Advanced Particle Physics 1FA355:

Brief notes on gauge theory

Rikard Enberg

April 1, 2014

1 The Standard Model

The Standard Model (SM) is a QFT that describes what the world is made of and how these
fundamental constituents interact with each other. The what was described in the first
lecture: a number of different matter particles (fermions) that build up matter, and a number
of force-carrying particles (gauge bosons) that mediated interactions.
The how of how these interactions happen is described by the following gauge theory
Lagrangian:
1 a a
LSM = F F /
+ i D + i ij j h + h.c. + |D h|2 V (h)
| 4 {z } | {z } | {z }
gauge sector flavor sector Higgs sector

with gauge group SU(3)SU(2)U(1). Thats all!

The first group of terms describes the gauge sector of the SM. Its a nice and simple gauge
theory where all particles are massless. The second group of terms describes the flavor sector
of the SM, which after electroweak symmetry breaking describes the fermion masses. The
third group of terms describes the Higgs sector of the SM, which tells us about the gauge
boson masses and the Higgs dynamics.
Partly this course is about unpacking the above equation to understand what it is telling
us. To do that, we have to learn a few things about quantum field theories and in particular
about gauge theories.

2 Quantum field theory

Why quantum field theory? The answer is that in high energy physics we must combine
quantum mechanics (QM) with special relativity (SR). QM treats the physics of the very
small, such as electrons in an atom or in metals, when special relativity is not needed. SR on
the other hand treats things moving very fast when we dont need quantum mechanics. In
high energy physics, we need bothwe are dealing with very small things that are moving
very fast. Now in QM we have the uncertainty principle that states that the energy of a state
can fluctuate. In SR we have E = mc2 , so that particles with mass can be created out of
energy. This means that when we combine QM and SR, the number of particles is not a

1
constant, but particles can be created and annihilated. This gives us Quantum Field Theory.
QFT is very different from relativistic quantum mechanics (the Dirac equation) where the
number of particles is fixed.
In quantum mechanics, the electromagnetic field is treated as a field, with the associated
particle being the photon. Now we want to treat also matter particles such as the electron as
a field, e.g. the electron field.

2.1 Conventions
Before we start, we must take care of some preliminaries and definitions. I am going to use the
conventions used in the book by Peskin and Schroeder (which are different from Srednicki).

To start with, we will use natural units, meaning that we put c = ~ = 1. This means
that mass, momentum, and energy are measured in GeV. Length and time are measured
in GeV1 . A useful conversion factor is ~c = 1 = 197 MeV fm.

Relativity: Four-vectors (vectors with index up, or contravariant vectors) are written as
x = (x0 , x1 , x2 , x3 ) = (t, x, y, z) = (t, ~r) = (t, ~x) and p = (E, p~). The metric tensor is
g = diag(1, 1, 1, 1) = g , i.e. we are using the mostly minus convention. We
are always
P using the Einstein summation convention, summing over repeated indices:

x x x x .

Four-vectors with index down (dual vectors or covariant vectors) are given by lowering
the index: x = g x = (x0 , x1 , x2 , x3 ) = (t, ~x) and p = (E, p~).

Derivatives are written as x = t

, and x = t



, . The derivative
comes with the index naturally lowered, while the vector has it naturally raised. In
other words, the vector x is a vector of the manifold we are considering (Minkowski
space), while x is a vector in the dual space.

Quantum mechanics and relativity: Its well-known that in QM, we have the operator

relation E i~ t for energy and px i~ x or p~ i~ for momentum. In

relativistic QM we have (with ~ = 1) p i . You can check that this holds from the
QM relations, and that for a plane wave eikx this gives p eikx = k eikx .

Finally, the covariant form of electromagnetism is used. Maxwells equations in rational-

ized Heaviside-Lorentz units are
~ =
E
~ =0
B
B~
~ =
E
t
~
~ = ~j + E .
B
t

The potentials and A ~ = A~ , B

~ are defined through E ~ = A.
~ Then we
t
~ We also extract the charge from the
can define the four-vector potential A = (, A).

2
 current, j ej and define the field strength tensor F = D A A . Then the
relativistic form of Maxwells equations is

D F = ej (1)

 F = 0. (2)

3 Gauge theory
A gauge symmetry is a continuous, local, internal symmetry. A gauge theory is a quantum
field theory where the Lagrangian (L) is invariant under some gauge symmetry.
The set of possible gauge transformations form a group, known as the gauge group. This is
a Lie group, and its generators form a Lie algebra. In the Standard Model, we have the groups
U(1), SU(2), and SU(3). The group SU(N ) consists of all special unitary N N -matrices U ,
i.e., they fulfill U U = U U = 1 and det U = 1.

3.1 U(1) gauge theory

Let us start by considering a U(1) gauge theory. U(1) transformations consist of all unitary
1 1-matrices, i.e., the set of phase factors ei . The Dirac Lagrangian is given by

L = (x) (i m) (x). (3)

We demand invariance under the U(1) gauge transformation

(x) 0 (x) = ei(x) (x) U (x)(x). (4)

If is independent of x, then its easy to see that L is invariant, but this does not hold if
= (x). Then we instead get

L L0 = (x) (i m) (x) + i i (x). (5)

The additional term can be compensated for by introducing the covariant derivative

D igA (6)

where g is called the the gauge coupling and A is called the gauge field or the connection.
The gauge field must have the simultaneous gauge transformation
1
A (x) A (x) + (x), (7)
g
because then, under the gauge transformation (4),

D (x) ei(x) D (x) = U (x)D (x), (8)

so now D transforms in the same way as the field itself, meaning that the Lagrangian
will be invariant.
To get invariance under the gauge transformation, we are forced to introduce the new field
A (x). This is a physical new field, but what is it? To answer this question we write out the
new Lagrangian with the covariant derivative:

L = (x) (i D m) (x) = (x) (i m) (x) + g A . (9)

3
This is the original Dirac Lagrangian plus an additional term. The additional term is clearly
an interaction term since it contains a product of three fields. It is in fact the interaction
term of QED, if g is replaced by the electric charge e, and the field A is interpreted as the
photon field. The gauge field of QED is therefore the photon field.
Gauge fields are the force carriers of gauge theories, and in fact all force-carrying fields of
particle physics are gauge fields. We have seen how the requirement of gauge invariance on
the non-interacting Dirac Lagrangian dictates the form of the interaction.
The field A must also have a kinetic term. This is given by the Maxwell Lagrangian
1
F F (10)
4
where in U(1) theory, F A A . We also have
[D , D ] = igF . (11)
Note that from the gauge transformation of the field A in Eq. (7), it is easy to see that
the product A A is not gauge invariant. It is therefore not possible to add a term such as
m2 A A to the Lagrangian, and gauge fields must therefore be massless if the gauge symmetry
is exact. (This is the whole reason for introducing the Higgs mechanism to be discussed
later.) A product such as is however gauge invariant, and mass terms for fermions are
thus allowed in both QED and QCD. We will have a problem later when we consider the
electroweak theory, where mass terms are not allowed at all (we will look into why later in the
course). Since all fermions in the Standard Model carry electroweak charges, fermions must
also be massless in the Standard Model. This will be discussed later in the course.
So why did this happen? The reason we need to include the covariant derivative is that
when we take the partial derivative we are in effect trying to compare the field at two
different points in space-time, (x) and (x + dx), but because the gauge transformation os
local, i.e., it depends on the space-time point, the field at these two points have completely
different gauge transformation properties. The covariant derivative compensates for this
through the gauge field or connection. This is similar to the situation in general relativity
where space-time is curved and we introduce parallel transport.

3.2 Non-abelian Lie groups and Lie algebras

The above discussion can be generalized to more complicated gauge groups than U(1). All Lie
groups can be represented by matrices, and except U(1) they are all non-Abelian, meaning
that the commutator of two elements of the group is non-zero. The gauge groups of the
Standard Model are U(1), SU(2) and SU(3), and the gauge groups used in theories beyond
the Standard Model are mostly SU(N ) and SO(N ). I will from now on discuss only SU(N )
groups.
A general group element of SU(N ) is an N N -matrix with determinant equal to one. It
can be written as U = exp (ia T a ) where a is a set of N 2 1 parameters and T a is the set
of N 2 1 generators of the group. The generators are matrices and fulfill the commutation
relation h i
T a , T b = if abc T c (12)
where f abc are called structure constants. The commutation relation defines the Lie algebra
associated with the group and can be obtained by writing, for infinitesimal parameters a  1,
U ' 1 + ia T a . (13)

4
Thus the generators form a sort of basis for infinitesimal unitary transformations. Any group
element can then be obtained by exponentiation.
To be clear, there are infinitely many group elements U in SU(N ): they are all possible
matrices fulfilling the requirements of being unitary and having unit determinant. There is
a finite number of generators T a , a = 1, . . . , N 2 1, which fulfill the Lie algebra. Since the
group elements are unitary, the generators are Hermitian, i.e. T a = T a .
A representation of a group consists of a set of matrices that fulfill the same multiplication
law as the elements of the more abstract group, or in the case of the generators of the group,
they fulfill the same commutation relations. Any set of N 2 1 matrices T a that fulfills the
Lie algebra is a viable representation, and those matrices do not have to be N N matrices.
There is however always a representation in terms of N N matrices. This is called
the defining representation, or the fundamental representation. I.e., for a SU(N ) group,
the fundamental representation consists of all N N -matrices with unit determinant. The
fundamental representation of the corresponding SU(N ) Lie algebra then consists of a set of
N 2 1 different N N matrices that fulfill the commutation relations.
Since the gauge transformations are matrices acting on the fields, the fields must be
matrices too.

Example. SU(2) is the group of all unitary 2 2-matrices with unit determinant. The gauge
transformations are then in the fundamental representation the matrices U SU(2), such that

(x) U (x)(x). (14)

Then the field in the fundamental representation must be a column vector

 
1 (x)
(x) = (15)
2 (x)

We say that is a doublet of SU(2), or more precisely that it transforms as a doublet, or that
it transforms as a spinor under the spin-1/2 representation of SU(2).
It is also possible to have triplet fields. Then the generators are 3 3-matrices. This
corresponds to spin 1.

3.3 Non-abelian gauge theory

Let us now consider an SU(N ) gauge theory. Then the field transforms as

(1 + ia T a ) , (16)

or in component form
i i + ia (T a )ij j , (17)
which shows that the field is a column vector of the same dimension as the generators. This
dimension is different for different representations of the generators. This in turn depends on
what representation is chosen for the fields. In the SU(2) example given above, the generators
are 2 2-matrices and the field is a doublet. In QCD, the quark fields are in the triplet
representation of SU(3), meaning that they are 3-dimensional column vectors. The generators
acting on the quarks are then 3 3-matrices. In SUSY theories, there are fermions called
gluinos that are in the 8-dimensional, or octet, representation of SU(3). This is known as

5
the adjoint representation. Then the generators are 8 8-matrices. (The gluinos are the
supersymmetric partners of the gluons.)
We need a covariant derivative so that the covariant derivative of the field transforms
under the gauge transformations in the same way as the field itself. This turns out to be given
by
D = igAa T a (18)
where the sum runs over all the generators, a = 1, . . . N 2 1. There are therefore N 2 1
different gauge fields. (For example, in QCD, which is an SU(3) gauge theory, there are 8
gluons.)
The covariant derivative can be written in component form as

(D )ij = ij igAa (T a )ij , (19)

which shows that the covariant derivative is a matrix of the same dimension as the generators.
The gauge fields transform under the gauge transformation as
1 h i
Aa T a Aa T a + ( a )T a + i a T a , Ab T b , (20)
g
and you can check that simultaneously transforming and A gives the transformation

D (1 + ia T a ) D , (21)

just as we wanted.
The field-strength tensor of a non-abelian theory is more complicated than for U(1), where
we had [D , D ] = igF For SU(N ) this is generalized to
a
[D , D ] = igF T a, (22)

which gives the expression

h i
a
F T a = Aa T a Aa T a ig Aa T a , Ab T b , (23)

a T a , so one
for the field strength. Note that each term in this expression is a matrix, e.g. F
could write this in matrix form, with definitions such as F = Fa T a , as

F = A A ig [A , A ] . (24)
a in another useful form we can use the commutation relation T a , T b = if abc T c on
 
To put F
the commutator in Eq. (23), which gives
a
F = Aa Aa + gf abc Ab Ac . (25)

This form is useful to find the Lagrangian and Feynman rules for Aa .
The most important knowledge from all this are the forms of D and F a , since these give

all the interactions involving the gauge field.

The covariant derivative tells us how all other fields interact with the gauge field. In the
Standard Model, this means how the quarks interact with gluons, how the quarks and leptons

6
interact with the electroweak gauge bosons , W , Z 0 , and how the Higgs boson interacts
with the W and Z 0 . For example, the Dirac kinetic term is given by

i D = i + g Aa T a , (26)

which shows that the Feynman rule is that one should include a factor ig T a . Note that
and T a are both matrices, but they act in different spaces and are not multiplied with each
other as matrices. They therefore commute with each other.
The kinetic term for the gauge bosons tell us how the gauge bosons interact with each other.
This is a new phenomenon for non-abelian gauge theories and comes from the commutator in
Eq. (23). We have
1
Lkin = F a F
a
= (kinetic terms for A )
4
gf abc ( Aa )Ab Ac
1
g 2 f abc f ade Ab Ac Ad Ae , (27)
4
where I didnt write out the kinetic terms involving the partial derivative . The last two
lines are interaction terms for the gauge field, which arise from the kinetic term. We can
immediately see that there are three-particle and four-particle interactions. The Feynman
rule for the three-boson interaction includes momenta, since there is a derivative.
In fact, if the gauge group is SU(3), this gives exactly the interactions of QCD.