1

Introduction to Field Theory

The purpose of this book, is twofold. Here I will to introduce field theory as
a framework for the study of systems with a very large number of degrees
of freedom, N → ∞. And I will also introduce and develop the tools that
will allow us to treat such systems. Systems that involve a large (in fact,
infinite) number of coupled degrees of freedom arise in many areas of Physics,
notably in High Energy and in Condensed Matter Physics, among others.
Although the physical meaning of these systems and their symmetries are
quite different, they actually have much more in common than it may seem
at first glance. Thus, we will discuss, on the same footing, the properties of
relativistic quantum field theories, classical statistical mechanical systems
and condensed matter systems at finite temperature. This is a very broad
field of study and we will not be able to cover each area in great depth.
Nevertheless, we will learn that it is often that case that what is clear in one
context can be used to expand our knowledge in a different physical setting.
We will focus on a few unifying themes, such as the construction of the
ground state (the “vacuum”), the role of quantum fluctuations, collective
behavior, and the response of these systems to weak external perturbations.

1.1 Examples of fields in physics

1.1.1 The electromagnetic field
Let us consider a very large box of linear size L → ∞, and the electromag-
netic field enclosed inside it. At each point in space x we can define a vector
(which is a function of time as well) A(x, t) and a scalar A0 (x, t). These
are the vector and scalar potentials. The physically observable electric field
2 Introduction to Field Theory

E(x, t) and the magnetic field B(x, t) are defined in the usual way

(x, t) − !A0 (x, t) (1.1)

1 ∂A
B(x, t) = ! × A(x, t), E(x, t) = − c
∂t
The time evolution of this dynamical system is determined by a local La-
grangian density (which we will consider in section 2.6). The equations of
motion are just the Maxwell equations. Let us define the 4-vector field
A (x) = (A (x), A(x))
µ 0 0
A ≡ A0 (1.2)
where µ = 0, 1, 2, 3 are the time and space components. Here x stands for
the 4-vector
x = (ct, x)
µ
(1.3)
µ
To every point x of Minkowski spacetime M we associate a value of the
µ
vector potential A . The vector potentials are ordered sets of four real num-
4
bers and hence are elements of R . Thus a field configuration can be viewed
4
as a mapping of the Minkowski spacetime M onto R ,
µ 4
A ∶M↦R (1.4)
Since spacetime is continuous we need an infinite number of 4-vectors to
specify a configuration of the electromagnetic field, even if the box were
finite (which is not). Thus we have a infinite number of degrees of freedom
for two reasons: spacetime is both continuous and infinite.

1.1.2 The elastic field of a solid

Consider a three-dimensional crystal. A configuration of the system can be
described by the set of positions of its atoms relative to their equilibrium
state, (i.e. the set of deformation vectors d at every time t). Lattices are
labeled by ordered sets of three integers and are equivalent to the set
3
Z =Z×Z×Z (1.5)
whereas deformations are given by sets of three real numbers, and are ele-
3
ments of R . Hence a crystal configuration is a mapping
3 3
d∶Z ×R↦R (1.6)
At length scales ", which are large compared to the lattice spacing a but small
3
compared to the linear size L of the system, we can replace the lattice Z
by a continuum description in which the crystal is replaced by a continuum
3
three-dimensional Euclidean space R . Thus the dynamics of the crystal
 1.1 Examples of fields in physics 3
3 4
requires a four-dimensional spacetime R × R = R . Hence the configuration
space becomes the set of continuous mappings
4 3
d∶R ↦R (1.7)

In this continuum description, the dynamics of the crystal is specified in

terms of the displacement vector field d(x, t) and its time derivatives, the
velocities ∂d
∂t
(x, t), which define the mechanical state of the system. This is
the starting point of the theory of elasticity. The displacement field d is the
elastic field of the crystal.

1.1.3 The order-parameter field of a ferromagnet

Let us now consider a ferromagnet. This is a physical system, usually a solid,
in which there is a local average magnetization field M (x) in the vicinity of
a point x. The local magnetization is simply the sum of the local magnetic
moments of each atom in the neighborhood of x. At scales long compared
to microscopic distances (the interatomic spacing a), M (x) is a continuous
real vector field. In some situations, of interest, the magnitude of the local
moment does not fluctuate but its local orientation does. Hence, the local
state of the system is specified locally by a three-component unit vector n.
Since the set of unit vector is in one-to-one correspondence with the points
2
on a sphere S , the configuration space is equivalent (isomorphic) to the sets
2
of mappings of Euclidean three-dimensional space onto S ,
3 2
n∶R ↦S (1.8)

In an ordered state the individual magnetic moments become spontaneously

oriented along some direction. For this reason, the field n is usually said to
be an order parameter field. In the theory of phase transitions, the order
parameter field represents the important degrees of freedom of the physical
system,( i.e., the degrees of freedom that drive the phase transition).

1.1.4 Hydrodynamics of a charged fluid

Charged fluids can be described in terms of hydrodynamics. In hydrodynam-
ics, one specifies the charge density ρ(x, t) and the current density j(x, t)
µ
near a spacetime point x . The charge and current densities can be repre-
sented in terms of the 4-vector

j (x) = (cρ(x, t), j(x, t))

µ
(1.9)
4 Introduction to Field Theory

where c is a suitably chosen speed (generally not the speed of light!). Clearly,
the configuration space is the set of maps
µ 4 4
j ∶R ↦R (1.10)
In general we will be interested both in the dynamical evolution of such
systems and in their large-scale (thermodynamic) properties. Thus, we will
need to determine how a system that, at some time t0 is in some initial
state, manages to evolve to some other state after time T . In Classical Me-
chanics, the dynamics of any physical system can be described in terms of a
Lagrangian. The Lagrangian is a local functional of the field and of its space
and time derivatives. “Local” here means that the equations of motion can
be expressed in terms of partial differential equations. In other words, we
do not allow for “action-at-a-distance,” but only for local evolution. Simi-
larly, the thermodynamic properties of these systems are governed by a local
energy functional, the Hamiltonian. That the dynamics is determined by a
Lagrangian means that the field itself is regarded as a mechanical system
to which the standard laws of Classical Mechanics apply. Here, the wave
equations of the fluid are the equations of motion of the field. This point of
view will also tell us how to quantize a field theory.

1.2 Why quantum field theory?

From a historical point of view, quantum field theory (QFT) arose as an
outgrowth of research in the fields of nuclear and particle physics. In partic-
ular, Dirac’s theory of electrons and positrons was, perhaps, the first QFT.
Nowadays, QFT is used, both as a picture and as a tool, in a wide range of
areas of physics. In this course, I will not follow the historical path of the
way QFT was developed. By and large, it was a process of trial and error in
which the results had to be reinterpreted a posteriori. The introduction of
quantum field theory as the general framework of particle physics implied
that the concept of particle had to be understood as an excitation of a field.
Thus photons become the quantized excitations of the electromagnetic field
with particle-like properties (such as momentum), as anticipated by Ein-
stein’s 1905 paper on the photoelectric effect. Dirac’s theory of the electron
implied that even such “conventional” particles should also be understood
as the excitations of a field.
The main motivation of these developments was the need to reconcile,
or unify, Quantum Mechanics with Special Relativity. In addition, the ex-
perimental discoveries of the spin of the electron and of electron-positron
creation by photons, showed that not only was the Schrödinger equation
 1.2 Why quantum field theory? 5

inadequate to describe such physical phenomena, but the very notion of a

particle itself had to be revised.
Indeed, let us consider the Schrödinger equation

HΨ = ih̵
∂Ψ
(1.11)
∂t
where H is the Hamiltonian
2
+ V (x)
p̂
H= (1.12)
2m
and p̂ is the momentum represented as a differential operator
h̵
p̂ = ! (1.13)
i
acting on the Hilbert space of wave functions Ψ(x).
The Schrödinger equation is invariant under Galilean transformations,
provided the potential V (x) is constant, but not under general Lorentz
transformations. Hence, Quantum Mechanics as described by the Schrödinger
equation, is not compatible with the requirement that the description of
physical phenomena must be identical for all inertial observers. In addi-
tion, it cannot describe pair-creation processes since in the non-relativistic
Schrödinger equation, the number of particles is strictly conserved.
Back in the late 1920s, two apparently opposite approaches were proposed
to solve these problems. We will see that these approaches actually do not
exclude each other. The first approach was to stick to the basic structure of
“particle” Quantum Mechanics and to write down a relativistically invariant
version of the Schrödinger equation. Since in Special Relativity the natural
Lorentz scalar involving the energy E of a particle of mass m is E − (p c +
2 2 2

m c ), it was proposed that the “wave functions” should be solutions of the

2 4

equation (the “square” of the energy)

̵
[(ih̵ ) − (( !) + m c )] Ψ(x, t) = 0
∂ 2 hc 2
2 4
(1.14)
∂t i

This is the Klein-Gordon equation. This equation is invariant under the

Lorentz transformations,
x = (x0 , x)
µ µ,ν ′ µ
x =Λ xν (1.15)
provided that the “wave function” Ψ(x) is also a scalar (i.e. invariant) under
Lorentz transformations
Ψ(x) = Ψ (x )
′ ′
(1.16)
However, it soon became clear that the Klein-Gordon equation was not
6 Introduction to Field Theory

compatible with a particle interpretation. In addition, it cannot describe

particles with spin. In particular, the solutions of the Klein-Gordon equation
have the (expected) dispersion law
2 2 2 2 4
E =p c +m c (1.17)
which implies that there are positive and negative energy solutions
√
E = ± p2 c2 + m2 c4 (1.18)
From a “particle” point of view, negative energy states are unacceptable
since they would imply that there is no ground state. We will see in chapter
4 that in quantum field theory there is a natural and simple interpretation of
these solutions, and that in no way make the system unstable. However, the
meaning of the negative energy solutions was unclear in the early thirties.
To satisfy the requirement from Special Relativity that energy and mo-
mentum must be treated equally, and to avoid the “negative energy solu-
tions” that came from working with the “square” of the Hamiltonian H,
Dirac proposed to look for an equation that was linear in derivatives (Dirac,
1928). In order to be compatible with Special Relativity, the equation must
be covariant under Lorentz transformations,( i.e. it should have the same
form in all reference frames). Dirac proposed a matrix equation that is linear
in derivatives with a“wave function” Ψ(x) in the form of a four-component
vector, a 4-spinor Ψa (x) (with a = 1, . . . , 4)
̵ 3 ab
ih̵ (x) + ∑ αj ∂j Ψb (x) + mc βab Ψb (x) = 0
∂Ψa hc 2
(1.19)
∂t i
j=1

where αj and β are four 4 × 4 matrices. For this equation to be covariant

it is necessary that the 4-spinor field Ψ should transform as a spinor under
Lorentz transformations
Ψa (Λx) = Sab (Λ)Ψb (x)
′
(1.20)
where S(Λ) is a suitable matrix. The matrices αj and β have to be pure
numbers independent of the reference frame. By further requiring that the
iterated form of this equation (i.e. the “square”) satisfies the Klein-Gordon
equation for each component separately, Dirac found that the matrices obey
the (Clifford) algebra
{αj , αk } = 2δjk 1, {αj , β} = 0,
2 2
αj = β = 1 (1.21)

√
where 1 is the 4 × 4 identity matrix. The solutions are easily found to have
the energy eigenvalues E = ± p2 c2 + m2 c4 . (We will come back to this in
 1.2 Why quantum field theory? 7

chapter .2) It is also possible to show that the solutions are spin 1/2 particles
and antiparticles (we will discuss this later on).
However, the particle interpretation of both the Klein-Gordon and the
Dirac equations was problematic. Although spin 1/2 appeared now in a
natural way, the meaning of the negative energy states remained unclear.
The resolution of all of these difficulties was the fundamental idea that
these equations should not be regarded as the generalization of Schrödinger’s
equation for relativistic particles but, instead, as the equations of motion of
a field, whose excitations are the particles, much in the same way as the pho-
tons are the excitations of the electromagnetic field. In this picture particle
number is not conserved but charge is. Thus, photons interacting with mat-
ter can create electron-positron pairs. Such processes do not violate charge
conservation but the notion of a particle as an object that is a fundamental
entity and has a distinct physical identity is lost. Instead, the field becomes
the fundamental object and the particles become the excitations of the field.
Thus, the relativistic generalization of Quantum Mechanics is Quantum
Field Theory. This concept is the starting point of Quantum Field Theory.
The basic strategy to seek a field theory with specific symmetry properties
and whose equations of motion are Maxwell, Klein-Gordon and Dirac equa-
tions, respectively. Notice that if the particles are to be regarded as the
excitations of a field, there can be as many particles as we wish. Thus, the
Hilbert space of a Quantum Field Theory has an arbitrary (and indefinite)
number of particles. Such a Hilbert space is called a Fock space.
Therefore, in Quantum Field Theory the field is not the wave function of
anything. Instead the field represents an infinite number of degrees of free-
dom. In fact, the wave function in a Quantum Field Theory is a functional
of the field configurations which themselves specify the state of the system.
We will see below that the states in Fock space are given either by specify-
ing the number of particles and their quantum numbers or, alternatively, in
terms of the amplitudes (or configurations) of some properly chosen fields.
Different fields transform differently under Lorentz transformations and
constitute different representations of the Lorentz group. Consequently, their
excitations are particles with different quantum numbers that label the rep-
resentation. Thus,

1) The Klein-Gordon field φ(x) represents charge-neutral scalar spin-0 par-

ticles. Its configuration space is the set of mappings of Minkowski space
onto the real numbers φ ∶ M ↦ R, or complex numbers for charged spin-0
particles φ ∶ M ↦ C.
2) The Dirac field represents charged spin-1/2 particles. It is a complex 4-
8 Introduction to Field Theory

spinor Ψα (x) (α = 1, . . . , 4) and its configuration space is the set of maps

4
Ψα ∶ M ↦ C , while it is real for neutral spin-1/2 particles (such as
neutrinos).
3) The gauge field A (x) for the electromagnetic field, and its non-abelian
µ

generalizations for gluons (and so forth).

The description of relativistic quantum mechanics in terms of relativis-

tic quantum fields solved essentially all of the problems that originated its
initial development. Moreover, Quantum Field Theory gives exceedingly ac-
curate predictions of the behavior of quantized electromagnetic fields and
charged particles, as described by Quantum Electrodynamics (QED). Quan-
tum Field Theory also gives a detailed description of both the strong and
weak interactions in terms of field theories known as Quantum Chromody-
namics (QCD), based on Yang-Mills gauge field theories, and Unified and
Grand Unified gauge theories.
However, along with its successes, Quantum Field Theory also brought
with it a completely new set of physical problems and questions. Essentially,
any Quantum Field Theory of physical interest is necessarily a nonlinear the-
ory as it has to describe interactions. So even though the quantum numbers
of the excitations (i.e. the “particle” spectrum) may be quite straightforward
in the absence of interactions, the intrinsic non-linearities of the theory may
actually unravel much of this structure. Note that the equations of motion
of Quantum Field Theory are nonlinear, as they also are in Quantum Me-
chanics. However, the wave functional of a Quantum Field Theory obeys a
linear Schrödinger equation just as the wave function does in non-relativistic
Quantum Mechanics.
In the early days of Quantum Field Theory, and indeed for some time
thereafter, it was assumed that perturbation theory could be used in all
cases to determine the actual spectrum. It was soon found out that while
there are several cases of great physical interest in which some sort of per-
turbation theory yields an accurate description of the physics, in many more
situations this is not the case. Early on it was found that, at every order
in perturbation theory, there are singular contributions to many physical
quantities. These singularities reflected the existence of an infinite number
of degrees of freedom, both at short distances, since spacetime is a contin-
uum (the ultraviolet (UV) domain), and at long distances, since spacetime
is (essentially) infinite (the infrared (IR) domain). Qualitatively, divergent
contributions in perturbation theory come about because degrees of freedom
from a wide range of length scales (or wavelengths) and energy scales (or
frequencies) contribute to the expectation values of physical observables.
 1.2 Why quantum field theory? 9

Historically, the way these problems were dealt with was through the
process of regularization (i.e. making the divergent contributions finite),
and renormalization (i.e. defining a set of effective parameters which are
functions of the energy and/or momentum scale at which the system is
probed). Regularization required that the integrals to be cutoff at some
high energy scale (in the UV). Renormalization was then thought of as the
process by which these arbitrarily introduced cutoffs were removed from the
expressions for physical quantities. This was a physically obscure procedure,
but it worked brilliantly in QED and, to a lesser extent, in QCD. Theories
for which such a procedure can be implemented with the definition of only
a finite number of renormalized parameters (the actual input parameters
to be taken from experiment) are said to be renormalizable quantum field
theories. QED and QCD are the most important examples of renormalizable
quantum field fheories, although there are many others.
Renormalization implies that the connection between the physical observ-
ables and the parameters in the Lagrangian of a Quantum Field Theory is
highly non-trivial, and that the spectrum of the theory may have little to do
with the predictions of perturbation theory. This is the case for QCD whose
“fundamental fields” involve quarks and gluons but the actual physical spec-
trum consists only of bound states whose quantum numbers are not those of
either quarks or gluons. Renormalization also implies that the behavior of
the physical observables depends of the scale at which the theory is probed.
Moreover, a closer examination of these theories also revealed that they may
exist in different phases in which the observables have different behaviors,
with a specific particle spectrum in each phase. In this way, to understand
what a given Quantum Field Theory predicted became very similar to the
study of phases in problems in Statistical Physics. We will explore these
connections in detail later in this book when we develop the machinery of
the Renormalization Group in chapter 15. In this picture, the vacuum (or
ground state) of a quantum field theory corresponds to a phase much in the
same way as in Statistical (or Condensed Matter) Physics.
While the requirement of renormalizability works for the Standard Model
of particle physics it fails for Gravity. The problem of unifying Gravity with
the rest of the forces of Nature remains a major problem in contemporary
physics. A major program to solve this problem is String Theory. String
Theory is the only known viable candidate to quantize Gravity in a con-
sistent manner. However, in String Theory, Quantum Field Theory is seen
as an effective low energy (hydrodynamic) description of Nature, and the
Quantum Field Theory singularities are “regularized” by String Theory in
a natural way (but at the price of locality).