m 0  =  γ βγ βγ γ  It follows that √ E − E 2 − m2 q 1 − m2 /E 2 = β = γ = E 2 /m2 − 1 = βγ = s . To find the Lorentz transformation needed to go from the COM frame to the Lab frame. The total initial state 4-momentum is therefore p′µ = (E ′ + m.35) E′ + m . (2.30) are the 4-momenta of the two initial-state particles.25).36) E′ − m . corresponding to a fixed-target experiment. 0.32) Because we already found E > M . (2. E ′2 − m2 ). 0) 2 E ′2 − m2 ). p = (E 2 /m)  2 2   1 − M /E sin θ (m/E) p p 2 2 2 2 1 − m /E + cos θ 1 − M /E 17 (2. In the Lab frame.39) . m (2. 2m (2. (2. (2.31) This must be the same as eq. 0.29) (2. leading to a Lorentz invariant p′2 = (E ′ + m)2 − (E ′2 − m2 ) = 2m(E ′ + m). consider the 0.Now let us reconsider the problem in a frame where one of the initial-state particles is at rest. 0. (2. 0. 1 p p′µ = (m. For the first final-state particle: k1′µ  γ  0 =   0 βγ 0 1 0 0    E 0 βγ    0 0   √0   2 2 sin θ √E − M  1 0 0 γ cos θ E 2 − M 2 p p (2. 0. or ′ = E ′ > Ethresh 2M 2 − m2 . so the Lab frame energy is related to the COM energy of each particle by m(E ′ + m) = 2E 2 . and E ′ is the Lab frame energy of the moving √ particle.34) E′ − m .37) Now we can apply this Lorentz boost to the final-state momenta as found in the COM frame to obtain the Lab frame momenta. 0. p′µ = (E ′ .33) Let us also relate the Lab frame 4-momenta to those in the COM frame. 2m (2.38)  1 + cos θ 1 − m2 /E 2 1 − M 2 /E 2   0 . E′ + m 1 = E/m = 1 − β2 p q s  s (2.

) Then the pseudo-rapidity is defined by  1 |~ p | + pz η = ln 2 |~ p | − pz  = − ln [tan(θ/2)] .40) For fixed θ in the COM frame. the ordinary rapidity is given by: ρ=   E + |~ p| 1 ln . (The other final-state momentum is obtained by just flipping the signs of cos θ and sin θ. in fixed-target collisions the energy required scales like 2M 2 /m ≫ M .33) that while the production of a pair of heavy particles of mass M requires beam energies in symmetric collisions that scale like M . in general y does depend on the energy. where m is the beam particle mass. Note that the pseudo-rapidity depends only on the direction of the particle. not on its energy. (2. and they are very nearly equal for a particle whose energy is large compared to its mass. In collider applications. (2. while those at large |η| are said to be forward. A final state particle (or group of particles) emerging at an angle θ with respect to Beam 1 in general has a four-vector momentum given by: pµ = (E. not in the center-of-momentum frame of the scattering event.) The Lab-frame scattering angle with respect to the original collision axis (the z-axis in both frames) is determined by (m/E) sin θ p tan θ ′ = p . it is common to see the direction of a final-state particle with respect to the colliding beams described either by the pseudo-rapidity η or the longitudinal rapidity y. and E = p |~ p |2 + m2 is the energy.41) where pT = |~ p | sin θ is the transverse momentum.28) and (2.44) . pT cos φ. 2 E − |~ p| 18 (2. this four-vector is generally defined in the lab frame. regardless of the sign of cos θ. The longitudinal rapidity is defined somewhat similarly by y=  1 E + pz ln 2 E − pz  . Notice from eqs. However. depending on the situation) are said to be central.43) In fact. Suppose that the two colliding beams are oriented so that Beam 1 is going in the zˆ direction and Beam 2 is going in the −ˆ z direction. with m the mass and p~ the three-vector momentum. (2. the z-component of the momentum is always positive (in the same direction as the incoming particle in the Lab frame). |θ ′ | in the Lab frame decreases with increasing E/m. as the produced particles go more in the forward direction. pz ).42) Thus η = 0 corresponds to a particle coming out perpendicular to the beam line (θ = 90◦ ). pT sin φ. pz = |~ p | cos θ is the longitudinal momentum. which is often unknown. (2.Note that for M > m. 2 2 1 − m /E / 1 − M 2 /E 2 + cos θ (2. 180◦ ). This is why fixed-target collisions are no longer an option for frontier physics discoveries of very heavy particles or high-energy phenomena. Particles at small |η| (less than 1 or 2 or so. η = y in the special case of a massless particle. (In hadron colliders. For the same particle. while η = ±∞ correspond to the directions along the beams (θ = 0.

47) with gρκ . and is therefore known as time reversal: x′0 = −x0 x′1 = x1 x′2 = x2 x′3 = x3 . while ρ is the rapidity of the boost needed to move to the particle’s rest frame. it must be that: gµν Lµ ρ Lν σ = gρσ .51) This just flips the sign of the time coordinate. and among non-collider physicists it is common to see the letter η used to refer to the ordinary rapidity. = 0 0 −1 0  0 0 0 −1 19 (2. it has become a standard abuse of language among collider physicists to call y simply the rapidity. or space inversion: Lµ ν   1 0 0 0  0 −1  0 0 .52) Another “large” Lorentz transformation is parity.50) Let us now consider some more particular Lorentz transformations. (See homework problem. then gµν a′µ b′ν = gµν aµ bν .46) so that Since aµ and bν are arbitrary.48) Applying this to eqs.45) gµν Lµ ρ Lν σ aρ bσ = gρσ aρ bσ . (2.53) .2) and (2. In matrix form. we obtain Lν κ Lν σ = δσκ (2.19). called ρ here. (2. (2. it could be written as LT gL = g. (2. To begin.21) implies that.49) aν = a′µ Lµ ν (2. The Lorentz-invariance of equation (2. If we contract eq. det(L) = ±1. we note that as a matrix. if aµ and bµ are constant four-vectors.47) This is the fundamental constraint that a Lorentz transformation matrix must satisfy. we find that the inverse Lorentz transformation of any four-vector is aν = a′µ Lµ ν (2. Confusingly. 0 1 (2.The quantity y is the rapidity of the boost needed to move to a frame where the particle has no longitudinal momentum along the beam direction.) An example of a “large” Lorentz transformation with det(L) = −1 is: Lµ ν  −1  0 =  0 0 0 1 0 0 0 0 1 0  0 0 . (2. (2. Some care is needed to ensure that one is using and interpreting these quantities consistently. Now let us return to the study of the properties of Lorentz transformations.

σ = 0. states. . (2. (2.] Under a Lorentz transformation from coordinates xµ → x′µ . suppose that F (x) is a scalar function of xµ . σ = 1. including operators. .54) It was once thought that the laws of physics have to be invariant under these operations. Lorentz transformations obey the mathematical properties of a group. Likewise.56) gρσ + ωσρ + ωρσ + . with 4 · 3/2 · 1 = 6 independent entries. .)(δσν + ω ν σ + .55) where we agree to drop everything with more than one ω µ ν . and fields. So far we have considered constant four-vectors. all experiments up to now are consistent with invariance of the laws of physics under the subset of Lorentz transformations that are continuously connected to the identity. We’ll study these group representations in more detail later. particles.58) or Therefore is an antisymmetric 4 × 4 matrix. x1 . x3 ). 1 or 0. The subset of Lorentz transformations that can be built out of repeated infinitesimal boosts and rotations form a smaller group.47). at a given fixed point 20 . known as the Lorentz group.2 or 0. . called the proper Lorentz group. It is a mathematical fact that any Lorentz transformation can be built up out of repeated infinitesimal boosts and rotations.so that: x′0 = x0 x′1 = −x1 x′2 = −x2 x′3 = −x3 . specifically in the weak decays of the 60 Co nucleus and the K ± mesons. They can be built up out of infinitesimal Lorentz transformations: Lµ ν = δνµ + ω µ ν + O(ω 2 ). In the Standard Model of particle physics and generalizations of it. However. (2.3) and 3 boosts (ρ. combined with the operations of time-reversal and space inversion. according to eq. it was shown experimentally in the 1950’s that parity is violated in the weak interactions. 2 or 1. so F (x) really means F (x0 .) = gρσ . For example. all interesting objects. these are known as “proper” Lorentz transformations and have det(L) = +1. Then. = gρσ . x2 .3). However.57) ωσρ = −ωρσ (2. [It is usual to leave the index µ off of xµ when it is used as the argument of a function. gµν (δρµ + ω µ ρ + . one can also consider four-vectors that depend on position in spacetime. However. .3 or 2. These correspond to 3 rotations (ρ. (2. experiments in the 1960’s on the decays of K 0 mesons showed that time-reversal invariance is violated (at least if very general properties of quantum mechanics and special relativity are assumed). (2. . live in representations of the Lorentz group.

and transforms appropriately when one goes to a new reference frame. A tensor is an object that can carry an arbitrary number of spacetime vector indices.. c2 ∂t2 (2. (2. This is because: ∂ ∂xν ∂ ′ ′ F (x ) = F (x) = Lµ ν (∂ν F )(x).66) B µ = (0. The objects gµν and gµν and δνµ are constant tensors.67) C µ = (0.49) with aµ = xµ . it transforms so that in the primed coordinate system.] By raising the index... Four-vectors and scalar functions and 4-derivatives of them are also tensors..63) The object −∂ µ ∂µ F is a 4-dimensional generalization of the Laplacian.59) Then ∂µ F ≡ ∂F 1 ∂F ~ =( . Tν′µ1 ν12µ. (2. (2. 0).69) 21 . ∂x′µ ∂x′µ ∂xν (∂µ F )′ (x′ ) ≡ (2. 0. the defining characteristic of a tensor function Tνµ11νµ22.62) One can obtain another scalar function by acting twice with the 4-dimensional derivative operator on F .. dz).68) D µ = (0. Define 4 four-vectors which in a particular frame are given by the infinitesimal differentials Aµ = (cdt. ∇F ) µ ∂x c ∂t (2. one obtains a contravariant four-vector function ∂ µ F = gµν ∂ν F = ( 1 ∂F ~ ). [The second equality uses the chain rule and eq. −∇F c ∂t (2. 0)... 0. 0.in spacetime the value of the function F ′ reported by an observer using the primed coordinate system is taken to be equal to the value of the original function F in the original coordinates: F ′ (x′ ) = F (x).. 0.. 2 . (2.60) is a covariant four-vector.. 0. (x) is that under a change of reference frame. the last equality uses eq. 0).59). (2. (2. the corresponding tensor T ′ is: . .. (x′ ) = Lµ1 ρ1 Lµ2 ρ2 · · · Lν1 σ1 Lν2 σ2 · · · Tσρ11σρ22. (2. (2.65) One use for the Levi-Civita tensor is in understanding the Lorentz invariance of four-dimensional integration. In general. (x).. 0.61) showing that it transforms according to eq.. dx. contracting the indices on the derivatives: ∂ µ ∂µ F = 1 ∂2F − ∇2 F. (2.19). dy.64) A special and useful constant tensor is the totally antisymmetric Levi-Civita tensor: ǫµνρσ =    +1 if µνρσ is an even permutation of 0123 −1 if µνρσ is an odd permutation of 0123 otherwise   0 (2.

known as the action. since in the last expression it has no uncontracted four-vector indices. This is good because eventually we will learn to define theories in terms of such an integral. 22 .70) is Lorentz-invariant.71) is invariant under Lorentz transformations. It follows that if F (x) is a Lorentz scalar function of xµ . then the integral I[F ] = Z d4 x F (x) (2.Then the 4-dimensional volume element d4 x ≡ dx0 dx1 dx2 dx3 = Aµ B ν C ρ D σ ǫµνρσ (2.

treating p~2 as small compared to m2 c2 : i¯ h ∂ p~ 2 ~p 4 Φ = mc2 (1 + − + .1) Now.1.7) so that: − ∂2Φ = −∇2 Φ + m2 Φ.1. To make sense of this as an operator equation.1.4) where m is the mass of the particle. . It follows that 2 H 2 Φ = E 2 Φ = (c2 p~ 2 + m2 c4 )Φ = (c2 P~ + m2 c4 )Φ. proportional to the rest energy. However.2) Since H and P~ commute.8) .1. the presence of an infinite number of derivatives leads to horrible problems. In this section.)Φ ∂t 2m2 c2 8m4 c4 ¯h2 2 ¯h4 = [mc2 − ∇ − (∇2 )2 + .1 Klein-Gordon and Dirac equations Any realistic theory must be consistent with quantum mechanics. HΦ = EΦ.3) One can now turn this into a relativistic Schrodinger wave equation for free particle states. we consider how to formulate a theory of quantum mechanics that is consistent with special relativity. ∂t2 23 (3. (3.1. (3. (3. the first term mc2 is an unobservable constant contribution to the Hamiltonian.3 Relativistic quantum mechanics of single particles 3. (3. according to: HΦ = i¯h ∂ Φ ∂t (3. one can always take Φ to be one of the basis of wavefunctions for eigenstates with energy and momentum eigenvalues E and p~ respectively: P~ Φ = p~ Φ. including apparently non-local effects. the three-momentum operator is given by ~ P~ = −i¯h∇.6) If we keep only the first two terms.1. . and the second term is the usual non-relativistic kinetic energy. by using the fact that special relativity implies: E= q m2 c4 + p~ 2 c2 .5) (3. Instead. . avoiding the square root. A fundamental principle of quantum mechanics is that the time dependence of Φ is determined by a Hamiltonian operator.]Φ 2m 8m3 c2 (3.1. Suppose that Φ(x) is the wavefunction of a free particle in 4-dimensional spacetime. . then we recover the standard non-relativistic quantum mechanics of a free particle. one can consider the operator H 2 acting on Φ.1. we could try expanding it in an infinite series.

13) This relativistic generalization of the Schrodinger equation is known as the Klein-Gordon equation.1. It is then easy to check that this is an eigenstate of H and P~ with energy E = k0 and three-momentum p~ = ~k.1.1. p~ ) gives a solution. one can have |E| arbitrarily large. This would seem to lead to the release of an infinite amount of energy as the particle acquires a larger and larger three-momentum. These conventions greatly simplify the equations of particle physics. we only need to impose k2 = m2 to have a solution. Dirac suggested an alternative.9733 × 10−16 m. If we try: Φ(x) = Φ0 e−ik·x .Here and from now on we have set c = 1 and h ¯ = 1 by a choice of units.15) Therefore. then ∂µ Φ = −ikµ Φ and so ∂ µ ∂µ Φ = −kµ kµ Φ = −k2 Φ (3. (2. since they have more than one intrinsic degree of freedom.58212 × 10−25 sec. since relativity places t on the same † Apparently.1. without bound! In 1927.7827 × 10−24 g = 1. One can always recover the usual metric system units using the following conversion table for energy. Instead. . and velocity is dimensionless. based on the observation that the problem with the KleinGordon equation seems to be that it is quadratic in H or equivalently ∂/∂t.6022 × 10−3 erg = 1. there is a big problem with this. However. . If kµ = (E. This convention means that mass. (3. (1 GeV)−1 ¯ h = 6.1. n runs over the components. it will make transitions from higher energy states to lower energy states. If the particle can interact.10) (3.63). It is easy to guess the solutions of the Klein-Gordon equation.9) (3. he realized this only in hindsight.8) can be rewritten in a manifestly Lorentz-invariant way as (∂ µ ∂µ + m2 )Φ = 0 (3. 2. This is a disaster. respectively: 1 GeV = 1. p~ ).1. distance. and momentum all have the same units (GeV). while time and distance have units of GeV−1 . Dirac could also have been† motivated by the fact that particles like the electron have spin.6022 × 10−10 Joules. because the energy is not bounded from below. this leads to the sign ambiguity for E. the wave-equation eq. where the spinor index a = 1. (1 GeV)/c2 = 1. energy.7827 × 10−27 kg. .1. . The wave equation should be linear in ∂/∂t. (3. Dirac proposed to write a relativistic Schrodinger equation. satisfying E 2 = p~ 2 + m2 .12) Using eq. (3. for a multi-component wavefunction Ψa (x).11) (3. and time.9733 × 10−14 cm = 1. 24 . By increasing |~ p |.1. trying to explain them with a single wavefunction Φ(x) is doomed to failure. then so does kµ = (−E. mass. (1 GeV)−1 (¯ hc) = 1.14) where Φ0 is a constant and kµ is a four-vector.

it turns out that Dirac spinors would have 32 components. one finds that the two agree if. The matrices β. As we will see.16).1. αj β + βαj αj αk + αk αj (3. Equation (3.1. accounting for the doubling. is really just a coincidence.17) and (3. one finds − ∂2 Ψ = (−∇2 + m2 )Ψ. αy . αz .1.k=1 (3.16) is known as the Dirac equation.19) Then comparing eqs. This may be somewhat surprising. ∂t2 (3. It is easiest to write the solution in terms of 2 × 2 Pauli matrices: 1 σ =  0 1 1 0  . z.20) = 0. (3.22) The simplest solution turns out to require n = 4 spinor indices. Therefore.21) = 2δjk . To determine what α ~ and β have to be. so “0” actually denotes a 2 × 2 block of 0’s. 2 σ =   −i .k=1 ∂x ∂x (3. expressing H in terms of the right-hand side of eq. one can write: 3 X αj αk j. αj are written in 2 × 2 block form. it should also be linear in derivatives of the spatial coordinates.footing as x.18). if we lived in 10 dimensional spacetime. (3. (3. 3)  1 0 0 1  .1. just like ordinary spacetime. (3.k=1 3 ∂ ∂ ∂ ∂ 1 X (αj αk + αk αj ) j k = j k ∂x ∂x 2 j. by exactly the same reasoning as for the Klein-Gordon equation. (3. we find:   3 X X ∂ ∂ ∂ ∂2 αj αk j k − im (αj β + βαj ) j + β 2 m2  Ψ − 2 Ψ = − ∂t ∂x ∂x ∂x j j. and 0 σ = (j = 1.1. and β are n × n matrices acting in “spinor space”. 25 . k = 1. Note that the fact that Dirac spinor space is 4-dimensional.‡ One must be careful not to confuse the two types of 4-dimensional spaces! ‡ For example.17) On the other hand. 2.18) Since partial derivatives commute.1.1. 2. First.1. αj = 3 σ =   −σ j 0  1 0 0 . −1 0 σj  . y. for j.1.23) obey the required conditions. 3: β 2 = 1.16) where αx . since naively one only needs n = 2 to describe a spin-1/2 particle like the electron. There are two ways to evaluate the result. consider H 2 Ψ.1. the equation ought to take the form i ∂ Ψ = HΨ = (~ α · P~ + βm)Ψ ∂t (3. and the 4-component object is known as a Dirac spinor.1. 0 0 i Then one can check that the 4 × 4 matrices β=  0 σ0 σ0 0  .1. the Dirac equation automatically describes positrons as well as electrons.

3). The set we’ve given above is called the chiral or Weyl representation. 2.1. γ 0 γ µ† γ 0 = γ µ γ j† = −γ j (j = 1.It is convenient and traditional to rewrite the Dirac equation in a nicer way by multiplying it on the left by the matrix β. Then the Dirac equation implies: U (iγ µ ∂µ − m)U † U Ψ = 0. writing γ ′µ = U γ µ U † .33) Tr(γµ γν ) = 4gµν (3. Another popular choice used by some textbooks (but not here) is the Pauli-Dirac representation.1. there are an infinite number of different.1. (3.1.1.27) [The solution found above for the γ µ is not unique. the new γ ′µ matrices together with the new spinor Ψ′ are just as good as the old pair γ µ . (3.34) γ µ γµ = 4 (3.1. Then the Dirac equation takes the even more compact form: (i∂/ − m)Ψ = 0. The γ µ matrices are explicitly given. (3. To keep the notation from getting too bloated. it is often useful to use the Feynman slash notation: γ µ aµ = a / (3. (j = 1. 0 ∂x ∂x ∂x ∂x (3.31) Some important properties of the γ µ matricesare: γ 0† = γ 0 .1. and Ψ′ = U Ψ. by: γ0 =  0 σ0 σ0 0  γj = .36) 26 .32) (3.] Many problems involving fermions in high-energy physics involve many gamma matrices dotted into partial derivatives or momentum four-vectors.30) for any four-vector aµ . (j = 1.1. 2. equally valid choices.26)  . even more nicely: (iγ µ ∂µ − m)Ψ = 0.25) or.1.29) So. 3) (3. 2. suppose U is any unitary 4 × 4 matrix satisfying U † U = 1.35) γµ γν + γν γµ = {γµ . 3).1.28) from which it follows that. (iγ ′µ ∂µ − m)Ψ′ = 0.1. To see this. Ψ. γν } = 2gµν (3.24) The result is  i(γ 0  ∂ ∂ ∂ ∂ + γ 1 1 + γ 2 2 + γ 3 3 ) − m Ψ = 0. (3.1. and defining γ 0 = β.1.  0 −σ j σj 0 (3. in 2 × 2 blocks. γ j = βαj . (3.

after N iterations of the Lorentz transformation parameterized by ω µ ν : Lµ ν = (δνµ + Ωµ ν /N )N → [exp(Ω)]µ ν (3. one has in the same way: Ψ′ (x′ ) =  i 1 − Ωµν S µν /N 2 N   i Ψ(x) → exp − Ωµν S µν Ψ(x).37) where Λ is a 4 × 4 matrix that depends on the Lorentz transformation matrix Lµ ν .Note that on the right-hand sides of each of eqs.1. 2 27 (3. Letting Ωµν = N ωµν . (3. you will show for homework that for an infinitesimal Lorentz transformation Lµ ν = δνµ + ω µ ν . On the other hand.42) For the Dirac spinor. so it is not a tensor.1. (3.38) i S µν = [γ µ .35) and (3.1.1.1.39) where To obtain the result for a finite Lorentz transformation.1.) How does a Dirac spinor Ψa (x) transform under a Lorentz transformation? It carries no vector index. .37) corresponding to the Lµ ν that appears in eq. .1. there is an implicit 4 × 4 unit matrix. we might expect that the spinor reported by an observer in the primed frame is given by Ψ′ (x′ ) = ΛΨ(x). Instead. (3. (You will get some practice with this from the homework. (3.43) So. with N → ∞. the equations above can be used to derive identities needed in practical work. 2 (3.1.44) . . the fact that the Hamiltonian “mixes up” the components of Ψa (x) is a clue that it doesn’t transform like an ordinary scalar function either. we obtain. we can apply the same infinitesimal transformation a large number of times N .36). γ ν ]. (3. one has: i Ψ′ (x′ ) = (1 − ωµν S µν )Ψ(x).1. (3. In fact. Here we are using the identity: lim (1 + x/N )N = exp(x). 2 (3. Instead.40) as N → ∞.40):   i Λ = exp − Ωµν S µν .41) N →∞ with the exponential of a matrix to be interpreted in the power series sense: exp(M ) = 1 + M + M 2 /2 + M 3 /6 + . we have found the Λ that appears in eq.1.1. 4 (3.1. It turns out that one almost never needs to know the explicit form of the γ µ .

1. so: Ψ′† Ψ′ (x′ ) = Ψ† Λ† ΛΨ(x). (3. Since Ψ is not a scalar. consider a boost in the z direction: Ωµ ν  0  0 =  0 −ρ  0 −ρ 0 0  . Since this matrix is diagonal. . However. (3. it is particularly easy to exponentiate.50) a=1 However.1.1.)   0 0 0 0  − (ρ + ρ /6 + . as the example of eq. 0 0  0 0 0 0 0 0 etc.1. this is not true.1.40) and (3. Meanwhile. (3. =    0 0 0 0 − sinh ρ 0 0 cosh ρ in agreement with eq.42).7). γ 3 ] = 2 4 2  σ3 0 0 −σ 3 0 0 0 0 0 0 0 0  1 0  0 0 (3. 0 0  0 ρ2  0  0 Ω3 =   0 −ρ3  0 −ρ3 0 0  . in other words if Λ is a unitary matrix. Lµ ν    0 1 0 0 0    0 0 0 0 3 0  = (1 + ρ2 /2 + .49) eρ/2 Therefore.47)  (3. . and eq.1. . under a Lorentz transformation.45) Then  ρ2  0 Ω2 =   0 0 0 0 0 0  0 0 0 0  .44) gives:  ρ/2  0 Λ = exp   0 0 0 0 −ρ/2 0 0 −ρ/2 0 0   0 eρ/2  0  = 0  0  0 ρ/2 0 0 e−ρ/2 0 0 0 0 e−ρ/2 0  0 0  . so i ρ ρ − Ωµν S µν = − [γ 0 .1. (3.37). 0  (3. A tempting guess is to get rid of all the pesky spinor indices by Ψ† Ψ(x) ≡ 4 X Ψ†a Ψa . 0 0  0 0 0 0 0 0 (3.49) clearly shows! 28 .1.)  0 1 0 0 0 1   cosh ρ 0 0 − sinh ρ   0 0 0 0 . Ψ′ (x′ ) = ΛΨ(x) and Ψ′† (x′ ) = Ψ† (x)Λ† . (3.1. Ω03 = −Ω30 = −ρ.As an example.51) This will therefore be a scalar function if Λ† Λ = 1.1. (2. (3. it is natural to ask whether one can use it to construct a scalar quantity.46) so that from eqs. in eq.48) in 2 × 2 blocks. . this is the matrix that boosts a Dirac spinor in the z direction with rapidity ρ.1. (3.1.

56) One should think of Ψ as a column vector in spinor space. (3.1.33).1.1.1. (3. For now s just distinguishes between distinct solutions. labeled by the 4-momentum p and s. for any Dirac spinor Ψ.2 Solutions of the Dirac equation Our next task is to construct solutions to the Dirac equation.55) that is infinitesimally close to the identity. transforms as a scalar function under proper Lorentz transformations. Let us separate out the xµ -dependent part as a plane wave.1.1. The object u(p. by trying Ψ(x) = u(p. with a spacetime vector index µ hanging around.54) One can check that this is indeed true for the special case of eq.1. one can show that Ψγ µ Ψ (3. ΨΨ (3. Then their inner product. Ψ† γ 0 Ψ will transform as a scalar if: Λ† γ 0 Λ = γ 0 . More importantly. Motivated by this.58) as a (row vector)×(matrix)×(column vector) in spinor-index space. (3.1) Here pµ is a four-vector momentum. it can be proved for any i Λ = 1 − ωµν S µν 2 (3.53) Under a Lorentz transformation: Therefore. one defines.49). A solution to the Dirac equation must also satisfy the Klein-Gordon equation.1. with p0 = E > 0. and Ψ as a row vector. (3. s)e−ip·x .1. with amazing foresight.1. Ψ ≡ Ψ† γ 0 .1.2.57) with all spinor indices contracted. Similarly.58) transforms as a four-vector. so p2 = E 2 − ~p 2 = m2 . let us consider the object Ψ† γ 0 Ψ. Therefore. 3. s) is a spinor. One should think of eq. (3. but it will turn out 29 . using eqs. (3.52) Ψ′† γ 0 Ψ′ (x′ ) = Ψ† Λ† γ 0 ΛΨ(x).32) and (3.Instead. (3. it is true for any proper Lorentz transformation built out of infinitesimal ones. (3.

√ 30 me−ρ/2 = p E − pz .2.7) in a frame where the particle is at rest. (3.2.49):   e−ρ/2 √  0  −ip·x  .8) Now. we have called it the primed frame for convenience.2.6) As we will see.2.2. we can therefore write in 2 × 2 blocks:  −1 1 1 −1  u(p. In that frame. noting that from eq. √ meρ/2 = p E + pz . first consider this equation in the rest frame of the particle.44). so using the inverse of eq. these just correspond to spin eigenstates Sz = 1/2 and −1/2. Plugging this into the Dirac equation (3. consider the solution Ψ′ (x′ ) = u(p. in the unprimed frame: E = p0 = m cosh ρ. 0.10) .2. 2.1.1. (3. 1e 0 (3. Ψ(x) = m   eρ/2  e 0 (3. (3.2) To simplify things.1. s) = 0. 1)e−ip·x ′   1 √  0  −imt′  = m . (3. the particle has. and the √  χs χs  . For example. Now to construct the corresponding solution in any other frame. We suppose the primed frame is moving with respect to the unprimed frame with rapidity ρ in the z direction. where pµ = (m.1. (3. s): (/p − m)u(p.4) where each “1” means a 2 × 2 unit matrix. A particularly nice choice is: χ1 =   1 0 . The solutions are clearly √ u(p. pz = p3 = m sinh ρ.2. (3. (3. it is best to choose the χs orthonormal. satisfying χ†s χr = δrs for r. s = 1. s) = 0. (3.9) We can rewrite this. Ψ(x) = Λ−1 Ψ′ (x′ ) from eq. χ2 =   0 1 . s) = 0.8). 0. In practice. Thus. m(γ 0 − 1)u(p.3) Using the explicit form of γ 0 . we obtain a 4 × 4 eigenvalue equation to be solved for u(p.2.31).37).to be related to the spin.2. 0).5) m normalization is a convention. (3. one can just boost the spinor using eqs. (3. s) = m where χs can be any 2-vector. (3.2.

~ To do this. Similarly. spin let us now consider how to construct the spin operator S.11) Ψ(x) =   E + pz  e 0 so that √ in this frame. or in other words it must commute with the Hamiltonian: [H. where ~x and P~ are the three-dimensional position and momentum operators.13) (3.2. 31 (3.17) Using the Dirac Hamiltonian given in eq.2.16).Therefore. L] α · P~ + βm.2. (3. The total angular momentum must be conserved.14) in this frame. we have ~ = [~ [H. corresponding to the wavefunction for a particle moving in either the +z or −z directions.2.5) in the rest frame. one solution of the Dirac equation for a particle moving in the z direction.2. with energy E √ and three-momentum pz = E 2 − m2 .2.2. recall that by definition.15) ~ = ~x × P~ . (3.13) can have either sign.2. ~ J~ = L (3. L (3. In order to make a direct connection between spin and the various components of a Dirac spinor.2.18) . if we use instead χ2 = we find a solution: Ψ(x) = so that  E − pz   0  √ u(p. is:  √ E − pz  −ip·x   √ 0 . (3.2.   0 √ E − pz   √ 0  E + pz   u(p. is the difference between the total angular momentum operator J~ and the orbital angular momentum ~ operator L: ~ + S.12) e−ρ/2   √ 0  E + pz  −ip·x e = .2.16) Now. 0   eρ/2  −ip·x e 0  m  (3. and apply the same procedure. ~x × P~ ] = −i~ α × P~ . 2) =    √ 0 E − pz (3. 1) =   E + pz  0 √   0 1  in eq. (3.1. Note that pz in eqs.11) and (3. J~] = 0. (3.

where we have used the canonical commutation relation (with ¯h = 1) [Pj , xk ] = −iδjk . So, comparing
eqs. (3.2.15), (3.2.17) and (3.2.18), it must be true that:
~ ] = i~
[H, S
α × P~ = i 

−~σ × P~
0

0
~σ × P~ 

.

(3.2.19)

One can now observe that the matrix:
~=1
S
2 


0

0
~σ 

(3.2.20)

obeys eq. (3.2.19). So, it must be the spin operator acting on Dirac spinors.
In particular, the z-component of the spin operator for Dirac spinors is given by the diagonal matrix:

1
1
0
Sz = 

2 0
0

0
−1
0
0

0 0
0 0 
.
1 0 
0 −1

(3.2.21)

Therefore, the solutions in eqs. (3.2.11) and (3.2.13) can be identified to have spin eigenvalues Sz = +1/2
and Sz = −1/2, respectively. In general, a Dirac spinor eigenstate with Sz = +1/2 will have only the

first and third components non-zero, and one with Sz = −1/2 will have only the second and fourth
components non-zero, regardless of the direction of the momentum. Note that, as promised, Sz = 1/2

(−1/2) exactly corresponds to the use of χ1 (χ2 ) in eq. (3.2.5).
The helicity operator gives the relative orientation of the spin of the particle and its momentum.
It is defined to be:
h=

~
p~ · S
.
|~
p|

(3.2.22)

Like Sz , helicity has possible eigenvalues ±1/2 for a spin-1/2 particle. For example, if pz > 0, then

eqs. (3.2.11) and (3.2.13) represent states with helicity +1/2 and −1/2 respectively. The helicity is not

invariant under Lorentz transformations for massive particles. This is because one can always boost
to a different frame in which the 3-momentum is flipped but the spin remains the same. (Also, note
that unlike Sz , helicity is not even well-defined for a particle exactly at rest, due to the |~
p | = 0 in

the denominator.) However, a massless particle moves at the speed of light in any inertial frame, so
one can never boost to a frame in which its 3-momentum direction is flipped. This means that for
massless (or very energetic, so that E ≫ m) particles, the helicity is fixed and invariant under Lorentz
~ parallel has helicity h = 1/2, and a particle
transformations. In any frame, a particle with p~ and S
~ antiparallel has helicity h = −1/2.
with p~ and S

Helicity is particularly useful in the high-energy limit. For example, we can consider four solutions

obtained from the E, pz ≫ m limits of eqs. (3.2.11) and (3.2.13), so that |pz | = E:
Ψpz >0,Sz =+1/2

0
 0  −iE(t−z)
√ 
= 
 2E  e
0
32

~ ↑, h = +1/2]
[~
p ↑, S

(3.2.23)

Ψpz >0,Sz =−1/2

Ψpz <0,Sz =+1/2

Ψpz <0,Sz =−1/2

√0
 2E  −iE(t−z)

= 
 0 e
0

√
2E
 0  −iE(t+z)

= 
 0 e
0


0
 0  −iE(t+z)

= 
 0 e

2E

~ ↓, h = −1/2]
[~
p ↑, S

(3.2.24)

~ ↑, h = −1/2]
[~
p ↓, S

(3.2.25)

~ ↓, h = +1/2]
[~
p ↓, S

(3.2.26)

In this high-energy limit, a Dirac spinor with h = +1/2 is called right-handed (R) and one with
h = −1/2 is called left-handed (L). Notice that a high-energy L state is one that has the last two
entries zero, while a high-energy R state always has the first two entries zero.

It is useful to define matrices that project onto L and R states in the high-energy or massless limit.
In 2 × 2 blocks:
PL = 

1
0 

0
;
0

PR = 

0
0 

0
,
1

(3.2.27)

where 1 and 0 mean the 2 × 2 unit and zero matrices, respectively. Then PL acting on any Dirac spinor

gives back a left-handed spinor, by just killing the last two components. The projectors obey the rules:
PL2 = PL ;

PR2 = PR ;

PR PL = PL PR = 0.

(3.2.28)

It is traditional to write PL and PR in terms of a “fifth” gamma matrix, which in our conventions is
given in 2 × 2 blocks by:
γ5 = 

−1 0
0 1 

.

(3.2.29)

Then
PL =

1 − γ5
;
2

1 + γ5
.
2

(3.2.30)

{γ5 , γµ } = 0.

(3.2.31)

PR =

The matrix γ5 satisfies the equations:
γ52 = 1;

γ5† = γ5 ;

So far, we have been considering Dirac spinor wavefunction solutions of the form
Ψ(x) = u(p, s)e−ip·x

(3.2.32)

with p0 = E > 0. We have successfully interpreted these solutions in terms of a spin-1/2 particle,
say, the electron. However, there is nothing mathematically wrong with these solutions for pµ with
p0 < 0 and p2 = m2 . So, like the Klein-Gordon equation, the Dirac equation has the embarrassment
of negative energy solutions.
33

Dirac proposed to get around the problem of negative energy states by using the fact that spin-1/2
particle are fermions. The Pauli exclusion principle dictates that two fermions cannot occupy the same
quantum state. Therefore, Dirac proposed that all of the negative energy states are occupied. This
prevents electrons with positive energy from making disastrous transitions to the E < 0 states. The
infinite number of filled E < 0 states is called the Dirac sea.
If one of the states in the Dirac sea becomes unoccupied, it leaves behind a “hole”. Since a hole
is the absence of an E < 0 state, it effectively has energy −E > 0. An electron has charge† −e, so

the hole corresponding to its absence effectively has the opposite charge, +e. Since both electrons
and holes obey p2 = m2 , they have the same mass. Dirac’s proposal therefore predicts the existence

of “anti-electrons” or positrons, with positive energy and positive charge. The positron was indeed
discovered in 1932 in cosmic ray experiments.
Feynman and St¨
uckelberg noted that one can reinterpret the positron as a negative energy electron
~ → −S.
~ According to this interpretation, the
moving backwards in time, so that pµ → −pµ and S
wavefunction for a positron with 4-momentum pµ with p0 = E > 0 is
Ψ(x) = v(p, s)eip·x .

(3.2.33)

Now, using the Dirac equation (3.1.31), v(p, s) must satisfy the eigenvalue equation:
(p
/ + m)v(p, s) = 0.

(3.2.34)

We can now construct solutions to this equation just as before. First, in the rest (primed) frame of the
particle, we have in 2 × 2 blocks: 

m
m

m
m 

v(p, s) = 0.

(3.2.35)

So, the solutions are
Ψ′ (x′ ) =

m 

ξs
−ξs 

eimt

(3.2.36)

for any two-vector ξs .
One must be careful in interpreting the quantum numbers of the positron solutions to the Dirac
equation. This is because the Hamiltonian, 3-momentum, and spin operators of a positron described
by the wavefunction eq. (3.2.33) are all given by the negative of the expressions one would use for an
electron wavefunction. Thus, acting on a positron wavefunction, one has

Ψ,
∂t
~
P~ Ψ = i∇Ψ,  

1 ~σ 0
~
Ψ,

= −
2 0 ~σ

HΨ = −i

(3.2.37)
(3.2.38)
(3.2.39)

Here, e is always defined to be positive, so that the electron has charge −e. (Some references define e to be negative.)

34

~ are the operators whose eigenvalues are to be interpreted as the energy, 3where H, P~ , and S
momentum, and spin of the positive-energy positron antiparticle. Therefore, to describe a positron
with spin Sz = +1/2 or −1/2, one should use, respectively,
ξ1 = 



0
1

;

or

ξ2 = 



1
0

,

(3.2.40)

in eq. (3.2.36).
Now we can boost to the unprimed frame just as before, yielding the solutions:

√ 0

E + pz 
,
v(p, 1) = 


√ 0
− E − pz

 √
E − pz



√ 0
v(p, 2) = 
 − E + pz  .
0

(3.2.41)

(3.2.42)

Here v(p, 1) corresponds to a positron moving in the +z direction with 3-momentum pz and energy
E=

p

p2z + m2 and Sz = +1/2, hence helicity h = +1/2 if pz > 0. Similarly, v(p, 2) corresponds to a

positron with the same energy and 3-momentum, but with Sz = −1/2, and therefore helicity h = −1/2

if pz > 0.

Note that for positron wavefunctions, PL projects onto states that describe right-handed positrons
in the high-energy limit, and PR projects onto states that describe left-handed positrons in the highenergy limit. If we insist that PL projects on to left-handed spinors, and PR projects on to right-handed
spinors, then we must simply remember that a right-handed positron is described by a left-handed
spinor (annihilated by PR ), and vice versa!
Later we will also need to used the Dirac row spinors:
u(p, s) = u(p, s)† γ 0 ,

v(p, s) = v(p, s)† γ 0 .

(3.2.43)

The quantities u(p, s)u(k, r) and u(p, s)v(k, r) and v(p, s)u(k, r) and v(p, s)v(k, r) are all Lorentz
scalars. For example, in the rest frame of a particle with mass m and spin Sz = +1/2, one has
† 0

u(p, 1) = u(p, 1) γ =

m(1 0 1

0) 

0
σ0

σ0
0 


=( m

0

m 0),

(3.2.44)

so that

u(p, 1)u(p, 1) = ( m

0

m

√

m
 0 

0)
 √m  = 2m.
0

(3.2.45)

Since this quantity is a scalar, it must be true that u(p, 1)u(p, 1) = 2m in any Lorentz frame, in other
words, for any pµ . More generally, if s, r = 1, 2 represent orthonormal spin state labels, then the u and

35

v spinors obey:
u(p, s)u(p, r) = 2mδsr ,

(3.2.46)

v(p, s)v(p, r) = −2mδsr ,

(3.2.47)

v(p, s)u(p, r) = u(p, s)v(p, r) = 0.

(3.2.48)

Similarly, one can show that u(p, s)γ µ u(p, r) = 2(m, ~0 )δsr in the rest frame. Since it is a four-vector,
it must be that in any frame:
u(p, s)γ µ u(p, r) = 2pµ δsr .

(3.2.49)

But by far the most useful identities that we will use later on are the spin-sum equations:
2
X

s=1
2
X

s=1

u(p, s)u(p, s) = /p + m;

(3.2.50)

v(p, s)v(p, s) = /p − m.

(3.2.51)

Here the spin state label s is summed over. These equations are to be interpreted in the sense of a row
vector times a column vector giving a 4 × 4 matrix, like:

a1
 a2 
  ( b1
 a3 
a4

b2

b3

a1 b1
 a2 b1
b4 ) = 
 a3 b1
a4 b1

a1 b2
a2 b2
a3 b2
a4 b2

a1 b3
a2 b3
a3 b3
a4 b3

a1 b4
a2 b4 
.
a3 b4 
a4 b4

(3.2.52)

We will use eqs. (3.2.50) and (3.2.51) often when calculating cross-sections and decay rates involving
fermions.
As a check, note that if we act on the left of eq. (3.2.50) with p
/ − m, the left hand side vanishes
because of eq. (3.2.2), and the right hand side vanishes because of
(/
p − m)(/p + m) = /p/p − m2 = 0.

(3.2.53)

p//p = p2 ,

(3.2.54)

p/
p = pµ pν γµ γν = pµ pν (−γν γµ + 2gµν ) = −/p/p + 2p2 ,
/

(3.2.55)

This relies on the identity

which follows from

where eq. (3.1.36) was used. A similar consistency check works if we act on the left of eq. (3.2.51) with
/p + m.
The Dirac spinors given above only describe electrons and positrons with both momentum and
spin aligned along the ±z direction. More generally, we could construct u(p, s) and v(p, s) for states

describing electrons or positrons with any p~ and spin. However, in general that is quite a mess, and it
turns out to be not particularly useful in most practical applications, as we will see.
36

The two-component objects ψL and ψR are called Weyl spinors.] If we now write a Dirac spinor in its L and R helicity components.1.3.1. −σ 1 . respectively. (3. namely α ~ = ~σ or α ~ = −~σ .3) If we now define σ µ = (σ 0 .21) are not applicable. −σ 2 . as we will see in a moment.9) then the Dirac equation (3. So we have two possible quantum mechanical wave equations: i ∂ ~ ψ = ±i~σ · ∇ψ.3. (3.8) [Compare eq.3.3. σ 3 ). If we go back to Dirac’s guess for the Hamiltonian.1. (3. (3.5) then we can write the two possible equations in the form: iσ µ ∂µ ψL = 0.3.3.27). Therefore. (3. We can understand the relationship of the Dirac equation to the Weyl equations if we notice that the γ µ matrices can be written as γµ =  σµ 0 0 σµ  .1) and there is no need for the matrix β.1. (3. σ 2 .3.4) σ µ = (σ 0 .7) Here I have attached labels L and R because the solutions to these equations turn out to have left and right helicity.3. eqs.6) iσ µ ∂µ ψR = 0.10) . σ 1 .3 The Weyl equation It turns out that the Dirac equation can be replaced by something simpler and more fundamental in the special case m = 0. They are similar to the Dirac equation. ∂t (3. (3. Each of these equations is called a Weyl equation. Ψ=  ΨL ΨR  . and we have only the one requirement: αi αj + αk αj = 2δjk .3. and are 2 × 2 matrix equations rather than 4 × 4. (3. we now have just H=α ~ · P~ . (3.26) becomes: i  0 σµ σµ 0  ∂µ  ΨL ΨR  37 =m  ΨL ΨR  . (3.3.3. −σ 3 ).2) Now there are two distinct solutions involving 2 × 2 matrices.20) and (3. (3. but only apply to massless spin-1/2 particles.

(3. The modern point of view is that the Weyl equation is fundamental. we can indeed identify ΨR as a right-handed helicity Weyl fermion. so that this discussion has to be modified slightly. or equivalently as the left-handed part of a Dirac neutrino with no right-handed part.6).3. A Dirac fermion is then written as Ψ=  χ ξ†  .7) when m = 0. This is how neutrinos originally appeared in the Standard Model. (3.3.3. one can consistently set ΨR = 0 as an identity in the Dirac spinor without violating eqs.3.3.6) and (3. all fermions are massless until they are provided with a mass term by the spontaneous breaking of the electroweak symmetry. Then only the left-handed helicity fermion exists. Since a right-handed Weyl fermion is the Hermitian conjugate of a left-handed Weyl fermion. a massless neutrino corresponds to a left-handed Weyl fermion. it is still proper to treat a neutrino as a Weyl fermion. we this. for experiments in which neutrino masses can be neglected. To see ~ † = −∇ ~ and (σ µ )† = σ µ .3. and vice versa. a right-handed Weyl spinor is the Hermitian conjugate of a left-handed Weyl spinor. Note that if m = 0. However. suppose we take the Hermitian conjugate of eq.3.11) and (3.15) where χ and ξ are independent left-handed two-component Weyl fermions.6) and (3. The two versions of the Weyl equation. all one really needs is the single Weyl equation (3.12) Comparing with eqs. (3. (3. as described in section 13. eqs. (3.3.3.2. 3. (3.3.3.12). All fermion degrees of freedom can thus be thought of in terms of left-handed Weyl fermions.4. and ΨL as a left-handed helicity Weyl fermion. Recent evidence shows that neutrinos do have small masses.14) or: In other words. and all 4-component Dirac fermions can be thought of as consisting of two 2-component Weyl fermions coupled together by a mass. (3. This is done by the identification ψ ≡ ΨL = Ψ†R . This is because in the Standard Model.1) .or iσ µ ∂µ ΨR = mΨL .7).7). iσ µ ∂µ ψR (3.3.13) † = 0. are actually not distinct. Since (∇) obtain: i(σ 0 ∂ ~ † =0 − ~σ · ∇)ψ R ∂t (3. 38 (3.4 Majorana fermions A Majorana fermion can be obtained from a massive Dirac fermion by reducing the number of degrees of freedom by identifying the right-handed part with the conjugate of the left-handed part as a constraint.11) iσ µ ∂µ ΨL = mΨR .3.

3) iσ µ ∂µ ψ † − mψ = 0. (3.2) and obeys the same wave equation as a Dirac fermion. However. In the minimal supersymmetric extension of the Standard Model. (iγ µ ∂µ − m)ΨM = 0. the Majorana condition (3. in complex-conjugated form.11). As we will see in section 13. (3. which are predicted to be Majorana fermions.In four-component notation. a Majorana spinor has the form ΨM =  ψ ψ†  . (3.4. the experimental fact that neutrinos have small masses suggests that they are likely to be Majorana fermions (and thus their own antiparticles).4. it has only half as many degrees of freedom. one sees that in the two-component form a classical Majorana fermion obeys the wave equation: iσ µ ∂µ ψ − mψ † = 0.4. although this expectation is based partly on theoretical prejudice and it is also possible that they may be Dirac. 39 . there are new fermions called neutralinos and the gluino.4) or. (3. From eqs.12). (3.3.4.3.4.1) ensures that a Majorana fermion is its own antiparticle.

Aµ = (V.4) where ρ is the local charge density and J~ is the current density. (4.10) then eqs. (4. this yields the Law of eq. suppose we add together the equations obtained by taking ∂/∂t of ~ of eq. we can form a four-vector charge and current density: J µ = (ρ.11) (4.3) and (4. eqs.4 Maxwell’s equations and electromagnetism In the previous section.2). factored out. A more familiar set of relativistic wave equations are those governing electromagnetic fields. with the magnitude of the electron’s charge.7) so that eq.9) Now if we assemble the potentials into a four-vector: ~ ).2) = 0.6) ∂µ J µ = 0. (4. E ∂t ~ ~ ~ B = ∇ × A.5) To put this into a Lorentz-invariant form. These equations can be rewritten in a manifestly relativistic form. Recall that Maxwell’s equations can be written in the form: ~ ·E ~ ∇ ~ ~ ×B ~ − ∂E ∇ ∂t ~ ~ ∇·B ~ ~ ×E ~ + ∂B ∇ ∂t = eρ. (4. we worked out the relativistic wave equations governing the quantum mechanics of scalars and fermions. −Bx  0 (4. J~ ). (4. (4.12) . (4. A (4. (4.1) ~ = eJ.4) imply that we can write the electric and magnetic fields as derivatives ~ of the electric and magnetic potentials V and A: ~ ~ = −∇V ~ − ∂A . Since the divergence of a curl vanishes identically.8) (4. e.8) and (4.5) becomes Furthermore. (4. (4. (4. (4.3) = 0.9) mean that we can write the electric and magnetic fieldsas components of an antisymmetric tensor: Fµν = ∂µ Aν − ∂ν Aµ  0 Ex  −Ex 0 =   −Ey Bz −Ez −By 40 Ey −Bz 0 Bx  Ez By  . using the following observations.1) and ∇· Local Conservation of Charge: ∂ρ ~ ~ + ∇ · J = 0. ∂t (4. First.

Now the Maxwell equations (4.1) and (4.2) correspond to the relativistic wave equation
∂µ F µν = eJ ν ,

(4.13)

∂µ ∂ µ Aν − ∂ ν ∂µ Aµ = eJ ν .

(4.14)

or equivalently,

The remaining Maxwell equations (4.3) and (4.4) are equivalent to the identity:
∂ρ Fµν + ∂µ Fνρ + ∂ν Fρµ = 0,

(4.15)

for µ, ν, ρ = any of 0, 1, 2, 3. Note that this equation is automatically true because of eq. (4.11). Also,
because F µν is antisymmetric and partial derivatives commute, ∂µ ∂ν F µν = 0, so that the Law of Local
Conservation of Charge eq. (4.7) follows from eq. (4.13).
~ and B
~ as derived from
The potential Aµ (x) may be thought of as fundamental, and the fields E
it. The theory of electromagnetism as described by Aµ (x) is subject to a redundancy known as gauge
invariance. To see this, we note that eq. (4.14) is unchanged if we do the transformation
Aµ (x) → Aµ (x) + ∂ µ λ(x),

(4.16)

where λ(x) is any function of position in spacetime. In components, this amounts to:
∂λ
,
∂t
~ → A
~ − ∇λ
~ .
A

V

→ V +

(4.17)
(4.18)

~ and B)
~ unchanged. Therefore, the new Aµ is just
This transformation leaves F µν (or equivalently E
as good as the old Aµ for the purposes of describing a particular physical situation.

41

5

Field Theory and Lagrangians

5.1

The field concept and Lagrangian dynamics

It is now time to make a conceptual break with our earlier treatment of relativistic quantum-mechanical
wave equations for scalar and Dirac particles. There are two reasons for doing this. First, the existence
of negative energy solutions lead us to the concept of antiparticles. Now, a hole in the Dirac sea,
representing a positron, can be removed if the state is occupied by a positive energy electron, releasing
an energy of at least 2m. This forces us to admit that the total number of particles is not conserved. The
Klein-Gordon wavefunction φ(x) and the Dirac wavefunction Ψ(x) were designed to describe singleparticle probability amplitudes, but the correct theory of nature evidently must describe a variable
number of particles. Secondly, we note that in the electromagnetic theory, Aµ (x) are not just quantum
mechanical wavefunctions; rather, they exist classically too. If we follow this example with scalar
and spinor particles, we are lead to abandon φ(x) and Ψ(x) as quantum wavefunctions representing
single-particle states, and reinterpret them as fields that have meaning even classically.
Specifically, a scalar particle is described by a field φ(x). Classically, this just means that for every
point xµ , the object φ(x) returns a number. Quantum mechanically, φ(x) becomes an operator (rather
than a state or wavefunction). There is a distinct operator for each xµ . Therefore, we no longer have
a position operator xµ ; rather, it is just an ordinary number label that tells us which operator φ we
are talking about.
If φ(x) is now an operator, what states will it act on? The answer is that we can start with a
vacuum state
|0i

(5.1.1)

that describes an empty universe with no particles in it. If we now act with our field operator, we
obtain a state:
φ(x)|0i,

(5.1.2)

which, at the time t = x0 , contains one particle at ~x. (What this state describes at other times is a
much more complicated question!) If we act again with our field operator at a different point y µ , we
get a state
φ(y)φ(x)|0i,

(5.1.3)

which in general can be a linear combination of states containing any number of particles. [The operator
φ(y) can either add another particle, or remove the particle added to the vacuum state by φ(x). But,
in addition, the particles can interact to change their number.] In general, the field operator φ(x) acts
on any state by adding or subtracting particles. So this is the right framework to describe the quantum
mechanics of a system with a variable number of particles.

42

Similarly, to describe spin-1/2 particles and their antiparticles, like the electron and the positron,
or quarks and antiquarks, we will want to use a Dirac field Ψ(x). Classically, Ψ(x) is a set of four
functions (one for each of the Dirac spinor components). Quantum mechanically, Ψ(x) is a set of four
operators. We can build up any state we want, containing any number of electrons and positrons, by
acting on the vacuum state |0i enough times with the fields Ψ(x) and Ψ(x) = Ψ† γ 0 .

The vector field Aµ (x) associated with electromagnetism is already very familiar in the classical

theory, as the electrical and vector potentials. In the quantum theory of electromagnetism, Aµ (x)
becomes an operator which can add or subtract photons from the vacuum.
This way of dealing with theories of multi-particle physics is called field theory. In order to describe
how particles in a field theory evolve and interact, we need to specify a single object called the action.
Let us first review how the action principle works in a simple, non-field-theory setting. Let the
variables qn (t) describe the configuration of a physical system. (For example, a single q(t) could
describe the displacement of a harmonic oscillator from equilibrium, as a function of time.) Here n
is just a label distinguishing the different configuration variables. Classically, we could specify the
equations of motion for the qn , which we could get from knowing what forces were acting on it. A
clever way of summarizing this information is to specify the action
S=

Z

tf

ti

L(qn , q˙n ) dt.

(5.1.4)

Here ti and tf are fixed initial and final times, and L is the Lagrangian. It is given in simple systems
by
L = T − V,

(5.1.5)

where T is the total kinetic energy and V is the total potential energy. Thus the action S is a
functional of qn ; if you specify a particular trajectory qn (t), then the action returns a single number.
The Lagrangian is a function of qn (t) and its first derivative.
The usefulness of the action is given by Hamilton’s principle, which states that if qn (ti ) and qn (tf )
are held fixed as boundary conditions, then S is minimized when qn (t) satisfy the equations of motion.
Since S is at an extremum, this means that any small variation in qn (t) will lead to no change in S, so
that qn (t) → qn (t) + δqn (t) implies δS = 0, provided that qn (t) obeys the equations of motion. Here

δqn (t) is any small function of t that vanishes at both ti and tf , as shown in the figure below:

43

q(t2)
q(t)+δq(t)

q(t)
q(t)

q(t1)

t1

t2

t

Let us therefore compute δS. First, note that by the chain rule, we have:
X

δL =

δqn

n 

∂L
∂L
+ δq˙n
.
∂qn
∂ q˙n

(5.1.6)

d
(δqn ),
dt

(5.1.7)

Now, since
δq˙n =
we obtain
δS =

XZ
n

tf

ti 

δqn 

∂L
∂L
d
+ (δqn )
dt.
∂qn dt
∂ q˙n

(5.1.8)

Now integrating by parts yields:
δS =

XZ
n

tf

ti

δqn 

d
∂L

∂qn dt 

∂L
∂ q˙n 

dt +

X
n

δqn

∂L

.t=tf .

9) The last term vanishes because of the boundary conditions δqn (ti ) = δqn (tf ) = 0.11) ∂V ∂L =− = F.1. As a simple example. the position of some particle moving in one dimension in a potential V (x). 2 (5.1. ∂x ∂x (5.1.1. Since the variation δS is supposed to vanish for any δqn (t).12) from which there follows: 44 . ∂ q˙n t=ti (5. for each n.10) is the equation of motion for qn (t). Then 1 L = T − V = mx˙ 2 − V (x). it must be that ∂L d − ∂qn dt  ∂L ∂ q˙n  =0 (5. suppose there is only one qn (t) = x(t).

~x). ∂µ φ). (5. in other words. The action is obtained by summing contributions from each x: S= Z tf ˙ = dt L(φ. In quantum field theory.1. and d dt  ∂L ∂ x˙  = d (mx) ˙ = m¨ x. and qn (t) = φ(t.13) So the equation of motion is Newton’s second law: F = m¨ x. consider a scalar field φ(x) = φ(t.1.16) The object L is known as the Lagrangian density. So. δ is used rather than ∂ to avoid confusing between derivatives with respect to φ and spacetime partial derivatives with respect to xµ .1. By a similar argument as above.17) to mean the partial derivative of L with respect to φ.1.20) and  δ 1 αβ 1 1 δL = g ∂α φ∂β φ = gµβ ∂β φ + gαµ ∂α φ = ∂ µ φ. since this expression depends on φ. 2 2 (5. in the previous discussion. This just means that the form of the Lagrangian allows it to depend on the differences between the field evaluated at infinitesimally nearby points. how they decay. how they interact. a better way to write the action is: S= Z d4 x L(φ.which we recognize as the Newtonian force. φ) Z tf dt ti ti Z ˙ d3 ~x L(φ. Specifying a particular form for L defines the theory. upon which it depends.1. Now. we identify the label n with the spatial position ~x. dt (5. consider the choice: L= 1 1 ∂µ φ∂ µ φ − m2 φ2 . (5. As an example.15) ˙ it must also depend on ∇φ ~ in order to be Lorentz invariant. this implies the equations of motion: δL − ∂µ δφ Here. (5. for any small variation φ(x) → φ(x) + δφ(x). we must find φ(x) so that S is extremized. φ).1. ~x). As a first example of an action for a relativistic field theory. it will tell us what the particle masses are. or equivalently the Lagrangian L. Likewise. we must have δS = 0. δL δ(∂µ φ) means a partial derivative of L with respect to the object ∂µ φ. Then. (5.14) Everything we need to know about the dynamics of a physical system is encoded in the action S. and what symmetries provide selection rules on their behavior. To find the classical equations of motion for the field.19)  (5.1.18) It follows that: δL = −m2 φ δφ (5. we use δL δφ δL δ(∂µ φ) ! = 0.1. δ(∂µ φ) δ(∂µ φ) 2 2 2 45 .

The previous discussion for scalar fields can be extended to other types of fields as well.22) This we recognize as the Klein-Gordon wave equation for a scalar particle of mass m. (5. and over all of space.1.1. or several copies of any of these. This equation was originally introduced with the interpretation as the equation governing the quantum wavefunction of a single scalar particle. Weyl fields ψ(x) with two components. (5. compare to eq. For example.6).1.1.13). (Typically one evaluates S between two times ti and tf . and integrating the second term by parts (this is where the boundary conditions come in).1. Consider a general list of fields Φj (x).21) and so the equation of motion following from eq.1. let us consider how to make a Lagrangian density for a Dirac field Ψ(x). (5.24) If we require this to vanish for each and every arbitrary variation δΦj .1. classically it is a function on spacetime.25) for each j.Therefore. (5. or vector fields Aµ (x) with four components. (5. (5. δL δ(∂µ φ) ∂µ ! = ∂µ ∂ µ φ. so this means that δΦ(x) vanishes very far away from some region of interest. ∂µ Φj ) determines the classical equations of motion through the principle that S = R d4 x L should be stationary to first order when a deviation Φj (x) → Φj (x) + δΦj (x) is made.) The Lagrangian density L should be a real quantity that transforms under proper Lorentz transformations as a scalar. which could include scalar fields φ(x). (5. Similarly to eq. with boundary conditions that δΦj (x) vanishes on the boundary of the spacetime region on which S is evaluated. Ψ(x) transforms exactly like the wavefunction solution to the Dirac equation. we have by the chain rule: δS = Z d4 x X j " δL δL + δ(∂µ Φj ) δΦj δΦj δ(∂µ Φj ) !# . we obtain the Euler-Lagrange equations of motion: δL − ∂µ δΦj δL δ(∂µ Φj ) ! = 0. Under Lorentz transformations. Dirac or Majorana fields Ψ(x) with four components. Now it has reappeared with a totally different interpretation. (3. as the classical equation of motion for the scalar field.23) Now using δ(∂µ Φj ) = ∂µ (δΦj ). and quantum 46 . The Lagrangian density L(Φj . one obtains δS = Z 4 d x X δΦj j " δL − ∂µ δΦj δL δ(∂µ Φj ) !# . But now it is interpreted instead as a field.1.17) is: ∂µ ∂ µ φ + m2 φ = 0.

(5. For this. by plugging in to the general result eq.1.32) Plugging these into eq. (5. where we have multiplied δL δΨ† (5. This is. As we found in section 3.mechanically it is an operator for each point xµ . the Dirac equation. (5. we need: δL δΨ δL δ(∂µ Ψ) = −mΨ.1.1.1.29) The second equation just reflects the fact that the Lagrangian only contains the derivative of Ψ.27) which can also be written as: This is a Lorentz scalar.26) L = iΨ† γ 0 γ µ ∂µ Ψ − mΨ† γ 0 Ψ. it is just the Hermitian conjugate of eq.1. multiplied on the right by γ 0 and using eq. (5.1. (5.34) . it is useful to use Ψ(x) = Ψ† γ 0 as a building block. The Lagrangian density for an electromagnetic field is: 1 1 LEM = − Fµν F µν = − (∂µ Aν − ∂ν Aµ )(∂ µ Aν − ∂ ν Aµ ). (5. this is nothing new. if we want to build Lorentz scalar quantities. of course. One should actually treat Ψ(x) and Ψ† (x) as independent fields.31) = iΨγ µ .1.1.1. 4 4 47 (5.1. (5.25). we need: δL δΨ† δL δ(∂µ Ψ† ) = iγ 0 γ µ ∂µ Ψ − mγ 0 Ψ. (5.1. not the derivative of Ψ† .1.1.33) However. (5. so that when integrated R d4 x it will give a Lorentz-invariant number. (3.25). (5. We can also find the equations of motion obtained from the Lagrangian by varying with respect to Ψ. in the same way that in complex analysis one often treats z = x + iy and z ∗ = x − iy as independent variables.28) = 0. the equation of motion is simply: iγ µ ∂µ Ψ − mΨ = 0. so Ψ† (x) is also a field. Let us now compute the equations of motion that follow from it. A good (and correct) guess for the Lagrangian for a Dirac field is: L = iΨγ µ ∂µ Ψ − mΨΨ.33). we obtain: −i∂µ Ψγ µ − mΨ = 0.30) = 0 on the left by γ 0 and used the fact that (γ 0 )2 = 1.1. Ψ(x) is a complex 4-component object. First. For that. let us find the equations of motion obtained by varying with respect to Ψ† . So. Now.30).

given the Lagrangian L(qn .1. (5.41) the classical equations of motion for the vector field become ∂µ F µν − eJ ν = 0. as we will see. The current density J µ may be regarded as an external source for the electromagnetic field of unspecified origin.1. To begin. (5.40) = 0. (4. In order to include the effects of a 4-current J µ . we compute: δLEM δAν = 0.1. Then. (4. (5.39) to the Lagrangian density. First.1. one defines the canonical momenta conjugate to each qn : pn ≡ ∂L . the equations of motion.To find the equations of motion that follow from this Lagrangian.1. (5.1.2 Quantization of free scalar field theory Let us now turn to the question of quantizing a field theory. ∂ q˙n 48 (5. let us recall how one quantizes a simple generic system based on variables qn (t).37) reduce to ∂µ F µν = 0. 5. and δLEM δ(∂µ Aν )  δ 1 = − (∂α Aβ − ∂β Aα )(∂ρ Aσ − ∂σ Aρ )gαρ gβσ δ(∂µ Aν ) 4 1h = − (∂ρ Aσ − ∂σ Aρ )gµρ gνσ − (∂ρ Aσ − ∂σ Aρ )gνρ gµσ 4 i +(∂α Aβ − ∂β Aα )gµα gβσ − (∂α Aβ − ∂α Aβ )gνα gµβ  = −∂ µ Aν + ∂ ν Aµ = −F µν .36) So.38) which we recognize as Maxwell’s equations in the case of vanishing J µ [see eq.2.1.13)].1.1) . since δLcurrent δAν δLcurrent δ(∂µ Aν ) = −eJ ν . (5.35) since Aµ doesn’t appear in the Lagrangian without a derivative acting on it. we can simply add a term Lcurrent = −eJ µ Aµ (5.42) in agreement with eq. (5.13). q˙n ). δLEM − ∂µ δAν δLEM δ(∂µ Aν ) ! =0 (5. or it can be built out of the fields for charged particles.

7) 1 ~ 2 − m2 φ2 ] d3 ~x [φ˙ 2 − (∇φ) 2 ~ 2 + m2 φ2 ].2.1. ~x) plays the role of qn (t).9) Notice that a nice feature has emerged: this Hamiltonian is a sum of squares.2.2.2. the qn and pn and H are reinterpreted as Hermitian operators acting on a Hilbert space of states.11) [π(~x). with ~x playing the role of the label n. φ(~y )] = 0.2.6) It should be emphasized that π(~x). pm ] = 0. the momentum conjugate to the field φ(~x). not a three-vector or a four-vector! The Hamiltonian is obtained by summing over the fields at each point ~x: H = = = Z Z 1 2 ˙ d3 ~x π(~x)φ(x) −L ˙ d3 ~x π(~x)φ(x) − Z Z (5. there is a different field at each point in space. (5. Notice that π(~x) is a scalar function. q˙n ). (5. pn ) ≡ X n pn q˙n − L(qn . Then φ(x) = φ(t.2) where the q˙n are to be eliminated using eq. it turns out to be profitable to analyze the system in a way similar to the way one treats the harmonic oscillator in one-dimensional nonrelativistic quantum mechanics. π(~y )] = 0.5): [φ(~x). The operators obey canonical equal-time commutation relations: [pn . To go to the corresponding quantum theory. In that system. (5.2. (5.12) As we will see.2.2. Let us now apply this to the theory of a scalar field φ(x) with the Lagrangian density given by eq.1).3) [qn . φ(~y )] = −i¯hδ(3) (~x − ~y ).2. (5.4) [pn . (5. (5. The momentum conjugate to φ is: π(~x) ≡ i δL δ h 1 ˙2 ˙ φ + . d3 ~x [π 2 + (∇φ) (5.18). unlike the case of the single-particle Klein-Gordon wave equation.2.The classical Hamiltonian is then defined by H(qn .10) [π(~x).5) and the time evolution of the system is determined by the Hamiltonian operator H. = φ. qm ] = 0. . (5.2. . (5.2. qm ] = −i¯hδnm . and is therefore always ≥ 0.2.3)-(5. At any given fixed time t. There are no dangerous solutions with arbitrarily large negative energy. = δφ˙ δφ˙ 2 (5. 49 . the field operators φ(~x) and their conjugate momenta π(~x) are Hermitian operators satisfying commutation relations exactly analogous to eqs. (5. is not in any way the mechanical momentum of the particle.8) (5.2.

(5.13) defines a distinct annihilation operator for each three-momentum p~.13) (5. compute the commutator: [ap~ . k (5. a~k ] = [ap†~ .2. [a. In a similar way. Similarly. one obtains: [ap~ . a] = [a† . The overall coefficient in front of eq. where Ep~ = q p~2 + m2 .20). (5. here we will define: ap~ = Z   d3 ~x e−i~p·~x Ep~ φ(~x) + iπ(~x) .22) .2. E~k φ(~y ) − iπ(~y )] ~ d3 ~ y e−i~p·~x eik·~y (E~k + Ep~ ) δ(3) (~x − ~y).2. using the fact that the delta function vanishes except when ~k = p~.2.21) Up to a constant factor on the right-hand side of eq. to obtain the final result: [ap~ .2.18) This can be further reduced by using the important identity: Z d3 ~x ei~q·~x = (2π)3 δ(3) (~q ). k (5. Taking the Hermitian conjugate yields: ap†~ = Z   d3 ~x ei~p·~x Ep~ φ(~x) − iπ(~x) .2.2.10)-(5. (5.20) Here we have put E~k = Ep~ . these results have the same form as the harmonic oscillator algebra [a. a~† ] = 0. We start with the vacuum state |0i.19) valid for any 3-vector ~ q . (5.17) where eqs.one defines “creation” and “annihilation” (or “raising” and “lowering”) operators a† and a as complex linear combinations of the position and momentum operators x and p.12) have been used.13) reflects an arbitrary choice (and in fact it is chosen differently by various books).15) To see the usefulness of these definitions. a† ] = 1. (5. Now we can understand the Hilbert space of states for the quantum field theory. a~† ] k = (E~k + Ep~ ) Z ~ d3 ~x ei(k−~p)·x . a~† ] = (2π)3 2Ep~ δ(3) (~k − ~p). Therefore. (5.2. (5.2. Now performing the ~y integral using the definition of the delta function. Equation (5. which is taken to be annihilated by all of the lowering operators: a~k |0i = 0.2. (5.16) (5.2.14) with the positive square root always taken. a† ] = 0 that are familiar from non-relativistic quantum mechanics. 50 (5. one associated with each three-momentum p~.2. a~† ] k = = Z Z 3 d ~x d3 ~x Z Z ~ d3 y~ e−i~p·~x eik·~y [Ep~ φ(~x) + iπ(~x) . the quantum mechanics of a scalar field behaves like an infinite collection of harmonic oscillators.2. one can check the commutators: [ap~ .2.2.

2. This is another way of saying that the multiparticle states obey Bose-Einstein ki kj statistics for identical particles with integer (in this case. renaming ~y → ~x.29) or. .19)]. The result is that eq. We begin by noting that: ap~ + a†−~p = Z d3 ~x e−i~p·~x 2Ep~ φ(~x). Acting multiple times with raising operators produces a state with multiple particles. because of the identity [a~† . Note that these states are automatically symmetric under interchange of any two of the momentum labels ~ki .2. ~kn . (5. ~k2 . .27) used often from now on.15) to find φ(~x) and π(~y ) in terms of the ap†~ and ap~ . ~kn i k1 k2 (5. . (5.25) Z d˜ p ei~p·~y .2.2.30) This expresses the original field in terms of raising and lowering operators. 0) spin. 51 (5. (5. (2π)3 (5. a~† |0i = |~k1 .31) . and p~ → −~ p in the second term on the right: φ(~x) = Z d˜ p (ei~p·~x ap~ + e−i~p·~x ap†~ ).2. one finds: π(~x) = −i Z d˜ p Ep~ (ei~p·~x ap~ − e−i~p·~x ap†~ ). .2.2. (5. . a~† ] = 0.2.23) k which describes a state with a single particle with three-momentum ~k (and no definite position). In order to check our interpretation of the quantum theory.2. Similarly.Acting on the vacuum state with any raising operator produces a~† |0i = |~ki. one obtains after performing the ~x integral: φ(~y ) = Z d˜ p ei~p·~y (ap~ + a†−~p ). for the conjugate momentum field. let us invert the definitions eqs.2. .13) and (5. (5.25) becomes Z i~ p·~ y d˜ pe (ap~ + a†−~p ) = Z 3 d ~x φ(~x) (Z ) d3 p~ i~p·(~y −~x) e . So a~† a~† .28) Since the p~ integral in braces is equal to δ(3) (~y − ~x) [see eq. . .2.2. . and then determine how H acts on the space of states. ~k2 . (5. (5. (5. First. .2.24) kn describes a state of the universe with n φ-particles with momenta ~k1 . let us now evaluate the Hamiltonian operator in terms of the raising and lowering operators.26) Now we act on both sides by: where we have introduced the very convenient shorthand notation: Z d˜ p≡ Z d3 p~ (2π)3 2Ep~ (5.

34) k 2 Adding up the pieces. (5.20). (5. using ap~ ap†~ = ap†~ ap~ + (2π)3 2Ep~ δ(3) (~ p − p~) (5.2. one finds: 1 H= 2 Z dk˜ Z d˜ p Z ( h ~ ~ d ~x (m2 − ~k · ~p − E~k Ep~ ) a~k ap~ ei(k+~p)·~x + a~† ap†~ e−i(k+~p)·~x 3 h +(m2 + ~k · ~p + E~k Ep~ ) k ~ a~† ap~ ei(~p−k)·~x k + ~ a~k ap†~ ei(k−~p)·~x i i ) . (5.36) (5.2. for the aa† and a† a terms one has ~k = ~ p from the delta function. The needed terms are: Z 1 d3 ~x π(x)2 = 2Z 1 ~ 2 = d3 ~x (∇φ) 2 Z m2 d3 ~x φ(x)2 = 2 Z Z Z  ~   1 ~ d3 ~x dk˜ d˜ p (−i)2 E~k Ep~ eik·~x a~k − e−ik·~x a~† ei~p·~x ap~ − e−i~p·~x ap†~ . so that (m2 + ~k · p~ + E~k Ep~ ) δ(3) (~k − p~) = 2Ep~2 δ(3) (~k − p~).2.Now we can plug the results of eqs.2.35) Now one can do the ~x integration using: Z Z ~ d3 ~x e±i(k+~p)·~x = (2π)3 δ(3) (~k + p~ ).41) where “∞” means an infinite. by a redefinition of the Hamiltonian. This is a simple example of the process known as renormalization. (5.37) As a result.2. and neglecting all contributions coming from momenta greater than 52 . (5.2.40) from eq. Since a uniform constant contribution to the energy of all states is unobservable and commutes will all other operators.2.2.2. the coefficient (m2 − ~k · p~ − E~k Ep~ ) of the aa and a† a† terms vanishes after plugging in ~k = −~ p as enforced by the delta function. (In a more careful treatment.39) Finally.38) Now performing the ~k integral in eq. (5. (5.2. Meanwhile. we can rearrange the second term. we are free to drop it.33) k 2 Z Z Z  2    m ~ ~ d3 ~x dk˜ d˜ p eik·~x a~k + e−ik·~x a~† ei~p·~x ap~ + e−i~p·~x ap†~ .30) and (5.31) into the expression eq. (5.32) k 2Z Z Z     1 ~ ~ d3 ~x dk˜ d˜ p i~keik·~x a~k − i~ke−ik·~x a~† · i~ pei~p·~x ap~ − i~ pe−i~p·~x ap†~ . but constant.35).2.2. (5. one could “regulate” the theory by quantizing the theory confined to a box of finite volume. one finds: H= 1 2 Z   d˜ p Ep~ ap†~ ap~ + ap~ ap†~ .2. (5. (5.2. The last term is infinite. ~ d3 ~x e±i(k−~p)·~x = (2π)3 δ(3) (~k − p~ ). so H= Z d˜ p Ep~ ap†~ ap~ + ∞.2. contribution to the energy.9) for the Hamiltonian. (5.

u(p. A sensible strategy is to expand the fields Ψ and Ψ† in terms of operators that act on states by creating and destroying particles with a given 3-momentum. we therefore obtain H|~ki = Ha~† |0i = [H. So we expand the Dirac field.2. H= Z d˜ p Ep~ ap†~ ap~ (5. one must expand it in a basis for the four-dimensional spinor space.2. (5. ~k2 . H|0i = 0. a~† ]|0i = E~k a~† |0i = E~k |~ki.45) k This proves that the one-particle state with 3-momentum ~k has energy eigenvalue E~k = expected. s)ei~p·~x bp~. Now.3 Quantization of free Dirac fermion field theory Let us now apply the wisdom obtained by the quantization of a scalar field in the previous subsection to the problem of quantizing a Dirac fermion field that describes electrons and positrons. This shows that the infinite constant we dropped from H is actually the infinite energy density associated with an infinite universe of empty space.2. . Then the infinite constant would be rendered finite. s) and v(p. from now on. you will show for homework that: [H. which removes an electron with 3-momentum 53 .s . . Note that it is not possible to construct a state with a negative energy eigenvalue! 5. .2. + E~kn .42) is the Hamiltonian operator. ~kn i = a~† a~† . But we’ve already agreed to ignore it.) So. filled with the zero-point energies of an infinite number of oscillators. .2. s). The op- erator bp~.1) Here s labels the two possible spin states in some appropriate basis (for example.46) kn is easily shown to be an eigenstate of H with eigenvalue E~k1 + E~k2 + .some cutoff |~ p|max .s will be interpreted as an annihilation operator. Acting on the vacuum state.3. a~† ] = E~k a~† . as (5. (5. Now. a multi-particle state |~k1 . one for each possible momentum 3-vector ~p. A convenient such basis is the solutions we found to the Dirac equation. we won’t bother doing this.s + v(p.44) k Acting with H on a one-particle state. as: Ψ(~x) = 2 Z X s=1 i h d˜ p u(p. since Ψ is a spinor with four components. . s)e−i~p·~x dp†~. k (5. Sz = ±1/2). More generally. . .43) since all ap~ annihilate the vacuum. a~† |0i k1 k2 q ~k2 + m2 . at a given fixed time t. so let it go. . k k (5. Since we are going to ignore the constant anyway.

s i = −|ep ~.3. b† and d.s . Taking ~k = p~ and r = s in eq.3.3. b act on states to create and destroy particles.s i (5. ep ~.s + v(p. (5. ep− ~. e~ i.3. we assume the existence of a vacuum state |0i.s i.3.s .r k. d create and destroy antiparticles.7) − which can only be true if |ep− ~. The operator dp†~.s i = 0.s b~† |0i.r k. (5. then you can substitute “particle” for electron and “antiparticle” for positron.s |0i = |e~− . with the corresponding 3-momentum and spin. etc. we have: b~† bp†~. Now.3.3. which describes a universe of empty space with no electrons or positrons present. k. k.s .s |0i = |ep− ~. electrons are fermions. ep ~. (5.s destroys a positron. In the present case.6) in terms of creation operators.s is a creation operator. s)ei~p·~x dp~.4) b~† bp†~. and dp~.3) for all p~ and s. s=1 (5.s |0i = dp~. they must obey Fermi-Dirac statistics for identical particles. that means that we cannot add to the vacuum two electrons with exactly the same 3-momentum and spin.s |0i = −bp†~.p~ and spin state s from whatever state it acts on.r 54 (5.1).s i = −|ep ~. a† in order to distinguish the fermion and antifermion creation and annihilation operators from the scalar field versions. To make a state describing a single electron with 3-momentum ~p and spin state s.6) A corollary of this is the Pauli exclusion principle.2) The operator bp†~.3.3. and multiplying by γ 0 on the right. which states that two electrons cannot be in exactly the same state. Writing eq.r is a state containing two electrons. we must get a minus sign: − − |e~− . More generally. (5. So b† .r (5.5). We are using b.s . we get: Ψ(~x) = 2 Z X h i d˜ p u(p. This means that if we interchange the momentum and spin labels for the two electrons in the state of eq. if the Dirac field describes some fermions other than the electron-positron system.3. ep− ~. (5. s)e−i~p·~x bp†~. The annihilation operators yield 0 when acting on the vacuum state: bp~. while d† .6): − − − |ep− ~.s creates an electron.3.8) . just act on the vacuum with the corresponding creation operator: bp†~. which adds a positron to whatever state it acts on.s |0i = 0 (5.s i.s . in other words there is no such state. Taking the Hermitian conjugate of eq. (5. ep ~.5) Similarly. k. Just as in the case of a scalar field. d† rather than a.r k.

let us construct the momentum conjugate to Ψ(~x). in the classical limit ¯h → 0.s ] = 0 (Wrong!) k. b† . So it must be true that b. Ψ(y)} = 0. and d† anticommute even classically. (5. bp†~. one must have {Ψ(x). (5.14) (5. bp~.3. k.3.s } = 0. we must have an anticommutation relation: b~† bp†~. h ¯ → 0.r (5.r (5. Now. Ψ† (y)} = 0.16) So the momentum conjugate to the Dirac spinor field is just i times its Hermitian conjugate. k. {Ψ† (x). Ψ(~y )] = −i¯hδ(3) (~x − ~y ).r . as classical fields. instead of the commutation relation [b~† .s } = 0.11) Similarly. (5. applying the same thought process to identical positron fermions.15). d. δ∂0 Ψ (5. naively following the path of canonical quantization.18) Now just rewriting P = iΨ† .12) Note that in the classical limit.3.3.r (5.3. In order to discover how the classical equations (5. these equations are unaffected.15) are modified when one goes to the quantum theory.17) However. dp~. one might expect the equal-time commutation relation: [P(~x).3. So.3.s } = 0. instead we postulate a canonical anticommutation relation for the Dirac field operator and its conjugate momentum operator: {P(~x).3. one must have: {d~† . eq.3.r k.r . Interchanging the order of any two Grassmann numbers results in an overall minus sign.3. Ψ† (y)} = 0.19) .9) that one might expect from comparison with the scalar field. we must also have: {b~k. dp†~.3. but rather an anticommuting or Grassmann number.3. this clearly cannot be correct. bp†~.10) Taking the Hermitian conjugate.17) disagrees with eq.3.13) and (5.So. (5.s + bp†~. So. (Wrong!) (5. (5. {Ψ(x).s b~† = {b~† .s } = {d~k. this becomes: {Ψ† (~x). It is: P(~x) = δL = iΨγ 0 = iΨ† γ 0 γ 0 = iΨ† .15) Evidently. since these are anticommuting fields.r k. the classical Dirac field is not a normal number. Ψ(~y )} = −¯hδ(3) (~x − ~y ). since ¯h doesn’t appear anywhere. Ψ(~y )} = −i¯hδ(3) (~x − ~y ).3. 55 (5.3.13) (5.

one can obtain: {bp~. a† operators. bp†~n . For homework.r1 .24) which describes a state with n electrons (particles) and m positrons (antiparticles) with the obvious 3-momenta and spins.21) s=1 in a way very similar to the way we found the Hamiltonian for a scalar field in terms of a. (5. + E~km . k.From this.3.s bp~.s ) (5. this time) which is unobservable because it is the same for all states. . .s † [H. + Ep~n + E~k1 + . d~† km . . b~† } = {dp~. using a strategy similar to that used for scalar fields.s .22) = (5.20) and all other anticommutators of b. b~† ] = E~k b~† .r k. you will show that: [H. In doing so. . d† operators vanish. Note that H again has energy eigenvalues that are ≥ 0. .sn d~† k1 .r (5.s dp~.rm |0i. b† .3.s . .s k. . and total energy Ep~1 + . 56 .3. d. one must again drop an infinite constant contribution (negative.s + dp†~. d~ ] k. One also can check that the Hamiltonian is H= 2 Z X d˜ p Ep~ (bp†~.23) k. k. d~† } = (2π)3 2Ep~ δ(3) (~ p − ~k) δsr . (5.3.s E~k d~† . .s1 .s (Note that these equations are commutators rather than anticommutators!) It follows that the eigenstates of energy and 3-momentum are given in general by: bp†~1 .3.

In field theory. with exact solutions that are not too hard to find. so one usually treats the non-quadratic part of the Hamiltonian as a perturbation on the quadratic part. and add to it an interaction term: L = L0 + Lint .1 Scalar field with φ4 coupling So far. so that the Euler-Lagrange equations obtained by varying L are linear wave equations in the fields.3) (6. this will lead to non-linear equations of motion that have to be solved approximately. except that now the Hamiltonian is H = H0 + Hint . The factor of 1/4! = 1/24 is a convention.1.6 Interacting scalar field theories 6. this nice feature shows up in the simple time evolution of the states. This gives only approximate answers. As an example. finding exact energy eigenstates of the Hamiltonian is not possible. To describe this.1.1. as given in eq. which is unfortunate. the state at some time t is just the same as the state at some previous time t0 .1. These are theories in which the Lagrangian density is quadratic in the fields. λ Lint = − φ4 . So nothing ever happens to the particles in a free theory. as in any quantum system.1. At the classical level. dt (6. but makes life interesting. and in which some particles can decay into other sets of particles. if we set λ = 0. At the quantum level.2) In other words. (6. a parameter of the theory known as a coupling. and the reason for it will be apparent later. (5. we have been dealing with free field theories. It governs the strength of interactions. the time evolution of a state |Xi is given in the Schrodinger picture by i¯h d |Xi = H|Xi. in the case of a multiparticle state with an energy eigenvalue E as described above. consider the free Lagrangian for a scalar field φ. (6. At the quantum level. We are interested in describing a more interesting situation where particles can scatter off each other.18). (6.1) So.1.6) . we would be back to the free theory in which nothing interesting ever happens. 24 (6. perhaps inelastically to create new particles.5) where Hint = − Z 3 d ~x Lint λ = 24 57 Z d3 ~x φ(~x)4 . their number does not change. up to a phase. the solution is just |X(t)i = e−i(t−t0 )H |X(t0 )i = e−i(t−t0 )E |X(t0 )i.1. and their momenta and spins remain the same. Now canonical quantization can proceed as before. we need a Lagrangian density that contains terms with more than two fields.4) Here λ is a dimensionless number.

1. and k1 k2 in the far future. giving rise to the factors of 4. and therefore not of interest. (6. The states that are simple two-particle states are |pa . (5. (5. This involves reordering the a’s and a† ’s.9) We will work in the Schrodinger picture of quantum mechanics. so the system is accurately described by the free Hamiltonian and its energy eigenstates. 6. and 4. because it just corresponds to the situation where a particle is “scattered” without changing its momentum at all. The result is: Hint λ = (2π)3 24 Z d˜ q1 Z d˜ q2 Z d˜ q3 Z " d˜ q4 aq†~1 aq†~2 aq†~3 aq†~4 δ(3) (~q1 + ~q2 + ~q3 + ~q4 ) +4aq†~1 aq†~3 aq†~3 aq~4 δ(3) (~q1 + ~q2 + ~q3 − ~q4 ) +6aq†~1 aq†~2 aq~3 aq~4 δ(3) (~q1 + ~q2 − ~q3 − ~q4 ) +4aq†~1 aq~2 aq~3 aq~4 δ(3) (~q1 − ~q3 − ~q3 − ~q4 ) +aq~1 aq~2 aq~3 aq~4 δ (3) (~q1 + ~q2 + ~q3 + ~q4 ) # . So.10) |k1 . To see how to use the interaction Hamiltonian. we can pretend that Hint is “turned off” in both the far past and the far future. we are interested in 58 . The same applies to the outgoing particles in the far future.1. Now. (6. These are built out of creation and annihilation operators just as before. by relabeling the momenta.11) in the far past. which is the same as no scattering.Let us write this in terms of creation and annihilation operators.8) Here we have combined several like terms. In doing so. This should not cause any worry. it is useful to tackle a specific process. Nevertheless. producing two scalar particles with 4-momenta k1 . so they are eigenstates of the free Hamiltonian H0 . k2 iOUT = a~† a~† |0i. pb iIN = ap†~a ap†~b |0i. For example.1.1. (6. using eq. (6.2. using eq. we have ignored the fact that a’s do not commute with a† ’s when the 3-momenta are exactly equal. pb that interact. we assume that the incoming particles were far apart.1.19). in the far past.1).7) Now we can perform the d3 ~x integration.30): Hint = λ 24 Z d3 ~x Z d˜ q1 Z d˜ q2 Z d˜ q3 Z  d˜ q4 aq~1 ei~q1 ·~x + aq†~1 e−i~q1 ·~x  aq~3 ei~q3 ·~x + aq†~3 e−i~q3 ·~x  aq~2 ei~q2 ·~x + aq†~2 e−i~q2 ·~x  aq~4 ei~q4 ·~x + aq†~4 e−i~q4 ·~x   . in which operators are time-independent and states evolve in time according to eq. k2 : pa pb → k1 k2 . consider a scattering problem in which we have two scalar particles with 4-momenta pa . (6.2.1. (6. but not of the full Hamiltonian.

21) . (6. We use the definition of the exponential to write: e−iT H = [1 − iHT /N ]N = [1 − i(H0 + Hint )T /N ]N .1.19) −i(T −t)H0 OUT hk1 . k2 |pa . Ef = E~k1 + E~k2 (6.computing the probability amplitude that the state |pa . pb iIN .1.1. let us consider the contribution linear in λ. k2 iOUT .1.1. pb iIN .10) and (6.15) for N → ∞.13) where T is the long time between the far past time when the initial state was created and the far future time at which the overlap is computed. OUT hk1 . pb iIN . say in the far future: OUT hk1 .1. For example. pb iOUT = e−iT H |pa .18) = (6. see eqs.1. So we have: = OUT hk1 . as uninteresting. the part of this that is linear in Hint can be expanded as: e−iT H = N −1 X n=0 [1 − iH0 T /N ]N −n−1 (−iHint T /N ) [1 − iH0 T /N ]n . Now. order by order in the coupling λ. (6.16) (Here we have dropped the 0th order part. k2 |e−iT H |pa . However. k2 |e −i(T −t)Ef . OUT hk1 . e−iT H cannot be written exactly in a useful way in terms of creation and annihilation operators.1.1.12) The state |pa . pb iOUT is the time evolution of |pa . (6.1.10) and (6. pb iIN . (6. (6. The complications are hidden in the operator e−iT H . (6.14) The states appearing on the right-hand side are simple. k2 |e where Ei = Ep~a + Ep~b . respectively. pb iOUT .1. k2 |e |pa . So we have: −iT H OUT hk1 . pb iIN = e−itEi |pa . k2 |Hint |pa . it just corresponds to the particles evolving as free particles.1. (6.20) are the energies of the initial and final states. k2 |pa .17) Next we can use the fact that we know what H0 is when acting on the simple states of eqs. According to the rules of quantum mechanics this is given by their overlap at a common time.11): e−itH0 |pa . In general. we can do it perturbatively. e−iT H0 . (6.11).1. pb iIN = −i Z 0 T dt e−i(T −t)Ef e−itEi 59 OUT hk1 . pb iIN from the far past to the far future: |pa . by letting t = nT /N and dt = T /N in the limit of large N : e−iT H = −i Z T 0 dt e−i(T −t)H0 Hint e−itH0 .) We can now turn this discrete sum into an integral. pb iIN evolves to the state |k1 .1. pb iOUT (6. (6.

1. so we can ignore them too. pb iIN .29) This can be done using the commutation relations of eqs.1.1. pb iOUT = −i(2π) λ 4 Z d˜ q1 Z d˜ q2 Z d˜ q3 Z d˜ q4 δ(4) (q1 + q2 − q3 − q4 ) † † OUT hk1 . (6. It remains to evaluate: † † OUT hk1 . k2 |aq~1 aq~2 aq~3 aq~4 |pa . The action of Hint on eigenstates of the free Hamiltonian can be read off from the different types of terms. k2 |Hint |pa .2.26) Therefore.1. which will just give 1 when we take the complex square of the probability amplitude): OUT hk1 .27) Here we have combined the three-momenta delta function from eq. we can use the integral identity Z ∞ dx eixA = 2πδ(A) (6. and commute aq†~1 and aq†~2 to 60 . and (dropping the phase factor e−i(Ef +Ei )T /2 . The aaaa-type term will remove four particles from the state.1. (6. (6.8). k2 |pa .1. the only term that will play a role in this example is the a† a† aa contribution: Hint = λ (2π)3 4 Z d˜ q1 Z d˜ q2 Z d˜ q3 Z d˜ q4 δ(3) (~q1 + ~q2 − ~q3 − ~q4 ) aq†~1 aq†~2 aq~3 aq~4 .28) which. we have: 4 OUT hk1 .25) to give a 4-momentum delta function. (6. (5.11). The terms of type a† a† a† a† and a† a† a† a create more than the two particles we know to be in the final state.2. pb iIN .1.1. Therefore.20) and (5.1. (6. (6. k2 |aq~1 aq~2 aq~3 aq~4 |pa . (6. As we take T → ∞.1.22) where we have redefined the integration variable by t = t′ + T /2. Clearly we don’t have to worry about that. k2 |pa .24) This tells us that energy conservation will be enforced.25) Now we are ready to use our expression for Hint in eq.1.26) with the energy delta function from eq. (6.23) −∞ to obtain: Z T 0 dt e−i(T −t)Ef e−itEi = 2π δ(Ef − Ei ) e−i(Ef +Ei )T /2 .21). according to eqs.1. because there were only two particles in the state to begin with! The same goes for the a† aaa-type term.1. (6.First let us do the t integral: Z 0 T dt e−i(T −t)Ef e−itEi = e−i(Ef +Ei )T /2 Z T /2 −T /2 ′ dt′ eit (Ef −Ei ) (6. so they can give 0 when acting on |0i. is equal to h0|a~k1 a~k2 aq†~1 aq†~2 aq~3 aq~4 ap†~a ap†~b |0i. (6.10) and (6. pb iIN . pb iOUT = −i 2π δ(Ef − Ei ) OUT hk1 . The strategy is to commute aq~3 and aq~4 to the right.

You can check that each of them gives a contribution equal to eq.2 Scattering processes and cross-sections In subsection 6. The two lines outgoing to the right represent the two final state scalar particles. called Feynman rules. we found that the matrix element corresponding to 2 particle to 2 particle scattering λ 4 in a scalar field theory with interaction Lagrangian − 24 φ is: OUT hk1 k2 |pa pb iOUT = −iλ(2π)4 δ(4) (k1 + k2 − pa − pb ). d3 ~ q3 . after relabeling momenta. pb iOUT = −iλ(2π)4 δ(4) (k1 + k2 − pa − pb ). (6. These rules. this cancels the factor of 1/4 in eq. q4 = pb .1.2. (6. and q2 = k2 in the remaining 4-momentum delta function that was already present in eq.1. This is just the simplest of many Feynman diagrams one could write down for the process of two particle scattering in this theory. and is associated with the factor −iλ. the factors of (2π)3 2Eq~3 etc. we are left with the remarkably simple result: OUT hk1 . one can instead summarize the procedure with a simple set of diagrammatic rules.31) Rather than go through this whole messy procedure every time we invent a new interaction term for the Lagrangian density. the Feynman diagram for the process is: initial state final state Here the two lines coming from the left represent the incoming state scalar particles.27).1. which are resurrected by the two creation operators in Hint . (6. (5. are useful both as a precise summary of a matrix element calculation. d3 ~ q2 .1.27). and as a heuristic guide to what physical process the calculation represents. Putting it all together. (6.30) There are 3 more similar terms.1) . 61 (6.2. [See eq. all neatly cancel against the corresponding factors in the denominator of d˜ q3 .the left so they can give 0 when acting on h0|. In the present case. It yields: (2π)3 2Eq~3 δ(3) (~ q3 − ~ pa ) (2π)3 2Eq~4 δ(3) (~ q4 − p~b ) (2π)3 2Eq~1 δ(3) (~q1 − ~k1 ) (2π)3 2Eq~2 δ(3) (~q2 − ~k2 ). Along the way.] The three-momentum delta functions then make the remaining d3 ~ q1 . they just set the four-vectors q3 = pa .1. q1 = k1 . But all other diagrams represent contributions that are higher order in λ. One contribution occurs when ~q3 = p~a and ~q4 = ~pb and ~q1 = ~k1 and ~q2 = ~k2 . k2 |pa . (6.27). (6.1. The vertex where the four lines meet represents the interaction itself. which get “destroyed” by the annihilation operators in Hint . and d3 ~ q4 integrations trivial. one picks up delta functions whenever the 3-momenta of an a and a† match.1. or every time we think of a new scattering process. 6.27).30) when put into eq. so if λ is small we can ignore them. Now. etc.

2. 1 fb = 10−39 cm2 = 2. NS = σ Z L dt. At the time of this writing (January 2011). namely the cross-section.7) The joke is that achieving an event with such a cross-section is “as easy as hitting the broad side of a barn”. and zero otherwise. Suppose that we have Na particles of type a and Nb of type b. Therefore. So clearly we must do some work to relate it to an appropriate physically measurable quantity.568 × 10−12 GeV−2 . However.5) (6. The rate at which the effective Na Nb /A is increasing with time in an experiment is called the luminosity L (or instantaneous luminosity). and the official unit is 1 barn = 10−24 cm2 = 2568 GeV−2 . and femto-: 1 nb = 10−33 cm2 = 2.3) The dimensions for cross-section are the same as area. The matrix element itself is infinite whenever 4-momentum is conserved.2.4) (6. 000 pb−1 .568 × 10−6 GeV −2 . (6. the total amount of data collected in Run II corresponds to an integrated luminosity of about: Z † Ldt = 10 fb−1 = 10.Now we would like to learn how to translate this information into something physically meaningful that could in principle be measured in an experiment. The equation NS = Na Nb σ A (6. ←− ~vb ~va −→ The two packets are assumed to have the same area A (shaded gray) perpendicular to their motion. 1 pb = 10−36 cm2 = 2.2. The cross-section is the observable that gives the expected number of scattering events NS that will occur if two large sets of particles are allowed to collide. (6. and the same quantity integrated over time.† so more commonly-used units are obtained by using the prefixes nano-.2) defines the cross-section σ. is called the integrated luminosity.2. and inversely proportional to the area A.2. The total number of scattering events occurring while the packets move through each other should be proportional to each of the numbers Na and Nb . (6. pico-. 62 . a barn is a very large cross-section.568 × 10−9 GeV −2 . formed into large packets of uniform density that move completely through each other as shown. the Tevatron collides protons (p) and antiprotons (p) with a center-of-momentum (CM) energy of ECM = 1960 GeV.2.6) As an example.

1 × 1032 cm−2 s−1 . A more interesting final state would be anything involving a top quark (t)and anti-top quark (t) pair. with a record integrated luminosity for one day of R Ldt = 6. At this writing. 2. (Unfortunately. as we will see in section 7. featuring final states of well-known and well-understood hadrons. and several hundred thousand will be within a year. the plan is to aim for 1 to 3 fb−1 by the end of the run starting in 2011. a total integrated luminosity of 48 pb−1 was delivered to each detector. in April 2010.2. (6. then the Tevatron cross-section to produce it in association with a Z 0 boson is roughly: σ(pp → hZ 0 + anything) = 75 fb. this estimate is quite fuzzy.5.04 × 1032 cm−2 sec−1 . so far.10) This implies 750 or so hZ 0 events in Run II. If the mass of the Higgs boson is mh = 130 GeV.075 barns. . Also. (6. In general.2. which depends on the final state. . The design luminosity to be eventually reached by the LHC is 1. for which the Tevatron cross-section is about: σ(pp → tt + anything) = 7. producing n final-state particles with masses mi and 4-momenta ki .11) so about 7700 tt pairs have already been produced.9) This means that about 75. To figure out how many scattering events one expects at a collider. where i = 1.0 pb−1 . A more speculative possibility is that the Tevatron will find a Higgs boson h. n: 63 .8) are extremely uninteresting. The total cross-section for any type of scattering at the Tevatron is quite large. by one estimate approximately σ(pp → anything) = 0.0 × 1034 cm−2 s−1 . . However. The record for a single week is R Ldt = 73 pb−1 . Let us consider a general situation in which two particles a and b with masses ma and mb and 4-momenta pa and pb collide. only a fraction of these are identified as such.) In the 7 TeV pp collision run at the LHC in 2010.The peak luminosity obtained so far in Run II of the Tevatron is about L = 4. .2.2. Much of this occurred during the last few days of running. (6.04 × 10−4 pb−1 sec−1 . (6. hopefully.5 pb. At 7 TeV collision energies. one needs to know the corre- sponding cross-section for that type of event. our task is to figure out how to relate the matrix element for a given collision process to the corresponding cross-section. if mh = 130 GeV.8) However.000 top pairs should have been produced so far in Run II. the cross-section actually becomes arbitrarily large due to the long-range nature of the Coulomb force. (6. there is also a large background from similar-looking events with a more mundane origin. For arbitrarily small momentum transfer in elastic scattering of charged particles. or 4. and a maximum peak instantaneous luminosity of L = 2. because it depends on the minimum momentum transfer that one requires in order to say that a scattering event has occurred. the vast majority of the scattering events reflected in eq. the cross-section for producing top-antitop pairs is about σ(pp → tt + anything) = 160 pb.2.

all of the particles could be different.16) . (6..14) Here Na Nb represents the total number of initial states. . kn . mn . mb iIN and the final state |f i = |k1 .12) i=1 Here Mi→f is called the reduced matrix element for the process.15) So. hf |f ihi|ii (6. we can always write: hf |ii = Mi→f (2π)4 δ(4) (pa + pb − n X ki ). provided that the process conserves angular momentum. If the particles are not scalars.2. mn iOUT .. They can be either fermions or bosons.k1 .. (6. so it is only known approximately. because of four-momentum conservation. (2π)3 (6.. It is computed order-by-order in perturbation theory. in general. ma As an abbreviation. the probability of a transition from the state |ii to the state |f i is: Pi→f = |hf |ii|2 . the total number of scattering events expected to occur is: NS = N a Nb X f Pi→f . mb kn .. p a . Now. m1 .2. M can be a non-trivial Lorentz-scalar function of the various 4-momenta and spin eigenvalues of the particles in the problem. P is over all possible final states f . then |ii and |f i should also carry labels that specify the spin of each particle. which are not unity. Now.2.2. .. To evaluate this. one for each possible pair of incoming particles. including a sum over the n final-state particles implies X f → n Y i=1 " V Z 64 # d3~ki . so that a different species of creation and annihilation operators might be used for each.2. then the density of one-particle states with 3-momentum ~k is The sum f density of states = V d3~k . ... they will be computed below. In general.1. recall that if one puts particles in a large box of volume V . (2π)3 (6. . According to the postulates of quantum mechanics. m1 p b . However. pb . the reduced matrix element we found to first order in the coupling λ was simply a constant: Mφφ→φφ = −iλ. In the example in subsection 6. ma . we can call the initial state |ii = |pa .13) The matrix element has been divided by the norms of the states.

2.1. For the initial state |ii of eq.22) Commuting ap†~b to the left in the same way yields the norm of the state |ii: hi|ii = (2π)6 4Ea Eb δ(3) (~ pa − p~a )δ(3) (~ pb − ~pb ).2.18) (Assume that the volume V and area A of each bunch are very large compared to the cube and square of the particles’ Compton wavelengths. (6.24) .2.10). (6. (6.2. so in the last term the ap†~a just commutes with ap~b according to eq. since the arguments of each delta function vanish! In order to successfully interpret it. V T |~va − ~vb | i=1 (2π)3 (6.20).2.2). we have for the differential contribution to the total cross-section: dσ = Pi→f A n Y i=1 " # d3~ki .Putting this into eq. one has: hi|ii = h0|ap~b ap~a ap†~a ap†~b |0i. One can write 3 (3) (2π) δ (~ p − ~p ) = Z 65 ~ d3 ~x ei0·~x = V. To see how that happens in the example of φφ → φφ scattering with a φ4 interaction.19) The total cross-section is obtained by integrating over 3-momenta of the final state particles. (5. (6.2.2. yielding 0 when acting on h0|. hi|ii = h0|ap~b ap†~b |0i (2π)3 2Ea δ(3) (~ (6. Note that the differential cross-section dσ depends only on the collision process being studied.2. So in the following we expect that the arbitrary volume V and packet collision time T should cancel out. (6.20) To compute this. let us first compute the normalizations of the states appearing in Pi→f .2.21) Now the incoming particle momenta p~a and p~b are always different. (5. A|~va − ~vb | (6.) It follows that the differential contribution to the cross-section is: # " n Y V d3~ki dσ = Pi→f .20): pa − p~a ).2. (6.2.2. (6. The first term can be simplified using the commutator eq. recall the origin of the 3-momentum delta functions.17) Let us now suppose that each packet of particles consists of a cylinder with a large volume V . V (2π)3 (6. one can commute the a† operators to the left: hi|ii = h0|ap~b [ap~a ap†~a ]ap†~b |0i + h0|ap~b ap†~a ap~a ap†~b |0i. Then the total time T over which the particles can collide is given by the time it takes for the two packets to move through each other: T = V .14) and comparing with the definition eq.23) This result is doubly infinite.

showing that a 3-momentum delta-function with vanishing argument corresponds to V /(2π)3 , where V
is the volume that the particles occupy. So we obtain
hi|ii = 4Ea Eb V 2

(6.2.25)

for the norm of the incoming state.
Similarly, the norm of the state |f i is:
hf |f i =

n
Y

(2π)3 2Ei δ(3) (~ki − ~ki ) =

i=1

n
Y

(2Ei V ).

(6.2.26)

i=1

In doing this, there is one subtlety; unlike the colliding particles, it could be that two identical outgoing
particles have exactly the same momentum. This seemingly could produce “extra” contributions when
we commute a† operators to the left. However, one can usually ignore this, since the probability that
two outgoing particles will have exactly the same momentum is vanishingly small in the limit V → ∞.
Next we turn to the square of the matrix element:
h

|hf |ii|2 = |Mi→f |2 (2π)4 δ(4) (pa + pb −

X

i2

ki )

.

(6.2.27)

This also is apparently the square of an infinite quantity. To interpret it, we again recall the origin of
the delta functions:
2πδ(Ef − Ei ) =
X

(2π)3 δ(
So we can write:

~k −

X

p~ ) =

Z

Z

T /2
−T /2

dt eit(Ef −Ei ) = T
P~ P
k−

d3 ~x ei~x·(

(2π)4 δ(4) (pa + pb −

n
X

p
~)

= V

(for Ef = Ei ),
(for

X

~k =

X

ki ) = T V.

(6.2.28)
p~ ).

(6.2.29)

(6.2.30)

i=1

Now if we use this to replace one of the two 4-momentum delta functions in eq. (6.2.27), we have:
|hf |ii|2 = |Mi→f |2 (2π)4 T V δ(4) (pa + pb −

n
X

ki ).

(6.2.31)

i=1

Plugging the results of eqs. (6.2.25), (6.2.26) and (6.2.31) into eq. (6.2.13), we obtain an expression
for the transition probability:
Pi→f = |Mi→f |2 (2π)4 δ(4) (pa + pb −

n
X

ki )

i=1 

n 
Y
T
1
.
4Ea Eb V i=1 2Ei V

(6.2.32)

Putting this in turn into eq. (6.2.19), we finally obtain:
dσ =

|Mi→f |2
dΦn ,
4Ea Eb |~va − ~vb |

(6.2.33)

where dΦn is a standard object known as the n-body differential Lorentz-invariant phase space:
4 (4)

dΦn = (2π) δ

(pa + pb −

n
X
i=1

66

ki )

n
Y

i=1

d3~ki
(2π)3 2Ei

!

.

(6.2.34)

As expected, all factors of T and V have canceled out of the formula eq. (6.2.33) for the differential
cross-section.
In the formula eq. (6.2.33), one can write for the velocities:
p~a
,
Ea

~va =

~vb =

p~b
.
Eb

(6.2.35)

Now, assuming that the collision is head-on so that ~vb is opposite to ~va (or 0), the denominator in
eq. (6.2.33) is:
4Ea Eb |~va − ~vb | = 4(Ea |~
pb | + Eb |~
pa |).

(6.2.36)

The most common case one encounters is two-particle scattering to a final state with two particles.
In the center-of-momentum frame, p~b = −~
pa , so that the 2-body Lorentz-invariant phase space becomes:
dΦ2 = (2π)4 δ(3) (~k1 + ~k2 )δ(Ea + Eb − E1 − E2 )
=

q

δ(3) (~k1 + ~k2 ) δ( ~k12 + m21 +
16π 2

where

q

q

q

d3~k2
d3~k1
(2π)3 2E1 (2π)3 2E2

~k2 + m2 − ECM )
2
2

~k2 + m2 ~k2 + m2
1
1
2
2

d3~k1 d3~k2 ,

ECM ≡ Ea + Eb

(6.2.37)
(6.2.38)

(6.2.39)

is the center-of-momentum energy of the process. Now one can do the ~k2 integral; the 3-momentum
delta function just sets ~k2 = −~k1 (as it must be in the CM frame). If we define
K ≡ |~k1 |

(6.2.40)

d3~k1 = K 2 dK dΩ = K 2 dK dφ d(cos θ),

(6.2.41)

for convenience, then

where Ω = (θ, φ) are the spherical coordinate angles for ~k1 . So
Z

. . . dΦ2 =

Z

...

q

δ( K 2 + m21 +
q

K2

+

q

K 2 + m22 − ECM )

m21

q

K2

+

m22

K 2 dK dφ d(cos θ),

(6.2.42)

where . . . represents any quantity. To do the remaining dK integral, it is convenient to change variables
to the argument of the delta function. So, defining
W =

q

K 2 + m21 +

q

K 2 + m22 − ECM ,

(6.2.43)

we find
dW =

q

K 2 + m21 +

q

q

q

K 2 + m22

K 2 + m21 K 2 + m22
67

KdK.

(6.2.44)

Noticing that the delta function predestines
write:
q

q

K 2 + m21 +

K 2 dK
q

K 2 + m21 K 2 + m22

=

q

K 2 + m22 to be replaced by ECM , we can

KdW
.
ECM

(6.2.45)

Using this in eq. (6.2.42), and integrating dW using the delta function δ(W ), we obtain:
dΦ2 =

K
dφ d(cos θ)
16π 2 ECM

(6.2.46)

for the Lorentz-invariant phase space of a two-particle final state.
Meanwhile, in the CM frame, eq. (6.2.36) simplifies according to:
4Ea Eb |~va − ~vb | = 4(Ea |~
pb | + Eb |~
pa |) = 4|~
pa |(Ea + Eb ) = 4|~
pa |ECM .

(6.2.47)

Therefore, using the results of eqs. (6.2.46) and (6.2.47) in eq. (6.2.33), we have:
dσ = |Mi→f |2

~

|k1 |
dΩ
2 |~
2
64π ECM
pa |

(6.2.48)

for the differential cross-section for two-particle scattering to two particles.
The matrix element is quite often symmetric under rotations about the collision axis determined by
the incoming particle momenta. If so, then everything is independent of φ, and one can do
leaving:
dσ = |Mi→f |2

|~k1 |
d(cos θ).
2 |~
32πECM
pa |

R

dφ = 2π,

(6.2.49)

If the particle masses satisfy ma = m1 and mb = m2 (or they are very small), then one has the further
simplification |~k1 | = |~
pa |, so that
dσ = |Mi→f |2

1
2 d(cos θ).
32πECM

(6.2.50)

The formulas eqs. (6.2.48)-(6.2.49) will be used often in the rest of these notes.
We can finally interpret the meaning of the result eq. (6.1.31) that we obtained for scalar φ4 theory.
Since we found Mφφ→φφ = −iλ, the differential cross-section in the CM frame is:
dσφφ→φφ =
Now we can integrate over θ using

R1

λ2
2 d(cos θ).
32πECM

−1 d(cos θ)

(6.2.51)

= 2. However, there is a double-counting problem that

we must take into account. The angles (θ, φ) that we have integrated over represent the direction of
the 3-momentum of one of the final state particles. The other particle must then have 3-momentum
in the opposite direction (π − θ, −φ). The two possible final states with ~k1 along those two opposite

directions are therefore actually the same state, because the two particles are identical. So we have

68

actually counted each state twice when integrating over all dΩ. To take this into account, we have to
divide by 2, arriving at the result for the total cross-section:
σφφ→φφ =

λ2
2 .
32πECM

(6.2.52)

In the system of units with c = h
¯ = 1, energy has the same units as 1/distance. Since λ is dimensionless,
it checks that σ indeed has units of area. This is a very useful thing to check whenever one has found
a cross-section!

6.3

Scalar field with φ3 coupling

For our next example, let us consider a theory with a single scalar field as before, but with an interaction
Lagrangian that is cubic in the field:
µ
(6.3.1)
Lint = − φ3 ,
6
instead of eq. (6.1.4). Here µ is a coupling that has the same dimensions as mass. As before, let us
compute the matrix element for 2 particle to particle scattering, φφ → φφ.

The definition and quantization of the free Hamiltonian proceeds exactly as before, with equal

time commutators given by eqs. (5.2.20) and (5.2.21), and the free Hamiltonian by eq. (5.2.42). The
interaction part of the Hamiltonian can be obtained in exactly the same way as the discussion leading
up to eq. (6.1.8), yielding:
Hint

µ
= (2π)3
6

Z


q1

Z


q2

Z

"


q3 aq†~1 aq†~2 aq†~3 δ(3) (~q1 + ~q2 + ~q3 )
+3aq†~1 aq†~2 aq~3 δ(3) (~q1 + ~q2 − ~q3 )

+3aq†~1 aq~2 aq~3 δ(3) (~q1 − ~q2 − ~q3 )
+aq~1 aq~2 aq~3 δ

(3)

(~q1 + ~q2 + ~q3 )

#

.

(6.3.2)

As before, we want to calculate:
~ ~

pa p~b iOUT
OUT hk1 , k2 |~

= OUT h~k1 , ~k2 |e−iT H |~
pa , p~b iIN

(6.3.3)

where H = H0 + Hint is the total Hamiltonian. However, this time if we expand e−iT H only to first
order in Hint , the contribution is clearly zero, because the net number of particles created or destroyed
by Hint is always odd. Therefore we must work to second order in Hint , or equivalently in the coupling
µ.
Expanding the operator e−iT H to second order in Hint , one has in the large-N limit [compare to
eq. (6.1.15) and the surrounding discussion]:
e−iT H

=
= 

1 − i(H0 + Hint )

N
−2 N −n−2
X
X 
n=0

m=0

T
N

1 − iH0 

N

T
N 

N −n−m−2

(6.3.4)
(−iHint 

T
T
) 1 − iH0
N
N 

m

(−iHint 

T
T
) 1 − iH0
N
N 

n

,
(6.3.5)

69

2 . Now we can convert the discrete sums into
where we have kept only terms that are of order Hint

integrals over the variables t = T n/N and t′ = T m/N + t with ∆t = ∆t′ = T /N . The result is:
−iT H

e

=

Z

T

dt

Z

T

T −t

0

dt′ e−iH0 (T −t ) (−iHint ) e−iH0 (t −t) (−iHint ) e−iH0 t .

(6.3.6)

When we sandwich this between the states h~k1 , ~k2 | and |~
pa , p~b i, we can substitute

e−iH0 (T −t ) → e−iEf (T −t ) ;

(6.3.7)

e−iH0 t → e−iEi t ,

(6.3.8)

where Ei = Ep~a + Ep~b is the initial state energy eigenvalue and Ef = E~k1 + E~k2 is the final state energy
eigenvalue. Now, (−iHint )|~
pa , p~b i will consist of a linear combination of eigenstates |Xi of H0 with
different energies EX . So we can also substitute

e−iH0 (t −t) → e−iEX (t −t)

(6.3.9)

provided that in place of EX we will later put in the appropriate energy eigenvalue of the state created
by each particular term in −iHint acting on the initial state. So, we have:
e−iT H = (−iHint ) I (−iHint )

(6.3.10)

where
I=

Z

T

dt

0

Z

T

T −t

dt′ e−iEf (T −t ) e−iEX (t −t) e−iEi t .

(6.3.11)

To evaluate the integral I, we can first redefine t → t − T /2 and t′ → t′ − T /2, so that
I = e−iT (Ei +Ef )/2

Z

T /2

−T /2

dt eit(EX −Ei )

Z

T /2

t

dt′ eit (Ef −EX ) .

(6.3.12)

Now e−iT (Ei +Ef )/2) is just a constant phase that will go away when we take the complex square of the
matrix element, so we drop it. Taking the limit of a very long time T → ∞:
I=

Z

−∞

it(EX −Ef )

dt e

Z

t

dt′ eit (Ef −EX +iǫ) .

(6.3.13)

The t′ integral does not have the form of a delta function because its lower limit of integration is t.

Therefore we have inserted an infinitesimal factor e−ǫt so that the integral converges for t′ → ∞; we

will take ǫ → 0 at the end. Performing the t′ integration, we get:
I =

i
Ef − EX + iǫ

= 2π δ(Ef − Ei )

!Z

−∞

dt eit(Ef −Ei )

i
Ef − EX + iǫ

!

.

As usual, energy conservation between the initial and final states is thus automatic.
70

(6.3.14)
(6.3.15)

Putting together the results above. and the left Hint contains a† a† a terms. (6.3.3.3. Taking these pieces from eq. p~b iOUT OUT hk1 .16).16) is: ~ ~ pa .3. p~b iOUT OUT hk1 . (6. (6. the contribution from eq.2). we have so far: ~ ~ pa . we can divide the calculation up into pieces. k2 |~  . k2 |~ i = 2πδ(Ef − Ei )h0|a~k1 a~k2 (−iHint ) Ef − EX + iǫ ! (−iHint )ap†~a ap†~b |0i. let us consider the contribution when the right Hint contains a† aa terms acting on the initial state.16) Let us now evaluate the matrix element eq. To do this. First. depending on how many a and a† operators are contained in each factor of Hint . (6.

.

.

21). a† pair.17) The factor involving EX is left inserted within the matrix element to remind us that EX should be replaced by the eigenvalue of the free Hamiltonian H0 acting on the state to its right. (6.3.2. This allows us to identify what EX is. Evaluating the quantity h0|a~k1 a~k2 a~r†1 a~r†2 a~r3 aq†~1 aq~2 aq~3 ap†~a ap†~b |0i 71 (6. because an a acting on |0i or an a† acting on h0| vanishes. The delta-function δ(3) (~q1 − ~q2 − ~q3 ) then insures that there will be a non-zero contribution only when ~ ~q1 = p~a + ~pb ≡ Q.20) and (5. The aq~2 and aq~3 operators must be contracted with ap†~a and ap†~b if a non-zero result is to be obtained. using eqs. we will get a non-zero contribution with a delta function whenever the 3-momentum of an a equals that of an a† with which it is commuted. (5. ap†~b ] and [aq~3 .3. There are two ways to do this: either pair up [aq~2 .3.18) So the energy eigenvalue of the state aq†~1 aq~2 aq~3 ap†~a ap†~b |0i must be replaced by EX = EQ~ = q |~ pa + ~pb |2 + m2 (6. ap†~a ].2.3. d˜ q3 (6. In doing so. (6. every a must “contract” with some a† in this way (and vice versa). the result can be non-zero only if ~q2 + ~q3 = p~a + ~ pb . removing that a. We commute a’s to the right and a† ’s to the left. or pair up [aq~2 .20) . ap†~b ]. The last line in eq.3. ap†~a ] and [aq~3 . In the end. In both cases.17) can be calculated using the following general strategy.19) whenever there is a non-zero contribution. 2 (a† a† a)(a† aa) part = Z Z µ r2 r1 d˜ −i (2π)3 2πδ(Ef − Ei ) d˜ 2 δ(3) (~r1 + ~r2 − ~r3 ) δ(3) (~q1 − ~q2 − ~q3 ) h0|a~k1 a~k2 a~r†1 a~r†2 a~r3 i Ef − EX ! Z d˜ r3 Z d˜ q1 Z d˜ q2 Z aq†~1 aq~2 aq~3 ap†~a ap†~b |0i.

the other three sets of contractions listed in eqs. ap†~b ] [a~r3 .23) [a~k1 . the various factors of (2π)2 2E just cancel the factors in the denominators of the definition of d˜ qi and d˜ ri .3. ~r1 .22)-(6. after a relabeling of momenta. ap†~a ] [aq~3 . as allowed by the delta function) we have: ~ ~ pa .24) For the first of these contributions to eq.3. ap†~b ] [a~r3 . ap†~b ] [aq~3 . corresponding to the following ways of contracting a’s and a† ’s: [a~k1 . One can do the ~ q1 . resulting in the following contribution to eq.20). 2EQ~ (6. ap†~b ] [aq~3 . ~r2 . ap†~a ] [a~r3 .3.3. (6. (6. ap†~a ] [aq~3 .3. a~r†2 ] [a~k2 . or (6. ~ q3 . (6. p~b iOUT OUT hk1 . a~r†1 ] [a~k2 . k2 |~ . a~r†1 ] [aq~2 . aq†~1 ].3. a~r†1 ] [a~k2 . a~r†2 ] [aq~2 .3.22) [a~k1 . or (6. using the 3-momentum delta functions. Now. aq†~1 ]. (6.now yields four distinct non-zero terms. or (6. ~ q2 .3.25) Now. ap†~a ] [a~r3 . a~r†2 ] [a~k2 .26) The two delta functions can be combined into δ(4) (k1 + k2 − pa − pb ). we get: (2π)3 2E~r1 δ(3) (~r1 − ~k1 ) (2π)3 2E~r2 δ(3) (~r2 − ~k2 ) (2π)3 2Eq~2 δ(3) (~q2 − p~a ) (2π)3 2Eq~3 δ(3) (~ q3 − p~b ) (2π)3 2E~r3 δ(3) (~r3 − ~q1 ).17):  µ −i 2 2 i Ef − EQ~ ! 1 (2π)4 δ(Ef − Ei )δ(3) (~k1 + ~k2 − p~a − p~b ). (6. aq†~1 ]. This gives a factor of 4. and ~r3 integrations trivially.24) are exactly the same. a~r†1 ] [aq~2 . aq†~1 ].3. so (replacing Ef → Ei in the denominator.3.21) [a~k1 . a~r†2 ] [aq~2 .

.

.

However. The ~ = ~pa + p~b = ~k1 + ~k2 . represented by the lines on the right. and represented by the vertex on the left in the figure) destroys the two particles and creates a virtual particle in their place. and the other Hint contains aaa operators. denoted by the lines on the left. (6. so momentum three-momentum carried by the intermediate virtual particle is Q is conserved at the two vertices. (6. Acting with the first factor of Hint (on the right in the formula.27) (Ei − EQ~ )(2EQ~ ) One can draw a simple picture illustrating what has happened in the preceding formulas: a† aa a† a† a initial state final state The initial state contains two particles.16) contains a† a† a† operators. (a† a† a)(a† aa) part = (−iµ)2 i (2π)4 δ(4) (k1 + k2 − pa − pb ). The second factor of Hint destroys the virtual particle and creates the two final state particles.3. Another one occurs if the Hint on the right in eq. there are other contributions that must be included. The corresponding picture is this: 72 .3.

a† a† a† initial state final state aaa Here the Hint carrying a† a† a† (the rightmost one in the formula) is represented by the upper left vertex in the figure. k2 |~   . p~b iOUT OUT hk1 . and the one carrying aaa is represented by the lower right vertex. The explicit formula corresponding to this picture is: ~ ~ pa .

.

.

3. (6.3. ap†~a ] [a~r2 . times the 6 ways of contracting ap†~a and ap†~b with any two of a~r1 . non-zero contributions arise only when each a is contracted with some a† .2. Finally. so we can just calculate one of them and multiply the answer by 36. aq†~2 .30) This leads to factors of (2π)3 2E and momentum delta functions just as before. we can do the ~q3 integral using one of the delta functions ~ with Q ~ the same as was already present in eq.31) . resulting in ~q3 = −~ pa − p~b = −~k1 − ~k2 = −Q. since the initial state particle and the final state particle would be exactly the same. aq†~2 ] [a~r1 .2 and ~r1. That would correspond to a situation with no scattering.3.28) As before. (6. (6.28). aq†~3 ]. In doing so. and aq†~3 . corresponding to the 6 ways of contracting a~k1 and a~k2 with any two of aq†~1 .18). all 36 of these contributions are identical under relabeling of momentum. However.3. a~r2 . (6. (aaa)(a† a† a† ) part = Z Z 2 µ −i (2π)3 2πδ(Ef − Ei ) d˜ r2 r1 d˜ 6 δ(3) (~r1 + ~r2 + ~r3 ) δ(3) (~q1 + ~q2 + ~q3 ) h0|a~k1 a~k2 a~r1 a~r2 a~r3 i Ef − EX ! Z d˜ r3 Z d˜ q1 Z d˜ q2 Z d˜ q3 aq†~1 aq†~2 aq†~3 ap†~a ap†~b |0i. In the end. (6.3 integrals using the delta functions. ap†~b ] [a~r3 . The result contains 36 distinct contributions. we should ignore any terms that arise whenever a final state state a~k is contracted with an initial state ap†~ . a~r3 .3. in the process setting ~q1 = ~k1 and ~ q2 = ~k2 and ~r1 = ~pa and ~r2 = p~b and ~r3 = ~q3 . This will neatly convert the factor of (−iµ/6)2 to (−iµ)2 .29) For example consider the term obtained from the following contractions of a’s and a† ’s: [a~k1 . 73 (6. So we can do the 3momentum ~q1. This allows us to identify in this case: EX = E~k1 + E~k2 + Ep~a + Ep~b + EQ~ = 2Ei + EQ~ . we can calculate this by commuting a’s to the right and a† ’s to the left. aq†~1 ] [a~k2 .3. defined in eq. namely: EX = Eq~1 + Eq~2 + Eq~3 + Ep~a + Ep~b . We also note that EX must be replaced by the free Hamiltonian energy eigenvalue of the state aq†~1 aq†~2 aq†~3 ap†~a ap†~b |0i.

k2 |~ .The end result is: ~ ~ pa . p~b iOUT OUT hk1 .

.

.

the second one just has a line that moves backwards. One hint that this is a good idea is the fact that the two cartoon figures we have drawn for them are topologically the same. we get: .3. and the final state is on the right. So if we just ignore the distinction between internal lines that move backwards and those that move forwards.32) (Ei + EQ~ )(2EQ~ ) = (−iµ)2 It is now profitable to combine the two contributions we have found. (6. The result of combining these two contributions is called the s-channel contribution. and the flow of 4-momentum is indicated by the arrows. we can draw a single Feynman diagram to represent both results combined: pa k1 pa + pb pb k2 The initial state is on the left. to distinguish it from still more contributions that we will get to soon. with 4-momentum conserved at each vertex. Using a common denominator for (Ei − EQ~ ) and (Ei + EQ~ ). (aaa)(a† a† a† ) part −i (2π)4 δ(4) (k1 + k2 − pa − pb ).

~ ~ pa . p~b iOUT .

.

(pa + pb )2 = Ei2 − |Q| ~ Q (6. k2 |~ s−channel = (−iµ)2 i (2π)4 δ(4) (k1 + k2 − pa − pb ). p~b iOUT OUT hk1 .] Ei2 i i .3. OUT hk1 .35) ~ So we can rewrite the term [Note that pµa + pµb is not equal to (EQ~ .34) (6.36) Q The final result is that the s-channel contribution to the matrix element is: ~ ~ pa . ~ 2 = Ei2 − E 2 + m2 . = 2 (pa + pb )2 − m2 − E~ (6. Q).33) Q If we now consider the four-vector q pa |2 + m2 + pµa + pµb = ( |~ then we recognize that q ~ |~ pb |2 + m2 . Q). p~a + p~b ) = (Ei .3.3.3. k2 |~ . Ei2 − E 2~ (6.

.

.

37) Note that one could just as well have put (k1 + k2 )2 in place of (pa + pb )2 in this expression. because of the delta function. and the leftmost Hint containing a† aa terms. One can draw Feynman diagrams that represent these terms. which look like: 74 . s−channel = (−iµ)2 i (2π)4 δ(4) (k1 + k2 − pa − pb ). Now one can go through the same whole process with contributions that come from the rightmost Hint (acting first on the initial state) consisting of a† a† a terms. (pa + pb )2 − m2 (6.3.

p~b iOUT OUT hk1 . Here we have combined all topologically-identical diagrams. p~b iOUT OUT hk1 .27) and (6. k2 |~ and ~ ~ pa .3. [The solid-line diagrams appearing between eqs. the contributions of the t-channel and u-channel Feynman diagrams can be found to be simply: ~ ~ pa . (6.3.] After much juggling of factors of (2π)3 and doing 3-momentum integrals using delta functions.pa pa k1 k1 pa − k2 pa − k1 pb pb k2 k2 These are referred to as the t-channel and u-channel contributions respectively.28) above are sometimes known as “oldfashioned Feynman diagrams”. but it is very rare to see them in the modern literature. k2 |~ . This is a standard procedure that is always followed. but using no new concepts. the diagrams we have drawn with dashed lines for the scalar field are the Feynman diagrams for the process.

.

.

t−channel .

.

.

(pa − k2 )2 − m2 (6.42) Mu (6. u−channel = (−iµ)2 i (2π)4 δ(4) (k1 + k2 − pa − pb ). suitable for plugging into the formula for the cross-section. (6.38) = (−iµ)2 i (2π)4 δ(4) (k1 + k2 − pa − pb ). is: Mφφ→φφ = Ms + Mt + Mu (6. So the total reduced matrix element. and u is because of the standard kinematic variables for 2→2 scattering known as Mandelstam variables: s = (pa + pb )2 = (k1 + k2 )2 . (pa − k1 )2 − m2 (6. 75 (6.2.3.3.3.46) .3.44) t = (pa − k1 )2 = (k2 − pb )2 . t. (6.3. = (−iµ)2 (pa − k1 )2 − m2 i .3.39) The reduced matrix element can now be obtained by just stripping off the factors of (2π)4 δ(4) (k1 + k2 − pa − pb ).3.3.45) u = (pa − k2 )2 = (k1 − pb )2 .40) where: i . as demanded by the definition eq.43) The reason for the terminology s. = (−iµ)2 (pa − k2 )2 − m2 Ms = (−iµ)2 (6. (6. (pa + pb )2 − m2 i .3.12).41) Mt (6.

the s. Evidently. but with more junk in the numerators coming from the appropriate reduced matrix elements. we should put a factor of i/(p2 − m2 + iǫ): ←→ p p2 i − m2 + iǫ This factor associated with internal scalar field lines is called the Feynman propagator.47) Mt (6.49) Typically. For the examples we have done with y = λ and y = µ. At each vertex.1) with coupling y. Then: • For each vertex appearing in a diagram.4 Feynman rules It is now possible to abstract what we have found. this turns out to be necessary for cases in which p2 become very close to m2 .4.(You will work out some of the properties of these kinematic variables for homework. one can draw a vertex at which n lines meet. with the understanding that ǫ → 0 at the end of the calculation. t-.and u. the Feynman rules are just: ←→ −iµ ←→ −iλ Note that the conventional factor of 1/n! in the Lagrangian eq. This 76 .48) Mu (6.3. 6.4.channel diagrams will have a similar form. we should put a factor of −iy. if one instead scatters fermions or vector particles or some combination of them. n! (6. s − m2 i . Here we have added an imaginary infinitesimal term iǫ. (6. with external lines corresponding to each initial state or final state particle.) The s-. and u-channel diagrams are simple functions of the corresponding Mandelstam variables: i .3. • For each internal scalar field line carrying 4-momentum pµ . to obtain the general Feynman rules for calculating reduced matrix elements in a scalar field theory. = (−iµ)2 t − m2 i . For each term in the interaction Lagrangian Lint = − y n φ .1) makes the corresponding Feynman rule simple in each case.3. the reduced matrix element M is the sum of contributions from each topologically distinct Feynman diagram. 4-momentum must be conserved. with m always the mass of the particle on the internal line.t. = (−iµ)2 u − m2 Ms = (−iµ)2 (6.

Typically. one needs the following additional rules: • For each closed loop in a Feynman diagram. where N is the number of distinct permutations of that type. (This is known as the “symmetry factor” for the loop diagram. In the φ3 theory. (2π)4 (6. (Here the gray blobs represent the rest of the Feynman diagram. then there is an additional factor of 1/N . without moving the external lines.) Some examples might be useful. One of them is shown below: k1 pa k1 + k2 pa + pb k2 pb k3 77 .) Those are all the rules one needs to calculate reduced matrix elements for Feynman diagrams without closed loops. because p2 = m2 is the equation satisfied by a free particle in empty space. there is an undetermined 4-momentum ℓµ . • For each external line. but it is useful to mention it because in the cases of fermions and vector fields. There are also Feynman rules that apply to diagrams with closed loops (loop diagrams) which have not arisen explicitly in the preceding discussion. or by other slimy tricks. because of the contribution from very large |ℓ2 |. The result is said to be a tree-level calculation.corresponds to the particle on the internal line being nearly “on-shell”.4. there are quite a few Feynman diagrams that will describe the scattering of 2 particles to 3 particles. • If a Feynman diagram with one or more closed loops can be transformed into an exact copy of itself by interchanging any number of internal lines through a smooth deformation. we just have a factor of 1: or ←→ 1 This is a rather trivial rule. The techniques of getting physically meaningful answers out of this are known as regularization (making the integrals finite) and renormalization (redefining coupling constants and masses so that the physical observables don’t depend explicitly on the unknown cutoff). the iǫ only makes a difference for propagators involved in closed loops (discussed below). but could be inferred from more complicated calculations. These loop momenta should be integrated over according to: Z d4 ℓ . also known as tree diagrams. which can make the integrals finite. external lines will turn out to carry non-trivial factors not equal to 1. This can be fixed by introducing a cutoff |ℓ2 |max in the integral. For them.2) Loop diagrams quite often diverge because of the integration over all ℓµ .

They might include real or complex scalars. without moving the external lines. according to the rules. The diagrams are built out of three types of building blocks: vertices ←→ interactions internal lines ←→ free virtual particle propagation external lines ←→ initial. which defines a particular theory. They follow from the Lagrangian density. The Feynman rules specify a mathematical expression for each of these objects. 78 . The index i runs over a list of all the fields. Dirac or Weyl fermions. and tremble with fear! Feynman rules are good.4) Again. An example of a Feynman diagram with a closed loop in the φ3 theory is: pa k1 ℓ pa + pb pb pa + pb ℓ − pa − pb k2 There is a symmetry factor of 1/2 for this diagram. and vector fields of various types. So the reduced matrix element for this diagram is: M=  1 i (−iµ)4 2 (pa + pb )2 − m2 2 Z  d4 ℓ i 4 (2π) (ℓ − pa − pb )2 − m2 + iǫ   i . final states. and over their spinor or vector indices.† To generalize what we have found for scalar fields.4. but extraordinarily unpleasant! In the future. this would only be sufficient to describe phenomena that occur in a perturbative weak-coupling expansion. it is always possible to obtain the Feynman rules by writing an interaction Hamiltonian and computing matrix elements. deriving this result starting from the creation and annihilation operators is possible. let us consider a set of generic fields Φi . the contribution to the reduced matrix element is just 3 M = (−iµ)  i (pa + pb )2 − m2   i . because one can smoothly interchange the two lines carrying 4-momenta ℓµ and ℓµ − pµa − pµb to get back to the original diagram. Alternatively. (k1 + k2 )2 − m2 (6. The general procedure for doing this is rather simple (although the proof is not). which can include both commuting bosons and anticommuting fermions.4. Now. A Feynman diagram is a precise representation of a contribution to the reduced matrix element M for a given physical process.For this diagram. However. and is outlined below. 2 ℓ − m2 + iǫ (6. one can use powerful path integral techniques that are beyond the scope of this course to derive the Feynman † It is tempting to suggest that the Feynman rules themselves should be taken as the definition of the theory. we will simply guess the Feynman rules for any theory from staring at the Lagrangian density.3) Imagine having to calculate this starting from scratch with creation and annihilation operators.

and an arrow coming out representing Ψ. ΦiN . However. . . .6) then there are Feynman rules: ←→ −iλ1 ←→ −iλ2 Here the longer-dashed lines correspond to the field φ. . .. say.4. . The lines for distinguishable fields among i1 .5) where P is the product of n! for each set of n identical fields in the list Φi1 . and Xi1 i2 . . more rigorous derivations can be found in field theory textbooks. In these notes.. (6. Then the mathematical expression assigned to this vertex is −iXi1 i2 .iN . Lint = − λ1 2 2 λ2 3 φ ρ − φ ρ. The corresponding Feynman rule attaches N lines together at a vertex. For Dirac fermions. consider a theory in which a real scalar field φ couples to a Dirac fermion Ψ according to: Lint = −yφΨΨ.rules.. consider a theory with two real scalar fields φ and ρ. .. i2 .4. .iN is the coupling constant that determines the strength of the interaction.iN Φi1 Φi2 . and the shorter-dashed lines to the field ρ. Φi2 . For interactions.. More generally. 4 6 (6. we must distinguish between lines for all three fields.7) In this case. For example. one draws solid lines with an arrow coming in to a vertex representing Ψ in Hint . in the end the rules can be summarized very simply in a way that could be guessed from the examples of real scalar field theory that we have already worked out. . As another example. we will follow the technique of guessing.. ΦiN . or otherwise distinguished by drawing them differently from each other. If the interaction Lagrangian includes terms.4. P (6. because Ψ = Ψ† γ 0 is independent of Ψ. So the Feynman rule for this interaction is: a ←→ b 79 −iyδa b . consider an interaction Lagrangian term: Lint = − Xi1 i2 . iN should be labeled as such. we have now found in two cases that the Feynman rule for n scalar lines to meet at a vertex is equal to −i times the coupling of n scalar fields in the Lagrangian with a factor of 1/n!.

4.4.4. 2 i. Recall that for a scalar field. and writes simply −iy for the Feynman rule.4.26).4. the propagator is just proportional to i divided by the inverse of the coefficient of the quadratic piece of the Lagrangian density.7) is called a Yukawa coupling. These are determined by the free (quadratic) part of the Lagrangian density. (6. This is a because the interaction Lagrangian can be written −yδa b φΨ Ψb . 80 (6. Let us turn next to the topic of internal lines in Feynman diagrams. with the identity matrix implicit.13) .1.4. with the replacement ∂µ −→ −ipµ . This theory has a real-world physical application: it is precisely the type of interaction that applies between the Standard Model Higgs boson φ = h and each Dirac fermion Ψ.12) This corresponds to an internal line in the Feynman diagram labeled by i at one end and j at the other.11) and (6.4.10) L0 = X (6.9) The free Lagrangian density for generic fields Φi can always be put into either the form L0 = 1X Φi Pij Φj . To accomplish this. Often. the Feynman propagator connecting vertices with spinor indices a and b should be: b i[(/p − m)−1 ]a . throwing away a total derivative in L0 which will not contribute to S = R d4 x L. Here Pij is a matrix that involves spacetime derivatives and masses. (6. one may need to integrate the action by parts.j for complex fields (including.Note that this Feynman rule is proportional to a 4 × 4 identity matrix in Dirac spinor space. or the form (Φ† )i Pij Φj . (6. consider the free Lagrangian for a Dirac spinor Ψ. (6. We will return to this interaction when we discuss the decays of the Higgs boson into fermionantifermion pairs. up to the iǫ factor.4. (5.8) This corresponded to a Feynman propagator rule for internal scalar lines i/(p2 − m2 + iǫ). one just suppresses the spinor indices. The interaction Lagrangian eq. for example. with the coupling y proportional to the mass of that fermion. As an example. we can write the free Lagrangian after integrating by parts as: 1 L0 = φ(−∂µ ∂ µ − m2 )φ.11) for real fields. where a is the Dirac spinor index for Ψ and b for Ψ. According to the prescription of eqs. Then it turns out that the Feynman propagator can be found by making the replacement eq. as given by eq.12).4. (6. i.9) and taking i times the inverse of the matrix Pij : i(P −1 )ij . Dirac spinors). 2 (6. So.4.j (6.

the 4momentum pµ appearing in the propagator is also assigned to be in the direction of the arrow on the internal fermion line. (3.m on the left. the coefficient for a final state antiparticle (positron) is v(p.4. (6. (5. this means that the arrow on the propagator points in the direction of the flow of negative charge. n) are expansion coefficients. Hint will therefore k. we see from eq. As indicated.14) In the last line we have put in the iǫ factor needed for loop diagrams as a prescription for handling the possible singularity at p2 = m2 .m produce a factor of i(~k. n) ei~p·~x ap~. where a is a spinor index. when acting on a final state h0|b~k. the interaction Hamiltonian will produce a factor of f (~k.4. n) e−i~p·~x bp†~. initial and final state lines just correspond k. Similarly. p (p p − m2 + iǫ /−m / − m)(p / + m) (6. we find that i(~ p. m) after commuting (or anticommuting. then multiply numerator and denominator by (p / + m).n .4. s)a .15) where ap~. (6. m).n operator in Φ to the right.n are annihilation and creation operators (which may or may not be Hermitian conjugates of each other). we build an interaction Hamiltonian out of the fields Φ. At a fixed time t = 0.n and bp†~. When acting on an initial state a~† |0i on the right.2. Therefore. For example. So the Feynman rules for these types of external particle lines are: initial state electron: a p→ ←→ 81 u(p. with particles (anti-particles) moving with (against) the arrow. we can write it as a fraction.2. Likewise. n) are both just equal to 1. Next we turn to the question of Feynman rules for external particle and anti-particle lines. n) and f (~ p. (5.1) that the coefficient for an initial state particle (electron) carrying 4-momentum pµ and spin state s is u(p.m to the appropriate coefficient of annihilation and creation operators in the Fourier mode expansion for that field. n is an index running over spins and perhaps other labels for different particle types.3. s)a . s)a . So the Feynman rule for a Dirac fermion internal line is: b p→ a ←→ b b i([p /]a + mδa ) p2 − m2 + iǫ Here the arrow direction on the fermion line distinguishes the direction of particle flow. For Dirac fermions. and i(~ p.30) in the scalar case. a generic field Φ is written as an expansion of the form: Φ(~x) = XZ n h i d˜ p i(~ p. comparing eq. n) and f (~ p.In order to make sense of the inverse matrix.n + f (~ p. For electrons and positrons. for fermions) the ap~. removing the a~† . In general.15) to eq. and use the fact that p /p / = p2 from eq.54): i(p i(p i / + m) / + m) = = 2 .

the pµ label ofan external state is always the physical 4-momentum of that particle or anti-particle. s)a Note that in these rules.2). So the Feynman rules for these external states are: initial state positron: a p→ p→ final state electron: a ←→ v(p. s)a . this is in the opposite direction to the arrow on the fermion line itself. For v(p. Similarly. s). s)a and that for a final state particle (electron) is u(p. the pµ associated with each of u(p.p→ final state positron: a ←→ v(p. s). s)a ←→ u(p. u(p. s) and v(p. we see that the coefficient for an initial state antiparticle (positron) is v(p. s).3. s) and v(p. (5. s)a Here the blobs represent the rest of the Feynman diagram in each case. 82 . s) is always taken to be pointing to the right. this means that with the standard convention of initial state on the left and final state on the right. considering the expansion of the field Ψ in eq. v(p.

r + v(p. the appropriate value is a little larger.) The free Lagrangian for the theory is: 1 L0 = − F µν Fµν + Ψ(iγ µ ∂µ − m)Ψ. 2Ep~ 83 (7.39)]. and −1/3 for down.1 QED Lagrangian and Feynman rules Let us now see how all of these general rules apply in the case of Quantum Electrodynamics.2) The interaction Lagrangian density for a fermion with charge Qe and electromagnetic fields is therefore: Lint = −eQΨγ µ ΨAµ . it is a running coupling constant. we can guess that: J µ = QΨγ µ Ψ. strange and bottom quarks and their antiquarks.313. (7.3) The value of e is determined by experiment. which means that its value has a logarithmic dependence on the characteristic energy of the process. (7.s + v † (p.1.1. and one of the momentum integrations using the resulting delta function. However.1. where e is the magnitude of the charge of the electron. Consider the total charge operator: b= Q Z d3 ~x ρ(~x) = Z d3 ~x J 0 (~x) = Q Z d3 ~x Ψγ 0 Ψ = Q Z d3 ~x Ψ† Ψ. 4π (7. Let us take a small detour to check that eq.7 Quantum Electro-Dynamics (QED) 7.3.5) Plugging in eqs. (5. one finds: b = Q Q 2 X 2 Z X s=1 r=1 d˜ p 1 [u† (p. r)dp†~.1. (7. Since J µ must be a four-vector built out of the charged fermion fields Ψ and Ψ. s)dp~. +2/3 for up.1. 4 (7. The fermions in the theory are represented by Dirac spinor fields Ψ carrying electric charge Qe. assigned a single value of Q. r)bp~.2).6) . This is the quantum field theory governing photons (quantized electromagnetic waves) and charged fermions and antifermions. s)bp†~. corresponding to the experimental result for the fine structure constant: α≡ e2 ≈ 1/137. the numerical value is e ≈ 0.1. J~ ) by a term in the Lagrangian −eJ µ Aµ [see eq. For very low energy experiments. more like e ≈ 0.036. charm and top quarks and their anti-quarks. in section 4 we found that the electromagnetic field Aµ couples to the 4-current density J µ = (ρ.1. Thus Q = −1 for electrons and positrons.3028. (7.s ] [u(p. (5.2) really has the correct form and normalization to be the electromagnetic current density.3.1) Now.1) and (5. and doing the ~x integration.r ]. is used to describe both particles and their anti-particles. (Recall that a single Dirac field.1.4) For experiments done at energies near 100 GeV.

3. (7. and (7. (7. (7.20). much like we had to do in getting to eqs.1.s bp~. this verifies that the charge density ρ is indeed the time-like component of the four-vector QΨγ µ Ψ. (5. r) = 0.1. 84 (7. (7.10) k.7) u† (p.r k.1. s)v(p. ~ ~ (7.s − dp†~. The full QED Lagrangian is invariant under gauge transformations: 1 Aµ (x) → Aµ (x) − ∂µ θ(x). From eq. [Q.1. (7. s)u(p.13) b d† |0i = −Q d† |0i Q ~ ~ (7. b b† ] = Q b† .17) where θ(x) is an arbitrary function of spacetime [equal to −eλ(x) in eq.16)].r for single-particle states. Ψ(x) → e−iQθ Ψ(x). the result is: b=Q Q 2 Z X s=1 h i d˜ p bp†~.1. The result is that the total charge eigenvalue of the vacuum vanishes: b Q|0i = 0. r) = v † (p.11) b b† |0i = Q b† |0i Q ~ ~ (7. One can now check that Therefore.16) (7.r −Q d~† . ~k. More generally.1.21).14) k.s dp~.1. d† satisfy the anticommutation relation eqs.42) and eqs. k.r (7. s)v(p.r † b d ] [Q.12) b acting on a state with N particles for single anti-particle states.5).r = k. which must therefore be equal to J µ .1.1.2. e iQθ Ψ(x) → e Ψ(x).1.3. This represents a uniform and constant (and therefore unobservable) infinite charge density throughout all space.This can be simplified using some spinor identities: u† (p. (4.15) (7.18) . the eigenvalue of Q and N antiparticles is (N − N )Q.r k.1. (5.9) Here we have dropped an infinite contribution.8) Taking into account that the operators d. r) = v † (p. (5.r k.s . r) = 2Ep~ δsr . A nice way to see this invariance is to define the covariant derivative: Dµ Ψ ≡ (∂µ + iQeAµ )Ψ. s)u(p.1.1.

1. Note that the covariant derivative actually depends on the charge of the field it acts on. This means that 85 . From classical electromagnetism.1. To find the Feynman rules for initial and final state photons. λ)e−i~p·~x ap†~.20) The point of the covariant derivative is that it transforms under the gauge transformation the same way Ψ does. Now one can write the full Lagrangian density as 1 L = L0 + Lint = − F µν Fµν + iΨγ µ Dµ Ψ − mΨΨ. not the Lorentz transformation symmetry as it did when we introduced covariant four-vectors.λ + ǫ∗µ (p.21) Here the contribution from the transformation of Aµ in Dµ cancels the extra term in eq. 3. by acquiring a phase: Dµ Ψ → eiQθ Dµ Ψ.λ and ap~. (7.λ . we can now identify the Feynman rule for QED interactions. (6.1. Returning to the interaction term eq. consider the Fourier expansion of the vector field at a fixed time: Aµ = 3 Z X λ=0 i h d˜ p ǫµ (p. (7. (7. 1. Using eqs. by following the general prescription outlined with eq. we know that electromagnetic waves are transversely polarized.19) The ordinary derivative of the spinor transforms under the gauge transformation with an “extra” term: ∂µ Ψ → ∂µ (eiQθ Ψ) = eiQθ ∂µ Ψ + iQΨ∂µ θ.λ act on the vacuum state by creating and destroying photons with momentum p~ and polarization vector ǫµ (p.1.Here the term “covariant” refers to the gauge transformation symmetry. (7.23) The operators ap†~.20).1. not all of the four degrees of freedom labeled by λ can be physical. (7. λ)ei~p·~x ap~.4.15)-(7.1. 2. 4 (7. (7. λ).3). it is easy to see that L is invariant under the gauge transformation.1. λ)ǫ∗µ (p. However. λ′ ) =    −1 +1   0 for λ = λ′ = 1.1.17) and (7.21).22) Here ǫµ (p. since the multiplicative phase factors just cancel. λ) is a basis of four polarization four-vectors labeled by λ = 0. or 3 for λ = λ′ = 0 for λ 6= λ′ (7.1. 2. They satisfy the orthonormality condition: ǫ µ (p.1.5): a µ ←→ −iQe(γ µ )a b b This one interaction vertex governs all physical processes in QED.

1. 1. not all of the physical states we are attempting to propagate are really physical.1. due to Fermi. 0. (6. After imposing these two conditions. 0) y-polarized. it appears that we ought to find the propagator by finding the inverse of the 4 × 4 matrix Pµν = −p2 gµν + pµ pν . only two of the four λ’s will survive as valid initial or final states for any given pµ .and right-handed circular polarizations that carry definite helicities λ = R. −i.~ǫ). obtained by integrating L0 . Therefore.1. L: ǫµ (p. suppose that a state contains a photon with 3-momentum ~p = pˆ z . (7. 2 in some convenient basis of choice. ~ǫ · p~ = 0). we will sum or average over the polarization labels λ.34): can be rewritten as: 1 L0 = Aµ (gµν ∂ρ ∂ ρ − ∂µ ∂ν ) Aν .1. 0. equivalent to the last condition.1. this matrix is not invertible. (The action. Then we can choose a basis of transverse linearly-polarized vectors with λ = 1. 2: ǫµ (p.28) left-handed. λ). however. does not depend on total derivative terms.4. p). For example. The reason for this can be traced to the gauge invariance of the theory.30) where we have dropped a total derivative.31) Unfortunately. 2) = (0.the electric and magnetic fields are perpendicular to the 3-momentum direction of propagation.29) In general.26) ǫµ (p. 2 (7.1. 1. 0. Often.27) However. following the prescription of eq.12). λ) and outgoing photon lines have a Feynman rule ǫ∗µ (p. (7. L) = 1 √ (0. (7. (7. we can add a term to the Lagrangian density proportional to (∂µ Aµ )2 : (ξ) L0 = L0 − 1 (∂µ Aµ )2 . 0) x-polarized. 0. incoming photon lines have a Feynman rule ǫµ (p. 1) = (0. In terms of the potentials. The free Lagrangian density given in eq. 1. called “gauge fixing”. This problem can be avoided using a trick. 2ξ 86 (7. so the ǫµ (p. 1. R) = ǫµ (p. As long as we agree to stick to Lorenz gauge. (7.) Therefore.1.24) p µ ǫµ = 0 (7. i. (7.25) (or. 0) 2 right-handed.32) . physical electromagnetic wave quanta corresponding to the classical solutions to Maxwell’s equations Aµ = ǫµ e−ip·x with p2 = 0 can be taken to obey: ǫµ = (0. (5. where λ = 1.1. it means that one can always choose a gauge in which A0 = 0 and the Lorenz gauge condition ∂µ Aµ = 0 is satisfied.1. 0) 2 1 √ (0. ∂µ Aµ = 0. in high-energy physics it is usually more useful to instead use a basis of left. λ) will not need to be listed explicitly for a given momentum. Let us next construct the Feynman propagator for photon lines.1. so pµ = (p.

34) Pµν (P −1 )νρ = δµρ . + 1− ξ (7.In Lorenz gauge.1. providing a consistency check.38) Another common choice is ξ = 0. not only does the extra term vanish.32).1. this propagator corresponds to an internal wavy line. labeled by µ and ν at opposite ends.) In a Feynman diagram. We have now encountered most of the Feynman rules for QED. (7.1. one notices that as a tensor. (7.36) It follows that the desired Feynman propagator for a photon with momentum pµ is:  i pµ pν −gµν + (1 − ξ) 2 2 p + iǫ p  (7. Now the matrix to be inverted is: 2 Pµν = −p gµν   1 pµ pν . The new term in the modified free Lagrangian L0 is called the gauge-fixing term.1. Even though this means having to calculate more terms. these can be understood by carefully 87 . for which the Feynman propagator is:  pµ pν −i gµν − 2 p2 + iǫ p  (Landau gauge). it can be picked at will. (p2 )2 (7. The most popular choice for simple calculations is ξ = 1. Then the Feynman propagator for photons is simply: −igµν p2 + iǫ (Feynman gauge). p2 C2 = 1−ξ .35) and requiring that one finds the solution C1 = − 1 . the payoff is that in the end one can demand that the final answer for the reduced matrix element is independent of ξ.1. but also its contribution to the equations of motion vanishes. So.1.33) To find the inverse. (7.37) (Here we have put in the iǫ factor in the denominator as usual. called Feynman gauge.1. Sometimes it is useful to just leave ξ unspecified. The gauge-fixing parameter ξ can be chosen at the convenience or whim of the person computing the Feynman diagram. (P −1 )νρ can only be a linear combination of terms proportional to gνρ and to pν pρ . which can make some calculations simpler (especially certain loop diagram calculations). known as Landau gauge.1. There are some additional rules having to do with minus signs because of Fermi-Dirac statistics.] The Landau gauge photon propagator has the nice property that it vanishes when contracted with either pµ or pν .39) [Comparing to eq. Here ξ is an arbitrary new gauge-fixing parameter. and carrying 4-momentum p. we see that this is really obtained as a formal limit ξ → 0. (7. making a guess: (P −1 )νρ = C1 gνρ + C2 pν pρ (7. Physical (ξ) results should not depend on the choice of ξ.

considering the effects of anticommutation relations for fermionic operators. In practice. Of course. one usually does not write out the spinor indices explicitly. That will be the subject of the next few subsections. 88 . All the rules are summarized in the following two pages in a cookbook form. the best way to understand how the rules work is to do some examples.

write: 89 . and interaction vertices.) 4. For external lines. s) final-state fermion: ←→ u(p. s) initial-state antifermion: ←→ v(p. and solid lines with arrows representing fermions. (Use ξ = 1 for Feynman gauge and ξ = 0 for Landau gauge. and spin polarization s or λ as appropriate): initial-state fermion: ←→ u(p. Draw all topologically distinct Feynman diagrams. write Feynman propagators: ←→ i(p / + m) − m2 + iǫ p2 with 4-momentum pµ along the arrow direction. λ) ←→ ǫ∗µ (p.Feynman rules for QED To find the contributions to the reduced matrix element M for a physical process involving charged Dirac fermions and photons: 1. For internal photon lines. The arrow direction is preserved when following each fermion line. s) ←→ ǫµ (p. using the rules below for external lines. with wavy lines representing photons. write (with 4-momentum pµ always to the right. write: µ ν ←→  pµ pν i −gµν + (1 − ξ) 2 2 p + iǫ p  with 4-momentum pµ along either direction in the wavy line. λ) initial-state photon: µ µ final-state photon: 3. For internal fermion lines. internal lines. 2. and m the mass of the fermion. s) final-state antifermion: ←→ v(p. Enforce fourmomentum conservation at each vertex. For interaction vertices.

After writing down the contributions from each diagram to the reduced matrix element M ac- cording to the preceding rules. following the line backwards until a u or v spinor is reached. or on an internal photon line propagator index. and follow the fermion line backwards until the original point is reached. For closed fermion loops.µ ←→ −iQeγ µ The vector index µ is to be contracted with the corresponding index on the photon line to which it is connected. v. 5. then there is an additional symmetry factor of 1/N . perform an integration Z d4 ℓ . assign an additional relative minus sign between different diagram contributions whenever the written ordering of external state spinor wavefunctions u. If a Feynman diagram with one or more closed loops can be transformed into an exact copy of itself by interchanging any number of internal lines through a smooth deformation without moving the external lines. v differs by an odd permutation. For fermion lines that go all the way through the diagram. To take into account suppressed spinor indices on fermion lines. take a trace over the gamma matrices in the closed loop. u. 7. write terms involving spinors as follows. (2π)4 Getting a finite answer from these loop integrations often requires that they be regularized by introducing a cutoff or some other trick. 8. 90 . This will be either an external photon line factor of ǫµ or ǫ∗µ . Put a factor of (−1) for each closed fermion loop. start at the end of each fermion line (as defined by the arrow direction) with a factor u or v. 9. 6. start at an arbitrary vertex on the loop. For each loop momentum ℓµ that is undetermined by four-momentum conservation with fixed external-state momenta. and write down factors of γ µ or (p / + m) consecutively. where N is the number of distinct permutations of that type.

These calculations will involve several thematic tricks that are common to many Feynman diagram evaluations. By contrast.2) The v(pb . A good first step is to label the momentum and spin data for the initial state electron and positron and the final state muon and anti-muon: Particle Momentum Spin e− pa sa e+ pb sb − µ k1 s1 µ+ k2 s2 Spinor u(pa .2 e− e+ → µ− µ+ In the next few subsections we will study some of the basic scattering processes in QED. this process requires a center-of-momentum energy of at least s = 211. s2 ) (7. sa ) part is obtained by starting at the end of the electron-positron line with the positron external state spinor. s1 ) (ieγ ν ) v(k2 .2.2.66 MeV. since the charge of the electron and muon is Q = −1. and following it back to its beginning.7.2. s1 ) (ieγ ν ) v(k2 . the u(k1 . The interaction vertex is −iQeγ µ = ieγ µ . The photon propagator is written in Feynman gauge. sb ) (ieγ µ ) u(pa . (7. and carries indices µ and ν that connect to the two fermion lines at their respective interaction vertices. sa )] (pa + pb )2 µ  [u(k1 . and following it backwards. there is only one Feynman diagram for this process.511 MeV. Here it is: pa k1 pa + pb µ ν pb k2 Applying the rules for QED to turn this picture into a formula for the reduced matrix element. s2 )] .1) At order e2 . 91 . sa ) v(pb .3 MeV. sb ) (ieγ ) u(pa . the mass of the electron is only about me = 0. We will calculate the differential and total cross-sections for this process in the center-of-momentum frame. Likewise. The error made in doing so is far less than the error made by not including higher-order corrections. which we can therefore safely neglect. Since the mass of the muon (and the anti-muon) is about √ mµ = 105. s2 ) part is obtained by starting at the end of the muon-antimuon line with the muon external state spinor. to leading order in the coupling e. sb ) u(k1 . we find:  −igµν M = [v(pb . for simplicity. using the Feynman rules found in the previous section and the general discussion of cross-sections given in section 6. s1 ) v(k2 . We begin with electron-positron annihilation into a muon-antimuon pair: e− e+ → µ− µ+ .

8) At this point.We can write this result more compactly by using abbreviations v(pb .50) and (3. Similarly.28) and (B. So.51) [also listed in Appendix B as (B. s2 |M|2 = (7. sa ) ≡ ua . .] Therefore we have: |M|2 = e4 (v b γµ ua )(ua γν vb ) (u1 γ µ v2 )(v 2 γ ν u1 ).2.) These spin sums and averages will allow us to exploit the identities (3. (xγµ γν .7) where x and y are any u and v spinors. Also. so that the cross-section can only depend on the angle θ between the collision 92 . sb ) = v b and u(pa . (7. there is only one initial state.29)]. (u1 γ ν v2 )∗ = v 2 γ ν u1 .4). . this is not very convenient. and fortunately it is not necessary either.2.2. and using the metric in the photon propagator to lower the index on one of the gamma matrices. so that the explicit forms for the spinors are never needed. In a real experiment. we should average over sa and sb . because sa and sb cannot simultaneously take on both spin-up and spin-down values. we should sum over s1 and s2 . This is because the only special directions in the problem are the momenta of the particles. which implies γν† γ 0 = γ 0 γν . not sum.8) into eq.5) The third equality follows from the identity (B.2. Writing the denominator of the photon propagator as the Mandelstam variable s = (pa + pb )2 = (k1 + k2 )2 . (7. s2 (7.2.48) to find the differential cross-section for any particular set of spins. Therefore. the final state spins of the muon and anti-muon are typically not measured. . taking care to write them in the reverse order. However. (7.4) Evaluating the complex conjugated terms in parentheses can be done systematically by taking the Hermitian conjugate of the Dirac spinors and matrices they are made of. to find the total cross-section for all possible final states.6) [Following the same strategy.2. (v b γν ua )∗ = (vb† γ 0 γν ua )∗ = u†a γν† γ 0 vb = u†a γ 0 γν vb = ua γν vb . we get: M = i e2 (v b γµ ua )(u1 γ µ v2 ). even if it is unknown. we could work out explicit forms for the external state spinors and plug eq.2. if the initial-state electron spin states are unknown. the differential cross-section must be symmetric under rotations about the collision axis. (One must average.2. γν γµ x (7. s (7. etc. (6.3) The differential cross-section involves the complex square of M: e4 (v b γµ ua )(u1 γ µ v2 )(v b γν ua )∗ (u1 γ ν v2 )∗ .2. . After doing the spin sum and average. γρ y)∗ = yγρ . for example. one can show that in general. over the initial-state spins.2.

we obtain: 1 X 1 XXX e4 X X µ k2 − mµ ]γ ν u1 ).10) v2 v 2 = k/2 − mµ . (7.2. |M|2 = (v b γµ p /a γν vb ) (u1 γ [/ 2 sa 2 sb s1 s2 4s2 sb s1 (7. X (7.2.12). (7.12) Now we apply another trick.2.28) and (B.2.29).13) b3 a4  b4 a4 Applying this to each expression in parentheses in eq. again neglecting me . is: e4 1 X µ k2 − mµ )γ ν (/ k1 + mµ )]. So.16) The reason this trick of rearranging into a trace is useful is that now we can once again exploit the spin-sum identities (B.2.2. (7.axis determined by the two initial-state particles and the scattering axis determined by the two finalstate particles. this time in the form: X sb X vb v b = /pb − me .15) so that 1 X e4 X X k2 − mµ )γ ν u1 u1 ]. we move the barred spinor (thought of as a row vector) to the end and take the trace over the resulting 4 × 4 Dirac spinor matrix.28) and (B.2. which in the present situation imply X sa ua ua = p / a + me . with the effects of initial state spin averaging and final state spin summing now included.49) to obtain: 1 X 1 XXX |~k1 | dσ = |M|2 d(cos θ) 2 sa 2 sb s1 s2 32πs|~ pa | (7. (7. s2 (7. (6.2.11) Neglecting me as promised earlier.17) u1 u1 = k/1 + mµ .14) u1 γ µ [/ k 2 − mµ ]γ ν u1 = Tr[γ µ (/ k 2 − mµ )γ ν u1 u1 ]. (B.29). So: v b γµ p /a γν vb = Tr[γµ p /a γν vb v b ]. We can now use eqs.2.2. we can apply eq. A dot product of two vectors is equal to the trace of the vectors multiplied in the opposite order to form a matrix: ( a1 a2 a3      b1 b1 b1 a1 h b  i  b2   b2 a1 2     a4 )   b3  = Tr  b3  ( a1 a2 a3 a4 ) ≡ Tr  b3 a1 b4 b4 b4 a1 b1 a2 b2 a2 b3 a2 b4 a2 b1 a3 b2 a3 b3 a3 b4 a3  b1 a4 b2 a4   .2.2. (7. Tr[γµ /pa γν vb v b ] Tr[γ µ (/ |M|2 = 2 4 spins 4s sb s1 (7. (7. |M|2 = 2 Tr[γµ p /a γν p/b ] Tr[γ (/ 4 spins 4s 93 (7.9) in the center-of-momentum frame.2.18) s1 The result.19) .

0. (B.2.2. We assume that the electron is initially moving in the +z direction.27) Then we have: pa = (P. Let us call P and K the magnitudes of the 3-momenta of the electron and the muon. 0.29) k1 = ( K 2 + m2µ .25) Our next task is to work out the kinematic quantities appearing in eq.31) q q Since (pa + pb )2 = (k1 + k2 )2 = s.2. 2 (7. and the muon makes an angle θ with respect to the positive z axis.2. making use of the fact that the trace of an odd number of gamma matrices is zero: Tr[γ µ (/ k2 − mµ )γ ν (/ k 1 + mµ )] = Tr[γ µ k/2 γ ν k/1 ] − m2µ Tr[γ µ γ ν ] = 4k2µ k1ν − 4gµν k1 · k2 + 4k1µ k2ν − 4gµν m2µ . −P ).2. (7. (7.2. K sin θ. 4 r s K = − m2µ . (7.2.25). k12 = k22 = m2µ .24) where eqs. Similarly.33) and pa · pb = 2P 2 = s . (B. (7.2.9). it follows that r s . and listed in eq.9) have been used. 0.7)-(B.2.22) where we have used the general result for the trace of four gamma matrices found in the homework.2.30) k2 = ( K 2 + m2µ . (7. one finds that the answer reduces to simply: i 1 X e4 h |M|2 = 2 8 (pa · k2 )(pb · k1 ) + (pa · k1 )(pb · k2 ) + (pa · pb )m2µ . The on-shell conditions for the particles are: p2a = p2b = m2e ≈ 0.2.2. (7.2.Next we must evaluate the traces.32) (7. −K cos θ).20) (7.2. K cos θ).2. (7. Tr[γµ p/a γν / pb ] = pαa pβb Tr[γµ γα γν γβ ] = pαa pβb (4gµα gνβ − 4gµν gαβ + 4gµβ gνα ) = 4paµ pbν − 4gµν pa · pb + 4pbµ paν .26) (7. respectively.28) pb = (P. First. 4 P = (7. Taking the product of the two traces.34) 94 . 0.23) (7. 0.21) (7. P ).2. within the yz plane. −K sin θ. 0. (7. 4 spins s (7.

going to increasing energy. σ quickly levels off because of the 1/s factors in eq. (7.2. (7.40) and (7. rewriting in terms of the fine structure constant α = e2 /4π.2. we find the total (7. recall that when c = h ¯ = 1. K 2 + m2µ − P K cos θ = 1 − cos θ 1 − 4 s 4m2µ s . (7. then s = ECM has units of mass2 or length−2 . and correctly predict the rate of production of muon-antimuon pairs at electron-positron colliders.2.41) becomes πα2 σ≈ 2m2µ s ∆E . eq.2.2. mµ (7.9): dσ d(cos θ) = = 1 X K |M|2 4 spins 32πsP e4 32πs s (7.38) " 4m2µ 4m2µ 4m2µ 1− + 1− 1+ s s s ! 2 # (7. Near the energy threshold for µ+ µ− production.40) cos θ . or. d(cos θ) = 2/3. ! cos2 θ . However.2. (7.25).2.2.2.2.pa · k1 = pb · k2 = P q pa · k2 = pb · k1 = P q  s   s  4m2µ s .41). Maximizing with 95 .2.2. dσ d(cos θ) = πα2 2s Doing the integral over cos θ using cross-section: s " 4m2µ 4m2µ 4m2µ + 1− 1+ 1− s s s R1 −1 d(cos θ) 4πα2 σ= 3s = 2 and s 4m2µ 1− s R1 −1 cos 2m2µ 1+ s ! 2θ .39) # (7. we get: 1 X |M|2 = 4 spins 8e4 s2 4 = e "  2  s 4 s 4m2µ 1 + cos θ 1 − s ! 4m2µ 4m2µ + 1− 1+ s s s 2  2  4m2µ 1 − cos θ 1 − s s + 4 2 # cos θ . 2 sm2µ + 2  (7. Let us examine some special limiting cases.43) The cross-section therefore rises like the square root of the energy excess over the threshold.2.41) have been tested in many experiments.42) To leading order in small ∆E.35) (7. Equations (7. K 2 + m2µ + P K cos θ = 1 + cos θ 1 − 4 s (7.41) 2 It is a useful check that the cross-section has units of area.2. one may expand in the quantity ∆E = √ s − 2mµ . (7.36) Putting these results into eq.37) Finally we can plug this into eq.2.

q √ √ respect to s. if we call nf the number of colors (or perhaps other non-spin degrees of freedom) of the fermion f .41) is reached for s = 1 + 21 mµ ≈ 2.2) where it is understood that mµ should be replaced by mf everywhere in |Me− e+ →f f |2 . and is about σmax = 0.2. In general.3. the cross-section decreases proportional to 1/s: σ≈ 4πα2 . top).36mµ .2. we calculated the cross-section for producing a muon-antimuon pair in e− e+ collisions. because there is another diagram in which the photon is replaced by a Z 0 boson.3.2. 3s (7. (7. there are three indistinguishable colors for each flavor (up. It follows that the differential and total cross-sections for e− e+ → f f are obtained from eqs.3 e− e+ → f f. down. one finds that the largest cross-section in eq. (7. (7.3. charm. In the case of quarks. In the high energy limit √ (7. The photon-quark-antiquark vertex is diagonal in color. bottom. then we have: |Me− e+ →f f |2 = nf Q2f |Me− e+ →µ− µ+ |2 . (7. 7. The Feynman diagram for this process is obtained by simply replacing the muon-antimuon line by an f -f line: e− f e+ f The reduced matrix element for this process has exactly the same form as for e− e+ → f f. (7.2.1) where Qf is the charge of the fermion f .41) by just multiplying by nf Q2f and replacing mµ → mf : σe− e+ →f f 4πα2 = nf Q2f 3s s 1− 96 4m2f s 2m2f 1+ s ! .44) s ≫ mµ .54α2 /m2µ = 1000 nb.45) However. so the three colors are simply summed over in order to find the total cross-section for a given flavor.3) . In the last subsection.1876 GeV. except that the photon-µ− -µ+ vertex is replaced by a photon-f -f vertex. strange.40) and (7. this formula is not good for arbitrarily high values of the center-of-momentum energy. We can easily generalize this to the case of production of any charged fermion f and antifermion f .2. with: ieγ ν −→ −iQf eγ ν . This effect becomes important √ when s is not small compared to mZ = 91.

5) = 1.3) to the asymptotic approximation (dashed line) given by eq. but the two √ already agree fairly well when s > ∼ 2. Qf = −1/3. This means that when several fermions contribute.4) The following graph compares the total cross-section for e+ e− → f f (solid line) as given by eq. (7. bottom quarks: muon.4).3. we have: σe− e+ →f f = nf Q2f 4πα2 . and there are three colors. (7. This means that the quark-antiquark production process e− e+ → QQ cross- section cannot easily be interpreted in terms of specific particles in the final state. charm quarks: down. one should view the quark production as a microscopic process. the up quark has charge Qu = +2/3. For example. tau leptons: Qf = 2/3. This always involves at least the further production of a quarkantiquark pair in order to make the final state hadrons color singlets. The prefactors for all the fundamental charged fermion types with masses less than mZ are: up. occurring at a distance scale much smaller than a typical hadron. nf = 3 −→ nf Q2f = 4/3 nf = 1 −→ nf Q2f nf = 3 −→ nf Q2f = 1/3 (7.3.54 nfQf α /mf σ 0 0 1 2 3 4 5 ECM/mf We see that the true cross-section is always less than the asymptotic approximation. Instead. 3s (7.In the high-energy limit √ s ≫ mf . However. strange. 2 2 2 0. Before we “see” them in macroscopic-sized detectors. A Feynman-diagram cartoon of the situation might looks as follows: 97 . the total cross-section well above threshold is just equal to the sum of nf Q2f for the available states times a factor 4πα2 /3s. the produced quarks then undergo further strong interactions that end up producing hadronic jets of particles with momenta close to those of the original quarks. so the prefactor indicated above is 3(2/3)2 = 4/3.3.5mf .3. free quarks are not seen in nature because the QCD color force confines them within color-singlet hadrons. Qf = −1.

and the ratio is Rhadrons = Finally. So. The traditional measure of the total hadronic cross-section is the variable Rhadrons . (7. one can just measure the total cross-section for producing hadrons. σ(e− e+ → µ− µ+ ) (7. one can make the approximation: σ(e− e+ → hadrons) ≈ X q σ(e− e+ → qq). down and strange quarks. c. b quarks). s. s quarks) (7.3. d.3. the charm quark contributes. which all have √ masses < 1 GeV. we get Rhadrons = 4 1 1 4 1 11 + + + + = .3. The threshold to produce charm-anticharm quarks occurs roughly when s > 2mc ≈ 3 √ GeV. for 4 1 1 + + = 2. 3 3 3 3 3 3 98 (u.3.9) s > 10 GeV. 3 3 3 3 3 (u. As each of these thresholds is passed. When s ≫ 1 GeV.10) . the calculation of the cross-section can only be trusted for energies that are significantly √ higher than this. 3 3 3 4 1 1 4 10 + + + = . so to test QED production of quarks.3. and that to produce bottom-antibottom quarks is at roughly s > 2mb ≈ 10 GeV. one gets a contribution to Rhadrons that is approximately a constant √ proportional to nf Q2f . d. one has Rhadrons = For 3 GeV< √ √ (u. s.8) s < 10 GeV. c quarks) (7.hadronic jet e− q q e+ hadronic jet Because the hadronic interactions are most important at the strong-interaction energy scale of a few hundred MeV. d.6) The final state can be quite complicated. defined as the ratio: Rhadrons = σ(e− e+ → hadrons) . for s < 3 GeV. (7.7) When the approximation is valid. one can always produce up.

but narrow.5 99 2 2.3. with a width of 0. production √ cross-sections when s is in just the right energy range to produce them. s 3-loop pQCD Naive quark model ρ ρ 1 Sum of exclusive measurements Inclusive measurements -1 10 0.10) provides a crucial test of the quark model of hadrons. including the charges of the quarks and the number of colors.5 1 1. 10 3 10 2 R J/ψ ψ(2S) Z φ ω 10 ρ ρ 1 S GeV -1 10 1 10 10 10 2 2 φ R ω 10 u. Schmitt. one gets the J/ψ particle resonance at s = 3. there are resonant contributions that come from e− e+ → hadronic bound states. Experimentally.3. The approximate agreement with the predictions for Rhadrons in eqs.096916 GeV. These bound states tend to have very large. from the online Review of Particle Properties. Janot and M. with corrections by P. being plagued by systematic detector effects. (7. Rhadrons is quite hard to measure. obtained from the COMPAS (Protvino) and HEPDATA (Durham) Groups.5 3 . Many of the older experiments at lower energy tended to underestimate the systematic uncertainties. For example. These resonances contribute very sharp peaks to the measured Rhadrons . Here are some plots of the data as of August 2007.8)-(7.Besides these “continuum” contributions to Rhadrons . when the bound √ state consists of a charm and anticharm quark. d.00093 GeV.

4 DHHM LENA MD-1 11 GeV Helicities in e− e+ → µ− µ+ Up to now. some e− e+ colliders can control the initial spin states.5 (3S) (2S) 7 (4S) 5 b 6 5 4 3 ARGUS Crystal Ball CLEO 2 9. we have computed the cross-section by averaging over the unknown spins of the initial state electron and positron. At a proposed future Linear Collider. it is a very important part of the experimental program to be able to run with at least the electron beam polarized. However.5 CLEO II CUSB DASP 10 10. Practical realities make it impossible to achieve 100%-pure polarized beams.5 8 4 (1S) 4.5 S 7.4. of course. using polarized beams.1) . Present estimates are that one might be able to get 90% or 95% pure polarization for the electron beam (either L or R). This terminology means that when the beam is operating in R mode. with perhaps 60% polarization for the positron beam.7 J/ ψ Mark I Mark I + LGW Mark II PLUTO DASP Crystal Ball BES 6 5 4 ψ4160 ψ(2S) ψ3770 c ψ4415 ψ4040 3 2 3 3. then a polarization of X% implies that PR − PL = X% 100 (7. This means that the beams are arranged to have an excess of either L-handed or R-handed helicity electrons and positrons.

PL acting on a v(p. because of the explicit expressions for PL and PR [see eq. s)γ 0 PR = v † (p.4. s)PR ←→ L final-state particle.9) PR v(p. (3. one can also measure the polarizations of outgoing particles. A nicer way is to keep the sum over spins. but eliminate the “wrong” polarization from the sum using a projection matrix from eq. then we need to use: u(p. s) ←→ L initial-state particle (7. s) ←→ R initial-state particle (7. s)PR = v † (p. (7.4. we must remember that the spin operator acting on v(p. the result must describe a R-handed positron. we can use: v(p.7) where ℓ is e or µ.4. If the polarization of a final-state fermion is fixed by measurement. s)PL γ 0 . Summing over the spin s will not change the fact that the projection matrix allows only L.4. which may be one of the final state fermions in a scattering or decay process of interest. then we would lose our precious trick of evaluating P s uu and P s vv. However. (7. (3. s) ←→ L final-state anti-particle. s)PL ←→ R final-state particle. (7.2.10) PL v(p. s) spinors is the opposite of the spin operator acting on u(p. in this case. s) and v(p. s)PR ←→ R initial-state anti-particle.8) u(p. We could redo the calculation of the previous subsections with particular spinors u(p.4) v(p. So if we form the object v(p. This experimental capability shows that one needs to be able to calculate cross-sections without assuming that the initial spin state is random and averaged over. Therefore. s) for the desired specific spin states s of initial state particles.4. (7.5) In some cases. for example.4.4. To get the equivalent rules for an initial state antiparticle. the bar on the spinor for an antiparticle “corrects” the handedness. s) ←→ R final-state anti-particle.30)].4.6) (7. we can use PL u(p. s) spinor projects onto a R-handed antiparticle. So.where PR and PL are the probabilities of measuring the spin pointing along and against the 3- momentum direction. s)PL ←→ L initial-state anti-particle (7. with the angular distributions of the final state directions depending on the spin of the τ . for an initial state antiparticle with either L or R polarization.30). for example by observing their decays. Tau leptons and anti-taus sometimes decay by the weak interaction processes: τ − → ℓ− ντ ν ℓ τ + → ℓ+ ν τ νℓ (7.or R-handed electrons to contribute to the cross-section. respectively.3) in place of the usual Feynman rules for an initial state particle.4.2) PR u(p. s) spinors.2.11) 101 . So. Now our traces over gamma matrices will involve γ5 . (7.4.

eq.R polarizations from each other in QED.4.4.17) + − + does not occur in QED. Similarly. we will √ assume the high energy limit s ≫ mµ . the process e− R eR → µ µ L L occur in QED. s (7. following from the same Feynman diagram as before.4. To simplify matters. the general rule is that the projection matrix in an initial state has the same handedness as the incoming particle or antiparticle. a projection matrix can be moved through a gamma matrix by changing L ↔ R: PR γ µ = γ µ PL . (7.) As an example. By the same type of argument. (7. is: M = i e2 (v b PR γ µ PR ua ) (u2 γµ v1 ).4.16) Therefore. because PR γ µ PR = γ µ PL PR = 0. µ− and µ+ in the final state must have opposite L. let us consider the process: + − + e− R eR → µ µ . s (7.20) . The reduced matrix element can be simply obtained from eq.4.4.4.4. let us therefore consider the process: + − + e− L eR → µL µR . To study a non-vanishing reduced matrix element. (7. e− e+ → µ− µ+ does not Therefore. The reduced matrix element for this process. (7.19) This can be simplified slightly by using the properties of the projections matrices: PR γ µ PL = γ µ PL PL = γ µ PL .15) This follows from the fact that γ5 anticommutes with γ µ : µ PR γ =    1 + γ5 1 − γ5 γµ = γµ 2 2  = γ µ PL . (7.13) As you should have discovered in a homework problem.18) in which we have now assumed that all helicities are perfectly known.2. (As a result of the barred spinor notation.in order to be able to calculate cross-sections to final states with specific L or R spin polarization states.12) where the helicities of the initial state particles are now assumed to be known perfectly. while the projection matrix in a final state has the opposite handedness of the particle being produced. 102 (7.4. (7.3) by just putting in the appropriate L and R projection matrices acting on each external state spinor: M = i e2 (v b PR γ µ PL ua )(u1 PR γµ PL v2 ). (7. (7.14) PL γ µ = γ µ PR .4.13) simply vanishes.

|M|2 = e4 (v b γ µ PL ua )(ua PR γ ν vb )(u1 γµ PL v2 )(v 2 PR γν u1 ).4.23) = u†a PL γ 0 γ ν vb (7.34) 103 . (7. s M = i (7.24) = u†a γ 0 PR γ ν vb (7.4. Let us do so.4.4. without changing anything.26) Here we have used the facts that (PL )† = PL . (u1 γν PL v2 )∗ = v 2 PR γν u1 .4. and PL γ 0 = γ 0 PR . (7. we are free to sum over the spin labels sa . s2 (7.4.4. s2 (7. and u†a γ 0 = ua . sb .27) Therefore. we compute: (v b γ ν PL ua )∗ = (vb† γ 0 γ ν PL ua )∗ = u†a (PL )† (γ ν )† γ 0 vb (7. and (γ ν )† γ 0 = γ 0 γ ν .4. and s2 .4. X s2 v2 v 2 = k/2 − mµ .4.4.32) This can be simplified by eliminating excess projection matrices. using: PL / pa PR = /pa PR PR = /pa PR .4.21) and so |M|2 = e4 (v b γ µ PL ua )(u1 γµ PL v2 )(v b γ ν PL ua )∗ (u1 γν PL v2 )∗ . In a similar way.33) PL k/2 PR = k/2 PR PR = k/2 PR .so that e2 (v b γ µ PL ua )(u1 γµ PL v2 ).4.4.25) = ua PR γ ν vb . (7. we therefore have |M|2 = = XXXX sa e4 s2 sb s1 XX s1 sb s2 |M|2 ν (v b γ µ PL p /a PR γ vb ) (u1 γµ PL k/2 PR γν u1 ). (7.30) Neglecting the masses because of the high-energy limit. (7. since it will allow us to apply the tricks X sa ua ua = /pa + me .28) Because the spin projection matrices will only allow the specified set of spins to contribute anyway.29) (7.31) (7. s1 .22) To evaluate this. (7.

4.18): Tr[γ µ / pa γ ν /pb γ5 ] = paα pbβ Tr[γ µ γ α γ ν γ β γ5 ] (7.4.35) Again using the trick of putting the barred spinor at the end and taking the trace [see the discussion around eq.to get |M|2 = e4 X X (v b γ µ /pa PR γ ν vb ) (u1 γµ k/2 PR γν u1 ).36) Doing the sums over s1 and sb using the usual trick gives: |M|2 = e4 Tr[γ µ /pa PR γ ν /pb ] Tr[γµ k/2 PR γν k/1 ]. Putting things together: h i Tr[γ µ p/a PR γ ν / pb ] = 2 pµa pνb − gµν (pa · pb ) + pµb pνa + ipaα pbβ ǫµανβ . Note that the cross-terms containing only one ǫ tensor vanish.40) = paα pbβ (4iǫµανβ ).45) . this becomes: |M|2 = e4 X X ν Tr[γ µ p /a PR γ vb v b ] Tr[γµ k/2 PR γν u1 u1 ]. from eq.4.2.4. (7. We have =   1 + γ5 γν p /b ] 2 1 1 ν µ ν Tr[γ µ p /a γ p/b ] + Tr[γ p /a γ p /b γ5 ].65).39) where we have used the fact that γ5 anticommutes with any gamma matrix to rearrange the order in the last term. (2. (7. contracting the indices µ and ν.2. The result is simply: Tr[γ µ p/a PR γ ν / pb ] Tr[γµ k/2 PR γν k/1 ] = 16(pa · k2 )(pb · k1 ). The first of these traces was evaluated in section 7. (B. s2 s1 sb (7.43) Finally.4. The trace involving γ5 is.44) which you can verify by brute force substitution of indices. (7.4. (7. (7. 2 2 ν µ Tr[γ µ p /a PR γ p /b ] = Tr[γ p/a (7.37) Now it is time to evaluate the traces. s2 s1 sb (7. we get Tr[γµ k/2 PR γν k/1 ] = 2 [k2µ k1ν − gµν (k2 · k1 ) + k1µ k2ν + ik2ρ k1σ ǫµρνσ ] .4. while the other terms are symmetric. s2 (7.4.41) where ǫµανβ is the totally antisymmetric Levi-Civita tensor defined in eq.38) (7. we have to multiply these two traces together.13)] for each quantity in parentheses. The term involving two epsilon tensors can be evaluated using the useful identity ǫµανβ ǫµρνσ = −2δρα δσβ + 2δσα δρβ . because the epsilon tensors are antisymmetric under µ ↔ ν.4.4. 104 (7.4.42) In exactly the same way.

The quantum mechanical overlap for two states with measured angular momenta in exactly opposite directions must vanish.52) Similarly.4. The total spin angular momentum of the final state is Snˆ = −1.4. 2s (7. If we describe the initial and final states as eigenstates of angular momentum with J = 1: Initial state: |Jzˆ = −1i.4. where n ˆ is the direction of the µ− . and so can be taken from the discussion in 7.4.4. 4 (7.51) then the reduced matrix element squared is proportional to: |hJnˆ = −1|Jzˆ = −1i|2 = (1 + cos θ)2 .47) Note that one does not average over initial-state spins in this case.48) (7. It follows that: dσe− e+ →µ− µ+ L R L R = d(cos θ) = e4 (1 + cos θ)2 32πs πα2 (1 + cos θ)2 . (7.46) This result should be plugged in to the formula for the differential cross-section: dσe− e+ →µ− µ+ L R L R d(cos θ) = |M|2 |~k1 | .4.so that |M|2 = 16e4 (pa · k2 )(pb · k1 ). This explains why the cross-section vanishes if cos θ = −1. because they have already been fixed. 2s corresponding to the picture: 105 (7.53) .4.50) Final state: |Jnˆ = −1i. s2 (7. (7. one can compute: dσe− e+ →µ− µ+ R L R L d(cos θ) = πα2 (1 + cos θ)2 .2 with mµ replaced by 0. 32πs|~ pa | (7. that corresponds to a final state with the total spin angular momentum in the opposite direction from the initial state.4.49) The angular dependence of this result can be understood from considering the conservation of angular momentum in the event. Drawing a short arrow to represent the direction of the spin: µ− L e− L θ e+ R µ+ R This shows that the total spin angular momentum of the initial state is Szˆ = −1 (taking the electron to be moving in the +z direction). The kinematics is of course not affected by the fact that we have fixed the helicities.

we get: " # dσe− e+ →µ− µ+ dσe− e+ →µ− µ+ dσe− e+ →µ− µ+ + − + 1 dσe− R L R L L R R L R L L R L eR →µL µR + + + + 12 · 0 4 d(cos θ) d(cos θ) d(cos θ) d(cos θ) 1 = 4 = in agreement with the √ πα2 2s ! h 2(1 + cos θ)2 + 2(1 − cos θ)2 πα2 (1 + cos2 θ). or a vector splits into a final state fermion and antifermion: or 106 .4.55) (7. as we have already seen. (7. Consider any field theory in which interactions are given by a fermion-antifermion-vector vertex with a Feynman rule proportional to a gamma matrix γ µ . the other 12 possible helicity combinations all vanish. and the sum of the possible final state helicities. If we compute the cross-sections for the final state muon to have the opposite helicity from the initial state electron.54) µ− R e− L µ− L e+ R θ e− R µ+ L θ e+ L µ+ R These are 4 of the possible 24 = 16 possible helicity configurations for e− e+ → µ− µ+ . However. − + + The vanishing of the cross-sections for e− L eL and eR eR in the above process can be generalized beyond this example and even beyond QED. 2s i (7.4. we get dσe− e+ →µ− µ+ L R R L d(cos θ) = dσe− e+ →µ− µ+ R L L R d(cos θ) corresponding to the pictures: = πα2 (1 − cos θ)2 . 2s (7.56) s ≫ mµ limit of eq. or µ− and µ+ with the same helicity. If an initial state fermion and antifermion merge into a vector.4.40).2. If we take the average of the initial state helicities. because they contain either e− and e+ with the same helicity.µ− R e− R θ e+ L µ+ L with all helicities reversed compared to the previous case.

s2 ) At order e2 . s1 ) v(k2 . Moreover. (7. the fermion and antifermion must have opposite helicities.2) . because of vPL γ µ PL u = vPR γ µ PR u = 0 and uPL γ µ PL v = uPR γ µ PR v = 0 and the rules of eqs.5 Bhabha scattering (e− e+ → e− e+ ) In this subsection we consider the process of Bhabha scattering: e− e+ → e− e+ . if an initial state fermion (or anti-fermion) interacts with a vector and emerges as a final state fermion (or anti-fermion): or then the fermions (or anti-fermions) must have the same helicity.4. because of the identities uPL γ µ PL u = uPR γ µ PR u = 0 and vPL γ µ PL v = vPR γ µ PR v = 0. summed over in the final state).5.4. as we will see.then by exactly the same argument as before. They are obviously useful when the helicities of the particles are controlled or measured by the experimenter.5. The conservation of angular momentum together with helicity conservation often allows one to know in which direction a particle is most likely to emerge in a scattering or decay experiment.11). 7.4. sa ) v(pb . with √ s = ECM ≫ me . the weak interactions only affect fermions with L helicity and antifermions with R helicity. These rules embody the concept of helicity conservation in high energy scattering.5) and (7. and we will consider all spins to be unknown (averaged over in the initial state. They are also useful because. sb ) u(k1 . (7.2)(7.4. This is true even if the interaction with the vector changes the fermion from one type to another. there are two Feynman diagram for this process: 107 (7. and in what cases one may expect the cross-section to vanish or be enhanced.8)-(7. Label the momentum and spin data as follows: Particle Momentum Spin e− pa sa e+ pb sb − e k1 s1 e+ k2 s2 Spinor u(pa .1) For simplicity we will only consider the case of high-energy scattering.

5. is: M = Ms + Mt = ie2   1 1 (v b γµ ua )(u1 γ µ v2 ) − (u1 γµ ua )(v b γ µ v2 ) .pa pa k1 µ pa + pb µ pb k1 and ν pa − k1 ν k2 pb k2 The first of these is called the s-channel diagram. a. contains a pure s-channel piece proportional to 1/s2 . We could have just as well assigned the minus sign to Ms instead. the corresponding contributions to the reduced matrix element for the process are:   −igµν Ms = [v b (ieγ )ua ] [u1 (ieγ ν )v2 ] . written in terms of the Mandelstam variables s = (pa + pb )2 and t = (pa − k1 )2 .8) The complex square of the reduced matrix element. |M|2 = M∗ M. b. It arises because the order of spinors in the written expression for Ms is b.5.1.5.5) Taking the complex conjugate of this gives: M∗ = M∗s + M∗t   1 2 1 ∗ ν ∗ ∗ ν ∗ = −ie (v b γν ua ) (u1 γ v2 ) − (u1 γν ua ) (v b γ v2 ) s t   1 ν ν 2 1 (ua γν vb )(v 2 γ u1 ) − (ua γν u1 )(v 2 γ vb ) . For organizational purposes.6) (7. 2. Using the QED Feynman rules listed at the end of subsection 7. The second one is called the t-channel diagram.1. a.7) (7.5. only the relative phases of terms in the matrix element are significant. and an interference piece proportional to 1/st.4) µ and Mt = (−1) [u1 (ieγ µ )ua ]  −igµν [v b (ieγ ν )v2 ] .5. it is exactly the same as the one we drew for e− e+ → µ− µ+ . Therefore the full reduced matrix element for Bhabha scattering.5.3)  (7. s t (7. and these differ from each other by an odd permutation. (pa − k1 )2 The additional (−1) factor in Mt is due to Rule 9 in the QED Feynman rules at the end of section 7. = −ie s t (7. but that in Mt is 1. 1. 108 . 2. (pa + pb )2 (7. it is useful to calculate these pieces separately. a pure t-channel piece proportional to 1/t2 .

5.12) in which we have turned the quantity into a trace by moving the u1 to the end. plagiarizing the result of eq. and P s1 u1 u1 = k/1 . 4t2 (7.11) (7.25): 1 X |Mt |2 = 4 spins 1 4 P 2 spins |Mt | by simply making the same 8e4 [(pa · k2 )(pb · k1 ) + (k2 · k1 )(pb · pa )] .5. t2 (7. The point of doing so is that now these traces have exactly the same form that we encountered in eq.5. (7.19).The pure s-channel contribution calculation is exactly the same as what we did before for e− e+ → µ− µ+ .25). t2 (7. M∗t Ms = − 4 spins 4st sb 109 (7.2. Therefore.17) . Therefore we can obtain replacements pa ↔ k2 and mµ → 0 in eq. consider the interference term: 1 X e4 X M∗t Ms = − (v b γµ ua )(ua γ ν u1 )(u1 γ µ v2 )(v 2 γν vb ).5. 4t2 s1 . 4 spins s (7.5. we have: 1 X 8e4 |Ms |2 = 2 [(pa · k2 )(pb · k1 ) + (pa · k1 )(pb · k2 )] .14) In the second line. (7.15) Next. but with pa ↔ k2 and mµ → 0.5. so that now we can immediately use the tricks P sa ua ua = /pa . the first trace has been rearranged using the cyclic property of traces.9) The pure t-channel contribution can be calculated in a very similar way. to obtain: e4 X 1 X (v b γµ /pa γ ν k/1 γ µ k/2 γν vb ).5.10) Taking the average of initial state spins and the sum over final state spins allows us to use the identities P sa P ua ua = /pa and sa vb v b = / pb (neglecting me ). We have: |Mt |2 = e4 (v 2 γν vb )(v b γµ v2 )(u1 γ µ ua )(ua γ ν u1 ).2. Now performing the sums over s1 .5. 4 spins 4st spins (7. except that now we can substitute mµ → me → 0.2. s2 gives: 1 X |Mt |2 = 4 spins = e4 Tr[γν /pb γµ k/2 ]Tr[γ µ /pa γ ν k/1 ] 4t2 (7.s2 1 X |Mt |2 = 4 spins e4 X Tr[γν p/b γµ v2 v 2 ]Tr[γ µ p/a γ ν u1 u1 ]. The result is: e4 X µ ν (v 2 γν p /b γµ v2 )(u1 γ p /a γ u1 ) 4t2 s1 .13) e4 µ ν Tr[γµ k/2 γν p /b ]Tr[γ p /a γ k/1 ].5.s2 = (7. (7.16) We chose to write the factors parentheses in that order. and P s2 v2 v 2 = k/2 .

26)- (7.2. Letting θ be the angle between the 3-momenta directions of the initial 110 . using eqs. This can be done by carrying over the kinematic analysis for the case e− e+ → µ− µ+ as worked out in eqs.5.19) 1 X M∗ Ms = 4 spins t 8e4 (pa · k2 )(pb · k1 ). the trace of any number of gamma matrices can be performed with the algorithm of eq. 4 spins 4 spins (7.18) (7. The procedure is to replace the trace over 2n gamma matrices by a sum over traces of 2n − 2 gamma matrices. However.27) It remains to identify the dot products of momenta appearing in the above formulas.36).24) in which the trace has finally been performed using eq.5.5. and repeat until all traces are short enough to evaluate using eqs. (7.10).22).7)-(B.23) (7. (7.14)-(B. So. with mµ .5.21) to write: γ ν /pa k/2 γν = 4pa · k2 .19) Now we are faced with the task of computing the trace of 8 gamma matrices.22) so that: Tr[γµ / pa γ ν k/1 γ µ k/2 γν p/b ] = −8(pa · k2 )Tr[/ k1 /pb ] = −32(pa · k2 )(k1 · pb ).21) which implies that This can be further simplified by now using eq.18). from eq. st (7. in many cases including the present one it is easier to simplify the contents of the trace first.26) This completes the contributions to the total i 1 Xh 1 X |M|2 = |Ms |2 + |Mt |2 + M∗t Ms + M∗s Mt .5.5. (7.5. (B. To evaluate the trace in eq.5.5.20) µ ν Tr[γµ p k1 γ ν p /a k/2 γν p /a γ k/1 γ k/2 γν p /b ] = −2Tr[/ /b ].5. (7. (7. In principle.5. st (7. (B.which can now be converted into a trace by the usual trick of moving the v b to the end: e4 X 1 X M∗t Ms = − Tr[γµ p/a γ ν k/1 γ µ k/2 γν vb v b ] 4 spins 4st sb = − e4 Tr[γµ /pa γ ν k/1 γ µ k/2 γν /pb ]. (7. 4st (7. (B. we also have: 1 X M∗ Mt = 4 spins s 8e4 (pa · k2 )(pb · k1 ). (B.8).19). (B. (B.2. (7.25) Taking the complex conjugate of both sides.10) and (B.22) to write: ν µ γµ p k1 γ ν p /a γ k/1 γ = −2/ /a .5. we first use eq. me → 0.19)-(B.

Jackson’s Classical Electrodynamics).5. because of the t’s in the denominator.34) 4e4 2 u .29) pa · k2 = pb · k1 = −u/2.33) 2e4 2 (u + s2 ) t2 (7. we obtain the spin-averaged differential cross-section for Bhabha scattering: dσ d(cos θ) = e4 16πs = πα2 2s u2 + t2 u2 + s2 2u2 + + s2 t2 st 3 + cos2 θ 1 − cos θ !2 .5.5. in which the electron and positron scatter off of each other in the forward direction (θ ≈ 0). It arises entirely from the t-channel diagram. Goldstein’s Classical Mechanics. J. although it may be extremely small.5. and (7.2.D.5. d(cos θ) (7. H.5.15). the differential cross-section for a nonrelativistic light particle with charge QA and a heavy particle with charge QB . we have: pa · pb = k1 · k2 = s/2.5.31) (7. d(cos θ) 2E 2 (1 − cos θ)2 111 (7.32) It follows from eqs.37) This result actually diverges for cos θ → 1.50).5. with center-of-momentum energy ECM to scatter through their Coulomb interaction is: πQ2A Q2B α2 dσRutherford = .36) (7.5. because the differential cross-section blows up quadratically near cos θ = 1. st (7.28) pa · k1 = pb · k2 = −t/2. (6.5. (7.5.38) The infinite total cross-section corresponds to the infinite range of the Coulomb potential between two charged particles. (7. 2 (7. This result is the relativistic generalization of the non-relativistic.35) Putting this into eq. since |~ pa | = |~k1 |. (7. classical Rutherford scattering problem.5.39) . As worked out in many textbooks on classical physics (for example.5. (7. in which an electron or alpha particle (or some other light charged particle) scatters off of the classical electric field of a heavy nucleus.state electron and the final state electron.9). (7.5.25). so σ= Z 1 −1 dσ d(cos θ) −→ ∞. (7.5.30) with s t = − (1 − cos θ). It simply reflects that an infinite-range interaction will always produce some deflection.26) that: 1 X |Ms |2 = 4 spins 1 X |Mt |2 = 4 spins 1 X (M∗t Ms + M∗s Mt ) = 4 spins 2e4 2 (u + t2 ) s2 (7. This is not an integrable singularity.5. 2 s u = − (1 + cos θ). ! (7.

one is limited by the angular resolution of detectors.5. Sometimes this is a real divergence with a physical interpretation. d(cos θ) (7. such as the finite life-time of the virtual particle. In other experiments.6. In a colliding-beam experiment. In other cases.1) (7. 7. we see that in both cases the small-angle behavior scales like 1/θ 4 . In terms of the Feynman diagram interpretation. which will give an imaginary part to its squared mass. any time that a virtual (internal line) particle can go on-shell.5. the divergence for small θ corresponds to the photon propagator going on-shell. the divergence is removed by higher-order effects. In general. removing the singularity in the Feynman propagator. For any scattering angle θ > 0. the true observable quantity is typically something more like: σexperiment = Z cos θcut − cos θcut dσ d(cos θ). this is usually dictated by the fact that detectors cannot be placed within or too close to the beamline.40) Of course. The relevant Feynman diagrams for these two processes are very similar: 112 (7. the classical value for a real massless photon. Therefore. there will be a divergence in the cross-section due to the denominator of the Feynman propagator blowing up. 2 (7.6. one has s (pa − k1 )2 = t = (1 − cos θ) > 0.(Here one must be careful in comparing results.) Comparing the non-relativistic Rutherford result to the relativistic Bhabha result. because the charge e used by Goldstein and Jackson √ differs from the one used here by a factor of 4π. as in the case of Bhabha scattering. in the Real World.6 Crossing symmetry Consider the two completely different processes: e− e+ → µ− µ+ e− µ+ → e− µ+ . the situation where the square of the t-channel virtual photon’s 4-momentum is nearly equal to 0. there is always some minimum scattering angle that can be resolved. the minimum resolvable angle is just one of many practical factors that have to be included.41) so that the virtual photon is said to be off-shell. and does not depend on the signs of the charges of the particles. in other words. In a real experiment.2) .

If the crossed particles have 4-momenta P1µ . as follows: . Pnµ . (7. (7.6) This similarity is generalized and made more precise by the following theorem.3) final µ− → initial µ+ .6. . Pnµ in M.5) .4) Two processes related to each other by exchanging some initial state particles with their antiparticles in the final state are said to be related by crossing.)|2 F = (−1) X spins 2 spins |M| P spins ′ |M can be obtained by substituting |M′ |2 . . 1 X |Me− e+ →µ− µ+ |2 = 2e4 4 spins 1 X |Me− µ+ →e− µ+ |2 = 2e4 4 spins u2 + t 2 s2 u2 + s 2 t2 ! ! . . by stretching and twisting.µ− e− e− e− µ+ µ+ and µ+ e+ In fact.6.6. the reduced matrix elements for √ these processes are also quite similar. Not surprisingly.6. . . . . For example. . Crossing Symmetry Theorem: Suppose two Feynman diagrams with reduced matrix elements M and M′ are related by the exchange (“crossing”) of some initial state particles and antiparticles for the corresponding final state antiparticles and particles. . in the high-energy limit s ≫ mµ . . one can turn the first process into the second by the transformation: initial e+ → final e− (7. then Pi′µ = −Piµ X spins P into the mathematical expression for |M(P1µ . (7.

.

.

. (P1′µ . . Pn′µ . . . . .)|2 . .

.

it is still perfectly valid as a mathematical relation. . since it has negative energy. So the relation between the two diagrams is a formal one rather than a relation between physical reduced matrix elements that would actually be measured. p~ ) is a valid physical 4-momentum. −~ p ) is never a physical 4-momentum.6. when one of the matrix elements involves the 4-momenta appropriate for a physical process. the other one does not. then p′µ = (−E. However. (7. Here F is the number of fermion lines that were crossed. If pµ = (E.7) Pi′µ → −Piµ with the other (uncrossed) particle 4-momenta unaffected. and therefore extremely useful. In 113 .

u′ = (p′a − k2′ )2 = (pa − k2 )2 = u.6.7. p′b = −k1 .general.2) by assigning primed momenta to the Feynman diagram for e− e+ → µ− µ+ : initial state final state ( ( e− ↔ p′a e+ ↔ p′b µ− ↔ k1′ µ+ ↔ k2′ . (7. crossing symmetry tells us that the formulas for the reduced matrix elements for these two processes are just related by the exchange of s and t. pb .1) and (7. So one only keeps the term with t (not s or u) in the denominator. (7. as illustrated in the high-energy limit in eqs. the point is that when the expression for |M′ |2 is analytically continued to unphysical momenta. (7.7) might look dubious at first. (7. the second result has been obtained for free.7.7. k1 .6.5) and (7. one can obtain the results for several “crossed” processes by merely substituting momenta. with no additional labor.2) (7. Then the Crossing Symmetry Theorem tells us that we can get the reduced matrix element for the process e− µ+ → e− µ+ as a function of physical momenta pa .6.1) into the formula for the reduced matrix element for the process e− e+ → µ− µ+ .6. t′ = (p′a − k1′ )2 = (pa + pb )2 = s. This means that we can identify: s′ = (p′a + p′b )2 = (pa − k1 )2 = t.6.6). Since we had already derived the result for the first process in section 7. Note that we could have obtained the particular result (7. it always flips sign for odd F . Equation (7. Label the momenta in the crossed Feynman diagram (for e− µ+ → e− µ+ ) without primes: initial state final state ( ( e− ↔ pa µ+ ↔ p b e− ↔ k1 µ+ ↔ k2 .3) (7. k1′ = −pb . since it may look like the right-hand side is negative for odd F . 7. However.6) even more easily just by noting that the calculation for the reduced matrix element of e− µ+ → e− µ+ is exactly the same as for Bhabha scattering. We will see this in some examples later.6. if one has calculated the cross-section or reduced matrix element for one process.4) In other words.2.7. since that corresponds to the t-channel diagram. except that only the t-channel diagram exists in the former case.7 e− µ+ → e− µ+ and e− µ− → e− µ− Let us apply crossing symmetry to the example of eqs. 114 . k2′ = k2 . k2 by substituting unphysical momenta p′a = pa .

11) Therefore. (7.7. (7.6) to eq.10) by doing the substitutions indicated by eqs.13) . k1′ = k2 .7.7.7.2. 2 s t = − (1 − cos θ). (7.7.We can carry this further by considering another process also related by crossing to the two just studied: e− µ− → e− µ− . the kinematics relating t and u to the scattering angle θ between the initial-state electron and the final state electron in these two cases is just the same as in Bhabha scattering: s u = − (1 + cos θ). (7. the charge of the muon does not matter at leading order. (7.7. from eq. (6. and the unprimed ones are for the desired process e− µ− → e− µ− .7. (7.7) (7.7. putting eq.5) with physical momenta: initial state final state ( ( e− ↔ pa µ− ↔ p b e− ↔ k1 .8) (7.12) (7. k2′ = −pb . Because we are neglecting the muon mass. 2s (1 − cos θ)2 u2 + s2 t2 Note that this again diverges for forward scattering cos θ → 1.9) Here the primed Mandelstam variable are the unphysical ones for the e− e+ → µ− µ+ process.10).6.5). µ− ↔ k2 This time.9). By comparing eq.10) into eq. We can therefore infer.7.7. 2 (7. that 1 X |Me− µ− →e− µ− |2 = 2e4 4 spins s 2 + u2 t2 ! .50) with |~ pa | = |~k1 | and using e2 = 4πα.7.6.6) so that s′ = (p′a + p′b )2 = (pa − k1 )2 = t. t′ = (p′a − k1′ )2 = (pa − k2 )2 = u. we obtain the differential cross-section for e− µ± → e− µ± : dσ d(cos θ) ! = πα2 s = πα2 5 + 2 cos θ + cos2 θ .7.7)-(7. 115 (7. (7. u′ = (p′a − k2′ )2 = (pa + pb )2 = s. p′b = −k1 . (7. the Crossing Symmetry Theorem tells us that we can identify the matrix element by again starting with the reduced matrix element for e− e+ → µ− µ+ and replacing: p′a = pa . (7. one sees that the spin averaged and summed squared matrix elements for the two processes are actually the same.7.

4) (7.1) There are two Feynman diagrams for this process: e− e− e− e− e− e− and e− e− Now. Making a table of the momenta: Bhabha Møller e− ↔ p′a e− ↔ pa e− ↔ k1′ e− ↔ k1 e+ ↔ p′b e− ↔ pb e+ ↔ k2′ e− ↔ k2 .8.7.8.3) (7.5) . (7.2) So. initial state electron) in Møller scattering. an easier way is to note that this is a crossed version of Bhabha scattering. which we studied earlier. and computing the Dirac traces. However. the Møller scattering reduced matrix element is obtained by putting s′ = (p′a + p′b )2 = (pa − k2 )2 = u t′ = (p′a − k1′ )2 = (pa − k1 )2 = t u′ = (p′a − k2′ )2 = (pa + pb )2 = s 116 (7. let us consider the case of Møller scattering: e− e− → e− e− . taking the complex square. final state positron) in Bhabha scattering with the (final state electron.8. (7. so that: p′a = pa p′b = −k2 k1′ = k1 k2′ = −pb . summing and averaging over spins. we see that crossing symmetry allows us to compute the Møller scattering by identifying the (initial state positron.8.8. nobody can stop us from getting the result for this process by applying the Feynman rules to get the reduced matrix element.8 Møller scattering (e− e− → e− e− ) As another example.

(7. we have a tricky and crucial factor of 1/2 in the total cross-section: σe− e− →e− e− = 1 2 Z cos θcut − cos θcut dσe− e− →e− e− d(cos θ). Instead of eq. because the two electrons in the final state are indistinguishable particles. This means that the final state with an electron coming out at angles (θ. we therefore get: s2 + t2 s2 + u2 2s2 + + u2 t2 ut 1 X |Me− e− →e− e− |2 = 2e2 4 spins ! .2.35).8. (As a check.9) Just as in the cases in the previous sections. (7. we find the differential cross-section: dσe− e− →e− e− d(cos θ) =   1 X 1  |Me− e− →e− e− |2  32πs 4 spins = πα2 s = 2πα2 s s2 + t2 s2 + u2 2s2 + + u2 t2 ut 3 + cos2 θ 1 − cos2 θ !2 (7. we would have   ie2 (pa + pb )µ (pa + pb )ν M = (v b γ ua )(u1 γ v2 ) −gµν + (1 − ξ) . note that eq. 117 (7. Using the results of eq. we had to also impose a cut on the minimum scattering angle θcut that we require in order to say that a scattering event should be counted.into the corresponding result for Bhabha scattering.8.) Therefore. (7.2. 7. This can be easily proved by observing that it contains the factor (v b γ µ ua )(pa + pb )µ = v b /pa ua + v b p/b ua = mv b ua − mv b ua = 0. or equivalently divide the total cross-section by 2. However.2) . We have been working with ξ = 1 (Feynman gauge). t = −s(1 − cos θ)/2 and u = −2(1 + cos θ)/2 where θ is the angle between the initial-state and final-state electrons.1) If the answer is to be independent of ξ. Consider what the matrix element for the process e− e+ → µ− µ+ would be if instead we let ξ remain unfixed. Applying eq. the part with t2 in the denominator).8.6) Again. (7. φ) is actually the same quantum state as the one with an electron coming out at angles (π − θ. then it must be true that the new term proportional to (1 − ξ) gives no contribution.5. if we keep only the t-channel part (that is. We have already integrated over the angle φ. (7.9.10) To obtain a finite value for the total cross-section.5.9 Gauge invariance in Feynman diagrams Let us now turn to the issue of gauge invariance as it is manifested in QED Feynman diagrams.2).7) ! (7.8. −φ). it contained a term that depended on an arbitrary parameter ξ.9) is invariant under cos θ → − cos θ. So.33)-(7. to avoid overcounting we must only integrate over half the range of θ.50). d(cos θ) (7.9.8. (6. Recall that when we found the Feynman propagator for a photon. (pa + pb )2 (pa + pb )2 µ ν (7. we recover the result for e− µ± → e− µ± in the previous section. in this case there is a special feature.8) .8.

(7. The polarization vector and momentum for a physical photon satisfy ǫ2 = −1 and ǫ · p = 0 and p2 = 0. λ) and ǫ∗µ (p. Now.9. P ). λ) to initial or final state photons. Consider a photon with momentum taken to be along the positive z axis. respectively.7) . (7.9. So. λ) = |M1 |2 + |M2 |2 .5) which defines Mµ . a gauge transformation is ǫµ (p. in terms of the polarization vector for the electromagnetic field.Here we have applied the Dirac equation. The reduced matrix element for a process with an external state photon with momentum pµ and polarization label λ can always be written in the form: M = Mµ ǫµ (p.23) and (B. note that if these relations are satisfied. Another aspect of gauge invariance involves a feature that we have not explored in an example so far: external state photons. (B. λ). with pµ = (P. Another nice consequence of the Ward identity is that it provides for a simplified way to sum or average over unmeasured photon polarization states. we have: 2 X λ=1 |M|2 = 2 X λ=1 Mµ M∗ν ǫµ (p. Gauge invariance implies that the reduced matrix element should also be unchanged after the substitution in eq. it follows from eq. 118 (7. λ) + apµ (7.6) This relation is known as the Ward identity for QED. one can show that the 1 − ξ term will always cancel when one includes all Feynman diagrams contributing to a particular process. 0. λ)ǫ∗ν (p. Recall that the Feynman rules associate factors of ǫµ (p.9.1.9.9. (7. In momentum space. in general.4) where a is any quantity. Summing over the two polarization vectors in eqs. as embodied in eqs.24). the derivative ∂ µ is proportional to pµ .9. (7.4) that Mµ pµ = 0.3) where Λ is any function. (7.1.4). 0. For any photon propagator connected (at either end) to an external fermion line.4). As a consistency check. the proof is similar. This is a nice consistency check on calculations.9. making a gauge transformation on the photon field results in: Aµ → Aµ + ∂ µ Λ (7. λ) → ǫµ (p. then they will also be obeyed after the gauge transformation (7. then the reduced matrix element should become 0. It says that if we replace the polarization vector for any photon by the momentum of that photon. which is usually ξ = 1.27).26)-(7. to write p /a ua = mua and v b p /b = −mp /b .9. Since M must be invariant under a gauge transformation. And. So we can choose the most convenient value of ξ.

(7.where Mµ = (M0 . according to eq. so it is true for any photon momentum direction. it must vanish when contracted with Mµ M∗ν . Gauge invariance. we can write 2 X λ=1 P2 2 λ=1 |M| = −|M0 |2 + |M1 |2 + |M2 |2 + |M3 |2 . not just momenta oriented along the z direction. M1 . therefore implies that we can always sum over a photon’s polarization states by the rule: 2 X λ=1 ǫµ (p.1) First let us assign the following labels to the external states: initial γ initial e− final γ final e− ǫµ (pa . so |M0 |2 = |M3 |2 . M3 ). or |M|2 = −gµν Mµ M∗ν . λa ) u(pb .8) The last equation is written in a Lorentz invariant form. s2 ).10.8). λ1 ) u(k2 . we obtain: Ms and Mu " # i(p /a + /pb + m) (ieγ µ )ub ǫ∗1ν ǫaµ = u2 (ieγ ν ) (pa + pb )2 − m2 (7.10. (7. sb ) ǫν∗ (k1 . λ) ǫ∗ν (p.3) i e2 h ν µ u γ ( p + p + m)γ u ǫ∗1ν ǫaµ / / 2 b a b s − m2 " # i(p /b − k/1 + m) = u2 (ieγ ) (ieγ ν )ub ǫ∗1ν ǫaµ (pb − k1 )2 − m2 (7.10. / 2 b 1 b u − m2 119 . as expressed by the Ward identity.9. consider Compton scattering: γe− → γe− .2) = −i (7. There are two Feynman diagrams for this process: and which are s-channel and u-channel.9.4) = −i (7. Applying the QED Feynman rules.10. (7.5) µ i e2 h µ ν / u γ ( p − k + m)γ u ǫ∗1ν ǫaµ . M2 . 7. respectively.10 Compton scattering (γe− → γe− ) As our first example of a process with external-state photons. Although the (irrelevant)µν part is non-zero.9. Therefore. (7. The Ward identity implies that pµ Mµ = P M0 + P M3 = 0.9) as long as we are taking the sum of the complex square of a reduced matrix element.10. λ) = −gµν + (irrelevant)µν .

16) s2 The result is a single spinor trace: 1 X e4 |M|2 = Tr 4 spins 4    1 1 µ ν µ (γ ν p (−γ µ k/1 γ ν + 2pνb γ µ ) (p /b + m) /a γ + 2pb γ ) + 2 s−m u − m2   1 1 k 2 + m) . .13) a 2 X λ1 =1 to obtain: 1 X e4 |M|2 = 2 λ .23). 2 b (7.10.10) and (7.10.14) Next we can average over sb .11) Now we multiply together eqs. γ µ }ub + 0 (7.7) = 2pbσ gσµ ub . using the usual tricks: 1X ub u b = 2 sb X 1 (p / + m).10. γ µ }ub − γ µ (/pb − m)ub (7. So we note that: (/ pb + m)γ µ ub = {/pb . (7. (/ k2 + m)]. using 2 1 1 X ǫaµ ǫ∗aρ = − gµρ + irrelevant. So the reduced matrix element is: M = −ie2 ǫ∗1ν ǫaµ u2     1 1 µ ν µ/ ν ν µ ν µ (−γ k γ + 2p γ ) ub .10. (7. and average over the initial photon polarization λa and sum over the final photon polarization λ1 .10.10. 2 s−m u − m2 (7. u2 = Tr[.10. (7. γ p γ + 2p γ + /a 1 b b s − m2 u − m2 (7.Before squaring the total reduced matrix element. and sum over s2 . The end result can be written compactly in terms of pa · pb = pb · k1 = 120 s − m2 .17) (γµ / (−γν k/1 γµ + 2pbν γµ ) (/ pa γν + 2pbµ γν ) + s − m2 u − m2 Doing this trace requires a little patience and organization.10) Taking the complex conjugate of eq.10.11).10.10. (7.10.18) (7.15) u2 .λ 2 a 1   u2 ub    1 1 (γ ν /pa γ µ + 2pµb γ ν ) + (−γ µ k/1 γ ν + 2pνb γ µ ) ub 2 s−m u − m2  1 1 (γµ p (−γν k/1 γµ + 2pbν γµ ) u2 /a γν + 2pbµ γν ) + s − m2 u − m2   . (7.10. . 2 (7. (7.10. . (7.8) = 2pµb ub .10.10. (B.19) . (7. 2 m2 − u .12) ǫ∗1ν ǫ1σ = −gνσ + irrelevant.10. . 2 λ =1 2 (7.9) where we have used eq.10) gives: ∗ 2 M = ie ǫ1σ ǫ∗aρ ub   1 1 (γ ρ /pa γ σ + 2pρb γ σ ) + (−γ σ k/1 γ ρ + 2pσb γ ρ ) u2 .10.6) = pbσ {γ σ . it is useful to simplify.

In the center-of-momentum frame. (7. as do the initial and final state electrons. 0. P cos θ) (7.10.10.27) pa · pb = Eγ (Ee + Eγ ). (7.49) to eq. we have: pµa = (P.10.26) Ee = Then we have: P 2 + m2 . 0. −P cos θ).23) k1µ p p k2µ = ( P 2 + m2 .2. |~k1 | = 1.20) Equation (7.25) p (7.10.10.22) = (P. (6. 0.10. and taking the initial state photon momentum to be in the +z direction and the final state photon momentum to make an angle θ with the z-axis. 0. one finds: "  1 1 pa · pb pb · k1 1 X + + 2m2 − |M|2 = 2e4 4 spins pb · k1 pa · pb pa · pb pb · k1  +m 4  1 1 − pa · pb pb · k1 2 # . applying eq.10.10.28) pb · k1 = Eγ (Ee + Eγ cos θ). We can now find the differential cross-section after choosing a reference frame.20) is a Lorentz scalar. the kinematics is just like in the case eµ → eµ studied in a homework problem. We will do this first in the center-of-momentum frame.29) (7. P sin θ. P ) (7. (7. 0. −P ) (7. Then using four-momentum conservation and the on-shell conditions p2a = k12 = 0 and p2b = k22 = m2 . |~ pa | (7. (7.10. we obtain: " πα2 Ee + Eγ cos θ Ee + Eγ dσ = + + 2m2 d(cos θ) s Ee + Eγ cos θ Ee + Eγ 121 cos θ − 1 (Ee + Eγ )(Ee + Eγ cos θ) ! . 1 (7. Call the magnitude of the 3-momentum of the photon in the initial state P . s = (Ee + Eγ )2 .10.After some calculation. −P sin θ. so define: Eγ = P. 0.10.24) θ a b 2 The initial and final state photons have the same energy.21) pµb = ( P 2 + m2 .10.10.30) So.20). and then redo it in the “lab” frame in which the initial electron is at rest. (7.

(7.) Just for fun.37) plus terms that vanish like (m2 /s2 )ln(s/m2 ) as m2 /s → 0.) 122 . In fact. (7. this term gives the dominant contribution to the total cross-section.10. so the energy of a photon is equal to its angular frequency. Ee + Eγ Ee + Eγ (7. starting from eq. However.34) Z (7.20).10. This corresponds to the virtual electron in the u-channel Feynman diagram going nearly on-shell. Integrating with respect to cos θ. but now working in the lab frame in which the initial state electron is at rest.36) The remaining two terms vanish as m2 /s → 0.10. The origin of this enhancement can be traced to the u-channel propagator. (Notice that s − m2 can never become small when s ≫ m2 . the energies in the center-of-momentum frame are much larger than the electron’s mass. (Recall that h ¯ = 1 in our units.10. where the denominator of the integrand becomes small.31). but with a logarithmic enhancement coming from back-scattered photons with angles θ ≈ π. − 1 = + O(m4 ). This is most important for the first term in eq. expanding in the small mass of the electron.10. for s ≫ m2 we have: σ= Z 1 −1 i πα2 h dσ d(cos θ) = 2 ln(s/m2 ) + 1 d(cos θ) s (7.10. since then the denominator factor Ee + Eγ cos θ can become large.31). (7. The cross-section falls at high energy like 1/s.10. one finds: Z 1 −1 Ee + Eγ cos θ 2Ee d(cos θ) = = 1 + O(m2 ). consider the high-energy limit Eγ ≫ m.35) Therefore. we can set Ee = Eγ and m = 0 in that limit. (7. . q Ee − Eγ = Eγ 1 + m2 /Eγ2 − Eγ = Eγ ! m2 m2 + . (7.10. let us redo the analysis of Compton scattering. coming from the cos θ ≈ −1 (back-scattering) region. So. Ee + Eγ cos θ with the dominant contribution coming from cos θ near −1. Naively.31) In a typical Compton scattering situation.33) 1+ 2Eγ2 2Eγ Ee + Eγ = 2Eγ + O(m2 ). Integrating the second term in eq. this requires some care for cos θ ≈ −1. which becomes large when u − m2 = 2pb · k1 = 2Eγ (Ee + Eγ cos θ) becomes small. (7. we find: Z 1 −1 Ee + Eγ Ee + Eγ Ee + Eγ d(cos θ) = ln Ee + Eγ cos θ Eγ Ee − Eγ ! .10.+m4 cos θ − 1 (Ee + Eγ )(Ee + Eγ cos θ) !2  .10.32) Now. So. Let us call the energy of the initial state photon ω and that of the final state photon ω ′ . (7. . 1 −1 Ee + Eγ d(cos θ) = 2 ln(4Eγ2 /m2 ) + O(m2 ) = 2 ln(s/m2 ) + O(m2 ).

10. |M|2 = 2e4 ′ + 4 spins ω ω 123 (7. 0. 0. so that   ω′ ω 1 X − sin2 θ .45) pb · k1 = ω ′ m. (7. ω − ω ′ cos θ). ω ′ sin θ.10. (7. ω ′ cos θ) (7. (7.20) "  1 1 ω′ 1 X ω + 2m − + |M|2 = 2e4 4 spins ω′ ω ω ω′  +m 2  1 1 − ω ω′ 2 # .40) k2 = (ω + m − ω ′ .10. 0. or vice versa: ω . (7. −ω ′ sin θ. 0. 1+ − cos θ) m(ω − ω ′ ) .10.46) so that. from eq.10.10.48) which follows from eq. 0.49) . (7. the 4-momenta are: pa = (ω.41) Here we have used four-momentum conservation and the on-shell conditions p2a = k12 = 0 and p2b = m2 . 0) (7.10.10. cos θ = 1 − ωω ′ ω′ = (7. (7. ω) (7. one has: pa · pb = ωm.44) In terms of lab-frame variables.10.42) This can be used to solve for ω ′ in terms of cos θ. Applying the last on-shell condition k22 = m2 now leads to: 2m(ω − ω ′ ) + 2ωω ′ (cos θ − 1) = 0.1 θ a b 2 Then. (7. in terms of the lab photon scattering angle θ.47) This can be simplified slightly by using m  1 1 − ′ ω ω  = cos θ − 1.10.39) k1 = (ω ′ .10. 0.38) pb = (m.43) ω m (1 (7.10. (7.10.10.42).

59) In order to do the ω ′ integral.58) Therefore. 16π 2 ω ′ E2 dΦ2 = (2π)4 δ(4) (k1 + k2 − pa − pb ) = (7.56) where now E2 = p ω ′2 − 2ωω ′ cos θ + ω 2 + m2 .) The two-body phase space in the lab frame is: d3~k2 d3~k1 (2π)3 2E1 (2π)3 2E2 d3~k1 δ(3) (~k1 + ~k2 − ω zb) δ(ω ′ + E2 − ω − m) d3~k2 . resulting in: dΦ2 = δ(ω ′ + E2 − ω − m) d3~k1 . 8πE2 (7.10.10.10. as defined in eq.10.10. as usual. to make a change of integration variables to the argument of the delta function. Evaluating the prefactors for the case at hand: Ea = ω.34).51) Eb = m.10.Now to find the differential cross-section.60) ω ′ − ω cos θ dK = 1 + . (7.10.2. (6.10. defining K = ω ′ + E2 − ω − m. (7.50) where dΦ2 is the two-body Lorentz-invariant phase space. for the final state particles in the lab frame.57) The phase space for the final state photon can be simplified by writing it in terms of angular coordinates and doing the integral R dφ = 2π: d3~k1 = dφ d(cos θ) ω ′2 dω ′ = 2π d(cos θ) ω ′2 dω ′ . dΦ2 = δ(ω ′ + E2 − ω − m) d(cos θ) ω ′ dω ′ .33): 1 dσ = 4Ea Eb |~va − ~vb |  1 |M|2 4  dΦ2 (7.10. (7. 16π 2 ω ′ E2 (7.2.53) (Recall that the speed of the photon is c = 1.10.55) q where ω ′ is now defined to be equal to E1 = |~k1 | and E2 is defined to be |~k2 |2 + m2 .52) |~va − ~vb | = 1 − 0 = 1. (7. Performing the ~k2 integral using the 3-momentum delta function just sets ~k2 = ω zb − ~k1 .61) we have 124 . it is simplest. we must apply eq. (7. (7. dω ′ E2 (7.10. So. (6.10.54) (7.

. .10. eq.10. when ω ≫ m.10.62) in eq.68) Equation (7.10.10. (7. using the dependence of ω ′ on cos θ as given in eq.66) In the high-energy (small m) limit.10.64)  ω′ ω + ′ − sin2 θ .2. 32πmω 2 (7. .37). (7.2). The numbers of particles NS . As in the center-of-momentum frame.43).10.62) Performing the dK integration sets K = 0.69) . this becomes: 2 σ = πα   2ω 1 ln ωm m   1 + .43) has been used to simplify the denominator.10. s (7. (7. ω ω (7.(7.. the largest differential cross-section is found for back-scattered photons with cos θ = −1.63) where eq. Notice that.10.65) This result is the Klein-Nishina formula. Na . The low-energy Thomson scattering limit is also interesting.10. + 2+ ω m(m + 2ω)2 (7. and so certainly do not depend on any choice of inertial reference frame. = E2 E2 + ω ′ − ω cos θ (7. Finally using this in eq.43).50) yields:  dσ =  1 X 4 spins so that. while the area A is invariant under Lorentz boosts along the collision axis. and Nb can be simply counted. so E2 = ω + m − ω ′ . according to eq. Using eq.Therefore.49). One can now integrate the lab-frame differential cross-section R1 −1 d(cos θ) to get the total cross-section. m + ω(1 − cos θ) 8π 8πmω (7.10.68) is the same result that we found in the center-of-momentum frame. d(cos θ) m 125 (7. The result is: 2 σ = πα    1 2 2m 2ω − − 3 ln 1 + ωm ω 2 ω m   2(m + ω) 4 . (7.10. so that dσ πα2 = 2 (1 + cos2 θ). + 2ωm (7.10. putting in eq. ω ≪ m implies ω ′ /ω = 1. σ= i πα2 h 2ln(s/m2 ) + 1 + .10. (7.67) We can re-express this in terms of the Mandelstam variable s = (ω + m)2 − ω 2 = 2ωm + m2 ≈ 2ωm. (7.59) gives dΦ2 = ω′ d(cos θ) ω ′2 = d(cos θ). . πα2 dσ = 2 d(cos θ) m   |M|2  ω′ ω 2  ω ′2 d(cos θ).10. (6. one need only look at the definition of the total cross-section given in eq. (7.10. ω ′ dK ω ′ dω ′ . one gets very low energies ω ′ when the photon is back-scattered. In the lab frame.. To see why. which corresponds to the smallest possible final-state photon energy ω ′ .10. This is an example of a general fact: the total cross-section does not depend on the choice of reference frame as long as one boosts along a direction parallel to the collision axis.

11.2) in the Compton scattering result (for small m): X spins 2 p′a · p′b p′b · k1′ + ′ ′ p′b · k1′ pa · pb 4 |Mγe− →γe− | = 8e obtained from eq.20).11.11. We could always compute the reduced matrix element starting from the Feynman rules.3) (7. with a factor of 2 enhancement in the forward (θ = 0) and backward (θ = π) directions compared to right-angle (θ = π/2) scattering.11. we have chosen e+ to be labeled by “a”.11.6) 126 . 2 4 pb · k1 = − (7.70) Unlike the case of high-energy Compton scattering.7). (7. so we can steal the kinematics information directly from eqs. since we have already done Compton scattering. For convenience. 2 4 t s = − = (1 − cos θ). (7. 7.11. k1′ = k1 . So.5) pb · k2 (7. Thomson scattering is symmetric under θ → π − θ. However.6. according to eq. Then according to the Crossing Symmetry Theorem. so that the initial-state e− can have the same label “b” in both processes. we can obtain P 2 spins |Me+ e− →γγ | by making the replacements p′a = −k2 . the result is: X spins 2 |Me+ e− →γγ | 4 = (−1)8e 4 = 8e   pb · k1 −k2 · pb + pb · k1 −k2 · pb   k2 · pb pb · k1 + . k2′ = −pa (7. p′b = pb . we can make the following comparison table: e+ e− → γγ Compton γ ↔ p′a e+ ↔ pa e− ↔ p′b e− ↔ pb γ ↔ k1′ γ ↔ k1 e− ↔ k2′ γ ↔ k2 . (7. it is easier to use crossing. 3m2 (7.28)-(7.1) ! (7.5. pb · k1 k2 · pb (7. The relevant facts for the present case are: s u = (1 + cos θ). The total cross-section in this limit is then σ= 8πα2 . Labeling the physical momenta for Compton scattering now with primes.which is independent of energy.10.32).5.4) The situation here is 2 → 2 massless particle scattering. there is also a factor of (−1)1 = −1. Because the crossing involves one fermion (a final state electron changes into an initial state positron).10.11 e+ e− → γγ As our final example of a QED process let us consider e+ e− → γγ in the high-energy limit.

for small m.11. The observable cross-section in one of these colliding beam experiments is something more like σcut = = Z 1 cos θcut dσ d(cos θ) 2 − cos θcut d(cos θ)     2πα2 1 + cos θcut − cos θcut . you will find that instead of diverging.11. (7. (7. In a real e− e+ → γγ experiment.12) into eq. 127 .9) (7. This is because we have set m = 0 in the kinematics.7) It is a useful check. which is not valid for scattering very close to the collision axis. 1 X |Me+ e− →γγ |2 = 2e4 4 spins  u t + u t  1 + cos2 θ sin2 θ = 4e4 ! .8) This cross-section is symmetric as θ → π − θ. In this sense. photons very close to the electron and positron beams will not be seen by any detector.8. and a vindication of the (−1)F factor in the Crossing Symmetry Theorem.50) for the differential cross-section. since the two final state photons are identical. (7. (7.) On the other hand. Redoing everything with √ m ≪ s but non-zero.11. ln s 1 − cos θcut (7.11. Note that this formula is just what you would have gotten by plugging in cos θcut = 1 − 4m2 /s (7. you can show:   2πα2 s σ= ln s 2m2   −1 . The integrand in eq.11.10). If you put back non-zero m. we get: dσ 2πα2 = d cos(θ) s 1 + cos2 θ sin2 θ ! . (6. the final state in which a photon is observed coming out at a (θ. −φ). if you are interested in the total cross-section for electron-positron annihilation with no cuts applied on the angle.11. the finite mass of the electron “cuts off” the would-be logarithmic divergence of the cross-section for small sin θ.8) diverges for sin θ = 0.11.Therefore.11. the total cross-section features a logarithmic enhancement ln(s/m2 ) coming from small sin θ.2.10) (You can easily check that this is a positive and increasing function of cos θcut . then you must take into account the non-zero electron mass. This is the same overcounting issue for identical final state particles that arose for Møller scattering at the end of section 7. however. we should therefore divide by 2 (or just integrate over half of the range for cos θ) to avoid overcounting. (7. That was inevitable. that this is positive! Now plugging this into the formula eq. φ) is actually the same as a state in which a photon is observed at an angle (π − θ. When we find the total cross-section.11) The logarithmic enhancement at large s in this formula comes entirely from the sin θ ≈ 0 region.

5) The quantity τ = 1/Γ 128 (8.1. So we can define the decay rate Γ [also known as the decay width. If the particle is moving with velocity β.6) .1. This is because in QED as a stand-alone theory. bound states can decay even in QED. (8. all at rest.1. So if there is only a muon in the initial state. then the number of particles remaining a short time later is therefore: N (t + ∆t) = (1 − Γ∆t) N (t). The weak interactions get around this by allowing interaction vertices that change the fermion type. the survival probability for a particular particle as a function of the laboratory time t is: √ 2 (Probability of particle survival) = e−Γt 1−β . so if only QED interactions are allowed. the total energy of the process is simply mµ . we have studied only 2 → 2 scattering processes. For any type of unstable particle. the probability that a decay will occur in a very short interval of time ∆t should be proportional to ∆t.1. but the energy of the final state would have to be larger than this.3) N (t) = e−Γt N (0). because they are massless. it is equal to the resonance width in eq. the fundamental particles are stable. (8.1) Suppose we observe the decays of a large sample of particles of this type. (1.1)] as the constant of proportionality: (Probability of decay in time ∆t) = Γ∆t. Going to the rest frame of the initial muon. Moreover.2) dN N (t + ∆t) − N (t) = lim = −Γ.1 Decay rates and partial widths So far.1. then because of relativistic time dilation. it is traditional and sensible to quote the result as measured in the rest frame of the particle. you can easily convince yourself that there must be at least one muon in the final state.8 Decay processes 8. If the number of particle at time t is denoted N (t). The photon interaction vertex with a fermion line does not change the type of fermion.1. (8. The electron is the lightest particle that carries charge.4) It follows that so that When one has computed or measured Γ for some particle. so it cannot decay because of charge conservation. ∆t→0 dt ∆t (8. Photons cannot decay into anything else (in any theory). (8. the muon must also be stable.

2. (8.37)-(6. (A. One can then compute or measure the decay rates into particular final states. Let us simplify the formula eq. of course. 0) (8. we should integrate the 3-momenta over all available ~ki .] There is often more than one final state available for a decaying particle.8) where 4 (4) dΦn = (2π) δ (p − X ki ) i n Y i=1 d3~ki (2π)3 2Ei ! (8.1.9) is the n-body Lorentz-invariant phase space. (1.is also known as the mean lifetime of the particle at rest. Consider a process in which a particle at rest with 4-momentum pµ = (M.1) K |M|2 dφ1 d(cos θ1 ). it is usually a very good bet that three-particle final states will lose. the decay rate for that final state tends to decrease.1. K dφ1 d(cos θ1 ). 2M (8. one can show by arguments similar to those in 6. 32π 2 M 2 (8. 16π 2 M (8. To find the total decay rate Γ.2.1.2).10) Two-body decays Most of the decay processes that one encounters in high-energy physics are two-particle or three-particle final states.2) dΦ2 = and dΓ = 129 . [Putting in the units recovers eq. The energies in this formula are defined by Ei = 8.2. As the number of particles in the final state increases.] To find the contribution to the decay rate for final-state particles with 3-momenta restricted to be in some ranges. Therefore. [Compare this to eq. you will see that the only difference is that the pa + pb in the 4-momentum delta function for a scattering process has been replaced by p for a decay process.2. so a particle will typically decay into few-particle states if it can. The rate Γ for a particular final state or class of final states is called a partial width.1. Given the reduced matrix element M for this process. with the simple replacement ECM → M . 0.34).2 for cross-sections that the differential decay rate is: dΓ = 1 |M|2 dΦn . (6. i i (8.2. If a two-particle final state is available.2 q ~k2 + m2 . The sum of all exclusive partial widths should add up to the total decay rate.8) for the case of two-particle final states with arbitrary masses. see eq.46).6).7) decays to several particles with 4-momenta ki and masses mi . (6. 0. The evaluation of the two-particle final-state phase space is exactly the same as in eqs. we should integrate dΓ over those ranges.1.

2. 2M M 2 − m21 . z) is −16 times the It is so-named because. in the present context M. † 130 . then the final state particles share the energy equally in the rest frame: E1 = E2 = M/2.2.14) √ √ √ is less than the sum of the other two. Mq 1 − 4m2 /M 2 . if M < m1 + m2 .6) .2.2.12) (8. 2M M 2 + m22 − m21 .5) (8. 1 64π 2 M (8.7) where λ(x. m1 = m2 = m. m21 . m2 = 0.13) This illustrates the general feature that since the final state particles have equal 3-momentum magnitudes. z. (8.2. then: E1 = E2 = K = M 2 + m21 . if each of x.√ y. then λ(x. m2 never form a triangle.2. The solution is: E1 = E2 = K = M 2 + m21 − m22 . z √ √ square of the area of a triangle with sides x. z) ≡ x2 + y 2 + z 2 − 2xy − 2xz − 2yz. 2M (8. the heavier particle gets more energy.3) One can solve this by writing it as E1 − M = q E12 − m21 + m22 (8. y. K = 2 (8. then the decay is forbidden.10) |M|2 q 1 − 4m2 /M 2 dφ1 d(cos θ1 ). dΓ =  |M|2  2 2 1 − m /M dφ1 d(cos θ1 ). y. y.2.It remains to solve for K.2. Energy conservation requires that E1 + E2 = q K 2 + m21 + q K 2 + m22 = M. (8.2.2. m22 ) 2M (8.2. 64π 2 M (8.9) (8. However. 2M q λ(M 2 . In this case.4) and squaring both sides.8) is known as the triangle function.† It is useful to tabulate results for some common special cases: • If the final-state masses are equal.11) and dΓ = • If one of the final-state particle is massless. (8.2. m1 .

• If the decaying particle has spin 0.3 Scalar decays to fermion-antifermion pairs.3) To turn M into a physically observable decay rate. Higgs decay Let us now consider a simple and very important decay process. Then the matrix element will be of the form OUT hk1 . s1 .2.4) . so the final state particles must be distributed isotropically in the centerof-momentum frame. There is an extra factor of 1/2 if they are identical. (6. (8.8 and 7.4: Lint = −yφΨΨ. and a final state containing a fermion and antifermion each with mass m and with 4-momenta and spins k1 .3.4. As a model. One then obtains the total decay rate from dφ1 d(cos θ1 ) → 4π.3. s2 |piOUT = −iM(2π)4 δ(4) (k1 + k2 − p).11]. s1 and k2 . as shown in the picture immediately after eq. 131 (8. (8. so they are the same as for QED. k2 .3. Therefore we can draw the Feynman diagram: 1 2 and immediately write down the reduced matrix element for the decay: M = −iy u(k1 . M∗ = iy v 2 u1 . From eq. The corresponding Feynman rule was argued to be equal to −iy times an identity matrix in Dirac spinor space.1) This type of interaction is called a Yukawa interaction.3). (8. and y is a Yukawa coupling. (8. (8. s2 ). then there can be no special direction in the decay. s1 ) v(k2 .3.2) The Feynman rules for fermion external states don’t depend on the choice of interaction vertex. or if its spin is not measured. s2 respectively.7). we need to compute the squared reduced matrix element summed over final state spins. 8. let us consider the Lagrangian already mentioned in section 6. namely a scalar particle φ decaying to a fermion-antifermion pair. Consider an initial state containing a single φ particle of mass M and 4-momentum p.15) provided that the two final state particles are distinguishable. to avoid counting each final state twice [see the discussions at the end of sections 7.3.

8) (8. (8. (8.so |M|2 = y 2 (u1 v2 )(v 2 u1 ).3.s2 |M|2 = y 2 Tr[(/ k 2 − m)(/ k1 + m)]   = y 2 Tr[/ k 2 k/1 ] − m2 Tr[1] = 4y 2 (k1 · k2 − m2 ). the Higgs boson h plays the role of φ. and eqs. (8. f 132 (8. using eq.3.14) Doing the (trivial) angular integrals finally gives the total decay rate: y2M Γ= 8π 4m2 1− M2 !3/2 .3. (8.3. and couples to each fermion f with a Lagrangian that is exactly of the form given above: Lint = − X yf hΨf Ψf .8).5) Summing over s2 .3. (B. so M 2 = (k1 + k2 )2 = k12 + k22 + 2k1 · k2 = m2 + m2 + 2k1 · k2 (8.15) In the Standard Model.3.2. y2M dΓ = 32π 2 4m2 1− M2 !3/2 dφ1 d(cos θ1 ).11).3.3.9) (8. X spins 2 2 |M| = 2y M 2 4m2 1− M2 ! .3. The fermion and antifermion have the same mass m.10) where we have used the fact that the trace of an odd number of gamma matrices vanishes. 2 k1 · k2 − m2 = (8.6) = y 2 Tr[(/ k 2 − m)u1 u1 ].12) Therefore.7) and (B. we have: X s2 |M|2 = y 2 u1 (/ k2 − m)u1 (8.3. (8.13) and.3.16) .11) implies that M2 − 2m2 .7) Now summing over s1 gives: X s1 .3. (8. (8.

19) → yτ ≈ 0.004. so we expect decays h → bb and h → τ − τ + and h → cc.017.3. since the top quark has a mass of about 173 GeV.3.24) This is an extremely narrow width for such a heavy particle.3.21) for a Higgs with mass of order Mh = 115 GeV. (8. the sum in eq. For quarks. and tau lepton. (8. At the Fermilab Tevatron Run II and the LHC.16) includes a summation over 3 colors.) Therefore. Γtotal 133 (8.017. (8. leading to an extra factor of nf = 3 in the decay rate.0 GeV mτ ≈ 1.3. the prediction is that bb final states win: Γ(h → bb) : Γ(h → τ − τ + ) : Γ(h → cc) :: 3(mb )2 : (mτ )2 : 3(mc )2 :: 1 : 0. it turns out that mτ > mc because of the large higher-order corrections for the charm quark. divided by the total decay rate. the decay rate to a particular fermion is approximately: Γ(h → f f ) = nf yf2 Mh ∝ nf m2f .3. so BR(h → bb) = Γ(h → bb) . charm quark.20) → yc ≈ 0. 115 GeV (8. Suppose for example that the mass of the Higgs is not too much larger than the current experimental limit of about 114 GeV. Since the kinematic factor (1 − 4m2f /Mh2 )3/2 is close to 1 for all allowed fermion-antifermion final states.25) . these corrections are quite large.7 GeV → yb ≈ 0.3. 16π (8. and mf tends to come out considerably smaller than the masses of the quarks quoted in Table 3. One can define the branching ratio to be the partial decay rate into a particular final state. Then. The next-lightest fermions in the Standard Model are the bottom quark. (8.The Yukawa coupling for each fermion is approximately proportional to its mass: yf ≈ mf .17) However. because of higher-order corrections.22) (8.23) and the partial width of the Higgs boson into bb can be estimated numerically as Γ(h → bb) = 2 MeV   Mh . the reason is the very small Yukawa coupling yb ≈ 0.18) Estimates of the mf from present experimental data are: mb ≈ 3.3. For quarks. the decay h → tt is clearly forbidden. the mf appearing in this formula is not quite equal to the mass. one of the major goals is to look for a Higgs boson through its decay modes.77 GeV mc ≈ 0.12 : 0.3.05.3.0102. (8. (Notice that even though the charm quark is heavier than the tau lepton. 174 GeV (8.3.

and increase the partial widths into quarks by roughly 10%. Second. Normally. such decays would be negligible. notably gluon-gluon (gg) and γγ. from eqs.0003 is so small.29) L-fermion. If we demanded that the final states have particular helicities.Then. (8.] So.4 GeV respectively.5. the last two are not kinematically allowed unless Mh > 160.3. that to get a L fermion or antifermion in the final state. (8. consider the helicities for the process h → f f . the rule here is that helicity is always violated by the scalar-fermion-antifermion vertex.28) However. the spins of the outgoing particles must therefore have opposite directions. R-antifermion: M = −iy u2 PR PL v1 = 0 (8.26) BR(h → τ − τ + ) ≈ 0. PL projection matrices: R-fermion. but they can still contribute if one or both of the weak vector bosons is off-shell (virtual).30) R-fermion. which both occur due to Feynman diagrams with loops.3. if no other decays contributed significantly. then we would have obtained for the matrix element. but they are competitive since the squared Yukawa coupling yb2 ≈ 0. R-antifermion: M = −iy u2 PL PL v1 = −iy u2 PL v1 6= 0 (8. L-antifermion: M = −iy u2 PR PR v1 = −iy u2 PR v1 6= 0 (8. We will return to the subject of the Higgs boson branching ratios in section 13. Since they have momentum in opposite directions.10. (8.27) BR(h → cc) ≈ 0. One may understand this result from angular momentum conservation.3. higher-order corrections are important.05. since the scalar must decay to a state with equal helicities.11).3.31) L-fermion.3. this means they must also have the same helicity. Naively. First. In the final state. while to get a R fermion or antifermion in the final state. this is not accurate for two reasons. one puts in a PL . The initial state had no spin and no orbital angular momentum.32) [Recall.4. Drawing a short arrow to represent the spin. Finally.4. one puts in a PR next to the spinor. (7.3. we would estimate BR(h → bb) ≈ 0. the allowed cases of RR helicities and LL helicities look like: R fermion h R anti-fermion L fermion h L anti-fermion 134 .3. using PR . there are other final states that can appear in h decays. L-antifermion: M = −iy u2 PL PR v1 = 0 (8.8 and 182.8)-(7.85. and W + W − and Z 0 Z 0 .

1) where dΦ3 = (2π)4 δ(4) (p − k1 − k2 − k3 ) d3~k2 d3~k3 d3~k1 (2π)3 2E1 (2π)3 2E2 (2π)3 2E3 (8. so there are several ways to present the result. k2 . the result for dΓ is a differential with respect to 5 remaining variables. so this effect may eventually be measured when a sample of h → τ − τ + decay events is obtained.The helicities of tau leptons can be (statistically) measured from the angular distributions of their decay products. Then one can show: dΦ3 = 1 dE1 dE2 dφ1 d(cos θ1 ) dα2 . 2M (8. If we think of the three 3-momenta within the decay plane as describing a rigid body. and k3 respectively: m1 . k1 M m3 . so its four-vector is pµ = (M.3) . 256π 5 135 (8. 0. then the energy of the third particle E3 = M − E1 − E2 is also known from energy conservation. k3 m2 . then the relative orientation can be described using three Euler angles. 0. In the rest frame of the decaying particle. Since there are 9 integrals to do. and an angle α2 that measures the rotation of the 3-momentum direction of particle 2 as measured about the axis of the momentum vector of particle 1.4. k2 In general. The remaining 3 variables just correspond to the orientation of the decay plane with respect to some fixed coordinate axis. m2 . the formula for a three-body differential decay rate is dΓ = 1 |M|2 dΦ3 . Specifying E1 and E2 also uniquely fixes the angles between the three particle momenta within this decay plane. Two of the 5 variables can be chosen to be the energies E1 and E2 of two of the final-state particles. These can be chosen to be the spherical coordinate angles φ1 and θ1 for particle 1. because of momentum conservation.4. The best choice of 5 variables depends on the problem at hand.4. and the four-vectors of the remaining particles are k1 . 8.4 Three-body decays Let us consider a generic three-body decay process in which a particle of mass M decays to three lighter particles with masses m1 . We will work in the rest frame of the decaying particle.2) is the Lorentz-invariant phase space. and m3 . 0). and 4 delta functions. the three final-state particle 3-momenta must lie in a plane.

9) A good strategy is usually to choose the label “1” for the particle whose energy we care the most about.4.4. for particular E1 and E2 .4. for spinless or spin-averaged initial states. m22 . θ1 or α2 . (8. 64π 3 M (8. (8.4.7) where the triangle function λ(x. and m223 = (k2 + k3 )2 = (p − k1 )2 = M 2 − 2E1 M + m21 (8.8) is the invariant (mass)2 of the combination of particles 2 and 3. dΦ3 = 1 dE1 dE2 .10) (8. one must find the limits of integration. the reduced matrix element cannot depend on the angles φ1 . 2 M . then there is no special direction to measure the orientation of the final state decay plane with respect to.min   q 1 2 2 2 = (M − E1 )(m23 + m2 − m3 ) ± (E12 − m21 ) λ(m223 .11) .5) and so. and one can do the integrals Z 0 2π dφ1 Z 1 −1 d(cos θ1 ) Z 2π 0 dα2 = (2π)(2)(2π) = 8π 2 .The choice of which particles to label as 1 and 2 is arbitrary. z) was defined by eq. 32π 3 (8.4. y. In the special case that all final state particles are massless (or small enough to neglect) m1 = m2 = m3 = 0.4. 2m223 (8.4. or if it is spinless. 2 (8. then these limits of integration simplify to: M − E1 < E2 < 2 0 < E1 < 136 M . 2M (8. Then the limits of integration for the final E1 integral are: m1 < E1 < M 2 + m21 − (m2 + m3 )2 . In that case. If the initial state particle spin is averaged over. m23 ) .2. we will be left with an expression for dΓ/dE1 .4) Then.4. Then after doing the dE2 integral. dΓ = 1 |M|2 dE1 dE2 . then by doing the kinematics one can show for any particular E1 that E2max.8). If one resolves to do the E2 integral first. and should be made to maximize convenience.6) To do the remaining energy integrals.

which in turn constantly convert 14 C.1. and antineutrino: n → p+ e− ν e (9.1.1. 14 N nuclei into Carbon-dioxide-breathing organisms.1 Weak nuclear decays In nuclear physics. so A nucleons in all. making it ideal for dating dead organisms.2) τ = 885.7 years). it dropped in the early 20th century as more ordinary 12 C entered the atmosphere because of the burning of fossil fuels containing the carbon of organisms † The lifetime of the neutron is an infamous example of an experimental measurement that has shifted dramatically over time.1.6 × 108 sec = 17. They are related by t1/2 = τ ln(2). tritium decays according to 3 H → 3 He + e− ν e (τ = 5. A nucleus with Z protons and A − Z neutrons. However.7) so that t1/2 = 5740 years for Carbon-14. the weak interactions are responsible for decays of long-lived isotopes. The simplest example of this is the decay of the neutron into a proton.7 ± 0. (9.4) where the quotes indicated that the neutron and proton are really not separate entities. complicating matters is the fact that this ratio is not constant. (9. cosmic rays produce energetic neutrons. it was thought that τn = 1010 ± 30 seconds.1.1.8 × 1011 sec = 8280 years).5) (τ = 1.8 sec. maintain an equilibrium with the carbon content of the atmosphere. (9. (9. If kinematically allowed. (9.1. (9. electron.6) and carbon-14 decays according to 14 C → 14 N + e− ν e Nuclear physicists usually quote the half life t1/2 rather than the mean lifetime τ . is denoted by A Z. As recently as the late 1960’s. at a level of roughly 14 C/12 C≈ 10−12 .9 Fermi theory of weak interactions 9. 137 . So for example.3) with a mean lifetime of† Decays of heavier nuclei can be thought of as involving the subprocess: “n” → “p+ ”e− ν e . or those that eat them. one can observe decays: A Z → A (Z + 1) + e− ν e . In the upper atmosphere.1) The existence of the antineutrino ν e was inferred by Pauli as a “desperate remedy” to save the principle of energy conservation. but part of the nuclear bound states.

with a mean lifetime of τπ± = 2.10) The long lifetimes of such decays are what originally gave rise to the name “weak” interactions.1.that have been dead for a very long time. in section 9. So. In any case.14) with a branching ratio 1. (9. This presents a puzzle: since the electron is lighter. simply because mp < mn . 138 . (9. dead organisms lose half of their 14 C every 5740±30 years.) The only other significant decay mode is π − → e− ν e . reaching a peak in the mid 1960’s from which it has since declined. unless it interacts (which it usually will in a collider detector).1.2 × 10−4 . We will calculate the reason for this later. (9.1.8 meters.1. (This includes submodes in which an additional photon is radiated away. and determining the historic atmospheric 14 C/12 C 14 C decay ratio as a function of time with control samples or by other means. yet the first decay dominates by almost a factor of 104 .8 m) 1−β /β . but under the right circumstances it is kinematically allowed when the proton and neutron are parts of nuclear bound states. (9.6. one can date the death of a sample of organic matter. One can also have decays that release a positron and neutrino: A Z → A (Z − 1) + e+ νe . The probability that a charged pion with velocity β will travel a distance L in empty space before decaying is therefore √ 2 P = e−(L/7.2 × 10−8 sec.13) with a branching ratio of 0.9) In free space.12) This means that a relativistic charged pion will typically travel several meters before decaying. and certainly do not regain it by breathing or eating. The relative abundance 14 C/12 C ≈ 10−12 then doubled after 1954 because of nuclear weapons testing. and corresponds to a proper decay length of cτ = 7. An example is 14 O → 14 N + e+ νe (τ = 71 sec). the proton cannot decay.8) These can be thought of as coming from the subprocess “p+ ” → “n”e+ νe .99988.1. (9. by measuring the rate of e− beta rays consistent with produced by a sample.1. The main decay mode is π − → µ− ν µ (9. (9. Charged pions also decay through the weak interactions. there is more kinematic phase space available for the second decay.1.11) This is till a very long lifetime by particle physics standards.

2. . νe for e+ . e. and the ellipses mean matrices in Dirac spinor space. νe ) (e .2. µ− → e− ν e νµ (9. ν τ for all other particles (9. (The only confirmed exceptions are neutrino oscillation experiments.2. . and electron neutrino. one needs clues from experiment. . we therefore expect that the Lagrangian should contain terms schematically of the form Lint = (ν µ .5) −1 for τ − . One clue is the fact that there are three quantum numbers. µ).2 × 10−6 sec). Muons do not undergo hadronic interactions like pions do. so that relativistic muons will usually penetrate at least the inner layers of particle detectors with a very high probability. µ) (e .2.2. The Feynman diagram for muon decay can be drawn as: e− µ− νe νµ This involves a 4-fermion interaction vertex. . To be more precise about the interaction Lagrangian.4) −1 for µ− .6)   0    +1   0    +1   0 We will later find out that this is not a true fundamental interaction of the theory. (9. νe ) or (ν µ . νµ for µ+ . . ν e for all other particles (9. 139 .1) This corresponds to a decay width of Γ = 3.) They are assigned as: Le = Lµ = Lτ † =    +1 −1 for e− .2. νµ . ν µ for all other particles (9. called lepton numbers. electron. . . νe mean the Dirac spinor fields for the muon.2) implying a proper decay length of cτ = 659 meters.9. that are additively conserved to a high accuracy in most experiments.0 × 10−19 GeV. muon neutrino. (9.2 Muon decay The muon decays according to (τ = 2. .† Because of the correspondence between interactions and terms in the Lagrangian. ντ for τ + .3) where the symbols µ. but rather an “effective” interaction that is derived from the low-energy effects of the W − boson.

However.11) Belle.2. (9.2. we will write the weak interactions for leptons in terms of building blocks with net Le = Lµ = Lτ = 0. .8) This is a remarkably strong constraint. and in reactors have given strong evidence for oscillations of neutrinos that are caused by them having small non-zero masses that violate the individual lepton numbers.2.0 × 10−8 BaBar.2. . ℓ). by accelerators. (9. More generally. these are very small effects for colliding beam experiments. µ. (9. νℓ ) or (ν ℓ . (9. (9. Now.2 × 10−11 .3) but not the second. with Lµ = Lτ = 0 trivially in each case.7) would be allowed. (9.2.13) for more on this. The (near) conservation of lepton numbers suggests that the interaction Lagrangian for the weak interactions can always be written in terms of fermion bilinears involving one barred and one unbarred Dirac spinor from each lepton family. like the first term in eq. The can be classified by their 140 .2. in the nuclear decay examples above.1). (9. . The muon decay mode in eq. τ . for example. experimental results from neutrinos produced in the Sun.12) Since 1998. (9. one always has Le = 0 in the initial state.2. then one might expect that decays like µ− → e− γ (9.10) BaBar.9) The BaBar and Belle experiments have put similar (but not as stringent) bounds on tau lepton number non-conservation: BR(τ − → e− γ) < 3.) However.2. and the most recent limit from the LAMPF experiment at Los Alamos National Lab is BR(µ− → e− γ) < 1. (It is still an open question whether they also violate the total lepton number L ≡ Le + Lµ + Lτ . If lepton numbers were not conserved. a basis for fermion bilinears involving any two fields Ψ1 and Ψ2 will have 4 × 4 = 16 elements. and can be ignored for almost all conceivable processes at Tevatron and LHC. we will want to use building blocks: (ℓ . and Le = 1 − 1 = 0 in the final state.4. see section 13. since each Dirac spinor has 4 components.3 × 10−8 BR(τ − → µ− γ) < 4. (9. (9.4 × 10−8 BR(τ − → e− π 0 ) < 8.2.6 × 10−30 GeV. Lµ ) = (0.So. this decay has never been observed. It implies that Γ(µ− → e− γ) < 3. the atmosphere.2.14) where ℓ is any of e.1) has (Le . 1) in both the initial and final states. So. for example.2. since this decay only has to compete with the already weak mode in eq. .

c. This can be explained consistently with angular momentum conservation if the electron produced in the decay is always polarized left-handed and the antineutrino is always right-handed. νℓ . with (−1)µ = ( +1 −1 for µ = 0 for µ = 1.15) The weak interaction Lagrangian for leptons could be formed out of any product of such terms with Ψ1 .c. Some other possibilities could have been that the building blocks were of type A: ρ LA int = −G(ν µ γ γ5 µ)(eγρ γ5 νe ) + c. 3. Using short arrows to designate spin directions.2. so that when cooled and placed in a magnetic field.16) Here “c. (9. or some combination of S and P . 2.2. or perhaps even T .18) has spin J = 4. The most important clue for determining the correct answer for the proper Lorentz and parity structure of the weak interaction building blocks came from an experiment on polarized 60 Co decay by Wu in 1957.transformation properties under the proper Lorentz group and the parity transformation ~x → −~x.” means complex conjugate. the nuclear spins ~ Wu then measured the angular dependence of the electron spin from the decay align with B. as follows: Parity (~x → −~x) Term Number Ψ1 Ψ2 1 Ψ1 γ5 Ψ2 1 +1 Scalar = S −1 Pseudo-scalar = P Axial-vector = A 2 4 (−1)µ Ψ1 γ µ γ5 Ψ2 4 −(−1)µ Ψ1 γµΨ i µ ν 2 Ψ1 [γ .2.2. 60 The nucleus 60 Ni Co → 60 Ni + e− ν e . (9. Fermi originally proposed that the weak interaction fermion building blocks were of the type V . so that muon decays would be described by LVint = −G(ν µ γ ρ µ)(eγρ νe ) + c. γ ] Ψ2 (−1)µ (−1)ν 6 Type Vector = V Tensor = T. (9. The 60 Co nucleus has spin J = 5. (9. The entry under Parity indicates the multiplicative factor under which each of these terms transforms when ~x → −~x.c. The observation was that the electron is emitted preferentially in the direction opposite to the original spin of the 60 Co nucleus.17) or some combination of V and A. this is necessary since the Dirac spinor fields are complex. so the net angular momentum carried away by the electron and antineutrino is 1. Ψ2 = ℓ. the most favored configuration is: 141 . Fermi’s original proposal of V for the weak interactions turned out to be wrong.

19) Since this is a complex quantity. Using the correspondence between terms in the Lagrangian and particle interactions.2. d 142 √ −i2 2GF (γ ρ PL )ba (γρ PL )cd .21) Here GF is a coupling constant with dimensions of [mass]−2 . For muon decay. a e. (9. b νe . (9. one must also have terms involving the complex conjugate of eq.J =5 −→ −→ e− 60 Co Before J =4 −→ 60 Ni −→ νe After The importance of this experiment and others was that right-handed electrons and left-handed antineutrinos do not seem to participate in the weak interactions. PR νe . known as the Fermi constant.20) The feature that was considered most surprising at the time was that right-handed Dirac fermion fields PR e. we can always put a PL to the left of the electron’s Dirac field. and a PR to the right of a ν e field. we therefore have two Feynman rules: µ. Its numerical √ value is most precisely determined from muon decay.19): ν e PR γ ρ e = ν e γ ρ PL e. The factor of 2 2 is a historical convention. This means that when writing the interaction Lagrangian for weak interactions. 2 (9. and the Lagrangian density must be real. This helped establish that the correct form for the fermion bilinear is V − A: 1 ePR γ ρ νe = eγ ρ PL νe = eγ ρ (1 − γ5 )νe . c ←→ ν µ. the relevant four-fermion interaction Lagrangian is: √ Lint = −2 2GF (ν µ γ ρ PL µ)(eγρ PL νe ) + c.2. and ν e PL never appear in any part of the weak interaction Lagrangian.2.c.2. and left-handed Dirac barred fermion fields ePL . (9.

there should be a PL next to each unbarred spinor in a matrix element. the only Feynman diagram for µ− → e− ν e νµ is: µ− e− νe νµ using the first of the two Feynman rules above.22) The reduced matrix element is obtained by starting at the end of each fermion line with a barred spinor and following it back (moving opposite the arrow direction) to the beginning.23) This illustrates a general feature. Let us label the momenta and spins of the particles as follows: Particle Momentum Spin Spinor − µ pa sa u(pa . (9. The result is: √ M = −i2 2GF (u3 γ ρ PL ua )(u1 γρ PL v2 ). At lowest order. that means starting with the muon neutrino and electron barred spinors.2. (The presence of the 143 . c µ.2. This weak interaction Lagrangian for muon decay violates parity maximally. d These are related by reversing of all arrows. In this case. We can now derive the reduced matrix element for muon decay. with neutrinos treated as fermions and antineutrinos as antifermions. helicity is conserved by this interaction Lagrangian. (9.2.21). because of the presence of one gamma matrix in each fermion bilinear. Comparing this to the experimentally measured result will allow us to find the numerical value of GF . The Feynman rules for external state fermions and antifermions are exactly the same as in QED.e. and use it to compute the differential decay rate of the muon. a ←→ νµ . corresponding to the complex conjugate in eq. b √ −i2 2GF (γ ρ PL )ab (γρ PL )dc ν e. sa ) = ua u(k1 . The slightly separated dots in the Feynman rule picture are meant to indicate the Dirac spinor structure. However. s2 ) = v2 u(k3 . s1 ) = u1 e− k1 s1 νe k2 s2 v(k2 . s3 ) = u3 νµ k3 s3 (9. just as in QED. since it treats left-handed fermions differently from right-handed fermions. and determine the energy spectrum of the final state electron. in the weak interactions. or equivalently a PR next to each barred spinor.

26) u1 u1 = k/1 .45).25) Therefore. (9.4.005. we have already seen a product of traces just like this one.2.2. 2 spins (9.31) Fortunately. (9. (9. (9. so that by substituting in the appropriate 4-momenta. s3 using the usual tricks: 1X ua ua = 2 sa X 1 (p / + mµ ). in eq.34) Therefore.2.2. (9. we immediately get: 1 X |M|2 = 64G2F (pa · k2 )(k1 · k3 ). (7.) Taking the complex conjugate.32). 0.35) .30) (9.28) u3 u3 = k/3 . 0. Eν e ).29) s1 X s2 X s3 to turn the result into a product of traces: 1 X |M|2 = 4G2F Tr[γ ρ PL (/pa + mµ )PR γ σ k/3 ]Tr[γρ PL k/2 PR γσ k/1 ] 2 spins = 4G2F Tr[γ ρ p/a PR γ σ k/3 ]Tr[γρ k/2 PR γσ k/1 ] (9. since me /mµ < 0.gamma matrix ensures the equivalence of these two statements. we can evaluate them in a frame where ~k2 is along the z-axis. (9.33) k2 = (Eν e . In the following. (9.2. 0.2.2. 144 (9. 2 a (9. we have: √ M∗ = i2 2GF (ua PR γ σ u3 )(v 2 PR γσ u1 ). Now we can average over the initial-state spin sa and sum over the final-state spins s1 . 0).4 to the example of muon decay.2. since PL ↔ PR when moved through a gamma matrix.2.27) v2 v 2 = k/2 . we can neglect the mass of the electron me . Then pa = (mµ .24) |M|2 = 8G2F (u3 γ ρ PL ua )(ua PR γ σ u3 ) (u1 γρ PL v2 )(v 2 PR γσ u1 ).2.2. pa · k2 = mµ Eν e . we need to evaluate the dot products pa · k2 and k1 · k3 . with M = mµ and m1 = m2 = m3 = 0. Applying the results of subsection 8.2. According to our result of eq. s2 . Since these are Lorentz scalars. (9.32) Our next task is to turn this reduced matrix element into a differential decay rate. 0.2.

0.40) The shape of this distribution is shown below as the solid line: _ e . mµ /4): 145 . −mµ /2).32). we get: dΓ = dEe dEν e G2F (m2 Eν − 2mµ Eν2e ). 2 (9. k1 · k3 = 12 [(k1 + k3 )2 − k12 − k32 ] = 12 [(pa − k2 )2 − 0 − 0] = 12 [p2a + k22 − 2pa · k2 ]. from eq. which are collinear.2 0.10). and k2µ = k3µ = (mµ /4.2. 2π 3 µ e (9. (9.2. (8.1 0. 0. 2 spins (9.2.4. and choosing E1 = Ee and E2 = Eν e .6) with M = mµ . we obtain: dΓ = dEe Z mµ 2 mµ −Ee 2 dEν e G2F G2F 2 2 ) = dE (m E − 2m E ν e µ µ ν e e 2π 3 π3 m2µ Ee2 mµ Ee3 − 4 3 ! .4 0.39) We have obtained the differential decay rate for the energy of the final state electron: G2 m2µ 4Ee dΓ = F 3 Ee2 1 − dEe 4π 3mµ ! .2. νµ νe dΓ/dEe 0 0 0.38) Doing the dEν e integral using the limits of integration of eq. (8. This corresponds to the situation where the electron is recoiling directly against both the neutrino and antineutrino.37) Therefore. 0.3 Ee/mµ 0.4. so 1 k1 · k3 = (m2µ − 2mµ Eν e ). (9.36) 1 X |M|2 = 32G2F (m3µ Eν e − 2m2µ Eν2e ). k1 = (mµ /2. (9. Plugging this into eq.Also.2.2.5 We see that the electron energy is peaked near its maximum value of mµ /2. 0. for example.

41) and the range of integration for Eν e is now: mµ mµ − Eνµ < Eν e < .2.4. Just for fun. by choosing E2 = Ee and E1 = Eν e in eqs. since that is what is directly observable in an experiment.4. (9.2.43) Therefore. We have done the most practically sensible thing by plotting the differential decay rate in terms of the electron energy.46) . To find dΓ/dEνµ . and ν e have well-defined L.42) so that dΓ = dEνµ Z mµ 2 mµ −Eνµ 2 dEν e G2F G2F 2 2 − 2m E ) = dE (m E ν µ νe e 3 2π 3 µ e π m2µ Eν2µ 4 − mµ Eν3µ 3 ! .11). Then dΓ = dEνµ dEν e G2F (m2 Eν − 2mµ Eν2e ).2. However.e− −→ µ− −→ νe ←− νµ The helicity of the initial state is undefined. let us pretend that we could directly measure the νµ and ν e energies.10)-(8. νµ . L.10)-(8.37). and compute the distributions for them.4. the Eνµ distribution of final states has the same shape as the Ee distribution: G2 m2µ 4Eνµ dΓ = F 3 Eν2µ 1 − dEνµ 4π 3mµ ! . as shown above. we can take E2 = Eν e and E1 = Eνµ in eqs.2. however. this tells us that the electron must move in the opposite direction to the initial muon spin in the limit that Ee is near the maximum.4. In the case of maximum Ee . The helicity of the electron is L. since this is dictated by the weak interactions.4. (9. (8. (8. and R helicities respectively. so its spin must be opposite to its 3-momentum direction. By momentum conservation.4. 2π 3 µ e (9.2. we know that the final state e− . (9. we can find dΓ/dEν e .44) Finally. which occurs when the neutrino and antineutrino move in nearly opposite directions.45) so that G2 m2µ 2Eν e dΓ = F 3 Eν2e 1 − dEν e 2π mµ 146 ! .11) with eq. (9. Then: dΓ = dEν e Z mµ 2 mµ −Eν e 2 dEe  G2F G2F  2 2 2 3 2 − 2m E ) = dE − 2m E (m E m E ν ν µ µ e νe νe . therefore. The smallest possible electron energies are near 0. 2 2 (9. the spins of νµ and ν e must be in opposite directions. since the muon is at rest. with the reduced matrix element from eq. (9.2.2.6) and (8. µ νe 2π 3 µ e 2π 3 (9. so that the 3-momentum of the electron recoiling against them is very small.2.37).6) and (8.

or eq. we see that dΓ/dEν e vanishes when Eν e approaches its maximum value of mµ 2 .2.2.2. (9. (9. (9. R helicities of e.2.50) (This determination also includes some small and delicate corrections reviewed below in section 9.2.This distribution is plotted as the dashed line in the previous graph. so the L. or eq. while m5µ has units of [mass]5 or [time]−5 . and the rate must vanish in that limit of maximal Eν e . We can understand this by noting that when Eν e is maximum. (9. so Γ indeed has units of [mass] or [time]−1 . It is also good to check units: G2F has units of [mass]−4 or [time]4 . we get: Γ = Z 0 mµ /2  dΓ dEe  dEe = Z mµ /2 0 dΓ dEνµ ! dEνµ = Z 0 mµ /2  dΓ dEν e  dEν e (9.3. and ν e tell us that the total spin of the final state is 3/2: νe µ− ←− ←− e ←− νµ Since the initial-state muon only had spin 1/2. L. The total decay rate for the muon is found by integrating either eq.48): GF = 1. (9.166364(5) × 10−5 GeV −2 .21) describes several other processes besides the decay µ− → e− ν e νµ that we studied in subsection 9.2. Unlike the distributions for Ee and Eν .49) so we obtain the numerical value of Fermi’s constant from eq. (9.44) with respect to Eνµ .48) 192π 3 It is a good check that the final result does not depend on the choice of the final energy integration = variable. (9. mµ = 0. (9. (9.1056584 GeV.40) with respect to Ee .46) with respect to Eν e . the ν e must be recoiling against both e and νe moving in the opposite direction.2.) The 4-fermion weak interaction Lagrangian of eq. Experiments tell us that Γ(µ− → e− νµ ν e ) = 2.47) G2F m5µ . In each case. νµ . the quantum states have 0 overlap. As the simplest example.99591(3) × 10−19 GeV.2.2.2.2. we can just replace each particle in the process by its anti-particle: µ+ → e+ νe ν µ .51) for which the Feynman diagram is just obtained by changing all of the arrow directions: e+ νe µ+ νµ 147 .

are left unchanged after one performs the combined operations of: • charge conjugation (C): replacing each particle by its antiparticle.2.2. X spins |M|2 = 128G2F (p′a · k2′ )(k1′ · k3′ ). This is actually a special case of a general symmetry relation between particles and anti-particles. the CPT Theorem implies that the mass and the total decay rate of a particle must each be equal to the same quantities for the corresponding anti-particle. It turns out to be impossible to write down any theory that fails to obey this rule. and is known as the CPT Theorem.c. let us label the 4-momenta for this process as follows: Particle Momentum µ+ p′a + e k1′ νe k2′ νµ k3′ Spinor va v1 u2 v3 . The statement of the theorem is that the laws of physics. as specified by the Lagrangian. and 9. which holds true in any local quantum field theory.53) As one might expect. as long as the Lagrangian is invariant under proper Lorentz transformations and contains a finite number of spacetime derivatives and obeys some other technical assumptions. of course. • time reversal (T): replacing t → −t. 148 . following from the “+c.54) is exactly the same as obtained in eq.2.” term in eq. The differential and total decay rates that follow from this are.2. is then √ M = −i2 2GF (v a γ ρ PL v3 )(u2 γρ PL v1 ).) We will study some other processes implied by the Fermi weak interaction Lagrangians in subsections 9. (It does not say that the differential decay rate to a particular final state configuration necessarily has to be equal to the anti-particle differential decay rate to the same configuration of final-state anti-particles. with the obvious substitution of primed 4-momenta. • parity (P): replacing ~x → −~x. (9. The four-fermion Fermi interaction for leptons does respect invariance under T. (9. (9.2.The evaluations of the reduced matrix element and the differential and total decay rates for this decay are very similar to those for the µ− → e− ν e νµ . For future reference. 9.52) The reduced matrix element. (9. (9. that stronger result holds only if the theory is invariant under T.32). but it is violated by a tiny amount in the weak interactions of quarks.5.6 below. the result for the spin-summed squared matrix element.4.21). Among other things. exactly the same as for µ− decay.

which must be “regularized” by using a high-energy cutoff. we derived the µ− decay rate in terms of Fermi’s four-fermion weak interaction coupling constant GF . there are loop diagrams involving virtual photons: e− µ− e− µ− νe νe νµ νµ e− µ− νe νµ Evaluating these diagrams is beyond the scope of this course. there is the dependence on me .3. Taking this into account using correct kinematics for me 6= 0 and the limits of integration in eq. These contributions to the reduced matrix element actually involve divergent loop integrals. one finds that the decay rate must be multiplied by a correction factor Fkin (m2e /m2µ ). which can then be absorbed into a redefinition of the mass and coupling parameters of the model.3 Corrections to muon decay In the previous subsection. There is then a logarithmic dependence on the cutoff energy.999813. There are also corrections coming from two types of QED effects. Fkin (m2e /m2µ ) = 0.1) Numerically.4.9). (9. it is worthwhile to note the leading corrections to it. First. However. (8.9. but could have included at the cost of a more complicated phase space integration. Since this decay process is actually the one that is used to experimentally determine GF most accurately. it should be clear that they give contributions to the reduced matrix element proportional to e2 GF . where Fkin (x) = 1 − 8x + 8x3 − x4 − 12x2 lnx. There are also QED contributions from diagrams with additional photons in the final state: 149 .4.7)-(8. The interference of the loop diagrams with the original lowest-order diagram then gives a contribution to the decay rate proportional to αG2F . First. by the systematic process of renormalization. which we neglected. since each contains two photon interaction vertices.

summing over final spins.995802.68. Therefore.3.3. 150 . the QED coupling α actually is dependent on the energy scale. After squaring.000001.5) The dominant remaining uncertainty in GF quoted in eq.γ γ e− e− µ− µ− νe νe νµ νµ The QED diagrams involving an additional photon contribute to a 4-body final state.50) comes from the experimental input of the muon lifetime. there are corrections involving the fact that the point-like four-fermion interaction is actually due to the effect of a virtual W − boson. and averaging over the initial spin.2) where the C2 contribution refers to even higher-order corrections from: the interference between Feynman diagrams with two virtual photons and the original Feynman diagram. and should be evaluated at the energy scale of interest for this problem. the square of the reduced matrix element for a Feynman diagram involving one virtual photon. This gives a correction factor FW = 1 + 3m2µ 2 ≈ 1.3) Finally. .3. (9. one usually just combines the two types of QED contributions into a total inclusive decay rate with one or more extra photons in the final state.3. one finds that the QED effect on the total decay rate is to multiply by a correction factor α FQED (α) = 1 + π !  m i m2 h 25 π 2 e − − 2e 9 + 4π 2 + 24 ln + 8 2 mµ mµ  2 α π C2 + . 5MW (9. (9. the contribution to the decay rate is again proportional to αG2F .9. which is naturally mµ . with a reduced matrix element proportional to eGF .4) using MW = 80. The predicted decay rate defining GF experimentally including all these higherorder effects is Γµ− = G2F m5µ Fkin FQED FW . the interference between Feynman diagrams with one virtual photons plus one final state photon and the original Feynman diagram. After a heroic calculation. and integrating over the 4-body phase space. 192π 3 (9. . which are difficult or impossible to resolve experimentally. A complicated calculation shows that C2 ≈ 6. Because of the renormalization procedure. (9. and two photons in the final state. At that scale. α ≈ 1/135.4 GeV. Much of this contribution actually comes from very soft (low-energy) photons. so numerically FQED (α) ≈ 0.2.

4. p′a µ− . and in the center-of-momentum frame.4.5.32).4. (9. k1′ = −pa . k1 .4. 151 (9.54).7) .28)-(7. resulting in: X spins |Me− νµ →νeµ− |2 = 128G2F (k2 · k1 )(pa · pb ).5) in eq. k3′ = −pb (9. k2 to e− . this scattering process is often known as inverse muon decay. k2′ = k1 . (9. νµ . νe . To apply the Crossing Symmetry Theorem stated in section 7. k2 e+ . In that case.4. pb . and X spins |Me− νµ →νe µ− |2 = 32G2F s2 .4.4 Inverse muon decay (e− νµ → νe µ− ) Consider the process of muon-neutrino scattering on an electron: e− νµ → νe µ− . pa νe . p b µ− .3) final ν µ → initial νµ . so that mµ can be neglected. k1 ν µ . we can assign momentum labels pa . and then multiplying by (−1)3 for three crossed fermions.4) In fact. as shown in the figure.2) final e+ → initial e− (9. pa νe . k1 νµ . k2 in which we see that the following particles have been crossed from the previous diagram for µ+ decay: initial µ+ → final µ− (9. (7. p b Therefore.4. (9. µ− respectively.6. so that pa · pb = k1 · k2 = s/2. we obtain the spin-summed.5.1) The Feynman diagram for this process is: e− . k3′ νµ . and then make the following comparison table: µ+ → e+ νe ν µ e− νµ → νe µ− µ+ .6) Let us evaluate this result in the limit of high-energy scattering.9. we can take the kinematics results from eqs. k1′ e− .2. all four particles being treated as massless. k2′ νe . (9. squared matrix element for e− νµ → νe µ− by making the replacements p′a = −k2 .

This is a general feature.7 × 10   Eνµ fb .4. and therefore necessarily isotropic. which are in turn produced by decaying pions.9 fb √ !2 s . The muon neutrinos might be produced from a beam of decaying µ− . they are all L helicity. GeV (9. (9.4.10) In a typical experimental setup. The isotropy of e− νµ → νe µ− scattering in the center-of-momentum frame can be understood from considering what the helicities dictated by the weak interactions tell us about the angular momentum. 152 .12) This is a very small cross-section for typical neutrino energies encountered in present experiments. as discussed later. dσ G2 s = F . µ− e− θ νµ νe We therefore see that the initial and final states both have total spin 0. † In the Standard Model with neutrino masses neglected. If we call the νµ energy in the lab frame Eνµ .4. we can evaluate this using eq.10) gives: −2 σe− νµ →νe µ− = 1.9) Numerically.4. the electrons will be contained in a target of ordinary material at rest in the lab frame. so it is trivial to integrate to get the total cross-section: G2F s . then the center-of-momentum energy is given by √ s = ECM = q 2Eνµ me + m2e ≈ q 2Eνµ me . and all antineutrinos are righthanded. so that the process is s-wave. GeV (9.8) This differential cross-section is isotropic (independent of θ). Since there is only one possible νµ helicity.2. (9. d(cos θ) 2π (9.2. it would be incorrect to average over the νµ spin.4. as long as they are being treated as massless.50): σe− νµ →νe µ− = 16.Including a factor of 1/2 for the average over the initial-state electron spin. one should never average over initial-state neutrino or antineutrino spins.50).† and using eq. Since this is a weak interaction process involving only fermions and not anti-fermions. all neutrinos are left-handed.11) Substituting this into eq. but it does grow with Eνµ . π σe− νµ →νe µ− = R1 −1 d(cos θ) =2 (9. (9. namely L.4. (6.

The result this time is: X spins |M|2 = 128G2F (k1 · pb )(pa · k2 ) = 32G2F u2 = 8G2F s2 (1 + cos θ)2 . we obtain: dσe− ν e →µ− ν µ G2 s = F (1 + cos θ)2 . (9.1) This process can again be obtained by crossing µ+ → e+ νe ν µ according to initial µ+ → final µ− (9.5. 3π (9. (9. Substituting this result into eq.2.5. pb ν µ . (7.5. with a factor of 1/2 to account for averaging over the initial e− spin. k1′ = −pa .8) This calculation shows that in the center-of-momentum frame.5.32) for 2→2 massless kinematics have been used.5) in eq.9. consider the process of antineutrino-electron scattering: e− ν e → µ− ν µ . Here θ is the angle between the incoming e− and the outgoing µ− 3-momenta. pa µ− . k3′ = k2 (9.54).5.30) and (7. (6. k2 Therefore.5.5. k1 ν e .7) Performing the d(cos θ) integration gives a total cross-section of: σe− ν e →µ− ν µ = G2F s . we obtain the spin-summed squared matrix element for e− ν e → µ− ν µ by making the substitutions p′a = −k1 . (9. d(cos θ) 8π (9.6) where eqs.5.2.5. k2′ = −pb .5 e− ν e → µ− ν µ As another example.50).3) final νe → initial ν e .5.2) final e+ → initial e− (9. (9.4) as can be seen from the Feynman diagram: e− . and multiplying again by (−1)3 because of the three crossed fermions. the µ− tends to keep going in the same direction as the original e− . This can be understood from the helicity-spin-momentum diagram: 153 .

µ− e− νe θ νµ Since the helicities of e− . The total cross-section for this reaction is 1/3 of that for the reaction e− νe → µ− νµ . and is given by: ′ ′ Jρ+ = (Jρ− )∗ = eγρ PL νe + µγρ PL νµ + τ γρ PL ντ + d γ ρ PL u + s′ γ ρ PL c + b γ ρ PL t. This is because the former reaction is an isotropic s-wave (angular momentum 0). which can only use one of the three possible J = 1 final states. (9. so one must correct for this when interpreting the distribution in the lab frame. The overlap between these two states is therefore maximized when the e− and µ− momenta are parallel. direct testing of the (V − A)(V − A) form of the weak interaction Lagrangian for quarks. µ− . ν µ are respectively L. and all barred fields are right-handed: Jρ− = ν e γρ PL e + ν µ γρ PL µ + ν τ γρ PL τ + uγρ PL d′ + cγρ PL s′ + tγρ PL b′ . (9. ν e .6. the total spin of the initial state must be pointing in the direction opposite to the e− 3-momentum. there is an obstacle to a detailed. while the latter is a p-wave (angular momentum 1). the one with J~ pointing along the ν µ direction. L.1) The weak-interaction charged current is obtained by adding together terms for pairs of fermions. and b′ appearing here are actually not quite mass eigenstates. 9. and the total spin of the final state must be pointing opposite to the µ− direction. Of course. because of mixing. This is because the quarks are bound into hadrons by strong interactions. (9. this reaction usually occurs in a laboratory frame in which the initial e− was at rest. More generally. and are V − A fermion bilinears. and vanishes when the µ− tries to come out in the opposite direction to the e− . with the constraint that the total charge of the current is −1. The complex conjugate of Jρ− has charge +1.6.2) Notice that we have included contributions for the quarks. this is the reason for the primes.6 Charged currents and π ± decay The interaction Lagrangian term responsible for muon decay and for the cross-sections discussed above is just one term in the weak-interaction Lagrangian. and all Dirac fermion fields involved in the current are left-handed.6. The quark fields d′ . R. we can write the Lagrangian as a product of a weak-interaction charged current Jρ− and its complex conjugate Jρ+ : √ Lint = −2 2GF Jρ+ J −ρ .3) Both J −ρ and J +ρ transform under proper Lorentz transformations as four-vectors. R. so that cross-sections and decays involving the weak interactions are subject to very large 154 . namely. s′ . Unfortunately.

In principle. is just a cartoon for the strong interactions. fπ is an experimentally measured quantity.) So we are simply parameterizing all of our ignorance of the bound-state properties of the pion in terms of a single constant fπ . by considering what we do know about the current-current Lagrangian. In principle we could compute it if we had perfect ability to calculate with the strong interactions. The point is that whatever the pion factor in the reduced matrix element is. (Recall that pions are spinless. we can write: √ M = −i 2GF fπ pρ (u1 γ ρ PL v2 ). The factor fπ pρ takes into account the part of the reduced matrix element involving the π − . so there is no spin dependence. called the pion decay constant. A π − consists of a bound state made from a valence d quark and u antiquark. one can still do a quantitative analysis of some aspects of weak decays involving hadrons. the factor (u1 γ ρ PL v2 ) just reflects the fact that leptons are immune from the complications of the strong interactions. However. we know that it is a four-vector in order to contract with the lepton part. we know that the external state spinors for the fermions are: P article M omentum Spinor µ− k1 u1 νµ k2 v2 (9. A Feynman diagram for π − → µ− ν µ decay following from the current-current Lagrangian might therefore look like: µ− .6. here pρ is the 4-momentum of the pion.5) In this formula.and very complicated corrections. the left-part of this Feynman diagram. k2 Unfortunately. k1 d π− u ν µ . First. with its value following most accurately from the π − lifetime that 155 . which are not amenable to perturbative calculation. To see how this works. by a method of parameterizing our ignorance. we can write down the general form of the reduced matrix element. The u antiquark and d quark do not even have fixed momenta in this diagram. together with many virtual gluons and quark-antiquark pairs.4) In terms of these spinors. and it must be proportional to pρ . since there is no other vector quantity in the problem that it can depend on. In practice. one can find some distribution for the u and d momenta. It is a quantity with dimensions of mass. and try to average over that distribution. but the strong interactions are very complicated so this is not very easy to do from the theoretical side. (9. However. involving the complications of the π − bound state.6. consider the process of charged pion decay. since they exchange energy and momentum with each other and with the virtual gluons and quark-anti-quark pairs.

11) (9. we have: |M|2 = 2G2F fπ2 pρ pσ (u1 γ ρ PL v2 )(v 2 PR γ σ u1 ).9) (9. we have: G2 f 2 m ± m2e Γ(π → e ν e ) = F π π 8π − − 156 m2 1 − 2e mπ ± !2 .6. (9. p · k1 = 2 2 1 1 2 2 2 p · k2 = [−(p − k2 ) + p + k2 ] = (−m2µ + m2π± ).6) Now summing over final state spins in the usual way gives: X spins k 1 + mµ )] = 2G2F fπ2 Tr[/pk/2 PR /pk/1 ].6. (9.6. except that me is substituted everywhere for mµ .16) .6. k22 = mν2 µ = 0. we get: G2 fπ2 mπ± m2µ dΓ = F 32π 2 m2µ 1− 2 mπ ± !2 dφ d(cos θ). 2 2 1 1 k1 · k2 = [(k1 + k2 )2 − k12 − k22 ] = (m2π± − m2µ ).7) that explicitly involves mµ does not contribute. Evaluating the trace. (8. (9.6.2.6.14) where (θ.756 is not a small number. (9. But it is useful to have the GF appear explicitly as a sign that this is a weak interaction.1396 GeV = 0.15) The charged pion can also decay according to π − → e− ν e . However.14). p2 = m2π± . then fπ is entirely a strong-interaction parameter. The factor √ 2GF is another historical convention.6. it could have been absorbed into the definition of fπ .6.6.6.we will compute below. Let us now compute the decay rate for π − → µ− ν µ .7) Note that we do not neglect the mass of the muon.10) (9. 1 1 [−(p − k1 )2 + p2 + k12 ] = (m2π± + m2µ ).6.5). and X spins 2 |M| = m2µ 1− 2 mπ ± 2G2F fπ2 m2π± m2µ ! .6. and: G2 fπ2 mπ± m2µ Γ(π → µ ν µ ) = F 8π − − m2µ 1− 2 mπ ± !2 .6. |M|2 = 2G2F fπ2 pρ pσ Tr[γ ρ PL k/2 PR γ σ (/ (9. we have: Tr[/ pk/2 PR p/k/1 ] = 4(p · k1 )(p · k2 ) − 2p2 (k1 · k2 ).12) (9.1056 GeV/0. (9. Therefore. since the pion is spinless.13) so that using eq. so the angular integration trivially gives dφd(cos θ) → 4π. Of course. the differential decay rate is isotropic. 2 2 Therefore. (9.8) The decay kinematics tells us that: k12 = m2µ . since mµ /mπ± = 0. The calculation of this decay rate is identical to the one just given. Tr[p /k/2 PR p /k/1 ] = m2µ (m2π± − m2µ ). (9. Taking the complex square of the reduced matrix element eq. since the trace of 3 gamma matrices (with or without a PR ) vanishes. the term in eq. (9. φ) are the angles for the µ− three-momentum.

2 × 10−4 .6.17) The dependence on fπ has canceled out of the ratio eq. Although the pion is a composite. with a rare decay to e− ν e occurring 0. The helicity suppression of this decay is therefore a good prediction of the rule that the weak interactions affect only L fermions and R antifermions. let us find a Lagrangian that would give rise to it involving a quantum field for the pion. bound-state particle. helicity conservation only holds in the high-energy limit in which we can treat all fermions as massless. we can draw a momentum-helicity-spin diagram.6. The result is: fπ = 0. a helicity flip for a fermion entails a suppression in the reduced matrix element proportional to the mass of the fermion divided by its energy. using the fact that the ℓ− and ν ℓ produced in the weak interactions are L and R respectively: −→ e J =0 π− −→ νe The π − has spin 0. The most striking feature of the π − decay rate is that it is proportional to m2µ . if helicity were exactly conserved. Previously. This is what leads to the strong suppression of decays to e− ν e (already mentioned at the end of section 9. meaning that the L-helicity fermion produced by the weak interactions has an amplitude to appear in the final state as a R-fermion. the charged lepton µ− or e− is said to undergo a helicity flip. which is therefore a robust prediction of the theory.1).5). but the final state predicted by the weak interaction helicities unambiguously has spin 1.6.17). (9. We can also use the measurement of the total lifetime of the π − to find fπ numerically. Having computed the decay rate following from eq. Therefore. using eq.18) It is not surprising that this value is of the same order-of-magnitude as the mass of the pion.6. since the only reason it can occur is because mµ and me are non-zero.and the ratio of branching ratios is predicted to be: Γ(π − → e− ν e ) BR(π − → e− ν e ) = = BR(π − → µ− ν µ ) Γ(π − → µ− ν µ ) m2e m2µ ! m2π± − m2e m2π± − m2µ !2 = 1. (9. we can still invent a quantum field for it. In the final state.128 GeV.012% of the time. This has been confirmed experimentally.15). (9. we recover exact helicity conservation and the reduced matrix element and the decay lifetime vanish. (9. We found this result just by calculating. in an approximate. the π − could not decay at all! However. This decay is said to be helicity-suppressed. with M proportional to mµ .6. Since there are no other kinematically-possible two-body decay channels open to π − . To understand it better. This explains why they should be proportional to mℓ and m2ℓ respectively. The π − corresponds to a charged spin-0 field. “effective” description. In general. (9. it should decay to µ− ν µ almost always. In the limit mℓ → 0. we studied spin-0 particles described by a real 157 .

1.19) We can construct a real free Lagrangian density from these complex fields as follows: L = ∂µ π + ∂ µ π − − m2π± π + π − .scalar field.24) One can now carry through canonical quantization as usual. ap†~. with an arrow to the right meaning a π + and an arrow to the left meaning a π − .21) d˜ p (ei~p·~x ap~. (9. Similarly.+ are taken to be independent.23) (9.+ ). and vice versa.− |0i = |π − . We can summarize this with the following mnemonic figures: 158 . The operators ap~. p~ i. In particular.6.− and ap~. it represents a final state pion. the single particle states are: ap†~.+ act on states by destroying and creating a π + particle with 3-momentum p~. This means that the π − particle should be described by a complex scalar field. However.− act on states by destroying and creating a π − particle with 3-momentum p~. Since a π − moving forward in time is a π + moving backwards in time. since the π − is charged.6. Likewise. the particle and antiparticle created by a real scalar field turned out to be the same thing. (9. if a pion line leaves the diagram to the right. the operators ap~. the π + and π − fields can be expanded in creation and annihilation operators as: π − (~x) = π + (~x) = Z Z d˜ p (ei~p·~x ap~. At any fixed time t = 0.6. eq.− and ap†~. and that they are each complex since ap~. one can derive the corresponding Feynman rules for the propagator and interaction vertices. (9.] This Lagrangian density describes free pion fields with mass mπ± . there is only one propagator for π ± fields.6. while a line entering from the left with an arrow pointing back to the left means a π − particle in the initial state. its antiparticle π + is clearly a different particle. A pion line entering from the left with an arrow pointing to the right means a π + particle in the initial state. (9. (5. It differs from the propagator for an ordinary scalar in that it carries an arrow indicating the direction of the flow of charge: ←→ i p2 − m2π± + iǫ The external state pion lines also carry an arrow direction telling us whether it is a π − or a π + particle.+ + e−i~p·~x ap†~.22) Note that these fields are indeed complex conjugates of each other.6.− ).− + e−i~p·~x ap†~.20) [Compare to the Lagrangian density for a real scalar field.+ and ap†~. Let us therefore define π − (x) to be a complex scalar field. (9. p~ i.18). Here we want something different. with its complex conjugate given by π + (x) ≡ (π − (x))∗ .+ |0i = |π + . Given an interaction Lagrangian.6.

28) .π− µνµ = − 2GF (µγ ρ PL νµ ) f (−∂ 2 )∂ρ π − . (9. When we computed the decay matrix element. The momentum-space factor f (p2 )pρ can be interpreted by identifying the 4-momentum as a differential operator acting on the pion field. one usually just works in momentum space where −∂ 2 = ∇ it is f (p2 ).initial state π + : initial state π − : final state π + : final state π − : In each case the Feynman rule factor associated with the initial. Returning to the reduced matrix element of eq.25) when the pion is on-shell. we must also include the complex conjugate of this term: Lint.6. (9.6.6.6.or final-state pion is just 1. (9. we conclude that the effective interaction describing π − decay is: √ Lint. the pion was on-shell. The pion decay constant fπ must therefore be generalized to a function f (p2 ). but in general this need not be the case.5).6. a π− ←→ νµ . we can interpret this as coming from a pion-lepton-antineutrino interaction vertex. with f (p2 )|p2 =m2 π± = fπ (9.27) Here f (−∂ 2 ) can in principle be defined in terms of its power-series expansion in the differential operator ~ 2 − ∂t2 acting on the pion field. using: pρ ↔ i∂ ρ . The Feynman rules for these effective interactions are: µ.π+ µν µ √ = − 2GF (ν µ γ ρ PL µ) f (−∂ 2 )∂ρ π + . b 159 √ −i 2GF f (p2 )pρ (γ ρ PL )ab (9. In practice. Since the Lagrangian must be real.26) Then reversing the usual procedure of inferring the Feynman rule from a term in the interaction Lagrangian.

like the K ± . and (5.1. this is equal to units of [mass]. we find that these types of fields must have dimensions of [mass]. Other charged mesons made out of a quark and antiquark. then our perturbative approximation e−iHt = 1−iHt represented by the lowest-order Feynman diagram must break down.1. 9. [mass]3/2 . and cross-sections must have dimensions of [mass]−2 . D ± . b In each Feynman diagram. it must be that the cross-section scales like the square of the characteristic energy of the process. or [mass]−3 .18). This behavior of σ ∝ s is not acceptable for arbitrarily large s. and the 4-momentum pρ is taken to be flowing in to the vertex. Therefore. and L = Z d3 ~x L. or by changing the physics of the weak interactions at some higher energy scale. This had to be true on general grounds just from dimensional analysis. We know that the Lagrangian must have the same units as energy. since the cross-section is bounded by the fact that the probability for any two particles to scatter cannot exceed 1. (5. When the pion is on-shell. the constraint is on the unitarity of the time-evolution operator e−iHt . in the highenergy limit in which all other kinematic mass scales are comparatively unimportant. fermions. and the W boson An important feature of weak-interaction 2 → 2 cross-sections following from Fermi’s four-fermion interaction is that they grow proportional to s for very large s. (9.7. have their own decay constants fK . renormalizability. Any reduced matrix element that contains one four-fermion interaction will be proportional to GF . fD . so the cross-section will have to be proportional to G2F .9) and (9.1) it must be that L has units of [mass]4 . (5. and vector fields found for example in eqs. see eqs. In quantum mechanical language. and fDs . Let us develop the dimensional analysis of fields and couplings further. from the kinetic terms for scalars. a spacetime derivative has units of inverse length. For example. and [mass] 160 . the arrow on the pion line describes the direction of flow of charge.26). or [mass].4. and Ds± . If the cross-section grows too large. Since d3 ~x has units of [length]3 .5.7 Unitarity. This fact allows us to evaluate the units of all fields and couplings in a theory. (9.1. In the standard system in which c = h ¯ = 1. and their decays can be treated in a similar way.34). s. a π+ ←→ √ −i 2GF f (p2 )pρ (γ ρ PL )ab µ. one can replace f (p2 ) by the pion decay constant fπ . The reduced matrix element found from just including this Feynman diagram will have to be compensated somehow by higher-order diagrams.and ν µ.8). Since this has units of [mass]−4 .

The effective coupling fπ for on-shell pions has dimensions of [mass]. or λn for n ≥ 5. like GF . because of the presence of a spacetime derivative together with a scalar field and two fermion fields in the Lagrangian. It is a necessary. is dimensionless. It is a general fact that theories with couplings with negative mass dimension.7. each of which must be redefined in order to absorb the momentum-cutoff dependence. Lint = − λn n φ n! (9.† In a renormalizable theory. in a non-renormalizable theory. For example. a coupling of n scalar fields. In contrast. v. condition. just because a theory has only couplings with positive or zero mass dimension does not guarantee that it is renormalizable. 161 . We can always use non-renormalizable theories as effective theories at low energies. u. This allows us to evaluate the units of various possible interaction couplings that appear in the Lagrangian density. v [mass]1/2 external-state spinors MNi →Nf [mass]4−Ni −Nf reduced matrix element for Ni → Nf particles σ [mass] Γ [mass] −2 fermion field cross-section decay rate. but not sufficient. although only in principle. Summarizing this information for the known types of fields and couplings that we have encountered so far: Object Dimension L [mass] ∂µ [mass] φ [mass] 4 Role Lagrangian density derivative scalar field 3/2 Ψ [mass] Aµ [mass] vector field λ3 [mass] scalar3 coupling λ4 [mass]0 scalar4 coupling y [mass]0 scalar-fermion-fermion (Yukawa) coupling e [mass] 0 photon-fermion-fermion coupling GF [mass] −2 fermion4 coupling fπ [mass] fermion2 -scalar-derivative coupling u. the divergences that occur in loop diagrams due to integrating over arbitrarily large 4-momenta for virtual particles can be regularized by introducing a cutoff. always suffer from a problem known as non-renormalizability. A vector-fermion-fermion coupling. This dependence on an infinite number of different coupling constants makes non-renormalizable theories non-predictive.respectively. as we have done in the case of the † The converse is not true. and then the resulting dependence on the unknown cutoff can be absorbed into a redefinition of the masses and coupling constants of the theory.2) implies that λn has units of [mass]4−n . one finds that this process requires introducing an infinite number of different couplings. like e in QED.

this coupling has negative mass dimension. we imagine “pulling apart” the two currents. However.3) . a nonrenormalizable theory will encounter related problems associated with the apparent failure of unitarity (cross-sections that grow uncontrollably with energy) and non-renormalizability (an uncontrollable dependence on more and more unknown couplings that become more and more important at higher energies). This is not a problem as long as we stick to scattering energies ≪ MPlanck .4 × 1018 GeV is the “reduced Planck mass”. This is the W − vector boson. where MPlanck = 2. W ± are complex vector fields. b νe . For example. |pρ | ≪ mW . which corresponds to all known experiments and directly measured phenomena. we do not know how unitarity is restored in gravitational interactions at energies comparable to MPlanck or higher. one of the current-current terms is: µ.four-fermion theory of the weak interactions. Unlike the case of the weak interactions. However. a e. The effective coupling constant for 2 Feynman diagrams involving gravitons is 1/MPlanck . and replacing the short line segment by the propagator for a virtual vector particle. a µ. so tree-level cross-sections for 2 → 2 scattering involving gravitons grow with energy like s. An example of a useful non-renormalizable theory is gravity. m2W (9. d νe . To do this. we note that the (V − A)(V − A) current- current structure of the four-fermion coupling could come about from the exchange of a heavy vector particle. By analogy with the charged pion. this propagator just becomes a constant: " pρ pσ i −gρσ + 2 2 2 p − mW + iǫ mW 162 # −→ i gρσ . when probed at sufficiently high energies. it is hard to conceive of an experiment with present technologies that could test competing ideas. For this reason. Like GF . c W ←→ ν µ. d Since the currents involved have electric charges −1 and +1. with an arrow on its propagator indicating the direction of flow of charge. To find a renormalizable “fix” for the weak interactions. the vector boson must carry charge −1 to the right. b ν µ. we are happier to describe physical phenomena using renormalizable theories if we can. c e.7. The Feynman rule for the propagator of a charged vector W ± boson carrying 4-momentum p turns out to be: ρ σ " pρ pσ i −gρσ + 2 2 2 p − mW + iǫ mW ←→ # In the limit of low energies and momenta.

4 GeV.65. leading to cross-sections that fall. there is at least a chance to make this into a renormalizable theory that is unitary in perturbation theory. we find that we must have:   √ −ig 2 −i2 2GF = √ 2 i m2W ! . These are the Feynman rules involving W ± interactions with leptons.In order to complete the correspondence between the effective four-fermion interaction and the more fundamental version involving the vector boson. a W. we need W -fermion-antifermion vertex Feynman rules of the form: ℓ. b √ Here g is a fundamental coupling of the weak interactions. The interaction Lagrangian for W bosons with standard model fermions corresponding to these Feynman rules is:  g  Lint = − √ W +ρ Jρ− + W −ρ Jρ+ . 4 2mW (9. and the 1/ 2 is a standard convention.4) Comparing the four-fermion vertex to the reduced matrix element from W -boson exchange. 2 (9.7. (9. rather than rise.2) and (9. ρ ℓ.7. there are similar rules for interactions with the quarks in the charged currents Jρ+ and Jρ− given earlier in eqs. (9. with a mass mW = 80.7) Since this is a dimensionless coupling.7. the W ± propagator will behave like 1/p2 . at very high √ s. (9. 163 . At very high energies.6.6. so we conclude that g ≈ 0.7. This “softens” the weak interactions at high energies. b ν ℓ.6) The W ± boson has been discovered. rather than the 1/m2W that is encoded in GF in the four-fermion approximation. ρ ←→ g −i √ (γ ρ PL )ab 2 ←→ g −i √ (γ ρ PL )ab 2 νℓ .5) so that g2 GF = √ 2 .3). a W.

λ)ǫ∗σ (p.When a massive vector boson appears in a final state. The weak interactions and the strong interactions are invariant under non-Abelian gauge transformations. The difference is that a massive vector particle V has three physical polarization states λ = 1. λ) = −gρσ + pρ pσ . This means that the gauge transformations not only multiply fields by phases. m2V (9. (9. like the W ± bosons. satisfying pρ ǫρ (p. with the result: 3 X λ=1 ǫρ (p. then an arrow is added to each line to show the direction of flow of charge. λ). 3. just like the photon did.7. 3).7. λ) ←→ ǫ∗µ (p. λ) If the massive vector is charged. which involve a generalization of the type of gauge invariance we have already encountered in the case of QED. λ) = 0 (λ = 1. known as Yang-Mills theories.8) One can sum over these polarizations for an initial or final state in a squared reduced matrix element. but can mix the fields. which have a non-Abelian gauge invariance. it has a Feynman rule given by a polarization vector ǫµ (p. 2. 2. This will enable us to get a complete theory of the weak interactions. In the next section we will begin to study the properties of field theories.9) Summarizing the propagator and external state Feynman rules for a generic massive vector boson for future reference: ρ σ ←→ " i pρ pσ −gρσ + 2 p2 − m2V + iǫ mV µ initial state vector: µ final state vector: # ←→ ǫµ (p. 164 .

Otherwise it is non-commutative or non-Abelian.1. called a gauge parameter. . e ′ iQθ Ψ → Ψ = e Ψ. . these gauge transformations are an example of a group. (10.1.1. 2) Associativity: gi (gj gk ) = (gi gj )gk . Now.7) Mathematically. 3) Existence of an Identity: There is a unique element I = g1 of the group. e e e Ψ → eiQθ2 (eiQθ1 Ψ) = eiQ(θ1 +θ2 ) Ψ. the result of doing one gauge transformation θ1 followed by another gauge transformation θ2 is always a third gauge transformation parameterized by the function θ1 + θ2 : 1 1 1 Aµ → (Aµ − ∂µ θ1 ) − ∂µ θ2 = Aµ − ∂µ (θ1 + θ2 ).1. (10.1.3) and a field strength as This Lagrangian is invariant under the local gauge transformation 1 Aµ → A′µ = Aµ − ∂µ θ. we will seek to generalize the idea of gauge invariance.1) Fµν = ∂µ Aν − ∂ν Aµ (10. with the properties: 1) Closure: If gi and gj are elements of the group G.1. then the group is commutative or Abelian. . g2 . L = − F µν Fµν + iΨDΨ 4 (10. Igi = gi I = gi . It may or may not be also true that the group also satisfies the commutativity property: gi gj = gj gi .8) If this is satisfied.1.1.6) (10.10 Gauge theories 10. then the product gi gj is also an element of G.1 Groups and representations In this section. 4) Inversion: For each gi . A group is a set of elements G = (g1 . such that for all gi in G. Recall that in QED the Lagrangian is defined in terms of a covariant derivative Dµ = ∂µ + iQeAµ (10.4) (10.) and a rule for multiplying them. there is a unique inverse element (gi )−1 satisfying (gi )−1 gi = gi (gi )−1 = I. 165 .2) 1 / − mΨΨ.5) where θ(x) is any function of spacetime. (10.

we will be interested in continuous Lie groups.9) Here θ labels the group elements. .In trying to generalize the QED Lagrangian. The matrices obey the same rules as the group elements themselves. it does not really transform under gauge transformations as any representation of the group U (1). A barred Dirac field transforms with the opposite phase from the original Dirac field of charge Q.1. The photon field Aµ has charge 0. For example.10) where Ui j is a representation matrix. dR . which always involve representations containing more than one field or state. muon. A Lie group is a continuous group that also has the desirable property of being differentiable with respect to the gauge parameters. and top quarks each live in a representation with charge Q = 2/3. so that i = 1. which turn out to be nothing other than the generalizations of the gauge parameter θ in QED.1. For example. (10. dR . . strange and bottom quarks each live in a representation with Q = −1/3. and tau Dirac fields each live in a representation of the group U (1) with charge Q = −1.5). unless θ is a constant function so that one is making the same transformation everywhere in spacetime. so that their action can be realized on the quantum Hilbert space by a unitary operator. . Objects that transform into themselves with no change are said to be in the singlet representation. the group of QED gauge transformations is the Abelian Lie group U (1). We can read off the charge of any field if we know how it transforms under the gauge group. Consider the subset of group elements that are infinitesimally close to the identity 166 . Let ϕi be a set of objects that together transform in some representation R of the group G. In general. This association of group elements with n × n matrices and the states or fields that they act on is said to form a representation of the group. each term of the QED Lagrangian carries no charge. the Lagrangian should be invariant under gauge transformations. and therefore has charge −Q. and so is a singlet of U (1). and is therefore usually said (by a slight abuse of language) to transform as a singlet representation of U (1). Under a group transformation.1. The number of components of ϕi is called the dimension of the representation. and therefore must be in the singlet representation. So we can say that the electron. We are especially interested in transformations that are represented by unitary matrices. because of the derivative term in eq.4). .1. According to eq. The action of group elements on physics states or fields can be represented by a set of complex n × n matrices acting on n-dimensional complex vectors. ϕi → ϕ′i = Ui j ϕj (10. charm.] Let us now generalize to non-Abelian groups. the Dirac fields for up. (10. the group is represented on Dirac fermion fields by complex 1 × 1 matrices: UQ (θ) = eiQθ . and the Dirac fields for down. [Technically. A continuous group has an uncountably infinite number of elements labeled by one or more continuously varying parameters. (10. and the charge Q labels the representation of the group.

T b ] = if abc T c (10. if necessary. . and the hermitian matrices T a are said to be generators of the Lie algebra for the corresponding representation.17).1. (10. one often picks a particular representation of matrices T a as the defining or fundamental representation.1. U (ǫ)† = U (ǫ)−1 = 1 − iǫa T a . dG . dG . For any given representation R. (10. By the closure property we can then form a new group element gǫ gδ gǫ−1 gδ−1 .1.) From eqs.1. (10. TRb ]TRc ). (10.13). A standard choice is that the index of the fundamental representation is 1/2. .1.1. called the structure constants of the group. (This can always be achieved by rescaling the T a . one can always choose the generators so that: Tr(TRa TRb ) = I(R)δab . and there is an implicit sum over a = 1. this corresponds to U (ǫ)U (δ)U (ǫ)−1 U (δ)−1 = (1 + iǫa T a )(1 + iδb T b )(1 − iǫc T c )(1 − iδd T d ) = 1 − ǫa δb [T a . The number of matrices T a is called the dimension of the group.1.element. which fixes their overall normalization.12) from which it follows that the matrices T a must be Hermitian. because we often only care about the subset of gauge transformations that are close to the identity. (10. (10. This determines the structure constants f abc once and for all. .13) Working with some particular representation.15) The closure property requires that this is a representation of the group element in eq.1. .17) The number I(R) is called the index of the representation.16) and (10. It follows that [T a .1. Since U (ǫ) is unitary. .1. Equation (10. Consider two group transformations gǫ and gδ parameterized by ǫa and δa . T b ] + .11) Here the Tiaj are a basis for all the possible infinitesimal group transformations. 167 (10. The ǫa are a set of dG infinitesimal gauge parameters (analogous to θ in QED) that tell us how much of each is included in the transformation represented by U (ǫ).1. (10.1.1. . (10. The set of matrices T a for all other representations are then required to reproduce eq. one obtains for any representation R: iI(R)f abc = Tr([TRa . In practice. which must also be close to the identity.16) defines the Lie algebra corresponding to the Lie group. Physicists have a bad habit of using the words “Lie group” and “Lie algebra” interchangeably.16).18) .16) for some set of numbers f abc . We can write these in the form: U (ǫ)i j = (1 + iǫa T a )i j .14) (10.

dG ). . Heuristically.. .19) which holds for any three matrices.  . .. . called the quadratic Casimir invariant.. . 0 . the dimension. from the cyclic property of the trace. 168 (10. 0 0 .20) Two representations R and R′ are said to be equivalent if there exists some fixed matrix X such that: XTRa X −1 = TRa′ . If we take the sum over a of eq.. [T b . Obviously. T b ]] = 0. T a ]] + [T c . A representation R of a Lie algebra is said to be reducible if it is equivalent to a representation in block-diagonal form.23) A representation that is not equivalent to a direct sum of smaller representations in this way is said to be irreducible. that f abc is totally antisymmetric under interchange of any two of a. where C(R) is another characteristic number of the representation R.1. . ⊕ r n ... [T a .21) for all a.1. (10.. if there is some matrix X that can be used to put all of the TRa simultaneously in a block-diagonal form: XTRa X −1  Tra1  0  = . . and the Casimir invariant are related to the dimension of the group by: dG I(R) = dR C(R). it is equal to the trace of eq.24) (with an implicit sum over a = 1. c. (10.1.25) . With the above conventions on the Lie algebra generators.24)..1. (10. One calls this a direct sum. (10. and writes it as R = r 1 ⊕ r2 ⊕ . b. .1.. in other words. (10. this requires that R and R′ have the same dimension.17). T c ]] + [T b .. (10. .It follows. . . [T a . .22) Here the Trai are representation matrices for smaller representations ri . Tran      for all a. one can show that for each irreducible representation R: (TRa TRa )i j = C(R)δij (10. From a physical point of view. [T c .1. (10. reducible representations are those that can be chopped up into smaller pieces that can be treated individually. It follows that for each irreducible representation R. equivalent representations are indistinguishable from each other.1. 0 0 Tra2 . By using the Jacobi identity.1.1. the index. one also finds the useful result: f ade f bce + f cde f abe + f bde f cae = 0.

20) that the matrices (T a )b c = −if abc (10. Suppose that we have some representation with matrices Tiaj . “positive-real” and “pseudo-real”. ⊕ R n (10. Note that. . with the same dimension as the group G.29) The representation R ⊗ R′ has dimension dR dR′ . and is typically reducible: R ⊗ R ′ = R1 ⊕ . the index of the adjoint representation is equal to its quadratic Casimir invariant: C(G) ≡ C(adjoint) = I(adjoint). R′ of the Lie algebra to get another representation: j. R′ ).1.1.1.1. they can.1.1. One can check from the identity eq. without proof.32) form a representation.33) I now list. (10. As a matter of terminology. .The simplest irreducible representation of any Lie algebra is just: Tiaj = 0.† and otherwise R is said to be complex. the quadratic Casimir invariant of the adjoint representation is also called the Casimir invariant of the group.1. so that there is some fixed matrix X such that XTRa X −1 = TRa .31) with This is a way to make larger representations (R1 . . Then one can show that the matrices −(Tiaj )∗ also form a representation of the algebra eq.y a ≡ (TRa )i j δxy + δij (TRa′ )x y . depending on whether the matrix X can or cannot be chosen to be symmetric. In a pseudo-real representation. and is often denoted R: TRa = −TRa∗ .16). called the adjoint representation.25). the T a cannot all be made antisymmetric and imaginary. (10. from eq.1. One can also form the tensor product of any two representations R.1.x (10. 169 . (10. .26) This is called the singlet representation.1.28) then the representation R is said to be a real representation. (10. (10. (10.30) dR⊗R′ = dR dR′ = dR1 + . (TR⊗R ′) i. (10. + dRn . This is called the complex conjugate of the representation R. in a positive-real representation.1. . .27) If TRa is equivalent to TRa . and given the symbol C(G). Rn out of smaller ones (R. some further group theory facts regarding Lie algebra representations: † Real representations can be divided into two sub-cases. (10.

• The adjoint representation is always real.1.• The number of inequivalent irreducible representations of a Lie group is always infinite.38) • As a corollary of the preceding rules.1. . .1. (10. • If the tensor product of two representations contains a third. . . .40) R1 ⊗ R 2 ⊗ R 3 = 1 ⊕ . .41) 170 .1. • Unlike group element multiplication. ←→ R1 ⊗ R 3 = R 2 ⊕ . . (10. . the tensor product multiplication of representations is both associative and commutative: (R1 ⊗ R2 ) ⊗ R3 = R1 ⊗ (R2 ⊗ R3 ). .34) R1 ⊗ R 2 = R 2 ⊗ R 1 . .39) Here the S and A mean that the indices of the two adjoints on the left are combined symmetrically and antisymmetrically respectively. . . .1. .1. (10.1. (10. . (10. (10. • The tensor product of a representation and its complex conjugate always contains both the singlet and adjoint representations: R ⊗ R = 1 ⊕ Adjoint ⊕ . . then R ⊗ R contains a singlet. .1. ←→ R 1 ⊗ R2 = R 3 ⊕ .36) • The tensor product of two real representations R1 and R2 is always a direct sum of representations that are either real or appear in complex conjugate pairs. (10. the tensor product of the adjoint representation with itself always contains both the singlet and the adjoint: Adjoint ⊗ Adjoint = 1S ⊕ AdjointA ⊕ . • The tensor product of two irreducible representations contains the singlet representation if and only if they are complex conjugates of each other: R1 ⊗ R 2 = 1 ⊕ . .35) • The tensor product of any representation with the singlet representation just gives the original representation back: 1 ⊗ R = R ⊗ 1 = R.37) It follows that if R is real. . (10. ←→ R2 = R 1 . . then the tensor product of the first representation with the conjugate of the third representation contains the conjugate of the second representation: R1 ⊗ R 2 = R 3 ⊕ .

One finds that the structure constants are f abc = ǫabc =    +1 if a. 1 or 3. The Lie algebra generators in the fundamental representation are: Ta = σa 2 (a = 1. part of the name). (10.44) Irreducible representations exist for any “spin” j = n/2.1. mi. (10.• If R1 ⊗ R2 = r1 ⊕ . 3. J b ] = iǫabc J c . 3 or 2. then the indices satisfy the following rule: I(R1 )dR2 + I(R2 )dR1 = n X I(ri ).50) where X=  0 i −i 0 171  .25) that the index of the spin-j representation is I(Rj ) = j(j + 1)(2j + 1)/3.47) or. mi = m|j.49) We therefore recognize from eq. (10.1. ⊕ rn .46) (10.1. in matrix-vector notation: ′ 3m Jm ϕm′ (J a J a )m m′ ϕm′ = mϕm .42) i=1 Let us recall how Lie algebra representations work in the example of SU (2). 2 −1 if a. The j = 1/2 representation is real. 2.51) . (10. (10.43) where the σ a are the three Pauli matrices [see. eq.1. 1.1. c = 1. (10. for special.1.45) These representation matrices can be chosen to act on states ϕm = |j.1.24) that the quadratic Casimir invariant of the spin-j representation of SU (2) is C(Rj ) = j(j + 1). 2 2 (10. where n is an integer. This group is familiar from the study of angular momentum in quantum mechanics.1. (3. It follows from eq. (10. c = 1. b. 3.48) = j(j + 1)ϕm . mi = j(j + 1)|j. and have dimension 2j + 1.1.down. the group of unitary 2 × 2 matrices (that’s the “U(2)” part of the name) with determinant 1 (that’s the “S”. (10. mi. mi. 2 or up. because X(− σ a∗ −1 σ a )X = . 2 or 3. 3   0 otherwise.1.1.1. 1.23)]. 3). The representation matrices J a in the spin-j representation satisfy the SU (2) Lie algebra: [J a . J a J a |j. . 1 or 2. and the defining or fundamental representation is the familiar spin-1/2 one with ϕi with i = 1. according to: J 3 |j. b. 2. for example. . 2. (10. (10.1.

⊕ (j1 + j2 ) (10. namely:   νeL . Making a table of the representations of SU (2): singlet fundamental adjoint spin dimension I(R) C(R) real? 0 1 0 0 yes 1/2 2 1/2 3/4 yes 1 3 2 2 yes 3/2 4 5 15/4 yes . under which the up and down quarks transform as a j = 1/2 doublet.54) (10. j 2j + 1 j(j + 1)(2j + 1)/3 j(j + 1) yes The tensor product of any two representations of SU (2) is reducible to a direct sum.More generally.. d′L   νµL . The condition that the matrices are Hermitian removes half of these. one must make the same transformation simultaneously on each of these representations. as: j1 ⊗ j2 = |j1 − j2 | ⊕ (|j1 − j2 | + 1) ⊕ . it serves as the Lie algebra of angular momentum operators. When one makes an SU (2)L gauge transformation. .1. The group SU (2) has many applications in physics.52) in which j is used to represent the representation of spin j. eL   uL . The Lie algebra generators of SU (N ) in the fundamental representation are Hermitian traceless N × N matrices. The irreducible j = 1/2 representations of SU (2)L are composed of the pairs of fermions that couple to a W ± boson. one can show that all representations of SU (2) are real. .. .1. First. and see more precisely how it ties into the weak interactions and QED.. µL   cL . We will come back to study the SU (2)L symmetry in more detail later.. a basis for the complex N × N matrices is 2N 2 dimensional. the single condition of tracelessness removes one from the basis. The weak interactions involve a still different SU (2).. Weak isospin is a gauge symmetry that acts only on left-handed fermion fields. That leaves dSU (N ) = N 2 − 1 172 . (10. .53) with a constant θ a that does not depend on position in spacetime. s′L   ντ L . τL   tL .1. One can generalize the SU (2) group to non-Abelian groups SU (N ) for any integer N ≥ 2. since each entry in the matrix is required to be the complex conjugate of another entry. and the primes mean that these are not quark mass eigenstates. . etc. In general.. or mass terms) is invariant under a different SU (2) isospin symmetry.. the transformation can be different at each point in spacetime. . Isospin is a global symmetry.. meaning that the same symmetry transformation must be made simultaneously everywhere:   u d a a → exp(iθ σ /2)  u d  ...55) Here eL means PL e. However. b′L (10.1. The strong force (but not the electromagnetic or weak forces. Finally.. since each matrix has N 2 entries with a real and imaginary part. known as weak isospin or SU (2)L .

.1.1. For all N ≥ 3.1. For example.1. 0    0 −i 0 λ2 =  i 0 0  .57) Note that each of these matrices is Hermitian and traceless.1. which has dimension dG = 8. 0 0 0  0 0 5  λ = 0 0 i 0  1 1  √ λ = 0 3 0 8 1 λ3 =  0 0   −i 0 .60) so that the quadratic Casimir invariant of the fundamental representation is C(F ) = 4/3. the fundamental representation is complex.63) 173 (10. (10. (10. is a gauge theory based on the group SU (3). 1 0 (10.58) and therefore the index of the fundamental representation is I(F ) = 1/2. the generators of the Lie algebra are given by: 1 T a = λa 2 (a = 1. . 3 i (10. (10. the theory of the strong interactions.64) . (10. √ f 458 = f 678 = 3/2.1.1. . In the fundamental representation. They have also been engineered to satisfy Tr(λa λb ) = 2δab . 0  0 −i  .56) where the λa are known as the Gell-Mann matrices:  0 1 λ1 =  1 0 0 0  0 0 4  λ = 0 0 1 0  0 0 7  λ = 0 0 0 i  0 0. one finds that the non-zero structure constants of SU (3) are: f 123 = 1.61) By taking commutators of each pair of generators. as required. Quantum Chromo-Dynamics (QCD).59) One can also check that (T a T a )i j = 4 j δ .as the dimension of the group and of the adjoint representation. (10.1. so that Tr(T a T b ) = 1 ab δ . 0 0  0 0 0 1. . 0 −2 0 6  λ = 0 0  0 0 −1 0  . 8). 2 (10.1. 0  0 0 1 0 .62) f 147 = −f 156 = f 246 = f 257 = f 345 = −f 367 = 1/2. 0  1 0.

1.70) 3 ⊗ 3 ⊗ 3 = 1A ⊕ 8M ⊕ 8M ⊕ 10S 8 ⊗ 8 = 1S ⊕ 8A ⊕ 8S ⊕ 10A ⊕ 10A ⊕ 27S .1.1. one relies on group-theoretic identities. For example.1.68) 3 ⊗ 3 = 3A ⊕ 6S (10. One can easily compute the latter by using eq. 174 (10. (10. 0  0  0 0 (10.β = (α + 1)(β + 1)(α + β + 2)/2. with. Instead.1.. (10. calculations of Feynman diagrams often involve the index or Casimir invariant of the fundamental representation.and those related to the above by permutations of indices.32).1. the representations can be classified by two non-negative integers α and β.1. Following is a table of the smallest few irreducible representations of SU (3): singlet fundamental anti-fundamental adjoint dimension I(R) C(R) real? 1 0 0 yes 3 1/2 4/3 no 3 1/2 4/3 no 6 5/2 10/3 no 6 5/2 10/3 no 8 3 3 yes 10 15/2 6 no 10 15/2 6 no .24) and (10.67) Some tensor products involving these representations are: 3⊗3 = 1⊕8 (10. following from the condition that the f abc are totally antisymmetric. β is dα.69) 3 ⊗ 3 = 3A ⊕ 6S (10..1. it is almost never necessary to actually use the explicit form of any matrix representation larger than the fundamental. and the Casimir invariant of the group.66) with the result C(G) = 3.1. (10. It is usual to refer to each representation by its dimension in boldface. The dimension of the representation labeled by α. In general.32): f abc f abd = C(G)δcd . for example:  0 0 0 0 0 0 1 0   0 −1 0 0  0 0 0 0  1 Tadjoint = −i  0 0 0 0  0 0 0 0  0 0 0 −1/2 0 0 0 0 0 0 0 0 0 1/2 0 0 0 0 0 0 −1/2 0 0 0 0 0 0 1/2 0 0 0 0  0 0  0  0  . From these one can find the adjoint representation matrices using eq.71) (10.1.65) etc. (10.72) . However.1.

the irreducible representations on the right side are labeled as A. . .1. For N = 3. . Under a gauge transformation. these are the 6 and 3 representations. we will construct the Lagrangian and Feynman rules for a theory of Dirac fermions and gauge bosons transforming under a non-Abelian gauge group. when we take the tensor product of two or more identical representations. . or M depending on whether they involve an antisymmetric.1.1. ⊗ N} . dR and a = 1. For SU (3).75) The first term in parentheses transforms as an adjoint representation.1.73) transforms under the tensor product representation N ⊗ . N N (10. Similarly. under SU (N ). and raised indices for the antifundamental. one can build any representation out of objects that carry only indices transforming under the fundamental N and anti-fundamental N representations. S. one can find all necessary tensor-product representation rules for any SU (N ) group. Then an object carrying n lowered and m raised indices: . and an N (N − 1)/2-dimensional antisymmetric tensor. Here i = 1. we can take an object that transforms under SU (3) as N × N. . dG .68).. symmetric. ⊗ N} ⊗ N ⊗ . and this decomposition corresponds to eq. .69).Here. So. In SU (N ). for example. and the second as a singlet. It is useful to employ lowered indices for the fundamental. where i is an index in some representation of the gauge group with generators Tiaj .1. . 10. . called a Yang-Mills theory.i n (10. an object that transforms under SU (N ) as N × N can be decomposed as ϕij = 1 1 (ϕij + ϕji ) + (ϕij − ϕji ). or mixed symmetry combination of the indices of the original representations on the left side. (10.1) .2 The Yang-Mills Lagrangian and Feynman rules In this subsection. . 2 2 (10. this corresponds to the rule of eq. respectively. By using this process of taking symmetric and anti-symmetric parts and removing traces. To reduce it.74) m times This is always reducible. (10. irreducible representations.76) The two terms on the right-hand side correspond to an N (N + 1)/2-dimensional symmetric tensor.. we have: Ψi → Ui j Ψj 175 (10.2. ..{z .{z . and write it as: ϕji = (ϕji − 1 1 j k δi ϕk ) + ( δij ϕkk ). Let the Dirac fermion fields be given by Ψi . one can decompose ϕ into parts that have different symmetry and trace properties. | | n times (10..jm ϕji11.1.

(10.2.3).2. while the Hermitian conjugate spinor carries a raised index. (10. Taking the Hermitian conjugate of eq. (10.2. and in the case of matrices.9) (10. So the Dirac spinor carries a lowered representation index. (10.2. (Notice that taking the Hermitian conjugate changes the heights of the representation indices.7) i Comparing with eq.2.3) Our goal is to build a Lagrangian that is invariant under this transformation.) Now we can multiply on the right by γ 0 .2) Specializing to the case of an infinitesimal gauge transformation θ a = ǫa . reverses their order. (10.10) . we find Ψ†i → Ψ†j (1 − iǫa T a )j i . so we get the transformation rule for the barred Dirac spinors: i j Ψ → Ψ (1 − iǫa T a )j i . (10. (10.2. under an infinitesimal gauge transformation.27). (10.5) We can rewrite this in a slightly different way by noting that Tjai = (Tjai )† = (T ai j )∗ .where U = exp(iθ a T a ).4) where we have used the fact that T a are Hermitian matrices. this establishes that Ψ transforms in the complex conjugate of the representation carried by Ψi . their tensor product must be a direct sum of representations that includes a singlet.2.8) As a check of this.2.2. The singlet is obtained by summing over the index i: i Ψ Ψi . The Dirac gamma matrices are completely separate from the gauge group representation indices. we have Ψi → (1 + iǫa T a )i j Ψj . (10.2. i Since Ψ and Ψi transform as complex conjugate representations of each other. First let us consider how barred spinors transform. 176 (10.1. this term becomes: i k j Ψ Ψi → Ψ (1 − iǫa T a )j i (1 + iǫb T b )i Ψk = i Ψ Ψi + O(ǫ2 ).6) so that i j Ψ → (1 + iǫa [−T a∗ ])i j Ψ .

because ∂µ Ψi does not gauge transform in the same way that Ψi does. (10.2. which we recall is the dimension of the gauge group dG . Now one can check that the Lagrangian i i Lfermions = iΨ γ µ Dµ Ψi − mΨ Ψi (10. (10. (10. The last term vanishes for Abelian groups like QED.2. giving an extra term: ∂µ Ψi → (1 + iǫa T a )i j ∂µ Ψj + i(∂µ ǫa )Tiaj Ψj . but it is necessary to ensure that the covariant derivative of a Dirac field transforms in in the same way as the field itself: Dµ Ψi → (1 + iǫa T a )i j Dµ Ψj .11) This shows that each component of the field Ψi must have the same mass. and is the direct analog of the coupling e in QED. we can fix this by writing a covariant derivative involving vectors fields. (10. known as a gauge coupling. the term i iΨ γ µ ∂µ Ψi (10. Notice that the definition of the covariant derivative depends on the representation matrices for the fermions.2.2. The number of such fields is equal to the number of generator matrices T a .17) If we limit ourselves to constant gauge parameters ∂µ ǫa = 0.2. The quantity g is a coupling.14) The vector boson fields Aaµ are known as gauge fields. Next we would like to include a derivative kinetic term for the fermions. The entries of the matrix Tiaj take the role played by the charges q in QED. we can include a fermion mass term in the Lagrangian: i Lm = −mΨ Ψi . (10.1. The problem is that the derivative can act on the gauge-parameter function ǫa .13): Dµ Ψi = ∂µ Ψi + igAaµ Tiaj Ψj .12) is not acceptable by itself. so there is really a different covariant derivative depending on which fermion representation one is acting on. provided that the gauge field is taken to transform as: 1 Aaµ → Aaµ − ∂µ ǫa − f abc ǫb Acµ . They carry an adjoint representation index a in addition to their spacetime vector index µ.13) By analogy with QED. (10. g (10.4).2. 177 .2. (10. Therefore.15) is invariant under infinitesimal gauge transformations.where the terms linear in ǫa have indeed canceled. see eq.2.2.16) is of the correct form for a field in the adjoint representation. which will also transform in such a way as to cancel the last term in eq. Just as in QED. It is dimensionless. then the transformation in eq.16) The term with a derivative acting on ǫa is the direct analog of a corresponding term in QED.

2. By analogy with QED.19) Then one can check. However.2. Lgauge = − F µνa Fµν 4 (10.2.Having introduced a gauge field for each Lie algebra generator. labeled by i. and one can take ξ = 1 for Feynman gauge and ξ = 0 for Landau gauge. we have found a gauge-invariant Lagrangian for Dirac fermions coupled to a non-Abelian gauge field: LYang-Mills = Lgauge + Lfermions . labeled by a. Fµν → Fµν − f abc ǫb Fµν (10. using eq.2.20) From this it follows that Lgauge transforms as: 1 1 1 a a c − F µνa Fµν → − F µνa Fµν − F µνa f abc ǫb Fµν + O(ǫ2 ). a ν. we have: µ. (10.16). Therefore we can immediately obtain the Feynman rules for vector and fermion propagators. Putting the above results together.2. Therefore. This is accomplished if we choose: a Fµν = ∂µ Aaν − ∂ν Aaµ − gf abc Abµ Acν .2. we must also require that this Lagrangian is invariant under gauge transformations. 4 (10.18) a is an antisymmetric field strength tensor for each Lie algebra with an implied sum on a. which is antisymmetric under the same interchange. because the part F µνa Fµν but its gauge indices are contracted with f abc . (10. eq. that a a c . First we identify the kinetic terms that are quadratic in the fields. For gauge fields. (10.22) Now we can find the Feynman rules for this theory in the usual way. we must now include kinetic terms for them.18) is a gauge singlet.2. Likewise. the Dirac fermion propagator is: j i ←→ 178 δij i(p / + m) − m2 + iǫ p2 . 4 4 2 (10. That part of LYang-Mills is: 1 i i Lkinetic = − (∂µ Aaν − ∂ν Aaµ )(∂ µ Aaν − ∂ ν Aaµ ) + iΨ ∂/Ψi − mΨ Ψi .23) These terms have exactly the same form as in the QED Lagrangian. the gauge field Lagrangian has the form: 1 a . where Fµν generator.21) c is symmetric under interchange of a ↔ c. b ←→ δab  i pµ pν −gµν + (1 − ξ) 2 2 p + iǫ p  where pµ is the 4-momentum along either direction in the wavy line. (10. The δab in the Feynman rule just means that a gauge field does not change to a different type as it propagates.2. and over dR copies of the fermion field. The extra term linear in ǫ vanishes. but with a sum over dG copies of the vector fields.

proportional to g and g2 respectively. The interaction Feynman rules follow from the remaining terms in LYang-Mills . a −igTiaj γ µ ←→ j This rule says that the coupling of a gauge field to a fermion line is proportional to the corresponding Lie algebra generator matrix. they can be written as LAAA = gf abc (∂µ Aaν )Aµb Aνc . there is a fermion-fermion-vector interaction coming from the covariant derivative in LΨ .26) The three-gauge-field couplings involve a spacetime derivative. from LAΨΨ = −gAaµ Ψγ µ T a Ψ (10. Since the matrices T a are not diagonal for non-Abelian groups. which we can treat according to the rule of eq.24) we get: i µ. δAaµ δAbν δAcρ (10. After combining some terms using the antisymmetry of the f abc symbol. 4 (10.2. c is obtained by taking functional derivatives of the Lagrangian density with respect to the corresponding fields: i δ3 LAAA . c 179 .6. g2 LAAAA = − f abe f cde Aaµ Abν Aµc Aνd . The Lagrangian density Lgauge contains three- gauge-field and four-gauge-field couplings. First. a p −gf abc [gµν (p − q)ρ + gνρ (q − k)µ + gρµ (k − p)ν ] ←→ k ρ. gauge) indices µ. Then the Feynman rule for the interaction of three gauge bosons with (spacetime vector. as i times the momentum of the field it acts on (in the direction going into the vertex). (9.2.2.27) resulting in: ν. b and ρ. this interaction can change one fermion into another.2.5)].4. a and ν. b q µ.25) (10.26). Again the factor of δij means the fermion does not change its identity as it propagates. and m the mass of the Dirac fermion. Identifying the Feynman rule as i times the term in the Lagrangian [recall the discussion surrounding eq. (6.with 4-momentum pµ along the arrow direction.

non-composite. like: LA-mass = m2V Aaµ Aaµ . massless vector field.29). a and ν. This sounds like a serious problem. c and σ. b.28) leading to: µ. q µ . (10. Therefore. Likewise. The external state particles carry a representation or gauge index determined by the interaction vertex to which that line is attached. The only problem with this is that the theory would be nonrenormalizable in that case. One way to proceed would be to simply keep the term in eq. we necessarily violate the gauge invariance of the Lagrangian. q). We will not do any loop calculations in this course.2. The ghost fields do not create and destroy real particles. (10. d is: i δ4 LAAAA . In that case. and kµ are the gauge boson 4-momenta flowing into the vertex.16). this is not quite the end of the story if one needs to compute loop diagrams. since the functional derivatives have a choice of several fields on which to act. as a related problem. b and ρ. (10. One way to fix this problem is by introducing “ghost” fields that only appear in loops. If one tries to get around this by introducing a mass term for the vector gauge fields. they are really just book-keeping devices that exist only to cancel the unphysical contributions of gauge fields in loops. b h −ig2 f abe f cde (gµρ gνσ − gµσ gνρ ) + f ace f bde (gµν gρσ − gµσ gνρ ) ←→ σ. [However. the massive W ± boson cannot be described by the Yang-Mills theory that we have so far. the Feynman rule for the coupling of four gauge bosons with indices µ. δAaµ δAbν δAcρ δAdσ (10.2.29) then one finds that this term is not invariant under the gauge transformation of eq.where pµ . External state fermions and gauge bosons are assigned exactly the same rules as for fermions and photons in QED. the photon. p) ↔ (ν. The above Feynman rules are all that is needed to calculated tree-level Feynman diagrams in a Yang-Mills theory with Dirac fermions. a. unitarity would be violated in scattering at very 180 . One is that the gauge fields are necessarily massless. and in particular never appear in initial or final states. if we put in such a term. and accept that the theory is not fully invariant under the gauge symmetry.] The Yang-Mills theory we have constructed makes several interesting predictions. for example (µ. In particular.2. d i + f ade f bce (gµν gρσ − gµρ gνσ ) ρ. so we will not go into more detail on that issue. one must take into account that not all of the gauge fields that can propagate in loops are actually physical. and the gauge symmetry will not be a symmetry of the theory.2. because there is only one known freely-propagating. a ν. Notice that these fields are invariant under the simultaneous interchange of all the indices and momenta for any two vector bosons. c There are more terms in these Feynman rules than in the corresponding Lagrangian.

as a consequence of the gauge invariance. Another nice feature of the Yang-Mills theory is that several different couplings are predicted to be related to each other. and what the three-gauge-boson and four-gauge-boson vertices should be. In particular. Once we have picked a gauge group G. a set of irreducible representations for the fermions. what the couplings of other fermions to the gauge fields should be (as long as we know their representations). if we know the coupling of one type of fermion to the gauge fields. 181 .high energies. This in turn allows us to predict. leading to a spontaneous breakdown in the gauge symmetry. we can explain the non-zero mass of the W ± boson by enlarging the theory to include scalar fields. then we know g. then the interaction terms are all fixed. and the gauge coupling g. Instead.

(11.1. etc. d → eiθ(x)Qd d. µ. − 31 ) e. c. c.3) with a = 1. τ (1. s.1.. d. no experiment can tell the difference between a red quark and a blue quark. there is a corresponding gauge vector boson called a gluon. ντ (1. (11. b. with the fermions and gauge bosons transforming as: SU (3)c × U (1)EM spin u.11 Quantum Chromo-Dynamics (QCD) 11. b (3. s. In fact. s transform into each other.4) where Qu = Qc = Qt = 2/3 and Qd = Qs = Qb = −1/3.1. The subscript c is to distinguish this as the group of invariances under transformations of the color degrees of freedom. c. .1. 0) 1 gluon (8. we can do a different SU (3)c transformation at each point in spacetime. + 23 ) d. s → eiθ a (x)T a s. b. so that: u → eiθ where T a = λa 2 a (x)T a d → eiθ u. −1) νe .. This symmetry is in addition to the U (1)EM gauge transformations: u → eiθ(x)Qu u. d. 0) 1 182 . s → eiθ(x)Qs s. (11. and each barred Dirac field u. 0) 1 2 1 2 1 2 1 2 γ (1. s. For each of the 8 generator matrices T a of SU (3)c .1 QCD Lagrangian and Feynman rules The strong interactions are based on a Yang-Mills theory with gauge group SU (3)c . a (x)T a d.1) while an anti-up quark is created in an initial state by any of the fields: u = ( ured ublue ugreen ) = ( u1 u2 u3 ) . represented by a field Gaµ carrying both a spacetime vector index and an SU (3)c adjoint representation index. .) Each of the quark Dirac fields u. For example.2.2) Since SU (3)c is an exact symmetry. As far as we can tell. as we saw on general grounds in subsection 10. . ugreen u3 (11. t therefore transforms as a 3. etc. (There is also an approximate SU (3)flavor symmetry under which the quark flavors u. νµ . . t (3. an up quark is created in an initial state by any one of the three color component fields:     ured u1    u = ublue = u2  . d. this is an exact symmetry of nature. t transforms separately as a 3 of SU (3)c . and the θ a (x) are any gauge parameter functions of our choosing. but simultaneously on each quark flavor. 8. One says that the unbroken gauge group of the Standard Model is SU (3)c × U (1)EM . isospin is an SU (2) subgroup of this symmetry. with quarks transforming in the fundamental 3 representation. so the labels are intrinsically arbitrary.

3 to run over the color degrees of freedom.2.1. t. to use “springy” lines for gluons. 183 . we know that the propagator for the gluon is that of a massless vector field just like the photon: µ. etc. to easily distinguish them from wavy photon lines. d. Using the general results for a gauge theory in subsection 10. c.1. in QCD. In addition to the propagator. The strength of the strong interactions comes from the fact that g3 ≫ e. a −ig3 Tiaj γ µ ←→ j Here the quark line can be any of u.1. The gluon interaction changes the color of the quarks when T a is non-diagonal.1. with g → g3 . The Lagrangian density also contains a “pure glue” part: 1 a Lglue = − F µνa Fµν 4 a Fµν = ∂µ Gaν − ∂ν Gaµ − g3 f abc Gbµ Gcν . but never changes the flavor of the quark line.2 in the general case.6) Here g3 is the coupling constant associated with the SU (3)c gauge interactions.and gauge-index structure are just as given in section 10. There are also quark-gluon interaction vertices for each flavor of quark: i µ. (10.5) (11. a ν. Using an index i = 1. s.1. the covariant derivatives are: 2 Dµ ui = ∂µ ui + ig3 Gaµ Tiaj uj + ieAµ ( )ui . 2. b ←→  i pµ pν δ 2 −gµν + (1 − ξ) 2 p + iǫ p ab  Note that it is traditional. a down quark remains a down quark. (11.1. so an up quark remains an up quark. b. 3 (11. this implies that there are three-gluon and four-gluon interactions: The spacetime. 3 1 aj Dµ di = ∂µ di + ig3 Gaµ Ti dj + ieAµ (− )di .The gluon appears together with the photon field Aµ in the full covariant derivative for quark fields.7) (11.64).62)-(10.8) where f abc are the structure constants for SU (3)c given in eqs.

In order to find the differential cross-section. we continue as usual by taking the complex square of the reduced matrix element: bj ∗ c 2 aj )(Tlbi Tm ) |M| . l p−k p′ . because the quarks in both the initial state and final state are parts of bound states.3) (11. We will see how to use the microscopic cross-section result to obtain the observable cross-section later.2. it does form the microscopic part of a calculation for the observable process hadron+hadron→jet+jet.2.11. spin.2. (11.2. j ν. in subsection 11. let us consider the example of quark-quark scattering.1) Let us assign momenta. and color to the quarks as follows: Particle Momentum Spin Spinor Color initial u p s1 u(p.5. (11. let us consider the process of an up-quark and down-quark scattering from each other: ud → ud.5) This illustrates that the “color charge matrix” g3 Tlai is analogous to the electric charge eQf . However. i u µ. To be specific. but with the QED squared coupling replaced by a product of matrices depending on the color combination: aj e2 → g32 Tlai Tm . b d k′ .2. This is not a directly observable process. s1 ) = u1 i initial d p′ s2 u(p′ . s2 ) = u2 j u(k.4) where t = (p − k)2 .6) . m The reduced matrix element can now be written down by the same procedure as in QED. s3 ) = u3 l final u k s3 final d k′ s4 u(k′ .2. One obtains. |M|2 = g34 (Tlai Tm 184 (11. This matrix element is exactly what one finds in QED for e− µ− → e− µ− . s4 ) = u4 m (11. using Feynman gauge (ξ = 1): M = h u3 (−ig3 γµ Tlai )u1 ih bj u4 (−ig3 γν Tm )u2 aj = ig32 Tlai Tm [u3 γµ u1 ] [u4 γ µ u2 ] /t # " i −ig µν δ ab (p − k)2 (11. there is only one Feynman diagram: p. There are 34 = 81 color combinations for quark-quark scattering.2 Quark-quark scattering (qq → qq) To see how the Feynman rules for QCD work in practice. a k.2) At leading order in an expansion in g3 .

(11. we find in the high energy limit of negligible quark masses.12) (11. which is 8 for SU (3)c . so we will sum over the colors of the final state quarks and average over the colors of the initial state quarks: 1 X 1 XXX |M|2 .14) In doing this.10) (11. the rest of |M|2 . and c = [u3 γµ u1 ] [u4 γ µ u2 ] /t. can be taken directly from the corresponding result for e− µ− → e− µ− in QED. one does not measure the colors of quarks in an experiment.7) It is not possible. we have used the definition of the index of a representation eq. the fact that the index of the fundamental representation is 1/2. Stripping off the factor e4 associated with the QED charges. (11. Meanwhile. m (with no implied sum yet).for each i. even with respect to some arbitrary choice. 1 X 1 XXX c 2 |M| = 2 2 s1 2 s2 s3 s4 s 2 + u2 t2 ! .2. the color factor is 1 X 1 X X X ai aj bj ∗ (T T )(Tlbi Tm ) = 3 i 3 j l m l m 1 X aj bm (T ai T bl )(Tm Tj ) 9 i. (11. bj ∗ (Tlbi Tm ) = (Tibl Tjbm ).2. because the gauge group generator matrices are Hermitian. that the incoming u-quark has color red= 1.2.9) Therefore.17). In practice. which we found by crossing symmetry in eq. even in principle.7.10). 3 i 3 j l m (11. However. (7. we note that.2.l. including a sum over final state spins and an average over initial state spins. by an arbitrary choice. j.11) (11.2. we have |M|2 1 X 1 X 4g4 |M|2 = 3 ≡ 9 colors 4 spins 9 185 s 2 + u2 t2 ! . to distinguish between colors.m l i 1 Tr(T a T b )Tr(T a T b ) 9 1 I(3)δab I(3)δab 9 1 1 2 ( ) dG 9 2 2 9 = = = = (11.16) .2.8) To do the color sum/average most easily.2. b of δab δab just counts the number of generators of the Lie algebra dG . (10. one can always imagine fixing. and the fact that the sum over a.1.2. M (11. then the colors of the other quarks can be distinguished up to SU (3)c rotations that leave the red component fixed. l.2.2.15) Putting this together with the factor of g34 and the color factor above.13) (11.j.

3 s 2 + u2 t2 ! . one can replace cos θ in favor of the Mandelstam variable t. which for a general process implies the appropriate sum/average over spin and color. Since we are neglecting quark masses. the kinematics for this process is the same as in any massless 2→2 process. Let us discuss this issue in a more general framework than just QCD. using d(cos θ) = 2dt .20) Renormalization Since the strong interactions involve a coupling g3 that is not small.32). (11. (11.5. for example as found in eqs. (7. we should worry about higher-order corrections to the treatment of quark-quark scattering in the previous subsection. The differential cross-section for this process is therefore: s 2 + u2 t2 1 2πα2s dσ = |M|2 = d(cos θ) 32πs 9s ! . In a general gauge theory. Therefore.2. s (11.2.17) where αs = g32 4π (11.The notation |M|2 is a standard notation.2.19) so dσ 4πα2s = dt 9s2 11.28)-(7.5.18) is the strong-interaction analog of the fine structure constant.2. the Feynman diagrams contributing to the reduced matrix element at one-loop order in fermion+fermion′ → fermion+fermion′ scattering are the following: 186 .

1) This involves a sum over all fermions that can propagate in the loop. and two from the propagators). As an example of what can happen. Let us call q µ = pµ − kµ the 4-momentum flowing through either of the vector-boson propagators. (11. Only the first two diagrams give a finite answer when one naively integrates d4 ℓ.4 × 1018 GeV (give or take an order of magnitude). but it could conceivably be much lower. there is a loop momentum ℓµ that is unfixed by the external 4-momenta. one must introduce a very high cutoff mass scale M . This is not surprising. It is generally thought that the highest this cutoff is likely to be is about MPlanck = 2. Now there are up to five powers of |ℓ| in the numerator (three from the d4 ℓ. For reasons that will become clear shortly. and replace the loop-momentum integral by one that kills the contributions to the reduced matrix element from |ℓµ | ≥ M .3. in part because there are actually several different ways to cutoff the integral at large M . we do not really know what physics is like at very high energy and momentum scales.) The d4 ℓ factor can be written as an angular part times a radial part |ℓ|3 d|ℓ|. we are calling the gauge coupling of the theory gˆ and the mass of each fermion species m ˆ f . but is clumsy to carry out and even clumsier to interpret. so we have no business in integrating over them. Therefore. We are being purposefully vague about what R |ℓµ |≤M d4 ℓ means. and must be integrated over. (A straightforward step-function cutoff will work. M should be the mass scale at which some as-yet-unknown new physics enters in to alter the theory. Physically. and a trace over the spinor indices of the fermion loop. and four powers 187 .In each of these diagrams. Then the part of the reduced matrix element associated with the fermion loop is: X (−1) f Z 4 |ℓµ |≤M d ℓ Tr ( h µ −iˆ g (Tf )aj i γ i " # " h i i(/ℓ + /q + m ˆ f) i(/ℓ + m ˆ f) bi ν −iˆ g (T ) γ f j 2 2 2 2 (ℓ + q) − m ˆ f + iǫ ℓ −m ˆ f + iǫ #) . consider the next-to-last Feynman diagram given above.

(11. . The arbitrariness in the precise definition of m can be absorbed into the “. Each of the three-vector couplings involves a factor of f abc . To be specific. . .) Therefore. so that Mgauge loop in gauge prop. inspired by eq. A similar sort of calculation applies to the last diagram involving a gauge vector boson loop. . ∝ gˆ4 X I(Rf )ln(M/m) + . (11. the group-theory factor. . (11. The m is a characteristic mass scale of the problem. .3. and a higher cutoff. there is a conspiratorial cancellation. represents a contribution that does not get large as M gets large. it is proportional to gˆ4 . It is again logarithmically divergent. . one might expect that the result of doing the integral will scale like M 2 for a large cutoff M . Specifically. we could just redo the calculation with that physics included. . The result is proportional to: gˆ2 (q 2 gµν − q µ q ν ) X Tr(Tfa Tfb ) [ln (M/m) + . . (11. So it must be that the loop part of the diagram make a contribution proportional to f acd f bcd = C(G)δab .3. .3.2) f where the . .6) . However. one defines a renormalized or running coupling g(µ) by writing:   gˆ = g(µ) 1 −  (g(µ))2 16π 2   4X 11 C(G) − 3 3 188 f    I(Rf ) ln(M/µ) .] (11. .5). one finds that the contributions to the differential cross-section is given by:       4X gˆ2 11 I(Rf ) ln(M/m) + . considered to be a function of gˆ.3. it is something with dimensions of mass built out of q µ and the m ˆ f . Let us ignore all the other diagrams for now.4) Doing everything carefully. the justification for this will be revealed soon.of |ℓ| in the denominator from the propagators. So naively. it is clear that the entire contribution must be proportional to: Mfermion loop in gauge prop. When one uses eq. and the large-M dependence on ln(M/m). .3.3. ∝ gˆ4 C(G)ln(M/m) + . The cutoff M may be quite large. .2) in the rest of the Feynman diagram. it is convenient to absorb our ignorance of M into a redefinition of the coupling. Furthermore.”. .3) f What we are trying to keep track of here is just the number of powers of gˆ.3. It must appear in the formula in the way it does in order to make the argument of the logarithm dimensionless.5) where dσtree (ˆ g ) is the tree-level result (which we have already worked out in the special case of QCD). or what the specific very-high-energy physics associated with it is.  (11. so that the large-M behavior is only logarithmic. (If we did. we typically do not know what it is. with two of the indices contracted because of the propagators. dσ = dσtree (ˆ g ) 1 + 2  C(G) −   4π 3 3 f (11.

(11. One can invert this relation to write the renormalized coupling in terms of the bare coupling:       gˆ2  11 4X  . we can now write. since we know that dσtree (ˆ g ) is proportional to gˆ4 .3. but with g(µ) in place of gˆ. The choice of renormalization scale eq. called the renormalization scale. = dσtree (g) 1 + 2  C(G) −   4π 3 3 f (11. C(G) − f   16π 2 3 3 f (11.3.6):       gˆ2 11 4X dσ = dσtree (g) (ˆ g /g)4 1 + 2  C(G) − I(Rf ) ln(M/m) + . but is still large. (11.3.9) is small. . . g(µ) = gˆ 1 + I(R ) ln(M/µ) + . (11.5).   4π 3 3 f       4X g2 11 I(Rf ) ln(µ/m) + .3. one should choose µ ≈ m.3.Here µ is a new mass scale. knowing that the one-loop correction from these diagrams is small. one can calculate using the tree-level approximation using a renormalized coupling g(µ). m = 0. suppose that M = MPlanck . These choice might be appropriate for experiments involving a non-relativistic electron and a TeV-scale collider.3.12) . . The factor dσtree (g) is the tree-level differential cross-section.511 MeV or m = 1000 GeV. This suggests that a really good choice for µ is to make the logarithm ln(µ/m) as small as possible.11) allows us to write: dσ ≈ dσtree (g(µ)). (11. except that we do not know what that is. 189 (11. (11. and perturbation theory would converge very slowly or not at all. Then ln(M/m) ≈ 50 or 35.3.8) (11. . using eqs. these are comparable to 2-loop contributions that we are neglecting anyway. say. For example. (11. respectively. dropping terms of order gˆ5 everywhere. that we get to pick. . the logarithm could then be very large.5) and (11.11) Then. and is replaced by the scale µ that we know.3. .9) Here we are again dropping terms that go like g4 .7) where we are treating g(µ) as an expansion in gˆ.3. and the characteristic energy scale of the experiment we are doing is. This formula looks very much like eq. because we get to pick it. The reason for this strategic definition is that. (It is not uncommon to see the renormalization scale denoted by Q instead of µ. Besides.) The original coupling gˆ is called the bare coupling.10) This logarithm typically gets multiplied by 1/16π 2 times g2 times a group-theory quantity. so that the correction term in eq.3. to a first approximation. . but with the crucial difference that the unknown cutoff M has disappeared. What should we pick µ to be? In principle we could pick it to be the cutoff M . Therefore.3.

(11. Finally.3. what is left in the expression for any cross-section after writing it in terms of the renormalized mass m(µ) and renormalized coupling g(µ) is a polynomial in ln(µ/m).3. there might be more than one characteristic energy scale in a given problem. one can show that by doing some redefinitions of the form: " gˆ = g(µ) 1 + m ˆf " L X bn g n=1 L X = mf (µ) 1 + 2n # pn (ln(M/µ)) . not just the ones that get large as M → ∞. they fall into three classes. rather than a single m. This has to be dealt with by fancier methods. It is an amazing fact that the two redefinitions eqs. but it can be absorbed into a similar redefinition of the mass. and if they are very different from each other. there are diagrams (the third through sixth diagrams) that are individually divergent like ln(M/m). and bn . there is no way around calculating all the one-loop diagrams. .3. but sum up to a total that is not divergent. Here pn (x) and qn (x) are polynomials of degree n.3.3.14) where C(Rf ) is the quadratic Casimir invariant of the representation carried by the fermion f . this is only good enough to get rid of the large logarithmic one-loop corrections. At any finite loop order.13) or mf (µ) = m ˆf ! g2 1 + 2 C(Rf )ln(M/µ) + . 190 . First.6) and (11.3. A clue to this is that they all involve sub-diagrams: The one-loop renormalized or running mass mf (µ) is defined in terms of the bare mass m ˆ f by ! g2 m ˆ f = mf (µ) 1 − 2 C(Rf )ln(M/µ) . no matter what µ is chosen. If you really want all one-loop corrections. with no M -dependence.15) (11. n=1 (11. 2π (11. This is what it means for a theory to be renormalizable at one loop order. these are to be made small by choosing† µ ≈ m.13) are enough to remove the cutoff dependence of all cross-sections in the theory up to and including one-loop order. one can calculate dσ for any process. What about the remaining diagrams? If we isolate the M → ∞ behavior. This is what it means for a theory to be renormalizable at all loop † Of course.16) one can simultaneously eliminate all dependence on the cutoff in any process up to L-loop order. and the index. If so. In Yang-Mills theories. In other words.Of course. and express it in terms of the renormalized mass m(µ) and the renormalized coupling g(µ). . Second. cn are some constants that depend on group theory invariants like the Casimir invariants of the group and the representations. there are diagrams that are not divergent at all (the first two diagram). keeping all the pieces. 2n # cn g qn (ln(M/µ)) . the seventh through tenth diagrams have a logarithmic divergence. 2π (11. then one may be stuck with some large logarithms. .

the results are first expressed in terms of observable quantities like cross-sections. Using this data.or 3. A situation that arises quite often is that one extracts running parameters from an experiment with a characteristic energy scale µ0 . which is an arbitrary one made by us.3.9) is independent of µ. we would also need to know the cutoff M . What we need is a way of taking a running coupling as determined in the first experiment at a renormalization scale µ0 . The bare coupling and the bare mass never enter into this process of comparing theory to experiment.3. We could guess at it.‡ As an example. this is not a group in the mathematical sense defined earlier. we see from eq. It does mean that we expect the theory to have trouble making predictions about processes at high energy scales. but with the non-logarithmic corrections included too. decay rates. . When one does an experiment in high energy physics. we find:  ‡     4 dg  d 1 g2 11 4X 0= I(Rf ) + . (dσ) = (dσtree ) + 2  C(G) −  g dµ  dµ 4π 3 3 f µ The use of the word “group” is historical.3. If we measure dσ in an experiment. we saw that the four-fermion theory of the weak interactions makes reliable predictions. Typically.loop order. using a theoretical prediction like eq. 191 (11.orders.17) . This tests both the theoretical framework. because the loop corrections involved in at least one of the two cases will be unnecessarily large. or the momentum exchanged between particles in a collision. the running mass is related to it by various corrections. (11. it does not necessarily mean that the theory is useless. . (The running mass is not quite the same thing as the physical mass. and physical masses of particles.9). However. we do not know what M is. and we still have no more predictive theory for gravity than Einstein’s relativity. If a theory is non-renormalizable.3. but this would usually be a wild guess. The physical mass can be determined from the experiment by kinematics. (11. It would be unwise to use the same renormalization scale when computing the theoretical expectations for both experiments. except in some special theories. or some suitable average of particle masses and exchanged momenta.or occasionally 4. Here µ0 and µ each might be the mass of some particle that is decaying. Since the differential cross-section dσ for fermion+fermion′ →fermion+fermion′ is an observable. devoid of practical significance. Therefore.) The running parameters can then be used to make predictions for other experiments. The change of the choice of scale µ is known as the renormalization group. .5) that in order to determine the bare coupling gˆ from the data. in principle it should not depend on the choice of µ. one extracts the value of the running couplings and running masses at some appropriately-chosen renormalization scale µ. and getting from it the running coupling at any other scale µ. and the specific values of the running parameters. we can require that eq. let us consider how g(µ) changes in a Yang-Mills gauge theory. We have seen that we can eliminate the dependence on the unknown cutoff of a theory by defining a renormalized running coupling g(µ) and mass mf (µ). the specifics of these redefinitions is only known at 2. (11. and one wants to compare with data from some other experiment that has a completely different characteristic energy scale µ. Remembering that dσtree ∝ g4 .

= C(G) + − dµ 16π 2 3 3 f (11. In QCD. from the equation we are about to write down. one can solve the RG equation explicitly. (11. . The right-hand side of the RG equation is known as the beta function for the running coupling g(µ). irrelevant to the process under study) are not included. so that: µ dg = β(g). so it is neglected as a higher-loop-order effect in the expansion in g2 . 0 1 16π 2 (16π 2 )2 (16π 2 )3 (11. and is written β(g). . it must be true that:   dg g3  11 4X µ I(Rf ) + . . The second term comes from the derivative acting on the lnµ one-loop correction term.22) b0 = − and. called the renormalization group equation or RG equation.19) In a Yang-Mills gauge theory.20) where we already know that 11 4X I(Rf ).23) you can check that g2 (µ) = g2 (µ0 ) 1− b0 g 2 (µ0 ) 8π 2 ln(µ/µ0 ) 192 . I(Rf ) + 4 C(G)2 + C(G) 3 3 f f (11.3.18) This differential equation.3. b1 = − etc.where we are dropping all higher-loop-order terms that are proportional to (dσtree )g4 .3.17) comes from the derivative acting on the g4 inside dσtree . including the effects of Feynman diagrams with more loops. tells us how to change the coupling g(µ) when we change the renormalization scale. β(g) = g3 g5 g7 b + b + b2 + .3.24) . (11. Other experiments then test the whole framework. as proportional to (dσtree )g5 . In the one-loop order approximation. So. .3. it is usually best to work within an effective theory in which fermions much heavier than the scales µ of interest are ignored. The sum over fermions then includes only those satisfying mf < ∼ µ.21) X X 20 34 C(Rf )I(Rf ). dµ (11. C(G) + 3 3 f (11. The first term in eq. The difference between this effective theory and the more complete theory with all known fermions included can be absorbed into a redefinition of running parameters. . The contribution from the derivative acting on the g2 in the one-loop correction term can be self-consistently judged. just to give you an idea of how it goes. Writing dg2 b0 = 2 g4 . and then we can solve the RG equation to find g(µ) at some other scale. after all. dlnµ 8π (11.3. In practical applications.3.3. An experimental result will provide a boundary condition at some scale µ0 . the coefficients up to b3 (four-loop order) have been calculated. The advantage of doing this is that perturbation theory will converge more quickly and reliably if heavy fermions (that are.

since two-loop effects are definitely big. once αs (µ) starts to get big. b0 = (−9. −7). A qualitative graph of the running of αs (µ) as a function of renormalization scale µ is shown below: αS(Q) ΛQCD Renormalization scale Q Of course. we can make αs blow up by choosing µ small enough. 6) quark flavors. In SU (3). b0.28) This shows that at the scale µ = ΛQCD . (11. 193 .3. −25/3. b0 ln(µ/ΛQCD ) (11.27) with dimensions of [mass].QCD is definitely negative for all accessible scales µ. we should no longer trust the one-loop approximation. and each quark flavor is in a fundamental 3 representation with I(3) = 1/2.26) Since b0 is negative. and so the beta function is definitely negative.3.3. 2 b0.To see how this works in QCD. eq. Therefore.3. b0 αs (µ0 ) ln(µ/µ ) 0 2π (11. 4. let us examine the one-loop beta function. The whole analysis has been extended to four-loop order. To make this more explicit.3. in terms of the running αs . −23/3. implying that αs (µ) = 2π . C(G) = 3.25) where nf is the number of “active” quarks in the effective theory. The crucial fact is that since there are only 6 quark flavors known.24). 5. usually those with mass < ∼ µ. the QCD gauge coupling is predicted to blow up. For an effective theory with nf = (3. we can define a quantity ΛQCD = µ0 e2π/b0 αs (µ0 ) . we have: αs (µ) = 1− αs (µ0 ) . Writing the solution to the RG equation. (11.QCD = −11 + nf 3 (11. according to the 1-loop RG equation.

This is remarkable. αs (mZ ) = 0. d. One can draw Feynman diagrams and make rough qualitative guesses. the force between two free color charges is large and constant as the distance increases. Heuristically. the two ideas are certainly compatible. called the MS scheme. Working in a theory with five “active” quarks u. there is some scale ΛQCD at which the gauge coupling is predicted to blow up in a theory with a negative beta function. Conversely. s. and have remained that way ever since. This theoretical prediction goes by the name of infrared slavery. one finds ΛQCD is about 210 MeV. Recently. Phys. or by quoting what ΛQCD is. This is not a sign that QCD is wrong. One can also work in an effective theory with only four active quarks u. which continuously varies the number of spacetime dimensions infinitesimally away from 4. in which case ΛQCD is about 300 MeV.0007. 194 . This nice property of theories with negative β functions is known as asymptotic freedom.1184 ± 0. These results hold when the details of the cutoff and the renormalization are treated in the most popular way. we see that for experiments conducted at characteristic energies much larger than ΛQCD . (2009) [arXiv:0908. b (the top quark is treated as part of the unknown theory above the cutoff).§ A summary of the experimental data on the QCD coupling is shown in the figures below (from S. Eur. and is getting smaller as µ gets larger. On the other hand. C64. the growth of the QCD coupling means that at very small energies or large distances. This means that perturbation theory becomes more and more trustworthy at higher and higher energies. Alternatively. after the temperature dropped below ΛQCD .1135]). the fact that the QCD gauge coupling becomes non-perturbative in the infrared means that we cannot expect to describe free quarks at low energies using perturbation theory. § In this scheme. but the qualitative effect remains: at any finite loop order. one cuts off loop momentum integrals by a process known as dimensional regularization. we can specify how strong the QCD interactions are either by quoting what αs (µ0 ) is at some specified µ0 . This trade of a dimensionless parameter for a mass scale in a gauge theory is known as dimensional transmutation. the gauge coupling is not large. In the early universe. c. If we want. Bethke. It agrees well with the fact that one does not observe free quarks outside of bound states. s. Instead. While it has not been proved mathematically that the infrared slavery of QCD necessarily requires the absence of free quarks. 689. while ΛQCD is a mass scale.with significant numerical changes. d. all quarks and antiquarks and gluons arranged themselves into color-singlet bound states. An important feature of the renormalization of QCD is that we can actually trade the gauge coupling as a parameter of the theory for the scale ΛQCD . this scheme is relatively easy to calculate in. J. it is a sign that perturbation theory is not going to be able to make good predictions when we do experiments near µ = ΛQCD or lower energy scales. and more complicated calculations show that they are plausibly linked. Although bizarre physically. The name refers to the fact that quarks in QCD are becoming free (since the coupling is becoming small) as we probe them at larger energy scales. since g3 (or equivalently αs ) is a dimensionless coupling. but the numbers cannot be trusted. rather than putting in a particular cutoff M . it has become standard to use this second way of specifying the QCD coupling strength. c.

31) . 3 9 9 3 (11. Also. s. then only the electron itself contributes. and nℓ is the number of charged leptons (e. 195 (11. For a U (1) group. since the generator of the group in a representation of charge Qf is just the 1 × 1 matrix Qf .11 0. so b0. (11.5 July 2009 αs(Q) Deep Inelastic Scattering e+e– Annihilation Heavy Quarkonia 0.QED = i 4h 4 4 16 3nu (2/3)2 + 3nd (−1/3)2 + nℓ (−1)2 = nu + nd + nℓ .30) This corresponds to a very slow running.EM = 0. c. the slower it will run. Q was used as the name of the renormalization scale instead of µ.29) where nu is the number of up-type quark flavors (u. Therefore.3. and the electron charge does not run at all: de =0 dlnµ (µ ≪ me ).) If we do experiments at characteristic energies that are much less than the electron mass. t).) The most accurate determinations come from lattice QCD calculations of the mass splittings in the Υ bottomonium system and from the hadronic branching ratio in τ decays. τ ) included in the chosen effective theory. there is no non-zero structure constant.3. and b0. so: e3 de = βe = dlnµ 16π 2   4 3 < (me < ∼ µ ∼ mµ ).EM = 4/3.78 GeV and 209 GeV. < If we do experiments with a characteristic energy scale me < ∼ µ ∼ mµ . then the relativistic electron is not included in the effective theory (virtual electron-positron pairs are less and less important at low energies).3. The four-loop renormalization group running of αS (µ) is then used to determine the reference value αS (mZ ). (Notice that the smaller a gauge coupling is. b). We can contrast this situation with the case of QED. (In the figure. the index for a fermion with charge Qf is I(Rf ) = Q2f .4 τ-decays (N3LO) Quarkonia (lattice) Υ decays (NLO) 0. so C(G) = 0.0007 1 10 Q [GeV] 0. b0.2 e+e– jets & shps (NNLO) electroweak fits (N3LO) e+e– jets & shapes (NNLO) 0.3 DIS F2 (N3LO) DIS jets (NLO) 0.0.12 100 0. and nd is the number of down-type quark flavors (d.13 αs (Μ Z) The data determine αS (µ) at a variety of renormalization scales µ between 1.1 QCD α s (Μ Z) = 0.1184 ± 0. µ.

since the effective electromagnetic coupling is perturbative. d The momenta flowing into the leftmost 3-gluon vertex are. one has the interesting process gg → gg even at tree-level. f p′ k′ σ. but quite tedious. −p − p′ . 196 . At extremely high energies. λ2 ) ρ∗ ǫρ∗ 3 = ǫ (k. but does occur at one loop. e ν. Because there are three-gluon and four-gluon interaction vertices. 11. λ3 ) σ∗ σ∗ ǫ4 = ǫ (k′ .1) Let the internal gluon line carry (vector.) The corresponding QED process of γγ → γγ does not happen at tree-level. a p k ρ. c p + p′ κ.This means that QED is not quite “infrared free”. so the apparent blowing up of α = e2 /4π much farther in the ultraviolet is just an illusion. more complete theory anyway at energy scales in the hundreds of GeV range. Also. and then skip directly to the final answer. λ4 ) Color a b c d (11.4. Furthermore. starting from the upper-left incoming gluon and going clockwise. In QCD. (It is traditional to represent the gluon particle name. f ) on the right. one. Fortunately. at very large distance scales.4 Gluon-gluon scattering (gg → gg) Let us now return to QCD scattering processes. because the QED beta function is always positive. e) on the left and (λ. we have: Particle initial gluon initial gluon final gluon final gluon Polarization vector ǫµ1 = ǫµ (p. QED is embedded in a larger. Choosing polarization vectors and color indices for the gluons. because e runs very slowly. the coupling e could in principle become very large. because of the three-gluon and four-gluon vertices. but not its quantum field.gauge) indices (κ. the momenta flowing into the rightmost 3-gluon vertex are. Just to get an idea of how this proceeds. p′ ). (p. this is predicted to occur only at energy scales far beyond what we can probe. λ1 ) ǫν2 = ǫν (p′ . let us write down the reduced matrix element for the first (“s-channel”) diagram. Labeling the Feynman diagram in detail: µ. b λ. there are four distinct Feynman diagrams that contribute at tree-level: The calculation of the differential cross-section from these diagrams is an important. but does not get arbitrarily small. by g.

and the Large Hadron Collider √ is now taking over the frontier with pp collisions up to. Because of asymptotic freedom. so one must deal with all of these possible subprocesses.2 to obtain: h  σ∗ Mgg→gg. (−k.2) After writing down the reduced matrix elements for the other three diagrams. and the process ud → ud. and “soft” QCD processes that involve exchanges and radiation of low energy virtual gluons. The scattering can be thought of as a factored into a “hard scattering” of one of the point-like partons. It is virtually certain that the most exciting discoveries in high-energy physics in the next decade will occur at the LHC. with the remaining partons as spectators.4. Suppose we scatter a hadron off of another particle (which might be another hadron or a lepton or photon) with a total momentum exchange much larger than ΛQCD . we will spend several sections here on the basic ideas behind hadron-hadron collisions. k + k′ ). In general. summing over final-state gluon colors and averaging over initial-state gluon colors according to 1 8 P 1 a 8 P P P b c d one finds: # " 9πα2s u2 + t2 s2 + u2 s2 + t2 dσgg→gg = + + +3 .starting from the upper-right final-state gluon and going clockwise. dt 4s2 s2 t2 u2 (11. antiquark. The characteristic size of a hadron. There is no way to separate the proton into simpler parts. this process.4.96 TeV. −k′ . like the proton or antiproton. So we can use the Feynman rules of section 10. and gluon parts of a hadron are called partons. since this is the scale at which the strongly-interacting particles are confined. 11. and the description of hadrons in terms of them is called the parton model. adding them together.5 The parton model. anti-quarks. The Tevatron has explored pp collisions at s = 1.s-channel = ǫµ1 ǫν2 ǫρ∗ −gf aeb gµκ (2p + p′ )ν + gκν (−p − 2p′ )µ + gνµ (p′ − p)κ 3 ǫ4 h  −gf cdf gρσ (k′ − k)λ + gσλ (−2k′ − k)ρ + gλρ (2k + k′ )σ " # −igκλ δef . (p + p′ )2 i i (11. are just two of many possible subprocesses that can occur. taking the complex square. is always roughly 1/ΛQCD ≈ 10−13 cm. parton distribution functions. a hadron is a QCD bound state of quarks. hopefully. The hard scattering subprocess takes place on a time scale much shorter than 1/ΛQCD expressed in seconds. We will consider the subprocesses of proton-(anti)proton scattering more systematically in the next subsection. at higher scattering energies it becomes a better and better approximation to think of the partons as individual entities that move collectively before and after the 197 . Therefore.3) When we collide a proton with another proton or an antiproton. the point-like quark. In general. and gluons. s = 14 TeV. and hadron-hadron scattering The last 10 years and at least the next 10 years of particle physics will be dominated by hadron-hadron √ colliders. summing over final state polarizations and averaging over initial state polarizations.

scattering, but are free particles at the moment of scattering. As a first approximation, we can consider
only the hard scattering processes, and later worry about adding on the various soft processes as part
of the higher-order corrections. This way of thinking about things allows us to compute cross-sections
for hadron scattering by first calculating the partonic cross-sections leading to a desired final state, and
then combining them with information about the multiplicity and momentum distributions of partons
within the hadronic bound states.
For example, suppose we want to calculate the scattering of a proton and antiproton. This involves
the following 2→2 partonic subprocesses:
qq → q ′ q ′ ,

qq ′ → qq ′ ,

qq′ → qq ′ ,

qg → qg,

qg → qg,

gg → gg,

(11.5.1)

where g is a gluon and q is any quark flavor and q ′ is a possibly different quark flavor. Let the center√
of-mass energy of the proton and antiproton be called s. Each parton only carries a fraction of the

energy of the energy of the proton it belongs to, so the partonic center-of-mass energy, call it sˆ, will

be significantly less than s. One often uses hatted Mandelstam variables sˆ, tˆ and u
ˆ for the partonic

scattering event. If sˆ ≫ ΛQCD , then the two partons in the final state will be sufficiently energetic

that they can usually escape from most or all of the spectator quarks and gluons before hadronizing
(forming bound states). However, before traveling a distance 1/ΛQCD , they must rearrange themselves
into color-singlet combinations, possibly by creating quark-antiquark or gluon-antigluon pairs out of
the vacuum. This hadronization process can be quite complicated, but will usually result in a jet of
hadronic particles moving with roughly the same 4-momentum as the parton that was produced. So,
all of the partonic process cross-sections in eq. (11.5.1) contribute to the observable cross-section for
the process:
pp → jj + X,

(11.5.2)

where j stands for a jet and the X includes stray hadronic junk left over from the original proton and
antiproton. Similarly, partonic hard scatterings like:
qq → q ′ q′ g,

qg → qgg,

qg → qgg,

gg → ggg,

(11.5.3)

and so on, will contribute to an observable cross-section for
pp → jjj + X.

(11.5.4)

The 2 × 2 hard scattering processes can also contribute to this process if one of the final state partons
hadronizes by splitting into two jets, or if there is an additional jet from the initial state.

In order to use the calculation of cross-sections for partonic processes like eq. (11.5.1) to obtain
measurable cross-sections, we need to know how likely it is to have a given parton inside the initial-state
hadron with a given 4-momentum. Since we are mostly interested in high-energy scattering problems,
we can make things simple, and treat the hadron and all of its constituents as nearly massless. (For
198

the proton, this means that we are assuming that

s ≫ mp ≈ 1 GeV.) Suppose we therefore take the

total 4-momentum of the hadron h in an appropriate Lorentz frame to be:
pµh = (E, 0, 0, E).

(11.5.5)

This is sometimes called the “infinite momentum frame”, even though E is finite, since E ≫ mp .

Consider a parton constituent A (a quark, antiquark, or gluon) that carries a fraction x of the hadron’s
momentum:
pµA = x(E, 0, 0, E).

(11.5.6)

The variable x is a standard notation, and is called the (longitudinal) momentum fraction for the
parton, or Feynman’s x. It is older than QCD, dating back to a time when the proton was suspected
to contain point-like partons with properties that were then obscure. In order to describe the partonic
content of a hadron, one defines:

Probability of finding a parton of type A with
 4-momentum between xpµ and (x + dx)pµ  = f h (x) dx .
A
inside a hadron h with 4-momentum pµ

(11.5.7)

The function fAh (x) is called the parton distribution function or PDF for the parton A in the hadron
h. The parton can be either one of the two or three valence quarks or antiquarks that are the nominal
constituents of the hadron, or one of an indeterminate number of virtual sea quarks and gluons. Either
type of parton can participate in a scattering event.
Hadronic collisions studied in laboratories usually involve protons or antiprotons, so the PDFs of
the proton and antiproton are especially interesting. The proton is nominally a bound state of three
valence quarks, namely two up quarks and one down quark, so we are certainly interested in the
up-quark and down-quark distribution functions
fup (x)

fdp (x).

and

The proton also contains virtual gluons, implying a gluon distribution function:
fgp (x).

(11.5.8)

Furthermore, there are always virtual quark-antiquark pairs within the proton. This adds additional
contributions to fup (x) and fdp (x), and also means that there is a non-zero probability of finding antiup,
antidown, or strange or antistrange quarks:
fup (x),

fdp (x),

fsp (x),

fsp (x).

(11.5.9)

These parton distribution functions are implicitly summed over color and spin. So fup (x) tells us the
probability of finding an up quark with the given momentum fraction x and any color and spin. Since
the gluon is its own antiparticle (it lives in the adjoint representation of the gauge group, which is
always a real representation), there is not a separate fgp (x).
199

[Although the charm, bottom and top quarks are heavier than the proton, virtual charm-anticharm,
bottom-antibottom, and top-antitop pairs can exist as long as their total energy does not exceed mp .
This can happen because, as virtual particles, they need not be on-shell. So, one can even talk about
the parton distribution functions fcp (x), fcp (x), fbp (x), fbp (x), ftp (x), ftp (x). Fortunately, these are small
so one can often neglect them, although they can be important for processes involving charm or bottom
quarks in the final state.]
Given the PDFs for the proton, the PDFs for the antiproton follow immediately from the fact that
it is the proton’s antiparticle. The probability of finding a given parton in the proton with a given
x is the same as the probability of finding the corresponding antiparton in the antiproton with the
same x. Therefore, if we know the PDFs for the proton, there is no new information in the PDFs for
the antiproton. We can just describe everything having to do with proton and antiproton collisions in
terms of the proton PDFs. To simplify the notation, it is traditional to write the proton and antiproton
PDFs as:
g(x) = fgp (x) = fgp (x),

(11.5.10)

fup (x),

(11.5.11)

d(x) = fdp (x) = fdp (x),

(11.5.12)

u(x) = fup (x) = fup (x),

(11.5.13)

d(x) = fdp (x) = fdp (x),

(11.5.14)

s(x) = fsp (x) = fsp (x),

(11.5.15)

s(x) = fsp (x) = fsp (x).

(11.5.16)

u(x) = fup (x) =

The PDFs are also functions of another parameter Q (sometimes denoted µ or µF ), known as the
factorization scale. The factorization scale can be thought of as the energy scale that serves as the
boundary between what is treated as the short-distance hard partonic process and what is taken to be
part of the long-distance physics associated with hadronization. The choice of the factorization scale
is an arbitrary one, and in principle the final result should not depend on it, just like the choice of
renormalization scale discussed in section 11.3 is arbitrary. It is very common to choose the factorization
scale equal to the renormalization scale, although this is not mandatory. (When they are distinguished
in the literature, some authors use Q for the renormalization scale and µ for the factorization scale,
and some use the reverse as we have here. Some use the same letter for both, especially when choosing
them to be the same numerically.) The PDFs have a mild logarithmic dependence on the choice of
factorization scale Q, which can be computed in perturbation theory by a set of equations known as the
DGLAP (Dokshitzer–Gribov–Lipatov–Altarelli–Parisi) equations. So they are really functions g(x, Q2 )
etc., although the Q-dependence is often left implicit for brevity, as we will mostly do below.
It is usual to choose the factorization scale Q to be comparable to some energy scale relevant to the
physical process of interest. For example, in deeply inelastic scattering of leptons off of protons, with
momentum transfer to the scattered quark q µ , the factorization scale is typically chosen as Q2 = −q 2 .
200

This Q2 is positive, since q µ is a spacelike vector. When producing a pair of heavy particles with some
mass m, it is common to choose Q = m or some fraction thereof. After having completed a calculation,
one often varies the renormalization and factorization scales (either together or independently) over a
range (say, from Q = m/4 to Q = 2m in the case just mentioned) to see how the cross-section or other
observables that resulted from the calculation vary. This is a test of the accuracy of the perturbation
theory calculation, since in principle if one could calculate exactly rather than to some low order in
perturbation theory, the results should not depend on either scale choice at all.
In the proton, antiquarks are always virtual, and so must be accompanied by a quark with the
same flavor. This implies that if we add up all the up quarks found in the proton, and subtract all the
anti-ups, we must find a total of 2 quarks:
Z

1

0

h

i

dx u(x, Q2 ) − u(x, Q2 ) = 2.

(11.5.17)

Similarly, summing over all x the probability of finding a down quark with a given x, and subtracting
the same thing for anti-downs, one has:
Z

1
0

h

i

dx d(x, Q2 ) − d(x, Q2 ) = 1.

(11.5.18)

Most of the strange quarks in the proton come from the process of a virtual gluon splitting into a strange
and anti-strange pair. Since the virtual gluon treats quarks and antiquarks on an equal footing, for
every strange quark with a given x, there should be† an equal probability of finding an antistrange
with the same x:
s(x, Q2 ) = s(x, Q2 ).

(11.5.19)

The up-quark PDF can be thought of as divided into a contribution uv (x) from the two valence quarks,
and a contribution us (x) from the sea (non-valence) quarks that are accompanied by an anti-up. (Here
and below the factorization scale dependence is left implicit, for brevity.) So we have:
u(x) = uv (x) + us (x),

us (x) = u(x);

(11.5.20)

d(x) = dv (x) + ds (x),

ds (x) = d(x).

(11.5.21)

There is also a constraint that the total 4-momentum of all partons found in the proton must be equal
to the 4-momentum of the proton that they form. This rule takes the form:
Z

1

dx x[g(x) + u(x) + u(x) + d(x) + d(x) + s(x) + s(x) + . . .] = 1,

(11.5.22)

0

or more generally, for any hadron h made out of parton species A,
XZ
A

0

1

dx xfAh (x) = 1.

(11.5.23)

Although QCD interactions do not change quark flavors, there is a small strangeness violation in the weak interactions,
so the following rule is not quite exact.

201

Each term xfAh (x) represents the probability that a parton is found with a given momentum fraction
x, multiplied by that momentum fraction. One of the first compelling pieces of evidence that the
gluons are actual particles carrying real momentum and energy, and not just abstract group-theoretic
constructs, was that if one excludes them from the sum rule eq. (11.5.22), only about half of the
proton’s 4-momentum is accounted for:
Z

0

1

dx x[u(x) + u(x) + d(x) + d(x) + s(x) + s(x) + . . .] ≈ 0.5.

(11.5.24)

If we could solve the bound state problem for the proton in QCD, like one can solve the hydrogen
atom in quantum mechanics, then we could derive the PDFs directly from the Hamiltonian. However,
we saw in section 11.3 why this is not practical; perturbation theory in QCD is not accurate for studying low-energy problems like bound-state problems, because the gauge coupling becomes very large at
low energies. Instead, the proton PDFs are measured by experiments including those in which charged
leptons and neutrinos probe the proton, like ℓ− p → ℓ− + X and νp → ℓ− + X. Several collabora-

tions perform fits of available data to determine the PDFs, and periodically publish updated result
both in print and as computer code. Two such collaborations are CTEQ (Coordinated TheoreticalExperimental Project on QCD) and MSTW (Martin, Stirling, Thorne, Watt), formerly MRST (Martin,
Roberts, Stirling, Thorne). In each case, the PDFs are given in the form of computer codes obtained
by fitting to experimental data. Because of different techniques and weighting of the data, the PDFs
from different groups are always somewhat different.
As an example, let us consider the CTEQ5L PDF set. Here, the “5” says which update of the PDFs
is being provided, this one from the year 2000, and the “L” stands for “lowest order”, which means
it is the set appropriate when one only has the lowest-order calculation of the partonic cross-section.
This set is somewhat old; I used it only because it was relatively easy to evaluate, since it is given in
parameterized function form rather than interpolation table form. There are more recent sets CTEQ6
and CT10 from CTEQ, and MSTW2008 from MSTW. Each of these has a version appropriate for
lowest-order work, and other versions appropriate when one has the next-to-leading order (NLO) or
next-next-leading order (NNLO) formulas for the hard scattering process of interest. At Q = 10 GeV,
the CTEQ5L PDFs for u(x), u(x), and the valence contribution uv (x) ≡ u(x) − u(x) are shown below,
together with a similar graph for the down quark and antiquark distributions:

202

6 0. since there is only one valence down quark to find in the proton. but the distribution dv (x) is of course only about half as big as uv (x).6 x f(x) 0.4 0.8 1 Here I follow tradition by graphing x times the PDF in each case. let us look at the strange and gluon PDFs: 1 CTEQ5L. and thus a smaller fraction x of the proton’s total momentum.8 d(x) d(x) = ds(x) 0.2 0.4 x 0.4 203 x 0.4 0.2 0. since they all tend to get large near x = 0.2.2 0 0 0.d(x) 0. the chance that a virtual particle can appear is greater when it carries a smaller energy. This is a general feature of sea partons. the sea quark distribution u(x) is strongly peaked near x = 0. Q=10 GeV CTEQ5L.4 x 0. In contrast.6 x f(x) x f(x) 0. We see from the first graph that the valence up-quark distribution is peaked below x = 0.8 0 0 1 0. The sea distribution d(x) is not very different from that of the anti-up. Next.8 1 . The solid curve shows the total up-quark PDF for this value of Q.8 uv(x) = u(x) . with a long tail for larger x (where an up-quark is found to have a larger fraction of the proton’s energy). Q=10 GeV u(x) u(x) = us(x) 0.2 0. There is even a significant chance of finding that an up quark has more than half of the proton’s 4-momentum.2 0.6 0. Q=10 GeV g(x) s(x) = s(x) 0.8 0.1 1 CTEQ5L.6 0.u(x) dv(x) = d(x) .4 0.2 0 0 0.6 0.

04 for quarks. More generally. Larger Q is appropriate for probing the proton at larger energy scales.2 0. and let the partonic differential cross-sections for the desired final state X be dˆ σ (ab → X) (11.8 0. The PDFs s(x) = s(x) are suppressed by the non-zero strange quark mass. The next two graphs show the CTEQ5L PDFs for Q = 100 and 1000 GeV: 1 1 CTEQ5L. the variation with Q can be made quantitative using the DGLAP equations.The parton distribution function for gluons grows very quickly as one moves towards x = 0. If X has two particles 1. then one defines partonic Mandelstam variables: sˆ = (pa + pb )2 . one will often be studying events with a much larger characteristic energy scale.6 0.8 1 0 0 0. This is because there are 8 gluon color combinations available. Consider scattering two hadrons h and h′ . The value of the factorization scale Q = 10 GeV corresponds roughly to the appropriate energy scale for many of the experiments that were actually used to fit for the PDFs. at the Tevatron and LHC.4 x 0. the PDFs become larger at very small x (although this > is hard to see from the graphs). or shorter distance scales.2 0.4 d 0.2 d u 0 0 CTEQ5L.5. but smaller for x > ∼ 0.4 x 0. Q=100 GeV g 0.6 0.2 0.25) for any two partons a (to be taken from h) and b (from h′ ).4 d 0. This explains why s(x) < d(x).8 0.2 in it. since this imposes a penalty on making virtual strange and antistrange quarks. like Q ∼ mt for top events and perhaps Q ∼ 1000 GeV for supersymmetry events at the LHC.6 x f(x) x f(x) u 0. and let us see how to use them to get a cross-section. and each virtual gluon can give rise to more virtual gluons because of the 3-gluon and 4-gluon vertex. 204 (11.6 0. Now suppose we have available a set of PDFs. Q=1000 GeV g u d u 0. However.8 1 As Q increases from Q = 100 to 1000 GeV.015 for gluons and x ∼ 0. The hat is used as a reminder that this is a partonic process.5. This means that the chance of finding a gluon gets very large if one requires that it only have a small fraction of the total 4-momentum of the proton.26) . which are built into the computer codes that provide the parton distributions as a function of x and Q.

but there is also a small but non-zero probability of finding an anti-quark in the 205 . let us consider top-antitop production. E). with s = (ph + ph′ )2 for the whole hadronic event.5. then integrate over all possible xa and xb . (11. The partonic differential cross-section dˆ σ (ab → X) depends on the momentum fractions xa and xb through pµa = xa pµh and pµb = xb pµh′ . using PDFs with Q chosen equal to some energy characteristic of the event. 0. −E). 0. 0.34) This integration is done by computer.5.28) Let us work in the center-of-momentum frame.27) u ˆ = (pa − k2 )2 . 0. uu → tt. ss → tt. h′ collisions.1) These are listed in the order of their numerical importance in contributing to the total cross-section for tt at the Tevatron. (11.5. first at the Tevatron. 0. Notice that the most likely thing is to find a quark in the proton and an antiquark in the anti-proton.5. (11.5. dd → tt.5.33) Here s is determined by the collider [(1960 GeV)2 at the Tevatron]. xa and xb .tˆ = (pa − k1 )2 . dd → tt. E).32) It follows that. with pµh and pµh′ controlled or known by the experimenter. so: pµa = xa (E.b 0 1 dxa Z 0 1 ′ dxb dˆ σ (ab → X) fah (xa ) fbh (xb ).5. with the first parton taken to be from the proton and the second from the antiproton: uu → tt. we should multiply the partonic cross-section by the probabilities of finding in h a parton a with momentum fraction in the range xa to xa + dxa and the same probability for parton b in h′ .30) = (E.31) pµb = xb (E. gg → tt. −E).5. Then we can define a Feynman x for each of the initial-state partons. 11. 0. and then sum over all the different parton species a and b. 0. to find the total cross-section to produce the final state X in h.6. (11. so that pµh = (E. (11. with approximately massless hadrons and partons. The result is: ′ dσ(hh → X) = XZ a. ss → tt. Now. while sˆ is different for each event.29) pµh′ (11. (11.6 Top-antitop production in pp and pp collisions As an example. 0. sˆ = (xa + xb )2 E 2 − (xa − xb )E 2 = 4xa xb E 2 = sxa xb . (11. The parton-level processes that can contribute to this are. (11.

4) Therefore. for each partonic process one has σ ˆ = Z tˆmax tˆmin dˆ σ ˆ dt .8) s = 1960 GeV. the range of xa is from sˆ/s to 1. corresponding to xa = xb = 1). Even these leading-order partonic differential cross-sections depend implicitly on the renormalization scale µ.6. In order to find the total cross-section. one can first integrate the partonic cross-sections with respect to tˆ.33) implies xb = sˆ . dtˆ (11.6. (11. The limits of integration on sˆ are from sˆmin = 4m2t (the minimum required to make a top-antitop pair) to sˆmax = s (the maximum available from the proton and antiproton. since they are related linearly by   q sˆ s . −1 + cos θˆ 1 − 4m2t /ˆ tˆ = m2t + 2 (11. All of the processes involving quark and antiquark in the initial state involve the same parton-level cross-section dˆ σ (qq → tt) .1082. Using the CTEQ5L PDFs and mt = 173 GeV. The gluon-gluon process has a partonic cross-section that is somewhat harder to obtain:  ˆ) m2 (ˆ s − 4m2 ) dˆ σ (gg → tt) πα2s 6(m2 − tˆ)(m2 − u − = 8ˆ s2 sˆ2 dtˆ 3(m2 − tˆ)(m2 − u ˆ) 2 2 2 2 4[(m − tˆ)(m − u ˆ) − 2m (m + tˆ)] 4[(m2 − tˆ)(m2 − u ˆ) − 2m2 (m2 + u ˆ)] + + 2 2 2 2 3(m − u ˆ) 3(m − tˆ)  2 2 2 2 2 ˆ) + m2 (tˆ − u ˆ)] 3[(m − tˆ)(m − u ˆ) + m (ˆ u − tˆ)] 3[(m − tˆ)(m − u − − .6. this is equivalent to integrating over the final-state top quark angle θˆ in the partonic COM frame. (11.5) where   q sˆ s .3) sˆ(m2 − u ˆ) sˆ(m2 − tˆ) where m is the mass of the top quark. (11. For a given sˆ. xa s dxb = dˆ s . xa s (11.3. and a quark in the anti-proton.26) starting from αs (mt ) = 0. tˆmax.5.6) It is also useful to note that eq.7) So instead of integrating over xb . through the renormalized coupling αS (µ).proton. we can integrate over sˆ. dtˆ (11.6.6. with √ (11.min = m2t + −1 ± 1 − 4m2t /ˆ 2 (11. and computing αs (µ) using eq.6.6. Relabeling xa as just x. we therefore have: σ(pp → tt) = Z Z  h 1 s/xs) + d(x)d(ˆ s/xs) + u(x)u(ˆ s/xs) dˆ s dx σ ˆ (qq → tt) u(x)u(ˆ xs 4m2t sˆ/s s 1 i  +d(x)d(ˆ s/xs) + 2s(x)s(ˆ s/xs) + σ ˆ (gg → tt)g(x)g(ˆ s /xs) . and working with the leading-order partonic cross-sections.2) which is left for you to compute as an exercise. the results as a function of the common factorization and renormalization scale Q = µ look like: 206 .

but there is instead a strong dependence of the leading-order prediction on Q = µ.1765 for mt = 173 GeV.9) So at least one of the momentum fraction x’s must be larger than 0.5. Accurate comparisons with experiment require a much more detailed and sophisticated treatment of the higher-order effects. and u. including at least a next-to-leading order calculation of the partonic cross-sections.0312.+ ss 0. s. so according to eq. The gluon-gluon contribution is suppressed in this case because most of the gluons are at small x and do not have enough energy to make a top-antitop pair. is clearly not very high. Ideally. (11. Finally. To produce a top-antitop pair. Also. xa xb > 4m2t /s = 4(173)2 /(1960)2 = 0.6. (11. Since both of the initial- 207 . s. the ratio of top-antitop events produced from up-antiup should be about 4 times that from down-antidown. the contributions from sea partons (u. we must have sˆ > 4m2t . the lines should be flat. s in the proton. Still. although a principled justification for this scale choice is hard to make. a pp collider.2 0.33). Experience has shown that evaluating the leading-order result at Q ∼ mt /2 gives a decent estimate of the total cross-section. the accuracy of the above results. and twice as many antiups as antidowns in the antiproton. d. The higher-order corrections to the quark-antiquark processes turn out to be of order 10 to 20%. Since there are roughly twice as many valence up quarks as down quarks in the proton for a given x. the relative sizes of the parton-level contributions can be understood qualitatively from the PDFs as follows. This means that the largest contributions come from the valence quarks. s in the antiproton) are highly suppressed for the same reason. obtained with only leading order partonic crosssections and PDFs. some useful information can be gleaned. Let us now consider tt production for the Large Hadron Collider.1 σ(tt) [pb] at Tevatron 10 Total uu - 0 dd 10 gg -1 10 - dd -uu -2 10 ss. by taking into √ account the larger s and the different parton distribution function roles. d. while the gluon-gluon process gets about a 70% correction from its leading-order value at Q = µ = mt .5 Q/mt 1 2 Unfortunately.

and is why 208 . the formula for the total cross-section is now: σ(pp → tt) = Z s 4m2t dˆ s Z 1 dx sˆ/s  h i 1 s/xs) + 2d(x)d(ˆ s/xs) + 2s(x)s(ˆ s/xs) σ ˆ (qq → tt) 2u(x)u(ˆ xs  +ˆ σ (gg → tt) g(x)g(ˆ s/xs) . but that is not the main reason. the top quark can be considered light (!) and so one can make them √ using partons with much lower x. (11.6.0247.+ -cc cc 0 100. the kinematic constraint on the longitudinal momentum fractions becomes xa xb > 4(173)2 /(14000)2 = 0. or the antiquark. with s = 14 TeV.+ -ss ss . This is actually a common feature. The most striking feature of the LHC result is that in contrast to the Tevatron situation. Numerically integrating the above formula with a computer using the CTEQ5L PDFs. and subject to large corrections at next-to-leading order and beyond. because the gluon distribution is identical in protons and in antiprotons.6. This is partly because all of the quark-antiquark contributions now require a sea antiquark PDF.+ uu uu . (11. The gluon-gluon contribution has the same form as in pp collisions. and they dominate over the quark and antiquark PDFs. then the contribution of the other proton is fixed.11) so that now the smaller one can be as low as 0.- dd + dd 1 10 .state hadrons are protons. one finds the leading order results shown below. one has plenty of gluons available with less than 1/10 of the protons’ total energy.10) The factors of 2 are present because each proton can contribute either the quark. we saw above that the gluon distribution function is very large.5 Q/mt 1 2 Again these leading-order results are highly dependent on Q = µ. For example. the gluon-gluon partonic contribution is dominant over the quark-antiquark contributions for the LHC. as a function of the common factorization and renormalization scale Q = µ: 3 σ(tt) [pb] at 14 TeV LHC 10 Total gg 2 10 . The really important effect is that at very high energies like at the LHC.2 0. At low x.000611.

and any other unobserved degrees of freedom.10) Tevatron LHC at 14 TeV -1 (1/σ)dσ(tt)/dMinv [GeV ] unperformed. To make things simple.8) and (11.002 0 300 400 500 600 700 800 tt Minv [GeV] 900 1000 The invariant mass distribution of the tt system is peaked not far above 2mt in both cases. There are some processes that do not rely on gluons at all. indicating that the top and antitop usually have only semi-relativistic velocities. so the most important contributions to the production cross-section occur when both x’s are not very far above their minimum allowed values. we should be able to compute the differential cross-section for the partonic event from its 209 . The resulting shape of the distribution normalized by the total cross-section. by leaving the sˆ integration in eqs. many processes get a large contribution from gluon-squark scattering as well. Those processes are dominated by sea quarks at the LHC. After doing a sum/average over spins. so what we are about to do does not work for tt in the final state (but could be generalized to do so). At the LHC. however. with a more substantial tail at high mass.6.12) = σ(tt) dMinv σ(tt) dˆ s √ is shown below for the Tevatron and the LHC with s = 14 TeV: 0. This is because the PDFs fall rapidly with increasing x. for example gluon-squark production in supersymmetry. the top and antitop are likelier to be produced with higher energy than at the Tevatron. (11. Also.7 Kinematics in hadron-hadron scattering Let us now consider the general problem of kinematics associated with hadron-hadron collisions with underlying 2 → 2 parton scattering. with so much energy available for the protons. colors.004 0. 11. (11. One can also look at the distribution of tt production as a function of the total invariant mass √ Minv = sˆ of the hard scattering process. We will see one example in subsection 11.8. many processes are dominated by the large gluon PDF at low x. we will suppose all of the particles are essentially massless.you sometimes hear people somewhat whimsically call the LHC a “gluon collider”.6.6. √ 1 dσ(tt) 2 s dσ(tt) .006 0.

ˆ pµa = (E.1) As we learned in subsection 11. 210 (11. It is equal to 0 if the final-state particles are back-to-back.5. xb . p2T . E).) The angular dependence about the collision axis is trivial. Working in the center-of-momentum frame of the partons. For example. (Once the magnitude of the momentum transverse to the beam for one particle is specified. tˆ).7. pT . we can use the following three variables: momentum of particle 1 transverse to √ the collision axis. Instead of the variables (xa . Y ). There are three significant degrees of freedom: two angles at which the final-state particles emerge with respect to the collision axis.3) This may look like an obscure definition. the other is determined. we can write: ˆ 0. sˆ. defined by 1 Y = ln(xa /xb ).b dˆ σ (ab → 12) ˆ dt dxa dxb . dtˆ (11. but it is the rapidity (see section 2) needed to boost along the collision axis to get to the center-of-momentum frame for the two-parton system.2) A picture of the scattering process in real space might look like: 1 k1 ph → h fah xa ph → a dˆ σ dtˆ ← xb p h ′ ′ fbh b ← ph′ h′ k2 2 There are several different ways to choose the kinematic variables describing the final state. 0. we can use the more directly observable variables (ˆ s. and one overall momentum scale. 2 (11.4) . dtˆ (11.7.7.7. the total center-of-momentum energy of the final state partons. which would occur in the special case that the initial-state partons have the same energy in the lab frame. we can then write: dσ(hh′ → 12 + X) = X ′ fah (xa )fbh (xb ) a. so we can ignore it.Feynman diagrams as: dˆ σ (ab → 12) . and the longitudinal rapidity of the two-parton system in the lab frame. or perhaps some subset of these with the others integrated over.

by taking the inverse of the determinant of its inverse. Making the change of variables (xa . −E). tˆ) to (ˆ s. sˆ sxa xb (11.6) q ˆ 0.7. (11. Y ) for a differential cross-section requires dtˆdxa dxb = J dˆ s d(p2T ) dY.ˆ 0. k2µ = (E.7.12) where J is the determinant of the Jacobian matrix of the transformation. − E ˆ 2 − p2 ).10) and so p2T = tˆu ˆ tˆ(sxa xb + tˆ) =− . q (11.7) with ˆ 2 = xa xb s. p2T .7. one finds: . pT . E T (11.9) (11. t = (pa − k1 ) = 1 + 1 − 4p2T /ˆ 2   q sˆ 2 2 u ˆ = (pa − k2 ) = s . sˆ = 4E (11. k1µ = (E.7.7.5) ˆ 2 − p2 ). T (11.7. xb . 0.11) where the last equality uses u ˆ = −ˆ s − tˆ = −sxa xb − tˆ for massless particles.7. Evaluating it. ˆ pµb = (E.7. −pT . 1 − 1 − 4pT /ˆ 2 (11. ˆ 0.8) from which it follows that   q sˆ 2 ˆ s .

.

−1 .

∂(ˆ .

2 xa xb .

s . Y ) . pT .

J =.

. .

= .

∂(tˆ. xb ) . xa .

it still depends on QCD.13) Therefore. eq. sˆ + 2tˆ (11. xb .8. or τ are singlets under SU (3)c color.3).7.8). (11. because to evaluate it we need to know the 211 .b dtˆ (11. (11. (11.14) where the variables xa .2) becomes: xa xb X h ′ dˆ σ (ab → 12) dσ(hh′ → 12 + X) = fa (xa )fbh (xb ) . let us consider the process of Drell-Yan scattering. (11.7. tˆ on the right-hand side are understood to be determined in terms of sˆ.8 Drell-Yan scattering (ℓ+ ℓ− production in hadron collisions) As an example. 2 ˆ dˆ s d(pT ) dY sˆ + 2t a. However. and (11.7.11).7. since the leptons ℓ = e.7. 11. µ.1) This does not involve QCD as the hard partonic scattering. p2T and Y by eqs.7. which is the production of lepton pairs in hadron-hadron collisions through a virtual photon: hh′ → ℓ+ ℓ− .

8. can be obtained by exactly the same methods as in 7. 9ˆ s (11. Y ) requires: .8.8. and to average over initial-state colors. The total partonic cross-section obtained by integrating eq.2) with the q coming from either h or h′ . in h: h i X 2πα2 tˆ2 + (ˆ dσ(hh′ → ℓ+ ℓ− ) s + tˆ)2 2 h h′ h h′ x x Q = f (x )f (x ) + f (x )f (x ) a b a q a q b b .3) is: σ ˆ (qq → ℓ+ ℓ− ) = 4πα2 Q2q . One finds that the differential partonic cross-section is: 2πα2 Q2q (tˆ2 + u ˆ2 ) dˆ σ (qq → ℓ+ ℓ− ) .8. the underlying partonic process is always: qq → ℓ+ ℓ− . 9ˆ s q (11. and the antiquark.PDFs for the quarks inside the hadrons.2 for √ e+ e− s ≪ mZ and not near a → µ+ µ− . there is no reaction if the colors do not match. for resonance. Alternatively. we just need to remember to use the charge of the quark Qq instead of the charge of the electron.5) Therefore we get: dσ(hh′ → ℓ+ ℓ− ) = dxa dxb i 4πα2 X 2 h h ′ ′ Qq fq (xa )fqh (xb ) + fqh (xa )fqh (xb ) . and Y .6) Now making the change of variables from (xa . pT .8.3) Therefore. (11. this will eliminate the integration over p2T .4) This can be used to make a prediction for the experimental distribution of events with respect to each of sˆ. The cross-section for this. xb ) to (ˆ s. we can choose to integrate the partonic differential cross-section with respect to tˆ first. using u ˆ = −ˆ s − tˆ for massless scattering. The latter effect leads to a suppression of 1/3.8. and writing separate contributions from finding the quark. = 3ˆ s4 dtˆ (11. q q q 3 dˆ s dp2T dY sˆ4 (ˆ s + 2tˆ) q (11. (11. Since gluons have no electric charge and do not couple to photons. Since tˆ determines the scattering angle with respect to the collision axis.

8.7) i 4πα2 X 2 h h ′ ′ dσ(hh′ → ℓ+ ℓ− ) Qq fq (xa )fqh (xb ) + fqh (xa )fqh (xb ) . (11. = dˆ s dY 9sˆ s q (11.8) .8.

we therefore have: .

Y ) . ∂(ˆ s.

.

dˆ s dY = dxa dxb = .−1 dˆ s dY .

.

.

xb ) s Still another way to present the result is to leave only sˆ unintegrated. and then trading one of the Feynman-x variables for sˆ. ∂(xa . by again first integrating the partonic differential cross-section with respect to tˆ. and do the remaining x-integration. This is how we wrote the top-antitop total cross-section. The Jacobian factor in the change of variables from (xa . xb ) → (xa . sˆ) is now: .

−1 .

dxa dˆ s .

s= . dxa dˆ .

x s a b .

.

∂ˆ s dxa dxb = .

.

8. ∂x 212 (11.9) .

the Drell-Yan process pp → µ− µ+ + X or pp → µ− µ+ + X (11. (11. π ± p. Furthermore. (11.495 GeV.8. 2 2 2 2 2 2 2 2 3 sˆ (ˆ s − mZ ) + mZ ΓZ (ˆ s − mZ ) + mZ Γ2Z q (11. Drell-Yan scattering has been studied in hh′ = pp. while the second comes from the virtual Z boson contribution.8. the CMS and ATLAS LHC detector collaborations released their measurements of the µ− µ+ and e− e+ invariant mass distributions. Since the PDFs run slowly with Q. the functions Fhh′ gives information about the PDFs.12) is often one of the first things one studies at a hadron collider to make sure everything is working correctly and understood.10) This version makes a nice prediction that is (almost) independent of the actual parton distribution functions. which depends on the mass and width mZ = 91. and will be specified explicitly at the end of section 13.Using this in eq.8. The resulting cross-section can be obtained from eq.11) to a very good approximation.8. It also provides a test of the PDFs. At the end of 2010. for low sˆ not near a resonance. The first term is from the virtual photon contribution to the matrix element.1. pp. and features a Breit-Wigner resonance denominator from the Z boson propagator. we see that it is predicted to scale like 1/ˆ s2 times some function of the ratio sˆ/s. √ For larger s.10) by replacing 4πα2 X 2 Q → 3ˆ s2 q q " # (Vq2 + A2q )(Vℓ2 + A2ℓ ) 2Qq Vq Vℓ (1 − m2Z /ˆ s) 4πα2 X Q2q + − . this is a reasonably good prediction. The dimuon plot from CMS is shown below: 213 . after replacing xa by x and integrating: Z i 4πα2 X 2 1 sˆ h h dσ(hh′ → ℓ+ ℓ− ) h h′ h′ = (ˆ s /xs) + f (x)f (ˆ s /xs) . one must take into account the s-channel Feynman diagram with a Z boson in place of the virtual photon.8. and in each case the results indeed satisfy the scaling law: sˆ2 dσ(hh′ → ℓ+ ℓ− ) s/s) = Fhh′ (ˆ dˆ s (11.19 GeV and ΓZ = 2.6) now gives.8. The right-hand side could have depended on both sˆ and s in an arbitrary way. but to the extent that the PDFs are independent of Q.13) where Vf and Af are coupling coefficients associated with the Z–fermion–antifermion interaction vertex. Because of the relatively clean signals of muons in particle detectors. The last term is due to the interference between these two amplitudes. especially at small x. Q dx f (x)f q q q q q dˆ s 3ˆ s2 q xs sˆ/s (11. and K ± p scattering experiments.

the √ Z boson resonance.Clearly visible are the effects of the η. and 3S meson resonances. 214 . 2S. J/ψ. ψ ′ . and Υ 1S. ω. ρ. and the general decrease of the cross-section with sˆ (the invariant mass of the final state). A similar plot is shown below for the e− e+ final state. φ. The distribution includes the effects of the intrinsic widths of the resonances as well as detector resolution effects and trigger and detector efficiencies. Note that the detector resolution and the trigger efficiency are evidently worse for electrons than for muons.

3) where α is any constant.25)]: δV δφ∗ δV ∂ µ ∂µ φ∗ + δφ ∂ µ ∂µ φ + = 0. The classical equations of motion for φ and φ∗ following from L are [see eq.5) Clearly. (12. rather than remaining randomized.12 Spontaneous symmetry breaking 12. There are two types of continuous symmetry transformations. like a stick balanced on one end. We will work out how spontaneous symmetry breaking works in each of these cases. The situation in which the laws of physics are invariant under some symmetry transformations.1.1) V (φ.6) . In this section we will study how this works in quantum field theory. then the minimum of the potential is at φ = 0. (5.1. (12.1. φ∗ ).4) = 0. (12. and local (or gauge) in which the transformation can be different at each point. the ground state of a ferromagnetic system involves a magnetization vector that points in some particular direction. This is because it is energetically favorable for the magnetic moments in the material to line up. The quantum mechanical counterpart of this statement is that the ground state of the system will be one in which the expectation value of φ(x) vanishes: h0|φ(x)|0i = 0.1. in which the transformation does not depend on position in spacetime. using the example of a U (1) symmetry. If m2 > 0 and λ > 0. but the vacuum state is not.1 Global symmetry breaking Not all of the symmetries of the laws of physics are evident in the state that describes a physical system. Consider a complex scalar field φ(x) with a Lagrangian density: L = ∂ µ φ∗ ∂µ φ − V (φ.1.1. is called spontaneous symmetry breaking. The state with randomized magnetic moments is unstable to small perturbations. in condensed matter physics. even though Maxwell’s equations in matter do not contain any special direction. there is a solution with ∂µ φ = ∂µ φ∗ = 0. This Lagrangian is invariant under the global U (1) transformations: φ(x) → φ′ (x) = eiα φ(x) (12. and will settle in the more energetically-favored magnetized state. φ∗ ) = m2 φ∗ φ + λ(φ∗ φ)2 .2) with potential energy where m2 and λ are parameters of the theory. global. For example. or even in the vacuum state with no particles.1. (12. where φ(x) is just equal to any constant that minimizes the potential. 215 (12. and then guess the generalizations to non-Abelian symmetries.

gaining an infinite amount of kinetic energy. However.1. Classically there would be runaway solutions in which |φ(x)| → ∞. canceling out of eq.11) (12. They have squared mass equal to m2 .1. then the potential V (φ.10) The VEV is not invariant under the U (1) symmetry operation acting on the fields of the theory. by an arbitrary choice. φ∗ ) is unbounded from below for arbitrarily large |φ|. with a local maximum at φ = 0.1. and the symmetry is spontaneously broken.1. there is nothing wrong with the theory if m2 < 0 and λ > 0. so it is impossible to unambiguously determine the phase of φ(x) at the minimum.1. One cannot restore the invariance by defining the symmetry operation to also multiply |0i by a phase.The scalar particles created and destroyed by the field φ(x) correspond to quantized oscillations of φ(x) about the minimum of the potential. Let us now consider what happens if the signs of the parameters m2 and λ are different. This cannot lead to an acceptable theory.7) −m2 /λ. and a degenerate set of minima with v2 . 2 (12. φ∗ ) has a “Mexican hat” shape. Therefore. 2 1 φ∗ (x) = √ [v + R(x) − iI(x)] .1. the system will have a ground state in which the expectation value of φ(x) is constant and equal to the classical minimum: v h0|φ(x)|0i = √ . it is convenient to rewrite the scalar field in terms of its deviation from its VEV. The quantum mechanical counterpart of this statement is that the expectation value of φ(x) will grow without bound.9) 2 The quantity v is a measurable property of the vacuum state. 2 216 (12. (One should think of m2 as simply a parameter that appears in the Lagrangian density. 2 2 (12. since h0| will rotate by the opposite phase. known as the vacuum expectation value. In order to further understand the behavior of this theory.10). In quantum mechanics. of φ(x). we can make Im(φmin ) = 0. (12. If we now ask what the VEV of φ is after performing a U (1) transformation of the form eq. If λ < 0. The sign of the parameter m2 is what determines whether or not spontaneous symmetry breaking takes place in the theory. and interact with a four-scalar vertex proportional to λ.) In that case.3).1. the potential V (φ.8) |φmin |2 = where we have defined: v = q The potential V does not depend on the phase of φ(x) at all. this reflects the fact that we had to make an arbitrary choice of phase. the vacuum state must not be invariant under the global U (1) symmetry rotation. However. One way to do this is to write: 1 φ(x) = √ [v + R(x) + iI(x)] . and not as the square of some mythical real number m. (12. (12.12) .1. (12. or VEV. we find: v v h0|φ′ (x)|0i = eiα h0|φ(x)|0i = eiα √ 6= √ .

The first term proportional to λv 2 is a mass term for R.1.20) instead of eqs. (5. but there is no term quadratic in I.12). this becomes: V (R.where R and I are each real scalar fields.1. 4 (12.1.1. (12. and plugging in eqs.1.1.19) (12.1.11). we can identify the physical particle masses: m2R = 2λv 2 = −2m2 . m2I = 0.22) (12.1.1.1. 4 (12. (12.21) Notice that the field G does not appear in V at all. Comparing to the Klein-Gordon Lagrangian density of eq. we compute: ∂µ φ = ∂µ φ∗ = i(v + h) 1 √ eiG/v [∂µ h + ∂µ G]. To find the derivative part of the Lagrangian. so it corresponds to a massless real scalar particle.18).15) λ 2 (R + I 2 )2 . I) = λ [(v + R)2 + I 2 − v 2 ]2 4 = λv 2 R2 + λvR(R2 + I 2 ) + (12. Again h(x) and G(x) are two real scalar fields. G does have interactions coming from the part of the Lagrangian density containing derivatives.17) (12. This is because G just corresponds to the phase of φ. v 2 217 (12. and the potential was chosen to be invariant under U (1) phase transformations. 2 1 √ [v + h(x)]e−iG(x)/v .1.14) Dropping the last term that does not depend on the fields.11) and (12. related to R(x) and I(x) by a non-linear functional transformation. (12.1.1. the potential is: V (h) = λv 2 h2 + λvh3 + λ 4 h .18) It is useful to redo this analysis in a slightly different way.13) The potential appearing in the Lagrangian can be found in terms of R and I most easily by noticing that it can be rewritten as V (φ. (12. as: 1 1 L = ∂ µ R∂µ R + ∂ µ I∂µ I. representing the real and imaginary parts of φ. 2 2 (12.1. However. In terms of these fields.1. v 2 1 i(v + h) √ e−iG/v [∂µ h − ∂µ G]. 2 (12. and IIII interaction vertices.1. we can interpret the terms proportional to λv as RRR and RII interaction vertices. φ∗ ) = λ(φ∗ φ − v 2 /2)2 − λv 4 /4. The derivative part of the Lagrangian can now be rewritten in terms of R and I.23) . and the last term proportional to λ as RRRR.12). RRII. by writing φ(x) = φ∗ (x) = 1 √ [v + h(x)]eiG(x)/v . (12.16) Comparing this expression with our previous discussion of real scalar fields in section 6.

g (12. 12.1.1.3) In order to make a gauge-invariant Lagrangian density.1. 218 (12.1) (12.2. This is an example of Goldstone’s theorem.25) with m2h = 2λv 2 . As a simple example.21) and (12.29) (12.24) The quadratic part of the Lagrangian. which determines the propagators for h and G.2 Local symmetry breaking and the Higgs mechanism Let us now consider how things change if the spontaneously broken symmetry is local.3. h → h′ = h. ψ(x) → e−iQθ(x) ψ(x). (12. (12. consider a U (1) gauge theory with a fermion ψ with charge Q and gauge coupling g and a vector field Aµ transforming according to: ψ(x) → eiQθ(x) ψ(x). The original U (1) symmetry acts on G by shifting it by a constant that depends on the VEV: G → G′ = G + αv.30) This explains why G only appears in the Lagrangian with derivatives acting on it. v 2v 4 (12. or gauged. the ordinary derivative is replaced by the covariant derivative: Dµ ψ = (∂µ + iQgAµ )ψ.24) is: Lint =   1 1 λ h + 2 h2 ∂ µ G∂µ G − λvh3 − h4 .28) The field and particle represented by G is known as a Nambu-Goldstone boson (or sometimes just a Goldstone boson). which we will state in a more general framework in subsection 12. (12. In general.1. (12.1. 1 Aµ (x) → Aµ (x) − ∂µ θ(x).1.2. (12. The interaction part of the Lagrangian following from eqs.26) m2G = 0. is Lquadratic = 1 µ 1 1 ∂ h∂µ h − m2h h2 + ∂ µ G∂µ G.1.2. 2 2 2 (12.1.2) (12.27) This confirms the previous result that the spectrum of particles consists of a massive real scalar (h) and a massless one (G). a broken global symmetry is always signaled by the presence of a massless Nambu-Goldstone boson with only derivative interactions.4) .1.2.so that: Lderivatives =  1 h 1 µ ∂ h∂µ h + 1+ 2 2 v 2 ∂ µ G∂µ G.

19). 2 1 √ [∂µ h − igVµ (v + h)] e−iG/v .7) where V (φ.9). 4 4 219 (12. (12.5) To make a gauge-invariant Lagrangian. If m2 > 0 and λ > 0. if m2 < 0. φ∗ ) is as given before in eq.2.1. we again replace the ordinary derivative acting on φ. 4 (12. It transforms under a gauge transformation like φ(x) → eiθ(x) φ(x). (12. However. (12. φ∗ by covariant derivatives: Dµ φ∗ = (∂µ − igAµ )φ∗ .2.8) this Lagrangian is easily checked to be gauge-invariant.2.12) Note also that since ∂µ Aν − ∂ν Aµ = ∂µ Vν − ∂ν Vµ . then the minimum of the potential for the scalar field brings about a non-zero √ p VEV h0|φ|0i = v/ 2 = −m2 /2λ. (12. (12. with self-interactions with a coupling proportional to λ. φ∗ ) − F µν Fµν .2.2.10) since this is the combination that appears in eq. gv (12.2. Dµ φ = (∂µ + igAµ )φ.1. in order to spontaneously break the gauge symmetry. (12.6) The Lagrangian density for the scalar and vector degrees of freedom of the theory is: 1 L = Dµ φ∗ D µ φ − V (φ. (12.9) It is convenient to define a new vector field: Vµ = Aµ + 1 ∂µ G. gv 2 (12.2).Now. just as in the global symmetry case. and interaction with the massless vector field Aµ . φ∗ (x) → e−iθ(x) φ∗ (x).2. then this theory describes a massive scalar.2.13) the vector field strength part of the Lagrangian is the same written in terms of the new vector Vµ as it was in terms of the old vector Aµ : 1 1 − F µν Fµν = − (∂µ Vν − ∂ν Vµ )(∂ µ V ν − ∂ ν V µ ).2. Then Dµ φ = Dµ φ∗ = 1 √ [∂µ h + igVµ (v + h)] eiG/v . we introduce a complex scalar field φ with charge +1.14) . Because the covariant derivative of the field transforms like Dµ φ → eiθ Dµ φ.2. 2 (12.11) (12. one finds:   1 1 Dµ φ = √ ∂µ h + ig(Aµ + ∂µ G)(v + h) eiG/v . Using the same decomposition of φ into real fields h and G as given in eq.2.

but since this is true for any θ. The words used to describe this are that the vector boson becomes massive by “eating” the would-be Nambu-Goldstone boson G. The original field φ(x) is called a Higgs field.2. (12. we have given a mass to the corresponding vector field: m2V = g2 v 2 .2. (12. each transverse to its direction of motion. 2 4 (12. (12. namely m2h = 2λv 2 .2. which becomes its longitudinal polarization degree of freedom.2. A massless spin-1 vector boson (like the photon) has only two possible polarization states. we see that it has the same squared mass as in the global symmetry case.17) This means that by spontaneously breaking the gauge symmetry.2.The complete Lagrangian density of the vector and scalar degrees of freedom is now: L= i 1 1h µ ∂ h∂µ h + g2 (v + h)2 V µ Vµ − F µν Fµν − λ(vh + h2 /2)2 . we can choose it to be proportional to the Nambu-Goldstone field itself: θ(x) = −G(x)/v.15) This Lagrangian has the very important property that the field G has completely disappeared! Reading off the part quadratic in h.19) behaves under a gauge transformation as: φ → eiθ φ.2. In contrast. and one longitudinal (parallel) to its direction of motion.20) Normally we think of θ in this equation as just some ordinary function of spacetime. The additional polarization state degree of freedom had to come from somewhere. Recall that φ = 1 √ (v + h)eiG/v 2 (12.21) . and the surviving real scalar degree of freedom h(x) is called by the generic term Higgs boson. a massive spin-1 vector boson has three possible polarization states. = − F µν Fµν + 4 2 (12. An alternative way to understand what has just happened to the would-be Nambu-Goldstone boson field G(x) is that it has been “gauged away”. the two transverse. This is called the Higgs mechanism. 220 (12. The Standard Model Higgs boson and the masses of the W ± and Z bosons result from a slightly more complicated version of this same idea.18) We can understand why the disappearance of the field G goes along with the appearance of the vector boson mass as follows.16) There is also a term quadratic in the vector field: LV V g2 v2 µ 1 V Vµ . so one real scalar degree of freedom had to disappear. as we will see.2.

In unitary gauge.22) Notice also that the gauge transformation eq. The propagators of the Higgs scalar and the massive vector are: µ ν " ←→ i pµ pν −gµν + 2 2 2 p − mV + iǫ mV ←→ i p2 − m2h + iǫ # From eq.15).21) gives exactly the term in eq.2. known as “unitary gauge”. (12. φ(x) = 1 √ [v + h(x)]. (12. However. eliminates G(x) completely. 2 (12.2.2.10). just as we saw in eq. in unitary gauge. (12.2. but the ease with which one can compute those results depends on picking the right gauge for the problem at hand. a fermion with charge Q inherits the same interactions with Vµ that it had with Aµ . The physical predictions of the theory do not depend on which gauge one chooses.This choice. (12.15). coming from the covariant derivative: 221 . The advantage of unitary gauge is that the true physical particle content of the theory (a massive vector and real Higgs scalar) is more obvious than in the version of the Lagrangian written in terms of the original fields φ and Aµ . it turns out to be easier to prove that the theory is renormalizable if one works in a different gauge in which the would-be Nambu-Goldstone bosons are retained.2. Let us catalog the propagators and interactions of this theory. there are also hV V and hhV V interaction vertices: µ µ ←→ 2ig2 vgµν ←→ 2ig2 gµν ←→ −6iλ ν ν and hhh and hhhh self-interactions: ←→ −6iλv Finally. so that Vµ is simply the unitary gauge version of Aµ .

without ruining renormalizability.2) Tiaj vj = 0 (12. The broken generators satisfy Tiaj vj 6= 0.3.3) Any group generators that satisfy correspond to unbroken symmetry transformations. each of the would-be Nambu-Goldstone bosons is eaten by the vector field with the corresponding index a. If one could succeed in concocting a theory in which a fermion spinor field or a vector field has a VEV: h0|Ψα |0i = 6 0 (?).] In the case of a local or gauge symmetry. so there was just one Nambu-Goldstone boson. First.3. without proof. The vector fields for the broken generators become massive.3.6) 222 . the unbroken global symmetry group is the one formed from the unbroken symmetry generators. (12. φ∗i ).4) Goldstone’s Theorem states that for every spontaneously broken generator.3. One might also ask whether it is possible for fields other than scalars to obtain vacuum expectation values. There are also Higgs boson(s) for the uneaten components of the scalar fields that obtained VEVs. with squared masses that can be computed in terms of the VEV(s) and the gauge coupling(s) of the theory. In general. of a global symmetry group. This of course depends on the parameters and couplings appearing in V . The fields φi will then have VEVs that can be written: vi h0|φi |0i = √ .3. (12. (12. how all of the considerations above generalize to arbitrary groups. (12.5) h0|Aµ |0i = 6 0 (?).3. 12.3 Goldstone’s Theorem and the Higgs mechanism in general Let us now state.µ ←→ −iQgγ µ This is a general way of making massive vector fields in gauge theories with interacting scalars and fermions. There is some potential V (φi . there must be a corresponding Nambu-Goldstone boson. and the vacuum state is invariant under this unbroken subgroup. [The group U (1) has just one generator. labeled by a. suppose we have scalar fields φi in some representation of a global symmetry group with generators Tiaj .1) which we presume has a minimum where at least some of the φi are non-zero. 2 (12.

s quarks are rotated into each other and right-handed u. For example. (12. s quarks are rotated into each other. in contradiction with experiment. these are a ubiquitous feature of technicolor models. or PNGBs. m2π ≪ m2p . and therefore transforms non-trivially under the Lorentz group. so these antifermion-fermion VEVs do not break SU (3)c symmetry.) Chiral symmetry breaking is actually the mechanism that is responsible for most of the mass of the proton and the neutron. since the alleged VEV carries an uncontracted spinor or vector index. but the Goldstone theorem successfully explains why they are much lighter than the proton.3. They are often called pseudo-Nambu-Goldstone bosons.then Lorentz invariance will necessarily be broken. (12. In fact. d. approximate symmetry by which left-handed u. However. Extensions of the Standard Model that feature new approximate global symmetries that are spontaneously broken generally predict the existence of heavy exotic PNGBs. with the “pseudo” indicating that the associated spontaneously broken global symmetry was only an approximate symmetry. in QCD the quark-antiquark composite fields do have vacuum expectation values: h0|uu|0i ≈ h0|dd|0i ≈ h0|ss|0i ≈ µ3 6= 0. When the chiral symmetry is spontaneously broken. and therefore most of the mass of everyday objects. the Nambu-Goldstone bosons that arise include the pions π ± and π 0 . 223 .8) where µ is a quantity with dimensions of [mass] which is set by the scale ΛQCD at which non-perturbative effects become important. This would imply that the broken generators would include Lorentz boosts and rotations.7) This is called a fermion-antifermion condensate. This is known as chiral symmetry breaking. The chiral symmetry is a global. (The objects qq are color singlets. there can be vacuum expectation values for antifermion-fermion composite fields.3. since they can form a Lorentz scalar: h0|ΨΨ|0i = 6 0. d. They are not exactly massless because the chiral symmetry was really only approximate to begin with.

Left-handed SM fermions are known to be doublets under SU (2)L :  νe eL  . A viable theory must explain the qualitative experimental facts that the W ± bosons couple only to L-fermions (and R-antifermions). and the masses of the leptons and most of the masses of the heavier quarks.  cL sL  . the gauge symmetry breaking is: SU (2)L × U (1)Y → U (1)EM . cR .4) and a coupling constant g.1.1 SU(2)L × U(1)Y representations and Lagrangian In this section. τR .3) with corresponding vector gauge boson fields: Wµa . uR . The Standard Model (SM) of electroweak interactions of Glashow. Weinberg and Salam successfully incorporates all of these features and tests into a spontaneously broken gauge theory.1. After spontaneous symmetry breaking. The weak hypercharge Y is a conserved 224 . 3) (13. The SU (2)L subgroup is known as weak isospin. that the Z boson couples differently to L-fermions and R-fermions.  νµ µL  . The right-handed fermions eR . (13. the U (1)Y subgroup is known as weak hypercharge and has a coupling constant g′ and a vector boson Bµ .13 The standard electroweak model 13. The electroweak interactions are mediated by three massive vector bosons W ± . µR . Also.2) Notice that the electric charge of the upper member of each doublet is always 1 greater than that of the lower member.  uL dL  .  tL bL  . tR .  ντ τL  . (13. dR . In the SM. there are very stringent quantitative experimental tests involving the relative strengths of fermion-antifermionvector couplings and the ratio of the W and Z masses. (a = 1. Meanwhile. (13. sR .1. the remaining unbroken gauge group is electromagnetic gauge invariance. (a = 1. which is responsible for the masses of W ± and Z bosons.1.5) are all singlets under SU (2)L . The SU (2)L representation matrix generators acting on these fields are proportional to the Pauli matrices: T a = σ a /2. Z and the massless photon γ. 2. 2. sometimes known as the hyperphoton. bR . 3) (13. but γ couples with the same strength to L-fermions and R-fermions. we will study the Higgs mechanism of the Standard Model.1) We will need to introduce a Higgs field to produce this pattern of symmetry breaking.1. The gauge group before spontaneous symmetry breaking must therefore have four generators.

1. Wµ1 − iWµ2 must carry electric charge +1 and Wµ1 + iWµ2 must † Some references define the weak hypercharge normalization so that Y is a factor of 2 larger than here.1.10) where YℓL and YℓR are the weak hypercharges of left-handed leptons and right-handed leptons. Lgauge = − Wµν 4 4 (13. A multiplicative factor can always be absorbed into the definition of the coupling g′ .2.3).19)].1. the pure-gauge part of the electroweak Lagrangian density is: 1 a aµν 1 W − Bµν B µν . (13. 2 2   g g = ∂µ eL + i g′ YℓL Bµ − Wµ3 eL + i (Wµ1 + iWµ2 )νe .  νe eL  . 2 2 ′ = ∂µ eR − ig Bµ eL . the covariant derivatives of the lepton fields can be summarized as: Dµ νe Dµ eL Dµ eR   g g = ∂µ νe + i g YℓL Bµ + Wµ3 νe + i (Wµ1 − iWµ2 )eL . For example.11) Therefore. (13.1.9). Both members of an SU (2)L doublet must have the same weak hypercharge in order to satisfy SU (2)L gauge invariance. (13. so without loss of generality.1. ′ (13. the covariant derivatives acting on the lepton fields are:   h i νe Dµ = ∂µ + ig′ Bµ YℓL + igWµa T a eL   Dµ eR = ∂µ + ig′ Bµ YℓR eR . it is traditional† to set YℓR = Qℓ = −1.13) (13. and Wµa self-interactions.9) (13.12) (13. Following the general discussion of Yang-Mills gauge theories in section 10. This Lgauge provides for kinetic terms of the vector fields.charge just like electric charge q. (13. 225 . but it is different for left-handed and right-handed fermions.2.1.1. The totally antisymmetric ǫabc (with ǫ123 = +1) are the structure constants for SU (2)L .8) where: are the SU (2)L and U (1)Y field strengths. then. Using the explicit form of the SU (2)L generators in terms of Pauli matrices in eq.7) Bµν = ∂µ Bν − ∂ν Bµ (13. The interactions of the electroweak gauge bosons with fermions are determined by the covariant derivative.14) The covariant derivative of a field must carry the same electric charge as the field itself.6) a Wµν = ∂µ Wνa − ∂ν Wµa − gǫabc Wµb Wνc . Evidently.1. in order for charge to be conserved.2 [see eqs.1.18) and (10. (13. and 2 × 2 unit matrices are understood to go with the ∂µ and Bµ terms in eq.1. the covariant derivative of left-handed leptons is: Dµ  νe eL  = ∂µ  νe eL   ′ + i g YℓL  Bµ 0 0 Bµ  g + 2  Wµ3 1 Wµ + iWµ2 Wµ1 − iWµ2 −Wµ3   νe eL  .1. The weak hypercharges of all other fermions are then fixed. (10. for each particle.

1. (9.6. we will find that they mix.3). As a result of spontaneous symmetry breaking. . one finds using eq.1. . .18) √ The 1/ 2 normalization agrees with our previous convention.14) can be written: Dµ eR = ∂µ eR − ig′ cos θW Aµ eR + ig′ sin θW Zµ eR .21) Comparing to Dµ eR = ∂µ eR − ieAµ eR from QED.4). known as the weak mixing angle or Weinberg angle:  =   =   Wµ3 Bµ  Zµ Aµ Zµ Aµ  . The covariant derivative of the right-handed electron field eq.1. Zµ . 2 2 (13.1. 2 (13. we find that to reproduce the weak-interaction Lagrangian of muon decay. . The vector bosons Bµ and Wµ3 are both electrically neutral. . (13.1. but are orthogonal linear combinations of these two gauge eigenstate fields. we must have: Wµ+ ≡ Wµ− ≡ 1 √ (Wµ1 − iWµ2 ). One can write the relation between the gauge eigenstate and mass eigenstate fields as a rotation in field space by an angle θW . (13. the real reason for it is so that the kinetic terms for W ± have a standard normalization: L = − 21 (∂µ Wν+ − ∂ν Wµ+ )(∂ µ W −ν − ∂ ν W −µ ). and the other is the massive Z boson vector field.20) sin θW cos θW  − sin θW cos θW  cos θW − sin θW with the inverse relation: cos θW sin θW We now require that the resulting theory has the correct photon coupling to fermions.22) Similarly. 2 (13. (9. Consider the interaction Lagrangian following from eL ) γ µ Dµ L = i ( νe  νe eL  (13. and (9. (13.1.1.19) Wµ3 Bµ  . (13.6. (13. (13.1.carry electric charge −1. the fields with well-defined masses (“mass eigenstates” or “mass eigenfields”) are not Bµ and Wµ3 . In other words.17) (13. One of the mass eigenstates is the photon field Aµ .7. so these must be identified with the W ± bosons of the weak interactions.2).19) that:  ′ Dµ eL = ∂µ eL + i g YℓL cos θW 226  g − sin θW Aµ eL + . 2 1 √ (Wµ1 + iWµ2 ).1.1.1.16) Comparing with eqs.23) . we conclude that: g′ cos θW = e.15) g g = − ν e γ µ eL (Wµ1 − iWµ2 ) − eL γ µ νe (Wµ1 + iWµ2 ) + . by requiring that the field Aµ appears in the covariant derivatives in the way dictated by QED.

24).1. The results can be summarized in terms of the SU (3)c × SU (2)L × U (1)Y representations:       νe νµ ντ .31) g′ = 0. . running parameters. −1) (3.1. tR .Again comparing to the prediction of QED that Dµ eL = ∂µ eL − ieAµ eL . (13. However.24) is the electromagnetic coupling.25) where the . In a similar way.29) so that g′ .1.32) e = 0. 1.652. . .1. (13.26) Now combining eqs.35) . (13.357. ←→ bR .1.1.30) These are requirements that will have to be satisfied by the spontaneous symmetry breaking mechanism. sin θW = p 2 g + g′2 g cos θW = p 2 . ←→ . cR . . (13.1. (13. g + g′2 (13. it must be that: g sin θW − g′ YℓL cos θW = e 2 (13.1.26).1.231. − 21 ) (1.27) (13. sR . . So. (13. µ . (13. 1 6) (3. . In the same way:  Dµ νe = ∂µ νe + i g′ YℓL cos θW +  g sin θW Aµ νe + .1. The numerical values from experiment are approximately: g = 0.1. evaluated at a renormalization scale µ = mZ = 91.22). (3.34) These are all renormalized. 2 (13. and therefore its covariant derivative cannot involve the photon. we learn that: YℓL = −1/2. 2 (13. represent W and Z terms. ←→ dR .1. 1.1.313. bL sL dL uR . 2. one can work out what the weak hypercharges of all of the other SM quarks and leptons have to be. τ . (13. in order to reproduce the correct electric charges appearing in the coupling to the photon field Aµ from the covariant derivative. (13.28) (13. the coefficient of Aµ in eq. and (13.1876 GeV in the MS scheme. ←→  R  R  R tL cL uL . 227 (1. 2 3) − 13 ). we know that the neutrino has no electric charge.33) sin2 θW = 0.1.1. .25) must vanish: g sin θW + g′ YℓL cos θW = 0. g + g′2 tan θW = g′ /g. ←→ eL µL τL e . 2.1. gg′ e = p 2 . 1.

44) BR(Z → hadrons) = 0. the electric charge of any field f is given in terms of the eigenvalue of the 3 component of weak isospin matrix.45) The “invisible” branching ratio matches up extremely well with the theoretical prediction for the sum over the three νℓ ν ℓ final states. LZf f = −Z µ f γµ (gVf − gA (13.1.1. (13.36) Here Tf3 is +1/2 for the upper component of a doublet.1. τ .37) where  g  3 TfL − sin2 θW Qf . while “hadrons” is due to quark-antiquark final states.40) with  g 1 f f gL + gR = 2 2 cos θW  1 f g f = gL − gR = 2 2 cos θW gVf = f gA    Tf3L − 2 sin2 θW Qf . νµ . This is unfortunate. b Tf3L 1 2 − 21 1 2 − 21 Y fL − 21 − 12 1 6 1 6 Y fR Qf 0 0 −1 −1 2 3 − 31 2 3 − 31 Equation (13. (11.2000 ± 0. since backgrounds for leptons are smaller than for hadrons or missing energy.42)  f The coupling parameters appearing in eq.13) are Vf = gVf /e and Af = gA /e.6991 ± 0. (13. T 3 .1.1. each) (13.43) BR(Z → invisible) = 0. µ. and 0 for an SU (2)L singlet. (13. and Z bosons can appear in many searches for new phenomena.1. cos θW  g  2 = − sin θW Qf . as: Qf = Tf3 + Yf .1. (13. 228 .8.1.41) Tf3L . s. and agree with the results from experiment: BR(Z → ℓ+ ℓ− ) = 0.1. cos θW gLf = g cos θW Tf3L − g′ sin θW YfL = (13. (13. −1/2 for the lower component of a doublet. c.39) with coefficients: fermion νe .1. and the weak hypercharge Y .38) f gR = −g′ sin θW YfR (13. It is an important fact that the Z branching ratio into charged leptons is small.37) can also be rewritten in terms of vector and axial-vector couplings to the Z boson: f γ5 )f.033658 ± 0. τ u.0006 (13. One finds for each SM fermion f : f LZf f = −Z µ f γµ (gLf PL + gR PR )f.1.0006. ντ e.In general. The partial decay widths and branching ratios of the Z boson can be worked out from these couplings (see homework).000023 (for ℓ = e. The couplings of the SM fermions to the Z boson then follow as a prediction. µ. t d.

but we will follow the simplest possibility. of the 0 potential. ).2.2. (13. (13.9) Without loss of generality.1.2. By definition.2. so the component 229 . Φ† ) = m2 Φ† Φ + λ(Φ† Φ)2 .2) (13. 2.4) (13.6) are gauge singlets. This will ensure the spontaneous symmetry breaking that we demand.2 The Standard Model Higgs mechanism Let us now turn to the question of how to spontaneously break the electroweak gauge symmetry in a way that satisfies the above conditions. rather than a minimum.8)   0 is a local maximum.3) The Hermitian conjugate field transforms as: Φ† → Φ′† = Φ† e−iθ SU (2)L : a σ a /2 Φ† → Φ′† = Φ† e−iθ/2 . the surviving U (1) gauge symmetry is U (1)EM .36). v= q −m2 /λ. (13.2. We can therefore build a gauge-invariant potential: V (Φ.13.5) It follows that the combinations Φ† Φ D µ Φ† Dµ Φ and (13. 2 ←→ (13.2. U (1)Y : (13. provided that m2 < 0. (13.2. U (1)Y : . Φ(x) → Φ′ (x) = eiθ(x)/2 Φ(x). (13. and entirely in the second (electrically neutral) component of the Higgs field: h0|Φ|0i =  0√ v/ 2  . we know that they carry electric charges +1 and 0 respectively from eq. we can choose the VEV of Φ to be real. There are degenerate Now. Under gauge transformations. Φ† ).10) This is a convention.7) and the Lagrangian density for Φ is: L = D µ Φ† Dµ Φ − V (Φ.2.2. There is actually more that one way to do this. then Φ = minima of the potential with Φ† Φ = v 2 /2.1) Each of the fields φ+ and φ0 is a complex scalar field. which can always be achieved by doing an SU (2)L gauge transformation on the field Φ to make it so. (13.2. Φ transforms as: Φ(x) → Φ′ (x) = eiθ SU (2)L : a (x)σ a /2 Φ(x). which is to introduce a complex SU (2)L -doublet scalar Higgs field with weak hypercharge YΦ = +1/2: Φ=  φ+ φ0  1 (1.

2.19) . +µ ′ µ 3µ 2gW g B − gW 1  (13. The covariant derivative of Φ in unitary gauge is: Dµ Φ = 1 √ 2  0 ∂µ h   i g′ g +√ Bµ + Wµa σ a 2 2 2  0 v+h  (13. an electrically neutral real scalar boson that has yet to be discovered experimentally. The Higgs field Φ has two complex.2. 2 2 2 2 (13. D Φ Dµ Φ = ∂µ h∂ µ h + 2 4 2 µ † 230 (13. The would-be Nambu-Goldstone fields can be removed by going to unitary gauge. 2.2. Therefore.2).14) where Ga (x) (a = 1. respectively.2. following the example of section 12.2.5).13) Comparing with the QED gauge transformation rule of eq.2. (13. (13. (13. we can write it as: iGa (x)σa /2v Φ(x) = e  0 v+h(x) √ 2  .2.12) φ0 → φ0 .2. (13. 1 (v + h)2 D µ Φ† Dµ Φ = ∂µ h∂ µ h + (0 2 8 √  2gWµ− g′ Bµ + gWµ3 √ 1) 2gWµ+ g′ B − gW 3 √ µ −µ µ    ′ µ 3µ 2gW 0 gB √ + gW .18) or. so four real. (10.17) Therefore.16) and its Hermitian conjugate is: Dµ Φ † =   g′ i g 1 √ ( 0 ∂µ h ) − √ ( 0 v + h ) Bµ + Wµa σ a . We can now plug this into the Lagrangian density of eq.2. The U (1)EM gauge transformations acting on Φ are a combination of SU (2)L and U (1)Y transformations:  ′ Φ → Φ = exp iθ  1 0 0 0  Φ.   (v + h)2 2 + −µ 1 1 g Wµ W + (gWµ3 − g′ Bµ )(gW 3µ − g′ B µ ) . with θ a = −Ga /v. (13. so that in the unitary gauge we have simply Φ(x) =  0 v+h(x) √ 2  .15) The field h creates and destroys the physical Higgs particle.1. to find interactions and mass terms for the remaining Higgs field h and the vector bosons.2. (13. in components. This completely eliminates the Ga from the Lagrangian. which means performing an SU (2)L gauge transformation of the form of eq. (13. 3) and h(x) are each real scalar fields. φ+ → eiθ φ+ . The Ga are would-be Nambu-Goldstone bosons.11) or.2. we see that indeed φ+ and φ0 have charges +1 and 0. corresponding to the three broken generators in SU (2)L × U (1)Y → U (1)EM .2. after simplifying.8). scalar field degrees of freedom.of Φ obtaining a VEV must be the one assigned 0 electric charge.

1876 ± 0. The specific prediction is: g2 v2 (g2 + g′2 )v 2 .2. (13. 4 4 which agrees with the experimental values provided that the VEV is approximately: √ v = 2hφ0 i = 246 GeV m2W = (13.21) are Higgs-vector-vector and Higgs-Higgs-vector-vector couplings.34). (13.2.This can be further simplified using: gWµ3 − g′ Bµ = q i h g2 + g′2 cos θW Wµ3 − sin θW Bµ = q g2 + g′2 Zµ .2. (13.2.2.2.24) All of the above predictions are subject to small. the present experimental values mW = 80.22) (13. (13. m2Z = . in agreement with the fact that U (1)EM gauge invariance remains unbroken. The remaining terms in eq.20). we have successfully shown that the photon remains massless.0005.2.1.25) which is significantly lower than the MS-scheme running value in eq. (13. (It should be noted that there is another extremely common convention √ in which v is defined to be a factor 1/ 2 smaller than here.0021 GeV yield: on−shell sin2 θW ≡ 1 − m2W /m2Z = 0.1. vindicating the earlier assumption of neutral vector boson mixing with the form that we took for the sine and cosine of the weak mixing angle.23) in the conventions used here. 2 4 2 µ † (13.399 ± 0. so that v = hφ0 i = 174 GeV in that convention. but measurable. So finally we have: LΦ kinetic   1 (v + h)2 2 + −µ 1 2 = D Φ Dµ Φ = ∂µ h∂ µ h + g Wµ W + (g + g′2 )Zµ Z µ .21) The parts of this proportional to v 2 make up (mass)2 terms for the W ± and Z vector bosons. (13. (13.20) where the first equality uses eq.023 GeV and mZ = 91. This part of the Lagrangian density implies the following unitary gauge Feynman rules: µ W± µ ν " i pµ pν −gµν + 2 2 2 p − mW + iǫ mW # Z ν " pµ pν i −gµν + 2 2 2 p − mZ + iǫ mZ Wν Zµ h h Wµ Zν igµν (g2 + g′2 )v/2 igµν g2 v/2 231 # .) A non-trivial prediction of the theory is that mW /mZ = cos θW . Since there is no such (mass)2 term for the photon field Aµ .1. loop corrections.30) and the second uses eq. For example.2226 ± 0.

ρσ Zµ Wρ Aµ Wρ Zν Wσ Zν Wσ −ig2 cos2 θW X µν.h Wµ h Zµ h Wν h Zν igµν (g2 + g′2 )/2 igµν g2 /2 with the arrow direction on W ± lines indicating the direction of the flow of positive charge. and also yields 3-gauge-boson and 4-gauge-boson interactions: q Aµ Zµ Wν p −ie[gµν (p − q)ρ + gνρ (q − k)µ + gρµ (k − p)ν ] k Wρ q Wν p −ig cos θW [gµν (p − q)ρ + gνρ (q − k)µ + gρµ (k − p)ν ] k Wρ Wµ Wν Aµ Wσ Wσ Wρ Aν Wρ ig2 X µν.2. Z propagators above. (13.ρσ = 2gµν gρσ − gµρ gνσ − gµσ gνρ .ρσ −ig2 sin θW cos θW X µν.1.6) provides the momentum part of the W . 232 (13. The field-strength Lagrangian terms of eq.26) .ρσ −ie2 X µν.ρσ where: X µν.

Therefore. 4 (13. Chiral fermions have the property that they cannot have masses without breaking the symmetry that makes them chiral.2.3) where.2.3. 13. There are indirect effects of the Higgs mass in loops that do affect some present experiments. to avoid any confusion between between (eL ) and (e)PL .27) just as in the toy model studied in 12.3 Fermion masses and Cabibbo-Kobayashi-Maskawa mixing The gauge group representations for fermions in the Standard Model are chiral.28) It would be great if we could evaluate this numerically using present data.2. e = PL eL + PR eR . For example. the Higgs potential V (Φ.3. we explicitly define eL ≡ e†L γ 0 . In unitary gauge: V (h) = λv 2 h2 + λvh3 + λ 4 h . This means that the left-handed fermions transform in a different representation than the right-handed fermions. there is no present experiment that directly sheds light on λ. but these bounds are not very strong. These experiments suggest that mh cannot be too large.4) . (13. especially if we consider that the SM is probably not the final story.Finally. Unfortunately.1) The Dirac spinor for the electron can be separated into left. eR ≡ e†L γ 0 . (13. suppose we try to write down a mass term for the electron: Lelectron mass = −me ee.2) and the corresponding barred spinor as: e = (e†L PL + e†R PR )γ 0 = e†L γ 0 PR + e†R γ 0 PL = eL PR + eR PL . the Higgs boson has self-interactions with Feynman rules: −6iλv −6iλ and a mass mh = √ 2λv.3. while we know what the Higgs VEV v is.and right-handed pieces.3. (13. 233 (13. (13. Φ† ) gives rise to a mass and self-interactions for h.

10) or.5) The point is that this is clearly not a gauge singlet. (13.3. and is not neutral under U (1)Y . the eL part of each term transforms as a doublet under SU (2)L . as does the Higgs field. (13. The field ( ν e (13. while the second term has Y = 1/2. (13.3.3. In the first place. the doublets transform under SU (2)L as:  ( νe φ+ φ0  −iθ a σa /2 → e eL ) → ( ν e  φ+ φ0 eL ) e+iθ  .9) so eq.8) .1) can therefore be written: Lelectron mass = −me (eL eR + eR eL ).12) . it becomes: ye LYukawa = − √ (v + h)(eL eR + eR eL ). for any Standard Model fermion f . Here ye is a Yukawa coupling of the type we studied in 8.7) is also an SU (2)L singlet. so each term is an SU (2)L doublet.7) eL ) carries weak hypercharge +1/2.3.15).3. 2 (13.3. 2 and.3. with Feynman rule: √ −iye / 2 234 (13. fermion masses can still arise with the help of the Higgs field. there is a gauge-invariant term: Lelectron Yukawa = −ye ( ν e eL )  φ+ φ0  eR + c.3. an electron-positron-Higgs interaction vertex. equal to ye v me = √ . (13. More generally. as a bonus. as required. the naive mass term Lf mass = −mf (f L fR + f R fL ) (13. so the whole term is a U (1)Y singlet. Moreover. and eR carries weak hypercharge −1. All terms in the Lagrangian must be gauge singlets in order not to violate the gauge symmetry. Furthermore. Going to the unitary gauge of eq.3.11) This can now be interpreted as an electron mass. and the eR is a singlet.Equation (13. and so is not allowed. reassembling the Dirac spinors without projection matrices: ye LYukawa = − √ (v + h)ee.c.6) is not an SU (2)L singlet.3. the first term has Y = YeR − YeL = −1/2. Fortunately.3. 2 (13. so the electron mass is disqualified from appearing in this form. a σ a /2 (13. For the electron.2.

The Higgs mechanism not only explains the masses of the W ± and Z bosons. For example. but also explains the masses of fermions.16) which transforms as an SU (2)L doublet in exactly the same way that Φ does: Φ → eiθ a σ a /2 iθ a σa /2 ˜ → e Φ Φ.13) Unfortunately. we have a b-quark mass and an hbb vertex: yb L = − √ (v + h)bb. the bottom quark mass comes from the gauge-invariant Yukawa coupling: L = −yb ( tL bL )  φ+ φ0  bR + c. b) in the Standard Model arise in exactly the same way. (13..19) Going to unitary gauge.c. it disappears in unitary gauge from that term. 246 GeV (13. Also.3.94 × 10−6 . It is convenient to define ˜≡ Φ  0 1 −1 0  Φ∗ =  φ0∗ −φ+∗  .15) The situation is slightly different for up-type quarks (u. (13. this is so small that we can forget about ever observing the interactions of the Higgs particle h with an electron.14) implying that. s.3. 2 (13. in unitary gauge.3. t). (The field φ+∗ has a negative electric charge. Φ one can write a gauge-invariant Yukawa coupling for the top quark as: L = −yt ( tL bL )  φ0∗ −φ+∗  tR + c. (13.3.) Therefore.3. and the down-type quarks (d.20) In all cases.3. Notice that all of the particles in the Standard 235 . Notice that although the neutrino participates in the Yukawa interaction.c. (13.17) ˜ Φ.511 MeV  2 = 2.3.3. 2 (13. we can compute the electron Yukawa coupling: ye = √  0. because the complex conjugate of the field Φ must appear in order to preserve U (1)Y invariance. c.3.Since we know the electron mass and the Higgs VEV already. 2 (13.18) ˜ has weak hypercharge Y = −1/2. the unitary-gauge version of the gauge-invariant Yukawa interaction is yf L = − √ (v + h)f f. (13.21) The mass and the h-fermion-antifermion coupling obtained by each Standard Model fermion in this way are both proportional to yf . one finds that the top quark has a mass: yt L = − √ (v + h)tt. Masses for all of the other leptons.

3. τL   dR dRj =  sR  .c.23) where yq is the running (renormalized) Yukawa coupling evaluated at a renormalization scale µ = mq .Model (except the photon and gluon.3. and must be taken into account in precision work. . and even charm quarks. for the W ± and Z bosons it depends on gauge couplings. They get only a few MeV of their mass from the Higgs field. for the up and down quarks. This was the reason for √ the notation mf (more nearly yf v/ 2 than mf ) that was used in 8. coming from nonperturbative QCD effects. tL (13.3. (13.s.3. bL    eL ℓLj =  µL  . .) are smaller but still significant.3.25) (13. bR uLj 236  νe νi =  νµ  . then these chiral symmetry breaking effects would do it anyway.28) . For the top.3.µ. it is actually the dominant effect. and for the Higgs particle itself. dL φ0  0∗  φ i  yu i j uRj + c. For each quark. which must remain massless because of SU (3)c × U (1)EM gauge invariance) get a mass from spontaneous electroweak symmetry breaking of the form: mparticle = kv. tR (13.c.c. for example when predicting the branching ratios of the Higgs boson decay. φ0  + i  φ yd i j dRj + c. The two-loop and higher corrections (indicated by .. dL −φ+∗ i ℓL  (13. The QCD gluon-loop correction increases mt by roughly 6%. not the Standard Model Higgs field.26) Here i. A second qualification is that there is actually another source of quark masses..c.τ Yukawas = − ν i Ld. If the Higgs field did not break SU (2)L × U (1)Y .3. the physical mass measured from kinematics is   4αs yq v mq = √ 1 + + . as has already been mentioned at the end of section 12. it is the Higgs self-coupling λ. gluon loops make a large modification to the tree-level prediction.b Yukawas = − uiL Lu.22) where k is some combination of dimensionless couplings. For the fermions. and has an even larger effect on mb because αs is larger at the renormalization scale µ = mb than at µ = mt . using an antifermion-fermion VEV rather than a scalar VEV.. the most general form of the Yukawa interactions is actually: Le. 3π 2 (13. so that: ℓRj dLj   eR =  µR  .24) (13.3. ντ   uL =  cL  . it is proportional to a Yukawa coupling.. This gives contributions to all quark masses that are roughly of order ΛQCD . j are indices that run over the three families.t Yukawas = − uiL   φ+ ye i j ℓRj + c. The Standard Model fermions consist of three families with identical gauge interactions. Therefore. this is relatively insignificant. the most important source of mass in ordinary nuclear matter is really chiral symmetry breaking in QCD. bottom. .27) uRj   uR  = cR  . Therefore. First. τR   dL =  sL  . There are two notable qualifications for quarks.3. However.

.32) where MD is diagonal with positive real entries.) It therefore appears that the masses of Standard Model fermions are actually 3× 3 complex matrices. to distinguish them from the basis we are about to introduce. 2 (13.30) v mf i j = √ yf i j . thanks to the following: Mass Diagonalization Theorem. L=− 1+ v (13. consider the following redefinition of the lepton fields: ℓ′Ri = LRi j ℓRj . the Yukawa interaction Lagrangian can be written as:  o h n ′i ′i ℓL mei j ℓ′Rj + dL mdi j d′Rj + u′iL mui j u′Rj + c. Any complex matrix M can be diagonalized by a biunitary transformation: UL† M UR = MD (13.3. Its is most convenient to work in a basis in which the fermion masses are real and positive. yd i j . mτ (13. v (13. L=− 1+ v 237 (13. so that the Feynman propagators are simple.3. the theorem just stated assures us that we can choose the matrices LL and LR so that:  me L†L me LR =  0 0 0 mµ 0  0 0 .3. To apply this in the present case.In a general basis.3. the Yukawa couplings ye i j . This can always be accomplished.3. ℓ′Li = LLi j ℓLj . and UL and UR are unitary matrices. In unitary gauge. The lepton mass term in the unitary gauge Lagrangian then becomes:  L=− 1+  j h i † ℓL (LL me LR )i ℓRj .35) So we can write in terms of the unprimed (mass eigenstate) fields:   h (me ee + mµ µµ + mτ τ τ ). yu i j (13.3.3. (13.36) .33) where LLi j and LRi j are unitary 3× 3 matrices.3.34) Now.31)  where (The fermion fields are now labeled with a prime.29) are complex 3 × 3 matrices in family space.c.

c. It might now seem that worrying about the possibility of 3 × 3 Yukawa matrices was just a waste of time.38) chosen in such a way that DL† md DR UL† mu UR  md 0  = 0 ms 0 0  mu 0 =  0 mc 0 0  0 0 .3. (13. The unitary redefinition matrices for quarks just commute with the SU (3)c generators. (13. for example. This is because the W ± interactions involve two different types of fermions. for right-handed up-type quarks: † j j / ′ / / iuj′ R ∂ uRj = i(uR UR ) ∂(UR uR )j = iuR ∂ uRj . In terms of the original primed fields: g L = − √ Wµ+ ν ′i γ µ ℓ′Li + c. There is one place in the Standard Model where the above argument does not work. L contains j′ j / Lj .3. (13. the form of the derivative kinetic terms is exactly the same for unprimed fields as for primed fields. the QED Lagrangian contains a term j′ j −eAµ ℓR γ µ ℓ′Rj = −eAµ (ℓR L†R )j γ µ (LR ℓR )j = −eAµ ℓR γ µ ℓRj . namely the interactions of W ± vector bosons.3.37) u′Li = UL i j uLj .3. one can do unitary-matrix redefinitions of the quark fields: d′Li = DLi j dLj . with different unitary redefinition matrices. u′Ri = UR i j uRj . d′Ri = DRi j dRj .41) This relies on the fact that the (constant) field redefinition matrix LL is unitary.In the same way. but one must now consider how these field redefinitions from primed (gauge eigenstate) to unprimed (mass eigenstate) fields affect the other terms in the Lagrangian. iℓL ∂/ℓ′Lj = i(ℓL L†L )j ∂/(LL ℓL )j = iℓL ∂ℓ (13. 2 238 (13. mt (13.. You can show quite easily that the same thing applies to interactions of all fermions with Zµ and the gluon fields. the unitary condition (this time L†R LR = 1) guarantees that the form of the Lagrangian term is exactly the same for unprimed fields as for primed fields.42) which relies on UR† UR = 1. consider the derivative kinetic terms. There are also interactions of fermions with gauge bosons.3. For example. L†L LL = 1. For the leptons. First. mb  0 0 .3.44) .3. So the redefinition has no effect at all here. The same thing works for all of the other derivative kinetic terms. Consider first the interactions of the W ± with leptons. for the right-handed leptons.43) Just as before.3. (13.39) (13.40) with real and positive diagonal entries. since they act on different indices.

3.51) where the dLj are mass eigenstate quark fields. it cannot be removed by going to some other basis using a further unitary matrix without ruining the diagonal quark masses. implying that V † V = V V † = 1. (13. But now we are free to do so. V = UL† DL . 2 (13. defining νi in the same way as the corresponding charged leptons. 2 (13.c.3. g L = − √ Wµ+ u′iL γ µ d′Li + c.c.53) .c. From eq. νi′ = LL νj .47) So once again the interactions of W ± bosons have exactly the same form for mass-eigenstate leptons. (13. 2 (13. But.3.3. So we have finally encountered a consequence of going to the mass-eigenstate basis. 2 (13. The CKM matrix V is itself unitary.c. and the d′Li are the quarks that interact in a simple way with W bosons: g L = − √ Wµ+ uiL γ µ d′Li + c. So we are stuck with it. since V † = (UL† DL )† = DL† UL .46) resulting in g L = − √ Wµ+ ν i γ µ ℓLi + c. 2 239 (13.49).3. However.3. consider the interactions of W ± bosons with quarks.48) g L = − √ Wµ+ uiL γ µ (UL† DL dL )i + c. we did not have to make a unitary redefinition for them..3.3.50) called the Cabibbo-Kobayashi-Maskawa matrix (or CKM matrix).45) Since we did not include a Yukawa coupling or mass term for the neutrinos. One can think of V as just a unitary rotation acting on the left-handed down quarks.. and in fact it is not. ν ′i = (νL†L )i . (13. The charged-current weak interactions contain a non-trivial matrix operating in quark family space. we get. (13.c. In terms of the original primed fields. 2 (13.52) and so in terms of mass eigenstate quark fields: g L = − √ Vi j Wµ+ uiL γ µ dLj + c.3.3.c. we can define d′Li = Vi j dLj .so that g L = − √ Wµ+ ν ′i γ µ (LL ℓL )i + c.49) which becomes: There is no reason why UL† DL should be equal to the unit matrix.

(13.6760 ± 0. This implies that the interactions of W + with the mass-eigenstate quarks are very nearly:   g L = − √ Wµ+ cos θc [uL γ µ dL + cL γ µ sL ] + sin θc [uL γ µ sL − cL γ µ dL ] + tL γ µ bL .53). and with reduced matrix elements that are proportional to sin2 θc = 0.23. Weak decays of mesons involving the W bosons allow the entries of the CKM matrix to be probed experimentally.3.3. This is an automatic consequence of the which implies in particular that Vud Vub cd cb td tb Standard Model. The numerical entries of the CKM matrix are a subject of continuing experimental investigation.This is the entire effect of the non-diagonal Yukawa matrices.3.3. Numerically.9743 Vcb  ≈  0. More precisely.3.004 0.3. and agree well with the experimental results: BR(W + → ℓ+ νℓ ) = 0.975 0.3.57) where the numerical values given are estimates of the magnitude only (not the sign or phase). (13.55) 2 The terms proportional to sin θc are responsible for strangeness-changing decays. For example.58) where B is a meson containing a bottom quark and D contains a charm quark.3. µ. (13.61) . The partial decay widths and branching ratios of the W boson can be worked out from eqs. To a first approximation. In fact.3.1080 ± 0.974.56) Strange hadrons have long lifetimes because they decay through the weak interactions.999 (13.041  .3. 0. can be used to extract |Vcb |.05. the CKM matrix contains one phase that cannot be removed by redefining phases of the fermion fields.60) (13. This phase is the only source of CP violation in the Standard Model.59) ∗ + V V ∗ + V V ∗ = 0.47) and (13.008 0. it turns out that the CKM mixing is just a rotation of down and strange quarks:  cos θc V =  − sin θc 0 sin θc cos θc 0  0 0 1 (13. (13. sin θc ≈ 0.2252 0. The very long lifetimes of B mesons are explained by the fact that |Vub | and |Vcb | are very small.04  0. then the weak interactions could appear to violate CKM unitarity.230 Vtb 0. each) (13. (13. One of the ways of testing the Standard Model is to check that the CKM matrix is indeed unitary: V † V = 1. the CKM matrix is:  Vud V =  Vcd Vtd Vus Vcs Vts   Vub 0. τ .54) where θc is called the Cabibbo angle. cos θc ≈ 0.0027. decays B → Dℓ+ νℓ .0009 BR(W + → hadrons) = 0. but if there is further unknown physics out there. 240 (for ℓ = e.

Also.where the “hadrons” refers mostly to Cabibbo-allowed final states ud and cs.3) just as in eq.3. 0).26). In the renormalizable version of the Standard Model given up to here.1) With these additional gauge-singlet right-handed neutrino degrees of freedom. The tb final state is of course not available due to kinematics.31) for the Standard Model charged fermions. The basic reason for this is the absence of right-handed neutrinos from the list in eq. (13. the very small magnitudes of Vtd and Vts imply that. several independent pieces of evidence indicate that neutrino masses are much smaller that any of the charged fermion masses. as seen in Table 2 in the Introduction. and reactors have now established that neutrinos do have mass. 1. (13. Although the magnitudes of the neutrino masses are not yet determined by experiment. 13.4. This neutrino mass matrix can be diagonalized to obtain the physical neutrino masses as the absolute values of its eigenvalues. (13. as the mass term couples together left-handed and righthanded degrees of freedom that are independent. this cannot be explained.2) Note the similarity of this with the up-type quark Yukawa couplings in eq. one obtains a neutrino mass matrix v mν i j = √ yν i j . (13. a conservative limit from the WMAP measurements of the cosmic background radiation. (13. 241 .3. The neutrinos in this scenario are Dirac fermions.3. implies that the sum of the three Standard Model neutrinos should be at most 1.1.4 Neutrino masses and the seesaw mechanism Evidence from observation of neutrinos produced in the Sun. Also.c. To remedy the situation. the atmosphere. (13. there are strong upper bounds. top quarks decay to bottom quarks every time: BR(t → W + b) ≈ 1. NR2 . to a very good approximation. 2 (13. Neutrino oscillation data do not constrain the individual neutrino masses. NR3 ←→ (1.62) This greatly simplifies the experimental identification of top quarks. Going to unitary gauge.35). interpreted within the standard cosmological model.4. we can add three right-handed fermions that are singlets under all three components of the gauge group SU (3)c × SU (2)L × U (1)Y : NR1 .4. this implies the useful fact that (up to a very small effect from CKM mixing) b quarks do not result from W decays in the Standard Model. it is now possible to write down a gauge-invariant Lagrangian interaction of the neutrinos with the Higgs field: Lν neutrino Yukawas = − νi i  ℓL  φ0∗ −φ+∗  yν i j NRj + c. accelerators. with a much smaller contribution from the Cabibbo-suppressed final states cd and us. So. but imply that the largest differences between squared masses should be less than 3 × 10−3 eV2 .7 eV.

ν2 . is encouraging. it can naturally be very large.3). νµ . Such small dimensionless couplings appear slightly unnatural. The seesaw mechanism is a way of addressing this problem. including both eq. Now the total mass matrix for the left-handed neutrino fields νLi and the right-handed neutrino fields NRi .3) and eq.4. The fact that the magnitude of M necessary to make this work is not larger than the Planck scale. the absolute values of the neutrino mass eigenvalues of M are approximately: v 2 yν 2 . taking M and yν to be 1 × 1 matrices.4. is: M= √v yν 2 0 √v yT 2 ν M ! .4) where Ni are the Majorana fermion fields (see section 3. One includes.4) that include NRi .1 and M = 3 × 1012 GeV.0 and M = 3 × 1014 GeV. besides eq.4). 2 then the smaller set of mass eigenvalues of M will be pushed down. (13. (13. ν3 (with masses m1 < m2 < m3 ) can be related to the left-handed parts of the neutrino weak-interaction eigenstates νe . ντ (which each couple to the corresponding charged lepton only.4. ντ L ν3L 242 (13. (13. the left-handed parts of the neutrino mass eigenstates ν1 .5) The point of the seesaw mechanism is that if the eigenvalues of M are much larger than those of the Dirac mass matrix √v yν . the eigenvalues of the neutrino Yukawa matrix yν i j would have to be extremely small. 2 (13. such that very small neutrino masses naturally occur. For example. or yν = 0.6) in the limit vyν ≪ M. then this Majorana mass term necessarily violates the total lepton number L = Le + Lµ + Lτ . one could have yν = 1.2). In any case. If the neutrino fields carry lepton number 1.To accommodate this within the Dirac mass framework of eq. in a purely subjective sense. There are also three extremely heavy Majorana neutrino mass eigenstates. For illustration. a new term in the Lagrangian: 1 L = − Mij N i Nj . The light neutrino states (corresponding to the lighter eigenvectors of M) are mostly the Standard Model νL and they are Majorana fermions.4. the ease with which the seesaw mechanism accommodates very small but non-zero neutrino masses has made it a favorite scenario of theorists. which decouple from present weak interaction experiments. In either of the two cases above. Since M does not arise from electroweak symmetry breaking. (13.1 eV. no larger than about 10−9 . and is very roughly commensurate with other scales that occur in other theories such as supersymmetry.4. 2M and M (13.7) .4. and Mij is a symmetric mass matrix. (13. and the W boson) by:     νeL ν1L  νµL  = U  ν2L  .4. even if the corresponding Yukawa couplings are of order 1. and this suggests that neutrino masses may have a different origin than quark and leptons masses. to get a neutrino mass of order 0.4.

h → gg.3. with the gluons eventually manifesting themselves in the detector as jets.1 Higgs decays revisited In section 8.4. we have already calculated the leading-order decays of the Higgs boson into fermionantifermion final states. which can proceed via the quark-level Feynman diagram shown below.8) (at the nucleon level nn → ppe− e− ). It is the subject of continuing searches. i. In the Dirac scenario. To tell the difference between these two scenarios.e. To recap. because of its large Yukawa coupling) go around the loop: 243 . where quarks (most importantly the top quark. there are two distinct scenarios for neutrino masses. However. it can only occur in the case of Majorana neutrinos. and the LHC will eventually be sensitive enough to discover it. First. 13. but does occur through the one-loop diagram below. The Tevatron continues to search for it. neutrinos are their own antiparticles. a nuclear transition A Z → A (Z + 2) + e− e− . both possibilities are consistent with the experimental data. there are other final states that are quite important besides h → f f . 13. (13. d u e− W− ν W− e− u d Since this process requires a violation of total lepton number in the neutrino propagator. This decay cannot happen at tree level. one can look for neutrinoless double beta decay.where U is a unitary matrix known as the Pontecorvo-Maki-Nakagawa-Sakata (PMNS) matrix. neutrinos and antineutrinos are distinct. if it exists.5 Search for the Standard Model Higgs boson The last remaining particle of the Standard Model that remains undiscovered is the Higgs scalar boson h. the Higgs boson can decay into two gluons. by introducing right-handed gauge-singlet neutrino decrees of freedom. In this section we will briefly review the processes that could lead to discovery of the Higgs boson or limits on its mass. At this writing. and total lepton number is conserved although the individual lepton numbers are violated. In the Majorana scenario. and total lepton number is violated along with the individual lepton numbers. Neutrino oscillations are due to the fact that U is not equal to the identity matrix.5.

) Even though one-loop graph amplitudes are usually not competitive with tree-level amplitudes. (13.) Another important effect is that tree-level decays into two vector bosons. when one of the vector bosons is virtual. it can still be important as a discovery mode because the corresponding backgrounds at colliders are also relatively tiny. The decay h → γγ also does not occur at tree-level. the decay can be thought of as really three-body.5. In fact. but does occur due to one-loop graphs where any charged particle goes around the loop. for leptonic final states of the off-shell vector boson: 244 . is any fermion-antifermion final state that couples to the off-shell Z.1) where the “(∗)” means that the corresponding particle may be off-shell.2 [below eq. (The decay h → Zγ is also mediated by similar one-loop graphs. but it will be neglected here because it turns out not to be as useful for searches. h → ZZ (∗) . One indicates this by writing h → W W (∗) . that is ¯ where f f¯′ could be any final state that couples to the off-shell W ∓ and f f¯ h → W ± f f¯′ or h → Zf f. For example. (13.2. they are even important for Mh < 2mW and 2mZ respectively.g q h q q g (One must also include the Feynman graph with the quarks going the other direction around the loop. notably the top quark again but also the W : γ f h h f f γ W W W γ γ Although the branching ratio to two photons is much smaller (less than 3×10−3 in the Standard Model). h → W + W − and h → ZZ.25)] are large. depending on the kinematics. corresponding to the Feynman rules found in section 13. this is an exception because the gluons have strong couplings and because yt is so much larger than yb (by a factor of roughly 55). If one of the vector bosons is off-shell.

suggest that h is probably lighter than roughly 240 GeV within the context of the Standard Model. as a function of Mh . since the tree-level two-body decays h → f f are suppressed by small Yukawa couplings. which depict the state-of-the-art computations of branching ratios and the total width Γtot for the Higgs boson.) 0 Higgs Branching Ratio 10 bb + − ττ -1 10 cc gg + W W -2 10 ZZ γγ -3 10100 120 140 160 180 200 Higgs mass [GeV] 245 220 240 − . The results of a careful computation (using the program HDECAY by A. 108. The closer the vector bosons are to being on-shell. and Z propagators. the decay h → tt becomes significant. and M. despite their kinematic suppression. although its branching ratio never exceeds about 0. Djouadi. so only that range is shown here. Spira. Comput. J. (Indirect constraints on the Higgs mass. the larger the amplitude will be. and already dominate for Mh > 135 GeV. 56. these three-body decays can be competitive and even win. Phys.2. Kalinowski. due to its effect on one-loop corrections to the γ. the main new important effect is that for Mh > ∼ 340 GeV. (1998). hep-ph/9704448) including all these effects are shown in the two graphs below. Commun. The decays h → W W (∗) become increasingly important for larger Mh . W . For larger masses.W+ h W −(∗) Z h ℓ− Z (∗) νℓ ℓ+ ℓ− Again.

The gg final state is useless because of huge QCD backgrounds to dijet production. or could greatly modify the branching ratios into Standard Model states because of radiative corrections or modified Higgs couplings. from h → bb and Z → qq. the γγ final state may be very important for the discovery of a light Higgs boson. despite its tiny branching ratio. 246 . again because it can lead to low-background signals if one or both of the Z bosons decays leptonically. with the total width dramatically increasing when both W bosons can be on shell. but it is important to keep track of because it reduces the branching ratios into more useful final states. It is also important to keep in mind that in some extensions of the Standard Model. 13.Higgs Total Width [GeV] 0 10 -1 10 -2 10 -3 10100 120 140 160 180 200 Higgs mass [GeV] 220 240 The largest decay modes are bb and W W (∗) over the entire range of Mh . Because it has very low backgrounds in colliders. the Higgs boson may decay into some new exotic particles.2 Limits from the LEP2 e− e+ collider LEP2 searched for Higgs bosons in e− e+ collisions at various energies up to √ s = 209 GeV. ending in the year 2000. The ZZ (∗) final state can be important for Mh > ∼ 130 GeV.5. The main process for possible Standard Model Higgs production at LEP2 was e− e+ → Zh due to the Feynman diagram shown below: e− h Z e+ Z Searches were conducted for the following signatures: • Four jets.

2) need not apply if the Standard Model is extended.5. However. (95% CL. (The roles of the initial state and the final state are simply exchanged. mostly from h → bb and Z → ℓ+ ℓ− . (13. or from h → τ + τ − and Z → qq. The process qq → Zh occurs due to a Feynman diagram just like the one given for LEP in the previous section.3 Searches at the Tevatron At hadron colliders. on average. • Two jets and missing energy. The end result is that the combination of the four LEP2 detector collaborations ruled out a Standard Model Higgs boson at 95% confidence unless Mh > 114. but also could not exclude a Higgs of that mass.3 GeV. and OPAL) saw no evidence of an excess.9 GeV compared to the expected bound.5.5. (13.5. The process qq ′ → W ± h is due to parton-level Feynman diagrams like the ones below: h u W d h d W W+ u 247 W− .5) The process gg → h cannot occur at tree-level.4) qq ′ → W ± h. due to the large gluon PDFs. except that e− and e+ are replaced by a quark and antiquark. by crossing. However. µ) leptons and two jets.) This is actually the largest source of Higgs bosons at both the Tevatron and the LHC. The ALEPH detector collaboration saw a 3σ excess at Mh = 115 GeV. either from h → bb and Z → τ + τ − .5. Thus the official impact of the ALEPH excess was to decrease the actual mass bound by about 0. L3. then if the entire LEP2 run were repeated many times the expected bound would have been.5.2) If indeed there is no Higgs boson within the kinematic reach of LEP2. It cannot be overemphasized that the mass exclusion (13. Mh > 115. 13. and favorable kinematics at the Tevatron. LEP2. from h → bb and Z → νν.4 GeV.5. Tagging of b jets was used to increase the sensitivity. • Two (e. the other three detectors (DELPHI. the largest parton-level production processes for the Higgs boson are: gg → h. (13. but does occur due to the same one-loop diagram mentioned above in section 13. Standard Model Higgs) (13. the other channels often provide better discovery prospects because the presence of the extra weak boson reduces backgrounds. near the kinematic limit of the collider.• Two tau leptons and two jets.3) qq → Zh.1 for the decay h → gg.

which is due to Feynman diagrams including the one below: t g h g t as well as weak vector boson fusion.as well as others related by d → s and/or u → c. or W h. One can expect this range to increase somewhat in the near future.5. using 6. eqs. for light Higgs bosons Mh < ∼ 135 GeV: • Two b jets with one lepton and missing energy. also important are pp → tth. from pp → h → W W (∗) → ℓ+ ℓ−′ νℓ ν ℓ′ . from pp → W ± h followed by W → ℓν and h → bb. (13. • Two leptons reconstructing a Z and two b jets. which affects the lepton kinematics and helps to distinguish the signal from the background of direct W + W − production (without a Higgs).5. the most important signature becomes • Two leptons and missing transverse energy.96 TeV. from pp → Zh followed by Z → ℓ+ ℓ− and h → bb. These signals retain some power up to Mh close to 150 GeV. qq → qqh. The backgrounds relevant for Higgs detection in Tevatron pp collisions are much more severe than at LEP2. 13. • Two b jets and missing energy.5). with h → τ + τ − .5. the important production processes for a Standard Model Higgs boson include the same ones as at Tevatron. • Two taus. The W + W − pairs in signal events have different spin correlations than in background events.3)-(13. and q q → q qh through Feynman diagrams like this: 248 .7 fb−1 of Tevatron pp collisions √ at s = 1. in both cases with h → bb. either from pp → Zh followed by Z → νν . probably by the time you are reading this. for Higgs bosons with Mh > ∼ 135 GeV. the combination of the CDF and D∅ detector collaborations have excluded a Standard Model Higgs boson with mass in a range from 158 to 175 GeV. from pp → h. The most important specific signatures that are being looked for at Tevatron are. which refers to the parton-level processes qq → qqh. However. or from pp → W h with W → ℓν and the lepton is missed.4 Searches at the LHC In pp collisions at LHC energies (up to √ s = 14 TeV). Zh. However. At this writing.

Z. Finding and definitively identifying the Higgs boson is arguably the most important result that is likely to appear from the LHC. 249 .q or q W or Z h W or Z q or q In the last process. • Four leptons ℓ+ ℓ− ℓ+′ ℓ−′ from inclusive Higgs production (mainly pp → h) followed by h → ZZ (∗) . • A lepton together with a peak in bb or γγ distributions. with the lepton coming from either a W . and with great precision in either the γγ or ZZ (∗) → ℓ+ ℓ− ℓ+′ ℓ−′ channels. Tagging events with these forward jets is a way to reduce backgrounds. • Forward jet tags from weak vector boson fusion together with h → τ + τ − or h → γγ or h → W W (∗) . This is appropriate for a higher mass Higgs boson. Determination of the Higgs mass is expected to be doable after discovery. but should include: • A diphoton peak. This is appropriate for a lower mass Higgs boson. the quark jets in the final state are usually found at small angles with respect to the beam. The most important signals at LHC for a Standard Model Higgs boson will depend on collider and detector performance. or t produced in association with the Higgs. from inclusive Higgs production (mainly pp → h) followed by the rare but low-background decay h → γγ.

1 sec = 1.14) . 250 (A.97326937 × 10−16 m.78266173 × 10−27 kg.89379201 × 108 pb = 3.60217646 × 10−3 erg = 1. (A.9) (A. the official definition of 1 meter is the distance traveled by light in a vacuum in exactly 1/299792458 of a second.58211814 × 10−25 GeV sec.8) (A. h = 1.) In units with c = h ¯ = 1.5) Conversions of particle decay widths to mean lifetimes and vice versa are obtained using: 1 GeV−1 = 6.13) (A. some other conversion factors are: 1 GeV = 1.4) (A.78266173 × 10−24 g = 1.99792458 × 1010 cm/sec = 2.11) and in reverse: 1 nb = 10−33 cm2 = 2. (A.89379201 × 1011 fb.89379201 × 105 nb = 3.56819059 × 10−12 GeV2 .51926778 × 1024 GeV−1 .05457148 × 10−34 J sec = 6.58211814 × 10−25 sec.12) (A. ¯ (A.99792458 × 108 m/sec. 1 pb = 10−36 cm2 = 2. 1 GeV = 1. (Since October 1983. by definition.Appendix A: Natural units and conversions The speed of light and h ¯ are: c = 2. (A.56819059 × 10−6 GeV2 .60217646 × 10−10 J.97326937 × 10−14 cm = 1.10) (A.89379201 × 10−4 barns = 3.2) The value of c is exact.7) Conversions of cross-sections in GeV−2 to barn units involve: 1 GeV−2 = 3.6) (A. 1 fb = 10−39 cm2 = 2.56819059 × 10−9 GeV2 . 1 GeV−1 = 1.3) (A.1) (A.

7) Tr(γµ γν ) = 4gµν (B.2) γ j† = −γ j (j = 1. 0 −1 σ 1 = −σ 1 = . Tr(1) = 4 (B. + (−1)k gµ1 µk Tr(γµ2 γµ3 . 3) (B. γ52 = 1. . 0 0 2 (B. . (γ 0 )2 = 1 (B. . γµ2n ) − gµ1 µ3 Tr(γµ2 γµ4 . . . .12) The matrix γ5 satisfies: γ5† = γ5 .16) Tr(γµ γν γρ γ5 ) = 0 (B. . . γµ2n−1 ) (B. γµ2n ) + .18) . {γ5 .6) The trace of an odd number of γ µ matrices is 0.1) where σ0 = σ0 =  σ 2 = −σ 2 = 1 0  0 1  0 −i i 0    0 1 .8) Tr(γµ γν γρ γσ ) = 4(gµν gρσ − gµρ gνσ + gµσ gνρ ) (B. γ µ } = 0 (B.Appendix B: Dirac Spinor Formulas In the Weyl (or chiral) representation): µ γ =  σµ 0 0 σµ  (B.13) Tr(γ5 ) = 0 (B. .9) Tr(γµ1 γµ2 . γ 0† = γ 0 . . .5) [γρ . [γµ .15) Tr(γµ γν γ5 ) = 0 (B.11) PR = 1 + γ5 = 2  0 0 0 1  (B.14) Tr(γµ γ5 ) = 0 (B.17) Tr(γµ γν γρ γσ γ5 ) = 4iǫµνρσ 251 (B. . γν ]] = 4(gρµ γν − gρν γµ ) (B. γµk−1 γµk+1 . . γµ2n ) .4) γµ γν + γν γµ = {γµ . +gµ1 µ2n Tr(γµ2 γµ3 . . in 2 × 2 block form:   −1 0 γ5 = 0 1   1 − γ5 1 0 = PL = . γν } = 2gµν (B. . γµ2n ) = gµ1 µ2 Tr(γµ3 γµ4 .3) γ 0 γ µ† γ 0 = γ µ (B. 2.10) In the chiral (or Weyl) representation. . 1 0   1 0 σ 3 = −σ 3 = . .

v(p.23) u(p. s)v(p.28) v(p. s)u(p.25) v(p. s)v(p.20) (B.19) (B.γ µ γµ = 4 γ µ γν γµ = −2γν γ µ γν γρ γµ = 4gνρ γ µ γν γρ γσ γµ = −2γσ γρ γν (B. s) = 0 (B.26) v(p.21) (B.27) X u(p. s)u(p. r) = u(p. s) = p /−m (B. r) = 0 (B. (p / + m)v(p. r) = −2mδsr (B.22) (p / − m)u(p.29) s X s 252 . s) = 0. s)v(p. s) = /p + m (B. r) = 2mδsr (B. s)u(p. s)(/p + m) = 0 (B.24) u(p. s)(/ p − m) = 0.

F. Schroeder. Phys. Perkins. B. Wiley-VCH Two excellent modern books on relativistic quantum field theory.D. are: • M.J. Cambridge University Press • J. Cambridge University Press • C.. Wiley • D. E. parton distribution functions. Gauge Theories of the Strong. Quantum Field Theory. Moore. Cambridge University Press Books that give good overviews of QCD. QCD and Collider Physics. B.R. Collider Physics. Quigg. Nakamura et al. Martin.J. see: http://zippy. but still useful. and Electromagnetic Interactions.H. Introduction to High Energy Physics. Halzen and A.R.gov/ is also handy. Barger and R. with different emphases. J.D. Golowich. Quarks and Leptons: An Introductory Course in Modern Particle Physics. Addison-Wesley 253 .V.E. Stirling. Ellis. and limits on violations of conservation laws and on new physics models. The Standard Model: A Primer. Westview Press • M.N. Cambridge University Press • V. Peskin and D. [K. Webber.lbl. Weak. Srednicki. Griffiths.) Books that give overviews of Standard Model phenomenology include: • C. and collider processes include: • R. An Introduction To Quantum Field Theory. The corresponding web site http://pdg.physics. Holstein.niu.Appendix C: Further Reading For an up-to-the minute list of corrections to these notes.edu/ppp/ The Review of Particle Properties by the Particle Data Group. (This book is out of date in some places. 075021 (2010)] is a very useful source on the latest high-energy particle physics data. Westview Press. the Standard Model and its interpretation. Donoghue. Dynamics of the Standard Model.K. W. Books that give general introductions to high-energy physics include: • F. Burgess and G. Cambridge University Press • D. Introduction to Elementary Particles. G 37. Phillips.

28 sum over. 8 Belle experiment. 177. 173–175. 14. 138 asymptotic freedom. 240. 31 Cabibbo angle.Index Abelian (commutative) group. 88 canonical commutation relations. 191 charge conservation. 250 charge operator. 170. 195. 197 boost. 168 ATLAS detector at LHC. 105–107. 54 240 anticommutation relations. 193. 97. 195 CMS detector at LHC. 9. 213 141 angular momentum operator. 40. 6. 193 closure property of group. 241 angular momentum conservation. 223. 133 active quarks. 83 barred Dirac spinor. 55 102 real scalar fields. 165 Carbon-14 dating. 251 (J = 3/2). annihilation operator. 185 commutation relations. 13. 5. 167 QED. 214 action. 49 associativity property of group. 81. 214 chiral (Weyl) representation of Dirac matrices. 8. Dirac fermion fields. 137. 169. 5. 112 Cabibbo-Kobayashi-Maskawa (CKM) mixing. Z boson. 137. 153 complex scalar field. 158 antiparticle. 10. 56. 45 branching ratios. 213 color. 53. 6–8 charmonium. 39. 196 chiral symmetry breaking. 191 charge density. 213 CDF detector at Tevatron. 140 charge conjugation. 40 barn (unit of cross-section). Higgs boson. 43. 154. 11. 157 adjoint representation. 194. 35. 11. 138. 55. 43. 32 antineutrino. 10. 204 W boson. 236 QCD. 236 lepton-number violating limits. 163 baryons. 148 bare coupling. 101. 237 confinement. 189. 192. 197 Casimir invariant. 134. 6. 140 chiral representations (of gauge group). 233 beta function. 42. 29 charged current. 22. 193. 62. 50 Bose-Einstein statistics. 63. 239. 7 Bhabha scattering. 20. 248 BaBar experiment. 194 charged pion. 5. 5. 165. 138 canonical quantization antineutrino-electron scattering. 133. 240 angular resolution. 51 254 . 228 Breit-Wigner lineshape. 49. 190. 166 bottomonium. 41 bare mass. 107 biunitary transformation. 7 26. 140 182 Altarelli–Parisi (DGLAP) equations. 54. 165. 50. 12. 200. (J = 1/2).

151. 188–191. 184. 224–233 Yukawa. 50 fermionic. 43. 65 complex scalar field. 167. 79 Dirac matrices. 64 140. 79 Yang-Mills gauge. 48. 177 equal-time commutators. 232 covariant derivative energy conservation. 169 bare. 83 cutoff. 161. Γ). 41 D∅ detector at Tevatron. 191 dimensional analysis. 153 confinement of color. 46. 157 defining representation. 194. 149. 57 Drell-Yan scattering. 242. 165 ǫµνρσ (Levi-Civita) tensor. 60. 26. 63 142–144. 80. 25 experimental value. 166 coupling dimension of tensor product representation. 169 differential. 64. 69.commutative (Abelian) group. 243 DGLAP equations. 89. 148 commutator of Lie algebra generators. 160. 47. 127. 167 conservation of lepton number (or not). 241. 224–227 dimensional transmutation. 105–107. 24–27 Fermi (GF ). 15 equivalence of representations of group. 6. 107. 194 134. 21 Yang-Mills. 83. 111 dimension of representation (dR ). 231. 211–213 scalarn . 165 CPT Theorem. 248 conservation of energy. neutrino. 126. 142. identical particles in scalar3 . 200. 133. 5. 85. cross-section. 108. 204 conservation of momentum. 197 CTEQ parton distribution functions. 227 double-counting problem. 251 QCD (g3 . 141 conservation of charge. 224–227 Dirac equation. 168 255 . 154 complex conjugate representation. 168 Coulomb potential. 167 creation operator. 147. 195 direct sum representation. 113–117. 77. 202 conservation of angular momentum. 76. electromagnetic. 193 Dirac spinor. 161 electromagnetic (e). 127. 183. 189. 177–181. 62. 84. 15. 137 decay rate (width. 186. 65 contravariant four-vector. current density. 168 renormalized. 90 differential cross-section. 54 complex conjugate of spinor expressions. 97. 21 dimension of group (dG ). 242 electroweak vector boson interactions. 163. 53–55 general. 49 covariant four-vector. 131 scalar4 . 53. 187–191. αs ). 117. 234–238 Salam) gauge theory. 68 Compton scattering. 131. 92. density of states. 70. 119–126 crossing. 194 electroweak (g′ ). 158 two particle → two particle. 134. 137 QED. 70. 40. 13–15. 218 electroweak Standard Model (Glashow-Weinberg- Yukawa. 149 Dirac field. 227 dimensional regularization. 90. 48. 68. 139. 40. 69 final state. 60. 128 conservation of helicity. 194 electroweak (g).

87 scalar. 183 free field theory. 87 Feynman rules. 43 Feynman propagator fine structure constant. 142. 13–15. 43–47. 81. 87 fermion external lines. 175–180 fermions with known helicity. 57 Higgs boson. 237–240 gauge fixing. 162 generic fields. 251 Yang-Mills vector fields. 41. electron. 142. 80. 221 fundamental representation. 26 158 traces. 163 ηc (scalar charmonium). 101. 77. 80 four-vector. 142 Majorana. 200 Feynman’s x. 242 Fermi-Dirac statistics. 143 gauge-fixing parameter (ξ). 42 numerical value. 178 gauge eigenstate fields. 234 Feynman slash notation. 173 massive vector bosons. 167 massive vector fields. 83. 46. 46–48 Feynman diagram. 102 general. 48 Fermi weak interaction theory. 43. 82 gauge invariance. 164 generators pion decay. 24 standard electroweak gauge theory. 40. 142. 39. 86. 61 Fermi weak interaction theory. 117. 26 factorization scale. 57 Yukawa coupling. 76. real. 86. 89 four-fermion interaction. 62 Feynman-St¨ uckelberg interpretation. 167 theory. 199 femtobarns (fb). 84. 165 Yang-Mills. 162 γ5 matrix. 79 GeV. 19 gluons. 76 W boson. 95 charged massive vector fields. 10 W -fermion-antifermion. 54. complex (π ± ). 45 fermion-antifermion condensate. 15 photon. 85. 171 QCD.ηb (scalar bottomonium). 86. 61. 6 Dirac fields. 46 Feynman gauge. 47. 33. 183. 173 QED. 85–90 SU (N ). 251. 172 scalar φn of Lie algebra. 252 scalar. 5. 160. 223 vector. 158–160 SU (2). 43. 87 scalar. 10 Yang-Mills. 178 field theory. 226. 34 Fermi constant (GF ). 42. 79. 184 SU (3). 47. 74–78 Weyl. 164 gµν (metric tensor). 78–81 Gell-Mann matrices. 143 field. 89. 165. 87. 80 Euler-Lagrange equations of motion. 163 electromagnetic. 214 Yukawa coupling. 46. 162 flavors of quarks. 147 Dirac. 231–233 256 . 81. 178–180 η meson. γ matrices. 53–55 relation to mW . 166. 43.

157 helicity flip. 168 single scalar particle. 243. 214 helicity conservation. 133. 157 Klein-Nishina formula for Compton scattering. 25. 155. 11 identical final state particle overcounting prob- gluon. 59 Goldstone boson. 197. 198 integrated luminosity. 167 Goldstone’s theorem. 247–249 257 . 165 gluon collider. 168 interaction. 81 half life. 157 Klein-Gordon equation. 15 free Dirac field. 140 Higgs field Landau gauge. 49 irreducible representation. 34 glueballs. parameterization of. 197–202. 11. 178 generic. 43 internal line. 199 hadron-hadron scattering. 57 Hamilton’s principle. 222 SU (2). 165 general classical system. 172 single Dirac particle. 184. 7. 107. 97. 51–53 inverse of group element. 162. 212 hard scattering of partons. 222. 125 233. 207–209. 99. Standard Model. 31 Jacobi identity. 24. 215–218 hole (absence of an electron). 74 Hamiltonian inverse metric tensor (gµν ). 8. 45 mass in Standard Model. 137 interaction Lagrangian. 69. 10. 231. 11 IN state. 219. 57. 81 isospin global symmetry. 20. 155. 76 gluon-gluon scattering. 230. 23 Jacobian. 45 interactions with Z. 233 LAMPF. 231–233 Lagrangian density. 209–213 infrared slavery. 63. see Nambu-Goldstone boson index of representations. 206–208 ignorance. 224 Large Hadron Collider (LHC). 7 interaction Hamiltonian. 56 inverse muon decay.global symmetry. 229–231. 58. 183 lem. 233 204. 151 free scalar field. 58. 209 iǫ factor in propagators. 131 propagator. 4. 68. 172. 198 helicity. Higgs boson. 29. 132–135. 55 SU (3). 188 gluonium. 46 helicity suppression. 62 hadrons. 219. 89. 171 Grassmann (anticommuting) numbers. 32. 213. 173 gravity. 87. 220. 43. 63. 245. 127. 58. 197 jet. 57. 33 J/ψ (vector charmonium). 211. 169 group. 80. 49. 183 identity element of group. 191 adjoint. 222. 246 Lagrangian. W. 6. 194 hadronization. 215 Higgs mechanism. 165 infinite momentum frame. 220. 182. h. 224 global symmetry breaking. 117.

39. 143. 100 Møller scattering. 241–243 2-body. 139. 6. 21 Lie algebra. 76. 242. would-be. 242 3-body. 216 linear collider. 140. 129 masses. 142–147 muon production in e− e+ collisions. 202 longitudinal polarization of massive vector bo- muon decay. 243 mean lifetime. 204 non-Abelian gauge invariance. 67. 215 momentum conservation. u ˆ). 139. 164. 90 natural units. 41. 10 (J = 0). 220–222. u). 10 leptons. 7 Majorana field. 40. 75. 220–222. 243 Majorana fermion. 166 Mexican hat potential. 241 luminosity. t. 164 projection matrix (PL ). 224. 20 neutrino. 9. 137 Mandelstam variables (s. 129 lepton numbers (individual). 6. 220 longitudinal rapidity. 230 180. 33 propagator. 137. 15 Lie group. 226. 129 mixing. 61–63 Weyl. 18.mass diagonalization. 166. 64. 210 Nambu-Goldstone boson. 237 left-handed fermions mass eigenstate fields. 86 LEP e− e+ collider experiment. 246. 191. 140 mesons. 243 neutron. 138 n-body. 242 lowering operator. 154. 242. 78. 135. 141. 222 loop corrections to muon decay. 64 left-handed polarization of photon. 66. 38 reduced. 218–222 longitudinal momentum fraction (Feynman’s x). 5 Levi-Civita (ǫµνρσ ) (J = 1). 81. 237–240 and weak interactions. 91–96 son. 224. 62 neutrinoless double beta decay. 108 Mandelstam variables. 13. 15. MS renormalization scheme. 62 fermion minus sign. 247 Maxwell’s equations. 51 Yukawa coupling. 165. 90. 227 199 MSTW parton distribution functions. 7. 198. 10 tensor. 180. 199 breaking. 35 matrix element. 50. 233 226 Dirac. 124. 250 Lorentz transformations. 38. 172. 242 neutron decay. 89. 90 momentum fraction (Feynman’s x). 149 pseudo. 33. 165. 175– 178 258 . 194. 223 loops in Feynman diagrams. massive vector boson. 167 metric tensor (gµν ). 68. 190. 161. 38 Lorentz-invariant phase space anti-. 236 nanobarns (nb). 225. 5. 186–188. 116 local (gauge) symmetry. 217. 77. 136 seesaw mechanism. 9. 48 lepton number (total). tˆ. partonic (ˆ s. 218.

183. 251 reduced matrix element. 28. 86. 11 polarized beams. 148 proper interval. 141–143 proper Lorentz transformations. 210 potential energy. 80 nuclear weak decays. 158 OUT state. 20. 183 nucleons. 83–85 photon propagator. 43. 184–186 massive vector boson. 164 quarkonium. 85 quark-antiquark condensate. 242. 197. 129 proton. 100. 20 partial width. 161. 83–87 182. 12. 30 trix. 97. 199–204 pseudo-scalar fermion bilinear. 34. 87 on-shell. 68. 194 Dirac fermion. 197 QCD coupling. 69. 87. 251 quadratic Casimir invariant. 199 pseudo-rapidity. 178 parity.non-renormalizable theories. 214 photon. 89 Feynman rules. 243 positron. 6. 25. 155. 34 longitudinal. 168 charged massive vector boson. 195 Pauli exclusion principle. 221 massive vector boson. 193 partonic subprocess. 19. 155–160 flavors.3 ). 5. 162 259 . 194. 223 and gauge transformations. 18. 164 regularization. 156 masses. 131 Yang-Mills vector fields. 25. 59 scalar. 5 pion (π ± ) decay. 198 experimental value. 223 parton. 162. 186. 77. 180. 162 particles. 214 Quantum Chromo-Dynamics (QCD). 81. 7. 191 generic. 183 photon polarization. 18 projection matrices for helicity (PL . 7 Particle Data Group. 51 Pontecorvo-Maki-Nakagawa-Sakata (PMNS) ma- rapidity. 164 ω meson. 193 Pauli matrices (σ 1. 33. 236 polarization vector. 197. 76 overcounting problem for identical final state W boson. 50. 187. 15 scalar. 118 quark-quark scattering. 253 pseudo-Nambu-Goldstone boson. 101 raising operator. 184. 7 Higgs boson. photon field (Aµ ). 162. 138 gluon. 18 parton distribution function (PDF). 62 quark. 4. complex (π ± ). 173. 85. 54 QCD scale (ΛQCD ). 215 pseudo-. 127. 85–90 picobarns (pb). real. 43. 168 φ meson. 137. 14. 14 parity violation. 89 renormalizable theories. 141.2. 52. 161. 162. 6 pion decay constant (fπ ). 86 Quantum Electro-Dynamics (QED). 141 parton model. 161. 44. 86. 64 propagator reducible representation. PR ). 117.

166–171 structure constants of non-Abelian group. 57. 225 4. partonic. 142. 215–218 of SU (N ). 37 spontaneous symmetry breaking. 108 right-handed projection matrix (PR ). 76. 198 supersymmetry. 191 Dirac. 63. 205. 197. 175 local (gauge). 200 spin operator. 166 Majorana. 247. 173–175 ρ meson. 63. 136 Schrodinger equation. 227 renormalized (running) mass. 223 rotations. 250 renormalization scale. 152 symmetry factor (of Feynman diagram). 20 Thomson scattering. 111. 42. 253 Rhadrons . 197. 214 SU (N ) Lie algebra. 209 seesaw mechanism for neutrino masses. 20. SU (2) Lie algebra. 21 Rutherford scattering. 7. 149. 23 time reversal (T). 207–209 sea quarks. 7. 13. 166 Standard Model. 201 Tevatron production cross-section. 199. 52. 25 representation of group. 36. 19. 7. 241 58 sea partons. 77. 224–233 of non-Abelian group. 191. 172 of SU (3). subprocess. 26 renormalization group. 203 LHC production cross-section. 171. 93 singlet representation. 77. 112 tensor product of representations. 175 right-handed antifermions and weak interactions. 78. 171. 166. 172. 158 scalar function. 204. 125. 191. 150 slash notation. 38 tau decays. scalar field. 209. 192 speed of light (c). right-handed fermions 90 Dirac. 6. 32 renormalized (running) coupling. 148 Schrodinger picture of quantum mechanics. 190. 62. 20 tensor. 209 141. 172. 133. 188–191 spin-sum identities. 224. 75. 45 248 complex. top quark. 39. 215 of SU (2). 167 Review of Particle Properties (Particle Data Group). 205– 207. 242 trace trick. 98–100 SU (3) Lie algebra. 126 s-channel. 201. 12 right-handed polarization of photon. 35 technicolor. 204. 172 SU (2)L (weak isospin). 169 Tevatron. 135. 7 decay. 189. 86 t-channel. 74–76 three-body phase space. 99. 63. 24 electroweak. 252 spinor. 173–175 global. 218–221 of U (1).renormalization. 169 260 . 38 representations Weyl. 33 Weyl.

225 transverse momentum (pT ). 210 weak mixing angle (Weinberg angle) θW . 50. 224. 7. 166 Weyl fermion. 252 weak isospin [SU (2)L ]. 250 Υ (vector bottomonium). 42. 4 branching ratios. W. 77 weak-interaction charged current. 27. 22 W boson. 231. 154. measured. 4 Standard Model numerical. 216 couplings to fermions. 228 vacuum expectation value (VEV). 229. 4 width. 241 couplings to fermions. 232 Standard Model. 166. 124 Weyl equation. 230 u-channel. 54 resonance at LHC. 214 Z boson. 229 mass. 80. 118. 4 volume element. 172. 232 mass. predicted. 231 transverse polarization. 119 weak hypercharge [U (1)Y ]. 86. 220 tree-level diagrams. 231. 4. 162. measured. 180 Yang-Mills theories. 155. 226– 228. 75. 230 230 Standard Model. 224. ξ (gauge-fixing parameter). would-be Nambu-Goldstone boson. 241 units. 231 vacuum state. Feynman (momentum fraction). 85. 38 U (1) for electromagnetism. 227. 225. 11. 133. 76 unitarity of CKM matrix. 26. 214 valence quarks. 164. 251 two-body phase space. 207 width. 220–222 Yukawa coupling.traces of γ matrices. 87 160. prediction. 24. 240 x. 225. 229. 240. 230–232 neutrino. 238–240 interactions with Z. prediction. 227. 228 antiquark-quark. 224. 251. 134. 220–222. 163 triangle function (λ). 234–238 Standard Model. 37 U (1) as a Lie group. 199. measured. 131. 37 U (1)Y (weak hypercharge). 231 width. 223 interactions with Z. h. 175–180 unitary gauge. 199 unitarity of quantum mechanical time evolution. 5 Ward identity. 4 mass. 182 Weyl spinor. 93. 230 261 . W. 130 Weyl (chiral) representation of Dirac matrices. 213 branching ratios. h. 201–203. 231 mass. 10.