The title claims it is an introduction, but don’t be misled – this book will take you

a lot further than that. Suffice to say, this is where most budding particle theorists
learn field theory these days.

• Gauge Theories in Particle Physics, Aitchison I J R and Hey A J G (4th edn 2 vols
IoP 2012) [4, 5].
These two volumes are designed for experimental particle physicists and offer a gentler
(if longer) introduction to the ideas of gauge theory. The canonical quantization
approach is followed and both volumes are needed to cover this course.

• An Invitation to Quantum Field Theory, Alvarez-Gaume L and Vazquez-Mozo M A

(Springer Lecture Notes in Physics vol 839 2011)[6].
At a similar level to these notes, but discusses other interesting aspects not covered
here. An earlier version can be found at [7].
The necessary group theory aspects of the course are covered in the above books, but
to learn it properly I would read

• Lie Algebras in Particle Physics, Georgi H (2nd edn Frontiers in Physics vol 54 1999)
[8].

3 Notation and conventions

To make the formulæ as streamlined as possible, we use a system of units in which there is
only one dimensionful quantity (so that we may still do dimensional analysis) – energy –
and in which ~ = c = 1. 2 Thus E = mc2 becomes E = m, and so on.
For relativity, we set x0 = t, x1 = x, x2 = y, x3 = z and denote the components of the
position 4-vector by xµ , with a Greek index. The components of spatial 3-vectors will be
denoted by Latin indices, e.g. xi = (x, y, z). We define Lorentz transformations as those
transformations which leave the metric η µν = diag(1, −1, −1, −1) invariant (they are said
to form the group SO(3, 1)). Thus, under a Lorentz transformation, xµ → x0µ = Λµν xν , we
must have that η µν → Λµσ Λν ρ η σρ = η µν . The reader may check, for example, that a boost
along the x axis, given by
 
γ −βγ 0 0
−βγ γ 0 0
 
Λµν =  , (3.1)
 0 0 1 0
0 0 01

with γ 2 = (1 − β 2 )−1 , has just this property.

Any set of four components transforming in the same way as xµ is called a contravari-
∂ ∂ ∂ ∂
ant 4-vector. The derivative ( ∂t , ∂x , ∂y , ∂z ) (which we denote by ∂µ ), transforms as the
(matrix) inverse of x . Thus we define, ∂µ → ∂µ0 = Λµν ∂ν , with Λµν Λµρ = δρν , where
µ

δ = diag(1, 1, 1, 1). Any set of four components transforming in the same way as ∂µ is
2
Unfortunately I have not been able to find a consistent set of units in which 2π = 1!

–3–
called a covariant 4-vector. We now make the rule that indices may be raised or lowered
using the metric tensor η µν or its inverse, which we write as ηµν = diag(1, −1, −1, −1).
Thus, xµ = ηµν xν = (t, −x, −y, −z). With this rule, any expression in which all indices are
contracted pairwise with one index of each pair upstairs and one downstairs is manifestly
Lorentz invariant. For example,3 xµ xµ = t2 − x2 − y 2 − z 2 → x0µ x0µ = xµ xµ .
When we come to spinors, we shall need the gamma matrices, γ µ , which are a set of
four, 4 x 4 matrices satisfying the Clifford algebra {γ µ , γ ν } ≡ γ µ γ ν +γ ν γ µ = 2η µν ·1, where 1
denotes a 4 x 4 unit matrix. In these lecture notes, we shall use two different representations,
both of which are common in the literature. The first is the chiral representation, given by
!
0 σµ
γµ = , (3.2)
σµ 0

where σ µ = (1, σ i ), σ µ = (1, −σ i ), and σ i are the usual 2 x 2 Pauli matrices:

! ! !
1 01 0 −i 1 0
σ = , σ2 = , σ3 = . (3.3)
10 i 0 0 −1

For this representation,

!
5 0 1 2 3 −1 0
γ ≡ iγ γ γ γ = . (3.4)
0 1

The other representation for gamma matrices is the Pauli-Dirac representation, in which
we replace
!
1 0
γ0 = (3.5)
0 −1

and hence
!
01
γ5 = . (3.6)
10

We shall often employ Feynman’s slash notation, where, e.g., a / ≡ aµ γ µ and we shall
often write an identity matrix as 1, or indeed omit it altogether. Its presence should always
be clear from the context.4
Finally, it is to be greatly regretted that the electron was discovered before the positron
and hence the particle has negative charge. We therefore set e < 0.

3
We employ the usual Einstein summation convention, xµ xµ ≡ 3µ=0 xµ xµ .
P
4
All this cryptic notation may seem obtuse to you now, but most people grow to love it. If you don’t,
sue me.

–4–
4 Relativistic quantum mechanics

4.1 Why QM does and doesn’t work

I promised, dear reader, that I would begin with the Schrödinger equation of non-relativistic
quantum mechanics. Here it is:

∂ψ 1 2
i =− ∇ ψ + V ψ. (4.1)
∂t 2m
For free particles, with V (x) = 0, the equation admits plane wave solutions of the form
p2
ψ ∝ ei(p·x−Et) , provided that E = 2m , corresponding to the usual Energy-momentum
dispersion relation for free, non-relativistic particles.
No doubt all of this, together with the usual stuff about |ψ(x)|2 being interpreted
as the probability to find a particle at x, is old hat to you. By now, you have solved
countless complicated problems in quantum mechanics with spinning electrons orbiting
protons, bouncing off potential steps, being perturbed by hyperfine interactions, and so on.
But at the risk of boring you, and before we leap into the weird and wonderful world of
relativistic quantum mechanics and quantum field theory, I would like to spend a little time
dwelling on what quantum mechanics really is.
The reason I do so is because the teaching of quantum mechanics these days usually
follows the same dogma: firstly, the student is told about the failure of classical physics at
the beginning of the last century; secondly, the heroic confusions of the founding fathers
are described and the student is given to understand that no humble undergraduate stu-
dent could hope to actually understand quantum mechanics for himself; thirdly, a deus ex
machina arrives in the form of a set of postulates (the Schrödinger equation, the collapse
of the wavefunction, etc); fourthly, a bombardment of experimental verifications is given,
so that the student cannot doubt that QM is correct; fifthly, the student learns how to
solve the problems that will appear on the exam paper, hopefully with as little thought as
possible.
The problem with this approach is that it does not leave much opportunity to wonder
exactly in what regimes quantum mechanics does and does not work, or indeed why it has
a chance of working at all. This, unfortunately, risks leaving the student high and dry when
it turns out that QM (in its non-relativistic, undergraduate incarnation) is not a panacea
and that it too needs to be superseded.
R
To give an example, every student knows that dx|ψ(x, t)|2 gives the total probability
to find the particle and that this should be normalized to one. But a priori, this integral
could be a function of t, in which case either the total probability to find the particle would
change with time (when it should be fixed at unity) or (if we let the normalization constant
be time-dependent) the normalized wavefunction would no longer satisfy the Schrödinger
equation. Neither of these is palatable. What every student does not know, perhaps, is that
this calamity is automatically avoided in the following way. It turns out that the current

i
j µ = (ρ, j) = (ψ ∗ ψ, − (ψ ∗ ∇ψ − ψ∇ψ ∗ )) (4.2)
2m

–5–
is conserved, satisfying ∂µ j µ = 0. (For now, you can show this directly using the Schrödinger
equation, but soon we shall see how such conserved currents can be identified just by looking
at the Lagrangian; in this case, the current conservation follows because a phase rotated
wavenfunction ψ 0 = eiα ψ also satisfies the Schrödinger equation.) Why conserved? Well,
integrating ∂µ j µ = 0 we get that the rate of change of the time component of the current
in a given volume is equal to (minus) the flux of the spatial component of the current out
of that volume:
Z Z
d
ρdV = − j · dS. (4.3)
dt ∂V

In particular, ψ ∗ ψ integrated over all space, is constant in time. This is a notion which
is probably familiar to you from classical mechanics and electromagnetism. It says that
ψ ∗ ψ, which we interpret as the probability density in QM, is conserved, meaning that the
probability interpretation is a consistent one.
This conservation of the total probability to find a particle in QM is both its salvation
and its downfall. Not only does it tell us that QM is consistent in the sense above, but it
also tells that QM cannot hope to describe a theory in which the number of particles present
changes with time. This is easy to see: if a particle disappears, then the total probability
to find it beforehand should be unity and the total probability to find it afterwards should
be zero. Note that in QM we are not forced to consider states with a single particle (like a
single electron in the Coulomb potential of a hydrogen atom), but we are forced to consider
states in which the number of particles is fixed for all time. Another way to see this is that
the wavefunction for a many-particle state is given by ψ(x1 , x2 , . . . ), where x1 , x2 , . . . are
the positions of the different particles. But there is no conceivable way for this wavefunction
to describe a process in which a particle at x1 disappears and a different particle appears
at some other x3 .
Unfortunately, it happens to be the case in Nature that particles do appear and dis-
appear. An obvious example is one that (amusingly enough) is usually introduced at the
beginning of a QM course, namely the photoelectric effect, in which photons are annihilated
at a surface. It is important to note that it is not the relativistic nature of the photons
which prevents their description using QM, it is the fact that their number is not conserved.
Indeed, phonons arise in condensed matter physics as the quanta of lattice vibrations. They
are non-relativistic, but they cannot be described using QM either.
Ultimately, this is the reason why our attempts to construct a relativistic version of
QM will fail: in the relativistic regime, there is sufficient energy to create new particles
and such processes cannot be described by QM. This particle creation is perhaps not such
a surprise. You already know that in relativity, a particle receives a contribution to its
energy from its mass via E = mc2 . This suggests (but certainly does not prove) that if
there is enough E, then we may be able to create new sources of m, in the form of particles.
It turns out that this does indeed happen and indeed much of current research in particle
physics is based on it: by building colliders (such as the Large Hadron Collider) producing
ever-higher energies, we are able to create new particles, previously unknown to science and
to study their properties.

–6–
 Even though our imminent attempt to build a relativistic version of QM will eventually
fail, it will turn out to be enormously useful in finding a theory that does work. That theory
is called Quantum Field Theory and it will be the subject of the next section. For now, we
will press ahead with relativistic QM.

4.2 The Klein-Gordon equation

To write down a relativistic version of the Schrödinger equation is easy - so easy, in fact,
that Schrödinger himself wrote it down before he wrote down the equation that made him
famous. Starting from the expectation that the free theory should have plane wave solutions
µ
(just as in the non-relativistic case), of the form φ ∝ e−iEt+ip·x = e−ipµ x and noting that
the relativistic dispersion relation pµ pµ = m2 should be reproduced, we infer the Klein-
Gordon equation

(∂µ ∂ µ + m2 )φ = 0. (4.4)

If we assume that φ is a single complex number, then it must be a Lorentz scalar, being
invariant under a Lorentz transformation: φ(xµ ) → φ0 (x0µ ). The Klein-Gordon equation is
then manifestly invariant under Lorentz transformations. The problems with this equation
quickly become apparent. Firstly, the probability density cannot be |φ|2 as it is in the
non-relativistic case, because |φ|2 transforms as a Lorentz scalar (i.e. it is invariant), rather
than as the time component of a 4-vector (the probability density transforms like the inverse
of a volume, which is Lorentz contracted). Moreover, |φ|2 is not conserved in time. To find
the correct probability density, we must find a conserved quantity. Again, we shall soon
have the tools in hand to do so ourselves, but for now we pull another rabbit out of the
hat, claiming that the 4-current

j µ = i(φ∗ ∂ µ φ − φ∂ µ φ∗ ) (4.5)

satisfies ∂µ j µ = 0 (exercise), meaning that its time component integrated over space,
R ∂ ∂ ∗
R ∂
dxi(φ∗ ∂t φ − φ ∂t φ ) is a conserved quantity. So far so good, but note that dxi(φ∗ ∂t φ−
∂ ∗ ∓ip x µ
φ ∂t φ ) is not necessarily positive. Indeed, for plane waves of the form φ = Ae µ , we
µ
obtain ρ = ±2E|A|2 . There is a related problem, which is that the solutions φ = Ae±ipµ x ,
correspond to pboth positive and negative energy solutions of the relativistic dispersion rela-
tion: E = ± p2 + m2 . Negative energy states are problematic, because there is nothing to
stop the vacuum decaying into these states. In classical relativistic mechanics, the problem
of these negative energy solutions never reared its ugly head, because we could simply throw
them away, declaring that all particles (or rockets or whatever) have positive energy. But
when we solve a wave equation (as we do in QM), completeness requires us to include both
positive and negative energy solutions in order to be able to find a general solution.

4.3 The Dirac equation

In 1928, Dirac tried to solve the problem of negative-energy solutions by looking for a wave
equation that was first order in time-derivatives, the hope being that one could then obtain

–7–
 p
a dispersion relation of the form E = + p2 + m2 directly, without encountering negative-
energy states. Dirac realised that one could write an equation that was linear in both time
and space derivatives of the form

(iγ ν ∂ν − m)ψ = 0 (4.6)

that implied the Klein-Gordon equation for ψ, provided that the 4 constants γ ν were ma-
trices. To wit, acting on the left with (iγ µ ∂µ + m), we obtain

(−γ µ γ ν ∂µ ∂ν − m2 )ψ = 0. (4.7)

Since ∂µ ∂ν = ∂ν ∂µ , we may symmetrize to get

1
(− {γ µ , γ ν }∂µ ∂ν − m2 )ψ = 0. (4.8)
2
Thus, (minus) the Klein-Gordon equation is recovered if the anticommutator is such that

{γ µ , γ ν } = 2η µν . (4.9)

The γ ν evidently cannot be simply numbers, since, for example, γ 0 γ 1 = −γ 1 γ 0 . In fact, the
smallest possible matrices that implement this relation are 4x4, as you may show by trial
and error. Any set of matrices satisfying the algebra will do, but some are more convenient
than others, depending on the problem at hand. We will either use the chiral representation
!
0 σ µ
γµ = , (4.10)
σµ 0

where σ µ = (1, σ i ), σ µ = (1, −σ i ), and σ i are the usual 2 x 2 Pauli matrices:

! ! !
0 1 0 −i 1 0
σ1 = , σ2 = , σ3 = (4.11)
10 i 0 0 −1

or we will use the Pauli-Dirac representation in which we replace

!
1 0
γ0 = . (4.12)
0 −1

Note that γ 0 is Hermitian in either representation, whereas γ i are anti-Hermitian.

This can be conveniently written as (γ µ )† = γ 0 γ µ γ 0 , but note that this equation (and the
hermiticity properties) are not basis-independent. Since the γ ν are 4 x 4 matrices, the
wavefunction ψ must have 4 components. It is not a 4-vector (and nor are the γ ν , despite
the suggestive notation, since they are constants and do not transform). It transforms in a
special way under Lorentz transformations (which we don’t have time to go through here,
sadly) and we call it a 4-component spinor. It is easy enough to show that Dirac’s equation
has a conserved current given by (one final rabbit, I promise)

j µ = (ψ † ψ, ψ † γ 0 γ i ψ), (4.13)

–8–
where ψ † is the Hermitian conjugate (transpose conjugate) of ψ. Note that the probability
density, ψ † ψ is now positive definite, so Dirac managed to solve one problem. But what
about the negative energy solutions? In the rest frame, with (E, p) = (m, 0), we find
solutions to (4.6) of the form A∓ e∓imt , provided that
   
A1 0
A  0
 2  
(±γ 0 − 1)A∓ = 0 =⇒ A− ∝   , A+ ∝   , (4.14)
0  A3 
0 A4

where we used the Pauli-Dirac basis. So there are four modes, two of which have positive
energy and two of which have negative energy. The two positive energy modes are inter-
preted (as we shall soon see) as the two different spin states of a spin-half particle. Dirac’s
proposal to deal with the negative energy states was as follows. Since the Pauli exclusion
principal for these spin-half fermions forbids multiple occupation of states, one can postu-
late that the vacuum corresponds to a state in which all of the negative energy states are
filled. Then, Dirac argued, if one has enough energy, one might be able to promote one of
these negative-energy particles to a positive-energy particle. One would be left with a ’hole’
in the sea of negative energy states, which would behave just like a particle with opposite
charge to the original particles. Thus Dirac came up with the concept of antiparticles. The
antiparticle of the electron, the positron, was duly found, bringing great acclaim to Dirac.
But this picture of the Dirac sea was soon rendered obsolete by the emergence of quantum
field theory.
It is not much harder to find the plane-wave solutions of the Dirac equation in any frame,
so we do it for completeness. For the positive-energy
! solutions of (4.6), write ψ = ue−ip·x ,
φ
such that (p / − m)u = 0. Writing u = implies
χ
!
φ
u=N σ·p . (4.15)
E+m φ

! !
1 0
Finally, taking the two states to be φ1 = and φ2 = , we obtain
0 1
   
1 0
 0   
1
   
u1 = N  pz  , u2 = N  px −ipy  . (4.16)
 E+m   E+m 
px +ipy −pz
E+m E+m

For the negative-energy solutions, write ψ = ve+ip·x , such that (p

/ + m)v = 0. Thus,
!
σ·p
v=N E+m χ , (4.17)
χ

–9–
such that
 px −ipy   pz

E+m E+m
 −pz   px +ipy 
 E+m   E+m 
v1 = N   , v2 = N  . (4.18)
 0   1 
1 0

We find it most convenient to normalize in such a way that that there is a number density
√
ρ = ψ † ψ = u† u = v † v of 2E particles per unit volume. This fixes N = E + m.
We end our treatment of the Dirac equation by showing that it does indeed describe a
spin-half particle. To do so, we show that there exists an operator S, such that J ≡ L + S
is a constant of the motion with S2 = s(s + 1) = 34 . First note that the orbital angular
momentum L does not commute with the Hamiltonian, defined, à la Schrödinger, to be
everything that appears on the right of the Dirac equation when i ∂ψ ∂t appears on the left.
Thus,

H = γ 0 (γ i pi + m). (4.19)

Then, for example

[L3 , H] = [x1 p2 − x2 p1 , H] = [x1 , H]p2 − [x2 , H]p1 = iγ 0 (γ 1 p2 − γ 2 p1 ) 6= 0. (4.20)

!
σi 0
The operator S that ensures [H, J i ] = 0 is given by S ≡ Σ
2, where Σi ≡ . As a
0 σi
check (in the chiral basis),
! !
1 σ3 0 −σ i pi m
[S3 , H] = [ , ] = −iγ 0 (γ 1 p2 − γ 2 p1 ) = −[L3 , H]. (4.21)
2 0 σ3 m σ i pi

Moreover, S2 = 41 σi σi = 43 , as required.

4.4 Maxwell’s equations

This is a convenient juncture at which to introduce Maxwell’s equations of electromag-
netism, even though we make no effort to make a quantum mechanical theory out of them
(since the number of photons is not fixed, it is doomed to fail). We shall need them for our
later study of QFT, however.
In some system of units, Maxwell’s equations may be written as

∇ · E = ρ, ∇ × E + Ḃ = 0 (4.22)
∇ · B = 0, ∇ × B = j + Ė. (4.23)

In terms of the scalar and vector potentials V and A we may solve the two homogeneous
equations by writing

E = −∇V − Ȧ, (4.24)

B = ∇ × A. (4.25)

– 10 –
All of this is more conveniently (and covariantly) written in terms of the 4-vector potential,
Aµ ≡ (V, A), the 4-current, j µ ≡ (ρ, j) and the antisymmetric field strength tensor, Fµν ≡
∂µ Aν − ∂ν Aµ ; indeed, Maxwell’s equations then reduce to the rather more compact form

∂µ F µν = j ν . (4.26)

This rendering makes it obvious that Maxwell’s equations are invariant (as are E and B
themselves) under the gauge transformation Aµ → Aµ + ∂µ χ, where χ is an arbitrary
function on spacetime. This ‘gauge’ is the same ‘gauge’ that appears in the title of these
lectures, so it behoves you to play close attention whenever you see the word from now on!
One way we can deal with the gauge freedom is to remove it (wholly or partially) by
gauge fixing. One common choice is the Lorenz (not Lorentz!) gauge ∂µ Aµ = 0. In this
gauge, each of the four components of the vector Aµ satisfies the Klein-Gordon equation
with m = 0, corresponding to a massless photon. We can find plane wave solutions of the
form Aµ = µ e−ip·x , with p2 = 0. Since we have fixed the gauge ∂µ Aµ = 0, we must have
that  · p = 0. Moreover, the residual gauge invariance implies that shifting the polarization
vector µ by an amount proportional to pµ gives an equivalent polarization vector. Thus,
there are only two physical degrees of polarization. These could, for example, be taken to
be purely transverse to the photon 3-momentum.5
Finally, we discuss how to couple the electromagnetic field to Klein-Gordon or Dirac
particles. The usual argument given in classical mechanics and non-relativistic QM is that
one should use the rules of minimal subtraction, replacing ∂ µ → Dµ ≡ ∂ µ + ieAµ .6 Thus,
the Klein-Gordon equation becomes

(∂ µ + ieAµ )(∂µ + ieAµ )φ + m2 φ = 0. (4.27)

It is interesting to note that, if we take a negative energy solution φ ∝ e+i(Et+p·x) with

charge +e, the complex conjugate field φ∗ ∝ e−i(Et+p·x) (which satisfies the complex con-
jugate of the Klein-Gordon equation) can be interpreted as a positive energy solution with
opposite momentum and opposite charge −e. This presages the interpretation of the neg-
ative energy solutions in terms of antiparticles in quantum field theory.
For the Dirac equation, the coupling to electromagnetism is even more interesting.
Blithely making the minimal substitution, we get

(iγ µ (∂µ + ieAµ ) − m)ψ = 0. (4.28)

Now, if we act on the left with (iγ µ (∂µ + ieAµ ) + m) we do not obtain the Klein-Gordon
equation (4.27). Instead, we find the equation (exercise – hint: use 2γ µ γ ν ≡ {γ µ , γ ν } +
5
The fact that there are two polarizations does not mean that the photon has spin one-half! In fact,
spin – which could be described as the total angular momentum of a particle in its rest frame – is not a
well-defined concept for massless particles, which do not have a rest frame. Massless particles can instead
be described by their helicity, which is defined as the angular momentum parallel to the direction of motion.
It can take just two values (±1 for the photon), leading to the two polarizations just found.
6
This is completely unmotivated. We shall , very shortly, have the means at hand to provide a satisfactory
discussion of how things should be done, but for now we beg the reader’s leniency.

– 11 –
[γ µ , γ ν ])
ie µ ν
(D2 + m2 + [γ , γ ]Fµν ) = 0, (4.29)
2
with the extra term ie µ ν i i j
2 [γ , γ ]Fµν . Now, in the Pauli-Dirac basis, 2 [γ , γ ] is given by
iijk Σk where, as we saw before, Σ2k represents the spin S k . Thus, in a magnetic field, with
Fij = ijk Bk , we get the extra term 2eS · B. This factor of 2 is crucial – if one works out
the D2 term (which is present even for a spinless particle), one will also find an interaction
between the orbital angular momentum L and B given by eL · B. Thus, Dirac’s theory
predicted that the electron spin would produce a magnetic moment a factor of two larger
that the magnetic moment due to orbital magnetic moment, as was observed in experiment.
In fact, increasing experimental precision eventually showed that the gyromagnetic ratio
of the electron is not quite two, but rather 2.0023193 . . . . In yet another heroic triumph for
theoretical physics, Schwinger showed in 1948 that this tiny discrepancy could be perfectly
accounted for by quantum field theory, to which we shortly turn.

4.5 Transition rates and scattering

Before we go further, we need to modify one more aspect of your quantum mechanics
education. QM has its hegemony in atomic physics, where one is interested in energy
spectra and so on. In particle physics, we are less interested in energy spectra. One reason
is that (as we shall see) we are not able to compute them. A more pragmatic reason is that
many of the particles in particle physics are very short-lived; we learn things about them by
doing scattering experiments, in which we collide stable particles (electrons or protons) to
form new particles, and then observe those new particles decay. The quantities of interest
(that we would like to compute using quantum field theory) are therefore things like decay
rates and cross sections. What a decay rate is should be obvious to you. A cross-section is
only a bit more complicated. Clearly, the probability for two beams of particles to scatter
depends on things like the area of the beams and their densities. The cross-section is a
derived quantity which depends only on the nature of the particles making up the beams
(and their four-momenta).
To derive formulæ for these, we start with something you should know from QM.
Fermi’s Golden rule decrees that the transition rate from state i to state f via a Hamiltonian
perturbation H 0 is given by

Γ = 2π|Tf i |2 δ(Ei − Ef ), (4.30)

where
hf |H 0 |nihn|H 0 |ii
Tf i = hf |H 0 |ii + Σn6=i + ... (4.31)
En − Ei
Let’s now try to apply this formula to the decay of a particle into n lighter particles,
a → 1 + 2 + · · · + n. There are n − 1 independent 3-momenta in the final state (momentum
must be conserved overall in the decay). Now, for states normalized such that there is one
particle per unit volume in position space, then we have one particle per h3 = (2π)3 volume

– 12 –