Professional Documents
Culture Documents
A Quantum Mechanics Tourist
A Quantum Mechanics Tourist
A Quantum Mechanics Tourist
Bijou M. Smith
Geonworld
Wellington, New Zealand
1 Introduction 1
1.1 The One and Only True Interpretation of Quantum Me-
chanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
iii
iv CONTENTS
4 Quantum Kinematics 83
4.1 The wave equations . . . . . . . . . . . . . . . . . . . . . . . 83
4.2 Probability and the Phase of the Wavefunction . . . . . . . 85
4.2.1 Complex Numbers are a Distraction . . . . . . . . . 86
4.2.2 Simple QM Preliminaries . . . . . . . . . . . . . . . 88
4.3 The Classical Action and Quantum Phase . . . . . . . . . . 93
4.4 Path Integrals and Phase . . . . . . . . . . . . . . . . . . . 94
4.4.1 Are Path Integrals Real? . . . . . . . . . . . . . . . . 96
4.5 Spin–Orbit Coupling . . . . . . . . . . . . . . . . . . . . . . 100
4.6 Probability Conservation . . . . . . . . . . . . . . . . . . . . 100
CONTENTS v
Bibliography 123
Index 125
vii
viii CONTENTS
structure.
• each chapter could be considered is it’s own isolated vignette, al-
though the whole books is a One,
• there is no attempt to take readers on a course covering topics com-
pletely. My aim is not to teach or instruct, but to free-form lay out
some ideas.
These are notes very much intended as a fairly random collection of topics
the reader could dip into and out of on a whim, very much like a tourists
scrapbook of travel highlights.
If at any time in the future I can do some serious mathematical physics
and restate many of these ideas in a coherent form, from some sort of
starting postulates to a fully fledged theory, then I hope to be able to
publish it all. At present (c.2021) that seems unlikely, because late in life
I have found activism in political economy to be afar more important for
humanity, and so the problems of elimination of poverty and justice for
the working class have over-riding importance in my life, and physics is
just a hobby I indulge in to keep myself from going mad.
Introduction
This book is a collection of loose notes, intended for “fun” reading. They
could be supplemental for an advanced undergraduate or beginning
graduate course on modern physics, with a focus on fundamental theory.
I am not going to pretend we will be preparing for practical applica-
tions. No! Most of this is all for fun, and we have no intention of ever
applying our studies in anger in the laboratory.
My philosophy, increasingly as I get older, is to do physics as an art.
This should not stop readers from joining laboratory teams to put their
learning into practical use, indeed such a motive would be highly laud-
able. But art too is a noble pursuit. The motive for art however has to
also be for the benefit of humanity in order to justify as noble in my opin-
ion, and that is the spirit of these chapters. I find this easy to justify: if I
can get good at this art of explaining physics, then it should be easier for
people to understand and apply to solve real problems, putting the art to
good practical use. There is no guarantee of this, but it is the motive.
One aspect of this fun are the diversions in algebraic geometry. For
a long time I wanted to understand in as simple terms as possible, but
no simpler, why 4-manifolds are so complicated. In Alexandru Scorpan’s
wonderfully titled The Wild World of 4-Manifolds the result is stated:
that dimension 𝑑 = 4 is the only dimension for which there exist infinitely
many distinct smooth structures on closed 𝑑-manifolds; and even more
wildly for open 4-manifolds, where for ℝ 𝑑 , 𝑑 ≠ 4 there is only one unique
smooth structure, whereas for ℝ4 there are uncountably many distinct
smooth structures! Closely related is the other unique result that only in
1
2 Introduction
I wrote this section a long time after beginning this book, it is really just
to lay out my current opinion on the matter.
If you just stare at the principles of QM you should immediately see
the biggest difference with classical mechanics is that QM is a theory of
measurement, and CM is not. All CM takes place without measurement,
the idea of us making measurements like Archimedes, Galileo, Kepler,
Faraday or Michelson and Morley, is quite incidental to the physics. Not
so with QM which cannot be written down or worked on without reference
to measurement. This is the fundamental interpretation of QM.
All the other philosophy that goes on under headings like Copenhagen,
Many Worlds, Pilot Wave, Qbism, and so forth are more in the vein of
attempting some very bad models that seek to explain QM (except I guess
Copenhagen, which is a big nothing). There is no good model of QM yet.
But the interpretation of QM is unambiguous and clear: it is the physics
we get when we want to describe measurements in the context of time
evolution of a system. The time evolution part is critical (and definitely
The One and Only True Interpretation of Quantum Mechanics 3
As with most of my writing, you should find this chapter long but gen-
tle. I aim for anti-terseness but not rigor. Anti-terseness + rigor = hun-
dreds more pages, which I am not yet prepared to go through.
In this chapter I have attempted (probably ‘begun to attempt,’ since I
might not ever finish) to give what I think is the nicest introduction to
quantum theory of particles using the concepts of preservation of space-
time symmetries. This is what I regard to be a prelude to geon theory.
Geon theory seeks to construct geometrical and topological models for the
fundamental particles or ‘quanta’ that are observed in nature, at least
at low energy. This chapter is more primitive — meaning more abstract,
less exact — and only enquires into the abstract group structure of geons.
We will ignore the topological structure which must exist at around the
Planck scale. We really have to ignore this topological structure because
it is not yet well understood. However, this higher level of abstraction
has proven to be adequate for almost all quantum theory so far, other
than extreme conditions that are rarely observed in nature (black hole
interiors and big bangs).
Considering that we will not be seeking topological models, we only
want the bare algebra for the fundamental particles. The organizing
5
6 Particles and Symmetries
principle is that introduced by Wigner [2]. This is the idea that elemen-
tary quanta — whatever else they are — should at a minimum be de-
scribed by symmetries of spacetime. I will expand more philosophically
on Wigner’s ideas in the introduction, before then going into the theory
proper.
Background optional reading:
In all sciences we seek invariants in nature, that is the basic task — in-
variants are what we generally call laws of nature. And for fundamental
theoretical physics we seek the most primitive invariants. Physically we
can identify invariants as properties that do not change unless there is
significant interaction. In our mathematical language we have a highly
formal structure for such ideas known as group theory. To explore the
fundamental invariants of physics we thus attempt a group theoretic de-
scription as a first type of foundational theory.
Thus when Wigner asks for the elementary quanta to be described by
“irreducible unitary representations of the Poincaré group,” he was asking
for a minimal way to describe the elementary particles. It is not asking
for a complete description. Wigner’s scheme is interested only in the basic
symmetry properties of the particles and the consequent Lie algebra that
they obey. This is not asking for much! But it turns out to get us a long
way towards classifying most of the elementary particles.
Introduction to particles and invariants 7
A few questions for students (answers are given but I encourage you
to think before reading):
1. Why the Poincaré group? Answer: because the only known sym-
metries of flat spacetime are the Lorentz symmetries of special rel-
ativity.
2. Why irreducible representations? Answer: because we want
fundamental particles. In group theory a reducible representation
is in a plain sense more structure, and an irreducible representa-
tion is plainly minimal structure. We can build physics models of
composite or non-elementary particles (indeed nuclear physicists
have to) but for basic quantum theory we explore only the elemental
quanta, and call these the “elementary particles,” and define these
to be the observed quanta that can be described by the irreducible
representations of the spacetime symmetries.
3. Why unitary representations? Answer: for quantum mechanics
we require conservation laws to hold, the most basic of which is
conservation of probability. In both the Hilbert space formalism
and wave-function formalism this means unitary operators. If the
time evolution is non-unitary then we get probabilities that do not
add up to 1.0. That means we would have an incomplete theory.
If the probabilities added up to more than unity then we’d have a
non-sensible theory (and we would just impose some normalization
rule in that case, so it would end up being unitary anyway).
theory extends Wigner’s scheme. But note this also means on the scale
of the topological geons spacetime is again not flat, but that is regarded
as close to the Planck scale, which is again far too small for us to detect
presently and so mostly irrelevant for applications of fundamental the-
ory. The problem is this length scale is enormously important for theory
of fundamental theory!
Hence with regard to question 1, why Poincaré? The full answer is that
we need extra local symmetries for fully describing spacetime topology
on the scale of the fundamental strings or geons. But provided we are
doing physics asymptotically far enough from the Planck scale then the
symmetries of spacetime are the main ones to care about.
old atomic theory: the atomic elements are not fundamental, but the lep-
tons, quarks and gauge bosons just might be fundamental, and so we’d
philosophically expect them to be like prime building blocks for matter.
As primes they cannot be factorized. That’s a metaphor with prime num-
bers in mathematics, but is nothing more than a metaphor. Perhaps the
deeper principle might be found in geon theory; there is a rough idea that
a geon is so fundamental that there is no way to rip up a geon without
tearing up or disconnecting spacetime. To my mind this would be a deep
enough principle that I would find thoroughly satisfying. You could ask,
“Why not allow spacetime to get ripped?” Sure, why not, but then I think
we are in the realm of scifi — where’s the evidence spacetime can be torn
apart? It is certainly not inside black holes which do the opposite, they
are extreme glue.
It is worth also noting that tearing a rip in spacetime that disconnects
a geon is precisely a loss of unitarity, because a quibit representing a geon
cannot just vanish without violating local unitary time evolution. This
was a puzzle in the early days of quantum gravity where baby universes
were thought to be allowed by topological fluctuation of the metric. In or-
thodox canonical Euclidean quantum gravity this is still an unresolved
problem. In geon theory there is no such problem because spacetime can-
not get disconnected using the mechanisms of metric fluctuations, be-
cause in geon theory there are no “quantum fluctuations” so-to-speak,
rather the metric merely appears to fluctuate due to the closed timelike
curve dynamical effects.
3. Later we will add scattering theory, which is a bit sketchy, but intu-
itively appealing, and with scattering theory we will be able to see
why electromagnetism, the other Yang-Mills forces, and gravity are
the only possible forces in Nature. None others can exist. (Assum-
ing scattering theory is well-founded.) I find this to be one of the
most remarkable discoveries in physics of all time. When we get to
this stage we will have finished, for my present purposes.
12 Particles and Symmetries
The story of particle physics continues beyond our story however, because
further symmetries have been discovered in the atomic nuclei, namely
the color symmetry governing the strong nuclear force, and the flavour
symmetries governing the electroweak force. These are all Yang–Mills
forces. Essentially the same methods can be used as for the Poincaré
algebra, but using so-called internal symmetry groups.
I would ask all students what can possibly be the source of the sym-
metries of these internal symmetry groups? It has to be physical phe-
nomenology, unless you believe in magic. No one has ever offered a phys-
ical grounding for these internal symmetries other than fine structure of
spacetime geometry, such as the Calibi-Yau compactifications of String
theory, or fictional constructs like gauge field fibre bundles.
The problem with gauge field theory here is that if you stick a lit-
eral, not mathematical, fibre bundle at each point in spacetime then you
have added geometry. The mathematical properties of those fibre bundles
are describing something. What are they describing? No one has ever
proffered a convincing answer different to string theory or some Kaluza–
Klein theory.
We would use the color symmetry, SU(3), to find physical invariants
and with the index labels for the new irreducible representations for the
SU(3) group we can label further elementary particles. Remembering,
when we do this, that the particles are not the labels. The particles are
not even the Hilbert space objects that have these labels. The labels are
just abstract symbols identifying the particles by discrimination firstly,
and secondly by how the particle transforms upon Poincaré or now color
symmetry transform operations. We are not identifying particles by any
ontology. We have no idea what the quanta particles are, the closest
we can say (at the present date) is they might be fundamental strings
and branes. We will not carry out this quantum electroweak or quan-
tum chromodynamics procedure because although it adds complexity and
richness, yet without topological models that go beyond the group theo-
retic model we can find no further insight into the nature of fundamental
particles.
For sure, in geometric algebra and geon theory context the electroweak
and color symmetries might eventually help identify some underlying
topology, but so far no one has come up with a good scheme other than
some basic algebraic ideals [7–17]. But those explorations are pure alge-
Introduction to particles and invariants 13
bra, they have not yielded insight into the spacetime topology of putative
geons. It might be easy once someone gets the basic idea, but no one
seems to have found a good principle yet. String theory is still the only
game in town.
Steven Weinberg gave the first account of the constraints on the types
of fundamental particles based on the orthochronous inhomogeneous
Lorentz group. Those are fancy words for ignoring the discontinuous
symmetries of space inversion ( 𝑃 -symmetry) and time reversal invari-
ance (𝑇 -symmetry) of the full Poincaré group.
It is a famous result that Nature does not actually respect those 𝑃
and 𝑇 symmetries anyway, so it was right to ignore them, however na-
ture does seem to respect 𝐶𝑃𝑇 -invariances, which means space inversion
combined with time reversal and charge conjugation. If all three are si-
multaneously applied then the physics is invariant. We are not going to
discuss 𝐶𝑃𝑇 in this chapter, but if you want a look ahead then chapter 3.3
of Weinberg Vol-I is a place to look.2
Also, as a teaser, a thread that I will leave hanging, doesn’t the fact
the combined 𝐶𝑃𝑇 symmetry get respected in nature tell you that electric
charge is probably not very mysterious, but might have some simple ge-
ometrical interpretation? It does to me, and I can guess what it is; since
charge is discrete, quantized, there is no doubt in my mind that electric
charge has something to do with topological indexes in gnarly wormhole
spacetime topology on the Planck scale. It cannot be too gnarly though,
since we only see three types of charge, ±1, ±1/3 and ±2/3. That’s where I
will leave this thread.
To dig in to Weinberg’s results we first need to wade through some
lengthy but fairly lightweight group theory. In the next few sections we
will go through the following pedagogic sequence:
By the last bullet point we will be done with studying the Poincaré sym-
metries and will have representations for both classical and quantum
physics for a huge class of “fundamental particles,” not all of which are
seen in nature. We will then have to go further into scattering theory
basics to see how to further reduce the number of physical elementary
particles down to spins ≤ 2.
Unfortunately I am not aware of any way to obtain the scattering the-
ory constraints without some quantum field theory formalism. We will
need to know a very small basic bit of information about the poles for the
scattering amplitudes. It does not require much QFT, just a little. Once
we have that then we will be able to show why we can get spins up to a
graviton (spin= 2) but no higher. This will tell us why in Nature we see
only gravity and the Yang–Mills forces, no other fundamental forces. (All
other forces arising from particle exchange will be derivative of Yang–
Mills, for example Pauli overlap forces in fermions, van der Waals forces,
and all the rest).
the actual physical space is a ray space in the Hilbert space, where all
vectors with the same “direction”3 are equivalent physical states, so the
physical space is denoted, H /∼, which just means the Hilbert space mod-
ulo equivalence by normalization. This is also called Projective Hilbert
space, sometimes denoted 𝑃 (H ).
I should make a philosophical note here: these physical states are not
complete models for physical reality, they are rather minimal information
states. They suffice to carry all the information we need to make prob-
abilistic statements about measurement outcomes on the system. The
way we describe the information propagation in time is not equivalent
to the actual physical stuff out there. Need an analogy? How about; the
Fourier transform of a radio message is not the actual words of the mes-
sage. Something a bit like that is what a choice of Hilbert space means,
it means we are choosing convenience over exact reality. We lose some-
thing physical when we choose to do quantum mechanics with a choice
of Hilbert space representation of systems. We lose forever the ability to
precisely trace fundamental particles.
To quantum theorists it is a moot point whether or not there is some
way to retrieve determinism, the fact is Nature herself tells us determin-
ism is impossible unless you can time travel and find out the result of a
measurement before starting the measurement. If you could time travel
like that then yes, you would have access to information that would mean
you do not need a Hilbert space representation, and you could essentially
do classical mechanics, albeit in some new formalism that takes into ac-
count time travel. That would be an action principle of some sort aug-
mented by data from the future. Good luck with that.
However, the fact time travel can be considered a metaphysical possi-
bility means the Hilbert space representation is incomplete, it is not de-
scribing physical stuff as physics is, it is a formalism that acknowledges
in practice we (humans) do not have time travel. If some other sentient
species in some other galaxy had time travel they’d (a) be like gods, and
(b) have no need for quantum mechanics, unless they were lazy and time
travel was say super costly.
3 The
term “direction” does not have the same meaning for a Hilbert space over a
complex scalar structure as it does for Euclidean vectors over the reals. All vectors
that can be got from one to another by multiplication by a complex number (or a rotor,
in geometric algebra language) are taken to be “in the same direction,” but this more
technically just means “in the same ray.”
Group theory preliminaries 17
Before proceeding with the group theory stuff, one other notation I will
use is the Dirac ket symbol for vectors in the Hilbert space, and when
I want to indicate a ray (an equivalence class of vectors) I will use an
underbar, so,
|𝜓i is a vector in H
|𝜓i is a ray in H /∼ in the equiv. class [𝜓] .
|𝑟i will be a ray, so is the same as | 𝑟 i .
𝐴ˆ |𝜓i = 𝑎 |𝜓i .
where h·|·i is the inner product on the Hilbert space, and we’ve
taken the squared norm.
18 Particles and Symmetries
This is the first clause of the postulate. Please note the notation
brevity: 𝑎𝑘 is a scalar number, | 𝑎𝑘 i is a Hilbert space vector (a pos-
sible state of the system, up to knowable information). The second
clause in the measurement postulate is that after actually perform-
ing such a measurement the system will be in the state | 𝑎𝑘 i, not in
the initial state |𝜓i.
Comment. Planck’s constant ℏ does not appear. We will see why later,
but the basic reason is that ℏ is just a unit conversion factor. All the
“quantumness” in QM comes from (i) superposition, and (ii) a further
postulate we have not yet written, which is the particle-postulate:
* * *
We use ordinary old linear algebra (on the Hilbert space, since it is a
vector space) to get all the usual theorems such as existence of complete
basis states, possibility of orthonormalization and so forth. Do you re-
member eigenvectors of a Hermitian operator form a basis? No matter
if you forgot, you remember now. I will not go through all that since you
can find it in any standard textbook, like Sakurai, or Ballentine.
We will use Wigner’s theorem to help classify the possible fundamen-
tal particles in QM. For this we need a definition of a symmetry trans-
formation.
Definition 2.1 (Symmetry transform). An operator 𝑇 on the Hilbert
space represents a physical symmetry of the QM system if the inner
product is preserved as follows,
h𝑇 𝜙|𝑇 𝜓i = h𝜙|𝜓i
Group theory preliminaries 21
Notice that I am using the ray representation here. Since unit scalars
disappear when we take a norm, up to scalar multiples a ray represents
a unique state, since only the transition probabilities and measurement
outcomes are observable. By this definition a symmetry transform 𝑇 will
preserve transition probabilities,
𝑇 : 𝜓 → 𝜓 0,
⇒ 𝑃 (𝜓1 → 𝜓2 ) = 𝑃 (𝜓10 → 𝜓20 ) .
Critical to finding unitary irreps. will be the result that unitary repre-
sentations can in fact be found. Without this theorem we might just get
lucky and find the Poincaré group does have a unitary representation,
but it is nice for the mathematicians to tell us one will exist!
𝑇 𝜓 = 𝑈ˆ 𝜓
𝑇 : H /∼ ↦→ H /∼
then,
with “a bit more work” a good mathematician can show with these prop-
erties a representation of the symmetry transform can be found which
satisfies,
(1) if a state |𝜓i is in the ray | 𝑟 i before the transformation, then after
the transformation the transformed state is in the transformed ray,
that is,
if |𝜓i ∈ | 𝑟 i then 𝑈ˆ |𝜓i ∈ | 𝑟0i
and,
(2) 𝑈ˆ † = 𝑈ˆ −1 .
The first requirements says 𝑈ˆ represents the symmetry, and the second
says it is (anti)unitary.
If you are feeling strong then have a go at proving this theorem. You
will need some more background in group theory and linear algebra than
I am prepared to provide, but you will not need too much. Also consider
just looking up the proof. This is the ethical hackers method for math-
ematical physics. You never write all your code from scratch, you read
clean code others have written, and only if you can follow it should you
copy it, but if you can follow it and it seems clean then you should copy
it and re-use.
Group theory preliminaries 23
* * *
What of the theorem itself? As part of the Theorem 2.2 we saw for
every physical symmetry transform of a QM system there is an associated
symmetry group. The theorem tells us for every element of the group
there is an (anti)unitary operator which represents that element (which
means that operator performs the transformation on the physical states,
just as the group element performs the precisely analogous transform on
the abstract group).
It does not tell us how to find a unitary representation! We did not
really need the mathematicians to tell us this theorem either, but it is
still nice to know, because it means we can use abstract group theory.
If unitarity was not guaranteed, then we would be stuck needing to find
a different set of algebraic tools to work out how to model fundamental
particles. What is nice about Theorem 2.2 is that it says group theory
will be useful to physics. It is a “lucky” break, because group theory is so
primitive — this means we can we can go a long way with it in theoretical
physics, for very little in we get a lot out. I hate to imagine what particle
physics would be like without applicability of group theory. It’d be like
zoology I suppose. We could do it as science, but it would not be pleasant.
irreps. for the Lorentz group. Once we have done that we will have
completed our objective, which is just to show that one suitable repre-
sentation of elementary particles in QM is in terms of momentum and
spin eigenstates, because the mass squared 𝑚2 and the spin will be
invariants, and moreover the spin numbers, 𝑗 , will be shown to be whole
number multiples of half-integers. These spin indices are labels for the
particles, they are not quantized units of rotational motion, this is im-
portant to keep in mind, because there is nothing inherently “quantum
mechanical” about what we are going to do. It is unusual to select a
Hilbert space for our states, but we do not get quantum mechanics out
of this, we could formulate classical mechanics instead as also a set of
states in a Hilbert space. The quantum mechanical things occur when
we include the postulates of superposition and entanglement.
It will be useful just to know how the connected covers of the discon-
nected Lie groups arise. This is the following First Homomorphism The-
orem. First a definition:
Definition 2.3 (kernel of a map). The kernel of a map 𝑓 is the set of ele-
ments in the domain that get mapped onto the identity in the codomain,
so for,
𝑓 : 𝑋 → 𝑌,
kern( 𝑓 ) = { 𝑥 ∈ 𝑋 | 𝑓 ( 𝑥) = 𝟙𝑌 }
𝑓 : 𝐺1 → 𝐺2
𝐺1 /kern( 𝑓 ) 𝐺2 .
most of which all follows just from using the fact 𝑓 is a homomorphism
(so also maps identity in 𝐺1 to the identity in 𝐺2 ). But you know any
homomorphism that is bijective will be an isomorphism, and we are given
𝑓 is surjective, so just show when you mod 𝑓 out by the kernel you get an
injection. To finish up then we need to show if [ 𝑎] ↦→ 𝑦, and [ 𝑏] ↦→ 𝑦, then
[ 𝑎] = [ 𝑏]. If 𝑎 ∈ kern( 𝑓 ) then 𝑦 = 𝟙 by definition, so 𝑏 ∈ kern( 𝑓 ) as well,
so [ 𝑎] = [ 𝑏]. Otherwise 𝑎 and 𝑏 will be either in the same or different
cosets of 𝐺1 by kern( 𝑓 ). (Remember, the equivalence classes here are the
cosets.) For ℎ ∈ kern( 𝑓 ), if 𝑎ℎ and 𝑏ℎ map to the same element of 𝐺2
they must be in the same coset, hence [ 𝑎] = [ 𝑏] again. So we’ve got an
isomorphism. □-ish.
Some background reading is necessary which you can get from almost
any textbook on Lie groups. The basic stuff we need is to know that a
Lie algebra is defined by the commutation relations for group generators
connected to the identity. That means if we are to use the simpler tool
of Lie algebra for our physics we need to know whether our Lie group
is connected or not, and if it is not connected then we need to find the
component that is connected to the identity. Once we know this, then
exponentiation of the Lie algebra generators will give us group elements
and group representations, but not only that — crucially — irreducible
representations.
Got it? Irreducible representations (irreps. for shorthand) can be
found from the Lie algebra generators, by exponentiation. We will start
with unitary groups, so unitarity will be taken care of, provided we study
physical symmetries as per Theorem 2.2.
26 Particles and Symmetries
Definition 2.5 (Connected Lie group). A connected Lie group can be de-
fined as follows:
(i) the group elements are parameterized by a set of real numbers, say
{𝜃 1 , . . . , 𝜃 𝑛 }, where 𝑛 is the dimension of the group. Abstractly then
the group is,
𝐺 = {𝑇 (𝜃 𝑎 )}, 𝑎 ∈ {1, . . . , 𝑛} .
The third property is critical for us: because we seek unitary representa-
tions (for quantum mechanics probability conservation) we must seek a
connected Lie group, not just any old Lie group for the Lorentz group.
The last few properties mean that the group will not have any anti-
unitary or anti-linear operators, since path-connectivity means all the
elements are smoothly connected to the unitary 𝟙𝐺 .
Note that we really should have a vector symbol on the 𝜃 since they
live in ℝ𝑛 , but in most physics courses the vector symbol is reserved for
Euclidean ℝ3 and so because we do not want to miss-prime the brain
we will just assume the Lie group parameters are understood to be 𝑛-
dimensional reals.
The unitary operators are linear so can be always written as matrices
if desired, we then get by definition a representation, because they satisfy,
𝑈ˆ (𝜃 1 ) ◦ 𝑈ˆ (𝜃 2 ) = 𝑈ˆ 𝑓 (𝜃 1 , 𝜃 2 ) .
𝑎 ∧𝑏
The last result is of course a nice linearization. I’ve always liked this def-
inition more than the Taylor series, since it shows us the iterated map-
ping aspect of all exponentials. It’s a very physical sort of definition even
though it is pure mathematics (due to the ∞ symbol).
The entire set of possible 𝑔 mapped onto by 𝑒 𝑋 is of course only the
component of 𝐺 connected to the identity. But since we can compute ir-
reducible representations from this, they will be irreps. for the parent
Lie group too. Since in QM we are only interested in the unitary reps. —
because we want conservation of probability — it is only the component
connected to the identity that matters for us. So that’s perfect.
Now for the formal definitions. First, a Lie algebra is a vector space.
This means it has a multiplication operation and an addition operation,
which form a closed group, and these are linear. (Linearity is another
reason why the Lie algebra is easier to work with than the parent Lie
30 Particles and Symmetries
End of definition.
We then can define a representation of 𝔤 for any Lie group 𝐺. It closely
follows how group representations were defined; by maps into the set of
invertible matrices. For quantum physics we will use a particular type
of representation, which is not the most general, which is matrices that
are operators on the Hilbert space of physical states. This is just for our
convenience so that our Lie algebra work is not too abstract.
Definition 2.7 (Lie algebra representation). A representation of 𝔤 is a
mapping,
𝑋ˆ : 𝔤 −→ GL(H )
which preserves the structure of the Lie algebra. So, for instance, for
every basis vector 𝑥1 in the algebra 𝔤, there is a mapping 𝑋ˆ 1 taking 𝑥1 to
some matrix in GL(H ).
Just as for groups, this mapping need not be bijective, if not then it is
not a faithful representation, but is still a representation.
In practice, to compute a representation we almost always use the defi-
nition that the Lie bracket operation is just the commutator product, and
then provided the maps 𝑋ˆ preserve the structure we will get a represen-
tation, thus for all 𝑣1 , 𝑣2 ∈ 𝔤;
Group theory preliminaries 31
(i) [ 𝑋1 , 𝑋2 ] ≡ 𝑋1 𝑋2 − 𝑋2 𝑋1 ,
(ii) [ 𝑋 ( 𝑣1 ) , 𝑋 ( 𝑣2 )] = 𝑋 [ 𝑣1 , 𝑣2 ] ,
whereas in Spin+ (1, 3) you have no need for that crutch, and in fact the
algebra over the reals is far more natural and intuitive, and you do not
lose any geometric insight by choosing the “correct” (or more natural)
group. The “correct” group is nearly always the one with an algebra over
the reals. Why is that? It is because whenever you employ the complex
numbers you are doing so for some geometric reason, so why not just use
the proper geometry instead? Sometimes the formula looks a bit shorter
on the line when you use the complex numbers, but that is often because
you are obscuring the full graded algebra, i.e., the full geometry. Which
do you prefer, compacter notation or full insight? Your answer to that
characterizes you as a type of physicist, the Julian Schwinger/Chen-Ning
Yang type (loves elegant formalism even when it is wrong) or the Gell-
mann/Feynman type (elegance is fine when you can get it, but being right
is overriding).
We will look at the isomorphism SL(2, ℂ) Spin+ (1, 3) later, if I have
the time and inclination. I might just skip it and go straight to using the
more natural Spin+ (1, 3) with the STA rotor representation.
SO(3) is the Lie group for rotation symmetries in ℝ3 , but abstractly in
its own right it has its own group theoretic properties that do not depend
upon physical space. Some properties;
• It is a smooth manifold.
To illustrate what this manifold looks like we need to consider some ba-
sic topology. This will lead us to see SO(3) is not simply connected, which
means there can be a simpler group from which irreducible representa-
tions of the rotations will be found.
Then in searching for a simply connected group that covers all rota-
tions we will discover SU(2). Although in these lectures I basically did
the reverse, and go to Spin(3) + first as the more natural group for rota-
tions
Group theory preliminaries 33
where 𝟙3 is the unit 3 × 3 matrix., and the reps. have unit determinant,
det 𝑅 = +1.
In the geometric algebra 𝐶ℓ (3) or G(3) picture, the reps. are rotors,
− 𝐵𝜃/2
𝑅=𝑒
for any unit bivector 𝐵 defining the plane of rotation, which have the
property,
𝑅 · 𝑅 = 1, and 𝑅𝑅˜ = +1.
𝐴0 = 𝑅𝐴 𝑅.
˜
𝑅𝑣 = 𝑣 𝑅˜
34 Particles and Symmetries
which means we can move the right rotor over to the left, so,
𝑣0 = 𝑅2 𝑣
𝑣 → 𝑣0 = (cos 𝜃 − 𝐵 sin 𝜃) 𝑣
Test driving this out with 𝑧 axis rotations means putting 𝐵 = 𝑒1 𝑒2 , and
expanding 𝑣 = 𝑣𝑖 𝑒𝑖 . The rest follows. You can reconstruct any rotation
matrix this way by acting on column vectors 𝑒 𝑖 , for 𝑖 = 1, 2, 3.
The other advantage of the rotors/spinors is that it is dead easy to
get the commutation relations for the generators of the Lie algebra from
them, and without having to insert a stupid unit imaginary to concoct
anti-Hermitian generators.
Physically SO(3) is the set of transformations that preserve lengths,
angles, and orientations, because distance is preserved it also preserves
volume. Whereas O(3) is the orthogonal group in 3 𝐷, the group which
preserves only the norm ||𝑇 ( 𝑥)|| = || 𝑥 ||, it does not preserve orientations,
e.g., with a single reflection a right-handed coordinate system { 𝑥, 𝑦, 𝑧 }
flips into a left-handed system { 𝑥0 = 𝑥, 𝑦0 = 𝑦, 𝑧0 = − 𝑧 } for one example.
SU(2) is the group of 2 × 2 unitary (Hermitian) matrices with unit
determinant. We will see SU(2) and SO(3) have the same Lie algebra, but
in searching for irreducible representations SU(2) is favoured because it
has a simply connected topology.
Now the cool thing here, is that we’ve got the rotors, and they form
a representation of SU(2). Most textbooks start with SO(3), we can by-
pass all that. If you want to study the generators for SO(3) and get the
commutation relations, be my guest, but why would you bother? When
you understand the geometric algebra 𝐶ℓ (3) you see Spin(3) is really the
“correct” group for rotations, which has a nice isomorphism with SU(2)
via Pauli spinors. Hence the first thing I want to do is show you how
Group theory preliminaries 35
to get the generators and commutation relations for Spin+ (3) or equiva-
lently SU(2). We will then work backwards to most textbooks and just
for fun show SO(3) has the same Lie algebra.
The formality is the abstract definition of SU(2), which is the group
represented by 2 × 2 Hermitian matrices with unit determinant:
𝑈 = 𝑎𝜎
ˆ 0 + 𝑖𝑏𝜎
ˆ 1 + 𝑖𝑏𝜎
ˆ 2 + 𝑖𝑐𝜎
ˆ3
the unit imaginary 𝑖 pre-multiplies because the Pauli matrices are all
anti-Hermitian. The bijective correspondence with the geometric algebra
is that with 𝐶ℓ (3) + Spin+ (3), which is the even sub-algebra of 𝐶ℓ (3),
these are the rotors, which are the closed algebra of the set,
{1, 𝜎𝑖 𝑗 }
Proof sketch. This can be seen from the fact every rotation is a compo-
sition of two reflections, and in any and all G 𝐴s the reflections are dead
easy to compute, you pick a unit normal vector to the plane of reflection,
say 𝑛 and to reflect a vector 𝑣 just do,
𝑣0 = −𝑛𝑣𝑛
This means every rotation is effected by a product of the form,
𝑣0 = 𝑅𝑣 𝑅˜
where 𝑅 = 𝑛𝑚 is a rotor, and 𝑅˜ = 𝑚𝑛 is it’s reverse. Since the geometric
product of two vectors is,
𝑛𝑚 = 𝑛 · 𝑚 + 𝑛 ∧ 𝑚
it will have scalar and bivector parts, hence is a rotor. The obstruction to
rotating bivectors and higher grade elements of the algebra is failure of
rotors to commute with them, the double-sided rotor transform takes care
of this. The rotor operation above is grade preserving since det 𝑅 = 1. It
thus takes 𝑟 -vectors to 𝑟 -vectors, is continuous under continuous change
in 𝜃, preserves the origin, preserves orientation and lengths and grade
preserves, for any multivector. We offer just an explicit demo for bivetors:
first note,
𝑎 ∧ 𝑏 = 12 ( 𝑎𝑏 − 𝑏𝑎)
with this,
( 𝑎 ∧ 𝑏) 0 = 𝑎0 ∧ 𝑏0 = 21 ( 𝑅𝑎 𝑅𝑅𝑏
˜ 𝑅˜ − 𝑅𝑏 𝑅𝑅𝑎 ˜ 𝑅˜ )
= 1 ( 𝑅𝑎𝑏 𝑅˜ − 𝑅𝑏𝑎 𝑅˜ )
2
˜
= 𝑅 ( 𝑎 ∧ 𝑏) 𝑅.
This concludes the proof. □
With this theorem under our belt, we can go straight on to finding the
infinitesimal generators of the Lie algebra 𝔰𝔲(2).
It is now time to take a little excursion in topology to see what the non-
simple connectivity of SO(3) means. What it ultimately means is that
there is a redundancy is the representations for rotations, meaning one
can get a projective representation rather than an ordinary representa-
tion. If we find a different simply connected Lie group with the same Lie
algebra, then it will have no projective representation, so will be better
for finding eigenvalues and hence labels for the quantum states.
First then, why is SO(3) non-simply connected? In naïve topology this
means the manifold has a “hole” in it, so some looped paths cannot be
shrunk (smoothly deformed) to the identity. But the manifold 𝑃 3 (ℝ (the
3-ball with antipodal points on the surface identified) does not appear to
have any obvious holes. Right. But in topology a “hole” can be a removal
of a single point. Still, it is not obvious 𝑃 3 (ℝ is a smooth structure with
a point removed.
The group of rotors, 𝐶ℓ + (3), restricted to the unit rotors makes the spin
group Spin(3). I think I have alreadt covered this isomorphism. You just
identify the Pauli spinors with the 𝐶ℓ + (3), but with a factor of 𝐼 = 𝝈1 𝝈2 𝝈3 ,
𝜎
ˆ 𝑖 ↔ 𝐼 𝝈𝑖
and the unit matrix goes over to the unit scalar. The old white due’s view
of rotations is SU(2), the vibrant young purist’s view is Spin(3).
Earlier I promised to show that the traditionally old white man Lorentz
group SL(2, ℂ) should really be thought of at first by innocent babies
straight out of the womb as the Spin group. Why? Because the Spin
group has a representation in the spacetime algebra (STA). Elements of
Spin+ (1, 3) are,
𝑅 = ± 𝑒 /2
−𝐵
38 Particles and Symmetries
for bivectors 𝐵. The bivectors 𝛾0𝛾𝑖 give the space-time rotations (Lorentz
boosts).
−𝛼𝛾0 𝛾3/2
𝑒
gives a boost in the 𝑧, i.e., the 𝛾3 direction, with rapidity 𝛼 given by,
tanh 𝛼 = 𝛽 .
Recall that the Symmetry Representation Theorem 2.2 tells us for ev-
ery element of the symmetry transforms on a QM system there is an
(anti)unitary operator which represents that element (which means
that operator performs the transformation on the physical states, just
as that group element performs the precisely analogous transform on
the abstract group). For the continuous (Lie) symmetries of special
relativity (the Poincaré group) we then want to restrict to connected Lie
(sub)groups, because these will have the identity and so all elements
of these subgroups will be unitary. This will ensure we find unitary
representations, as per our guiding philosophy from §2.1.
Recall that the reason we are not going to be able to use the Lie algebra
to find unitary irreps. with a disconnected group like SO(1, 3) + is because
such disconnected groups have elements that cannot be obtained from
exponentiating the Lie algebra generators. Only a connected Lie group
can be completely obtained from the Lie algebra alone.
We thus find ourselves needing to find a universal covering group,
which turns out to be SU(2), which was guaranteed to exist by the Sym-
metry Representation Theorem.
The general 2 × 2 Hermitian matrix is
𝑎0 − 𝑎3 − 𝑎1 + 𝑖𝑎2
− 𝑎1 − 𝑖𝑎2 𝑎0 + 𝑎3
which acts natural on Pauli spinors. This same matrix has a represen-
tation in the Clifford algebra 𝐶ℓ (3),
®𝑖
𝑎0 + 𝑎 𝑖 𝜎
® 𝑖2 = 𝟙,
𝜎 ® 1 𝜎®2 = 𝐼 𝜎3
𝜎 and cyclic,
40 Particles and Symmetries
𝜎® 1 𝜎®2 𝜎3 ≡ 𝐼
{𝜎𝑖 , 𝜎 𝑗 } = 0, for 𝑖 ≠ 𝑗.
That is exactly the Clifford algebra 𝐶ℓ (3). So by using Pauli spinors we
have not in fact gone into “quantum mechanics” we are still in classical
Euclidean coordinate geometry.
The same applies for the Dirac algebra which is the classical geomet-
ric algebra of 𝐶ℓ (3, 1), that is, the classical spacetime algebra. Things
only start to get “quantum” when we introduce discrete energy packets,
together with their superposition and entanglement. It is not sufficient
to just have wave packets with energy 𝐸 = ℏ𝜔, to get actual QM we must
have superposition and entanglement as well.
However, there is one subtlety of importance to physicists in these
representations, which is that in QM we only need the representation
property to hold up to a phase factor 𝑒 𝐼 𝜙 . Recall that the group 𝐺, we are
concerned with, is associated with some physical symmetry or invariance
transform 𝑇 , we wrote this,
𝐺 = {𝑇 : H /∼ → H /∼ such that 𝑃 (| 𝑟 i → | 𝑟 𝑛 i) = 𝑃 (| 𝑟0i → | 𝑟0𝑛 i)} .
This slightly indirect definition of the group from the transform 𝑇 is nec-
essary because the physical states in H are represented by rays |𝜓i. For
instance, suppose we have some symmetry group 𝑇𝑖 , with composition
law
𝑇1 ◦ 𝑇2 = 𝑇12
then an ordinary representation will have,
𝑈ˆ (𝑇1 )𝑈ˆ (𝑇2 ) = 𝑈ˆ (𝑇12 )
but then since the physical states are equivalent only up to their rays we
can have a looser requirement on our representations, namely,
𝑈ˆ (𝑇12 )|𝜓i ∈ | 𝑟00i
∴ 𝑈ˆ (𝑇2 )𝑈ˆ (𝑇1 ) = 𝑒 𝐼 𝜙(𝑇1 ,𝑇2 ) 𝑈ˆ (𝑇12 )
Lie algebra
First, lets cover soe motivatins. The reason we want to find the Lie al-
gebra for our physcial symmeteries is in order to be able to construct a
projective representaiton.
Due to analyticity of 𝑓 for our supposed connected Lie group, we can Tay-
lor expand about the identity, once for 𝑓 and also for 𝑈ˆ ,
𝑓 𝑎 (𝜃 1 , 𝜃 2 ) = 𝜃 1𝑎 + 𝜃 2𝑎 + 𝐶𝑏𝑐
𝑎 𝑏 𝑐
𝜃1𝜃2 + . . . (2.3a)
1 𝑎 𝑏ˆ
𝑈ˆ (𝜃) = 𝟙 + 𝑖𝜃 𝑎 𝑋ˆ 𝑎 + 𝜃 𝜃 𝑋𝑏𝑐 + . . . . (2.3b)
2 1 2
If we plug Eq.(2.3a) and Eq.(2.3b) and into (2.4), then equate terms (Tay-
lor series are equal term-by-term) we find the following relations,
𝑎 ˆ
[ 𝑋ˆ 𝑏 , 𝑋ˆ 𝑐 ] = 𝑖 𝑓𝑏𝑐 𝑋𝑎 ,
𝑎 𝑎 𝑎
where 𝑓 𝑏𝑐 = 𝐶𝑐𝑏 − 𝐶𝑏𝑐 .
When it comes time to do the Lorentz group (space rotations plus boosts)
we will see the boosts are just geometric rotations but with invariant
spacetime bivectors rather than invariant space bivectors. A beautiful
42 Particles and Symmetries
unifying bit of spacetime algebra. All the proper Lorentz group operators
are all just rotors!
This will mean we can drop the unit imaginary 𝑖 in the above expan-
sion of 𝑈ˆ , and in fact the operator 𝑈ˆ will be a simple G 𝐴 rotor, a geometric
object. This is also beautiful. This means we get Hermiticity automati-
cally, there is no need to plug in an 𝑖 like a kludge. With a rotor,
− 𝐵𝜃/2
𝑅=𝑒
the object 𝐵 is a unit bivector defining the plane of rotation, and this
works for space rotations as will as Lorentz boosts, beautifully. All unit
bivectors square to negative one,
𝐵2 = −1
and that will give us the replacement for the unit imaginary in the Lie
algebra generator expansions.
I am not sure if any of this is of much concern. The complex numbers
enter into QM in another way, which we will have to later on deal with in
another chapter. In this chapter the imaginary numbers disappear, since
the invariants are real numbers, which is only the correct physical thing
to get.
The complex numbers can re-appear if we choose a matrix represen-
tation for the operators 𝑈ˆ , however, little do most students know, we
can even get rid of these, since every complex valued matrix representa-
tion can be replaced by a higher dimensional real matrix representation.
That’s a pain if you have to write out a lot of matrices. So perhaps that
is why most textbooks go for the lower dimensional complex matrix rep-
resentation when they get down to computing wave functions. We will
indirectly show that is not needed, and the lowest dimension can be rep-
resented in the reals, using the bivector algebra of space/spacetime.
Although it is not the topic for this chapter, in case you were wonder-
ing, the complex numbers get squeezed back into quantum mechanics
because the probability amplitudes have to exhibit interference, and real-
valued probabilities cannot interfere, unless they are allowed to go neg-
ative. No one, except perhaps Feynman, figured out how to make sense
of negative probabilities, so everyone sticks with tradition and uses the
complex “square roots of probabilities” called amplitudes in QM.
Poincaré Symmetries 43
1. Space rotation.
and for inertial frames these require the metric be unchanged, hence,
𝜇
𝑔𝜇𝜈 Λ𝜌 Λ𝜈𝜎 = 𝑔𝜌𝜎 . (2.1c)
We are not quite done. To get the Poincaré group from the above defini-
tions we need to specify the group composition rule and show is satisfies
the group axioms. To compose two transforms,
𝑥 ↦→ 𝑥0 ↦→ 𝑥00
1. The full Lorentz group O(1, 3): preserves spacetime lengths. (The
full P (1, 3) does not, due to Lorentz contraction).
• det Λ = ±1,
• has all translations 𝑎𝜇 = 0,
• composition rule is: 𝐿 (Λ2 , 0) ◦ 𝐿 (Λ1 , 0) = 𝐿 (Λ2 Λ1 , 0),
• Λ00 ≥ 1 or Λ00 ≤ −1, so this group is still not simply connected.
O(1, 3)
SO(1, 3) +
𝑃 det Λ = −1
det Λ = 1
Λ00 ≥ 1
Λ00 ≥ 1
𝑇 𝑇
det Λ = −1 𝑃 det Λ = −1
Λ00 ≤ −1 Λ00 ≤ −1
The plan now is to find the generators of the Lie algebra for the
Poincaré group. Once we find the generators we can compute their
commutation relations, and use those to build a basis of unitary states
46 Particles and Symmetries
and find the eigenvalues, and those will be our unitary irreducible
representations, hence our “elementary particles.”
Notice also that the Lie algebra for the full Poincaré group will be the
same as the Lie algebra for the one connected sub-component containing
the identity. That will be SO(1, 3) + . So the problem of finding irreducible
representations reduces to studying the Lie algebra of SO(1, 3) + .
maining once we put aside space translations and the discrete symme-
tries of time and space inversion. One might worry this will miss some
elementary particles, but we can look ahead and say,
and so we should get all the quanta for the classical fields of gravity and
electromagnetism just by examining the homogeneous Lorentz group.
In section 2.3 we already went through the whole Poincaré group. Re-
call that we want to now look at the components connected to the identity
to study the Lie algebra. Restricting now to the connected component
SO(1, 3) + we introduce some symbols for the generators:
The last two symmetries, 𝐽 and 𝐾 , together comprise the Lorentz group.
We may also say the Poincaré group comprises the full set of isome-
tries of spacetime or ISO(1, 3) — meaning action by the group transforms
leaves the metric invariant (preserves spacetime vector lengths and an-
gles). Space lengths and time intervals are not preserved by the general
transform, as is familiar for students of special relativity (we get 3-vector
length contradiction and time dilation). Only the pure space rotation
and translations preserve 3-vector lengths. This is so far all physics, the
mathematical language of group theory is only that, a language. Without
Michelson–Morley we would not know to be using the Poincaré group.
In Minkowski space, 4 𝐷 flat spacetime with metric signature (− + ++),
there are ten degrees of freedom for the isometries, which may be thought
of as translation through time or space (four degrees, one per dimen-
sion); reflection through a plane (three degrees, the freedom in orienta-
tion of this plane); or a boost in any of the three spatial directions (three
48 Particles and Symmetries
We begin our study of the Lie algebra by considering the way spacetime
vectors transform under SO(1, 3) + . This is because our only tenuous con-
nection to physics, for now, is energy-momentum conservation. To study
the Lie algebra we need to consider infinitesimal transformation about
𝜇
the identity. To do so we introduce an infinitesimal tensor 𝜖𝜈 to param-
terize these infinitesimal Lorentz transforms,
𝜇 𝜇
Λ𝜈 ≈ 𝟙 + 𝜖𝜈
Then we can write the transform for a spacetime vector 𝑣, as
𝜇
𝑣𝜇 ↦→ 𝑣0𝜇 = Λ(𝜖)𝜈 𝑣𝜈 .
𝜇
NB: if you want tensor notion throughout then 𝟙 = 𝛿𝜈 . This definition for
𝜇 𝜇
the infinitesimals 𝜖𝜈 of course requires the 𝜖𝜈 satisfy certain conditions
𝜇
for yielding Lorentz transforms, we cannot just pick anything for 𝜖𝜈 . We
can see what the main condition is by remembering the Λ are required
to be metric preserving, hence,
𝜇
𝑔𝜇𝜈 Λ𝜌 Λ𝜈𝜎 = 𝑔𝜌𝜎
𝜇
plug the previous definition for 𝜖𝜈 into that condition and we find,
𝜇
𝜖𝜈 = −𝜖 𝜇𝜈
Invariants 49
2.4.2 Commutators
2.5 Invariants
[ 𝐽𝑧 , 𝐽± ] = ±𝐽 ±, (2.1)
[ 𝐽+ , 𝐽− ] = 2𝐽𝑧 . (2.2)
𝐽− 𝐽+ = 𝑂ˆ † 𝑂,
ˆ for some operator 𝑂ˆ
states, and the choice is 𝐽𝑧 , so there is some label 𝑠 say, for the eigenvec-
tors of 𝐽𝑧 (this is just a definition and convention, ok):
𝐽 𝑧 | 𝑠i = 𝑠 | 𝑠i .
( 𝐽𝑧 𝐽+ − 𝐽+ 𝐽𝑧 )| 𝑠i = 𝐽+ | 𝑠i
∴ 𝐽𝑧 𝐽+ | 𝑠i − 𝑠𝐽+ | 𝑠i = 𝐽+ | 𝑠i
∴ 𝐽𝑧 𝐽+ | 𝑠i = ( 𝑠 + 1) 𝐽+ | 𝑠i
⇒ 𝐽+ | 𝑠i ∝ | 𝑠 + 1i
and similarly,
𝐽− | 𝑠i ∝ | 𝑠 − 1i or |0i
. . . , 𝑠 − 1, 𝑠, 𝑠 + 1, . . .
Next little bit of jujitsu, although I guess you can come back to this point
and check it does not make any difference, but the “thing to do” is define
eigenvalues for 𝐽− and 𝐽+ as follows,
𝐽− | 𝑠i = | 𝑠 − 1i , 𝐽+ | 𝑠i = 𝜆 𝑠 | 𝑠i
𝐽+ 𝐽− | 𝑠i = 𝜆 𝑠−1 | 𝑠i , 𝐽− 𝐽+ = 𝜆 𝑠 | 𝑠i
[ 𝐽+ , 𝐽− ] = (𝜆 𝑠−1 − 𝜆 𝑠 )| 𝑠i
= 2 𝐽 𝑧 | 𝑠i = 2 𝑠 | 𝑠i
⇒ 𝜆 𝑠−1 − 𝜆 𝑠 = 2 𝑠. (2.3)
Invariants 51
Next a third little bit of jujitsu, we do one of these cheat look-ahead (“to
make things nice” the teacher always says. . . sure) and define an eigen-
constant 𝑗 as follows,
𝜆 𝑠 = 𝑗 ( 𝑗 + 1) − 𝑠 ( 𝑠 + 1) . (2.4)
Why? Well, just because this solves e.q(2.3). Check that it does! 𝑗 ( 𝑗 + 1)
is just a constant, so it disappears upon the subtraction, so yeah, it is a
solution. A difference equation like that always has an arbitrary constant
in the solution.
Now consider, if 𝑗 is a constant, and we suppose 𝑠 can get arbitrarily
large (positive or negative) then clearly at some worse case 𝑠 we’ll get
𝜆 𝑠 < 0. OK, but recall both 𝐽− 𝐽+ and 𝐽+ 𝐽− are both of the form 𝑂ˆ † 𝑂ˆ ,
which has a positive (or zero) eigenvalue. The eignenvalue is zero only if
𝐽− or 𝐽+ annihilate the state. This is inconsistent with 𝜆 𝑠 < 0, so it must
be true that there is a minimum and maximum eigenvalue for 𝑠 labelling
| 𝑠i, call these 𝑠min and 𝑠max .
Since, as we’ve defined them, 𝐽− lowers the 𝑠 value and 𝐽+ raises the 𝑠
value, we must have,
note that we get these from factorizing eq(2.4) in two different ways:
𝑥 ( 𝑥 + 1) − 𝑦 ( 𝑦 + 1) = ( 𝑥 − 𝑦)( 𝑥 + 𝑦 + 1) , or ( 𝑥 + 𝑦)( 𝑥 − 𝑦 + 1)
𝑗 ∈ {0, 21 , 1, 32 , 2, . . . }
52 Particles and Symmetries
* * *
Mathematicians must have known this since way back, but they were
never much doing quantum mechanics. Joe Numberson says: “If you
wish to rotate use a damn matrix on a space vector! Why go to a Hilbert
space?” Jane McMath says, “Hey, do not even use matrices, use | 𝐺 𝐴
rotors.”
Since the physicists go to a Hilbert space, and want these eigenstates
and eigenvalues, for the damn Measurement postulate P.2, we have these
half-integer labels possible. But they are not about quantization of any-
thing. They are just a choice of representation for measurements. What
are they measuring? The answer is anything to do with rotational sym-
metry invariance. What’s that? Angular momentum dumb ass! The
group is a smooth group of rotations, the representions of a basis in
Hilbert space have these discrete labels, and the two are not the same
thing, they are just related. They help classify elementary particles.
Got it? So the “discrete spin quantum numbers” are classifications
of possible robust (invariant) elementary particles, not quantization of
angular momentum or anything like that.
The same comments apply if we go and look at solutions to the
Schrödinger Equation for the Hydrogen atom or whathaveyou. We there
see orbital angular momentum, which is a rotational symmetry of the
electron orbits in the H atom. We should find a set of basis states in the
Hilbert space for the atom to represent this symmetry, and it will be the
same Lie algebra as these more fundamental particles. All rotationally
symmetry system everywhere will have these integer or half-integer
Invariants 53
something.
Yeah, I know about the Elitzer–Vaidman Bomb Test and all that, but
it’s the same thing. You set up a spacetime manifold so the photon must
go via the “upper” path to ever, ever, get to detector 𝐷. (See en.wikipedia-
.org/wiki/Elitzur-Vaidman_bomb_tester.) Here you are still disturbing
something, the detector at 𝐷 and the optic fibres and mirrors in the upper
path. The weirdness of QM is not because of anything mystical here, it
is the plain measurement postulate in action, along with superposition.
Remember 50% of the tested bombs that are not duds do explode. If an
iterated qubit system is used to reduce the live bomb explode rate to near
zero you will then have problems with stray photons setting the bomb off
just randomly. It’s a nice system, but it does not scale well. Same for
most awesome things about QM, except Josephson junctions, transistors,
superfuids and black holes, they have nice quantum effects at large scale
(I guess you could say the cosmos as a whole too). Or, you know what I
mean: all things show QM at large scale, or anything with an atom in
it at least. But the effect of superposition and entanglement at scale is
what I am talking about.
To conclude: there is always interaction and the interactions cannot
do what the experiment is not set-up to do, and quantum mechanics is
only a little bit weird.
After all this effort we have not really done a lot of physics. Group the-
ory is a very blunt instrument. All we’ve discovered is that in quantum
mechanics fundamental particles can be usefully labelled by momentum
and spin. All that really says is that elementary particles move around
in space by translations and rotations in Lorentz invariant manner. Big
deal!
Well, it is a kind of big deal. We had no right to think group theory
would tell us two labels suffice to characterize the elementary particles.
It turns out we were right to be suspicious, we do have no right, because
there are color and flavour symmetries we have not taken into account.
The Poincaré symmetries are only the most obvious.
The fact we now have labels for doing relativity in quantum mechanics
56 Particles and Symmetries
The current hints from the LHC are that there is physics “beyond the
Standard Model.” That said, in this chapter we have not even got up to
the Standard Model. But it is not hard to qualitatively describe how to get
to the Standard Model: we need to find other symmetries of fundamen-
tal particles, then repeat the same analysis — find the group, find the
unitary irreducible representations, and postulate “new particles” from
them. We’d find this is how we get the quarks and gluons, from flavour
and color4 symmetry. But what is the physical basis for those symme-
tries? There must be a physical basis otherwise we are doing magic, and
that is no good for me since I failed Wizardry 101 school.
The remarkable thing you “never get taught in school” is that the
quarks and gluons are not unexpected in Geon Theory. Of course, we had
to discover chromodynamics empirically, because the colors and flavours
are hidden in the composites, the protons and neutrons. We had to smash
them up to see.
Personally I do not think there was ever going to be any other way. The
experimentalists are the heroes, the theoreticians just get to tell stories,
and occasionally, if they are lucky, tell scifi stories that turn out to be
sci-fact, but that’s due to experimentalists finding the facts. However,
it is legitimate to tell fantasy stories of imagined alternative histories
where the theoretician predicted all the new fundamental particles in
advance, and the experimentalists were just confirming; firstly because,
as I just said, that’s rarely but sometimes how it happens for one or two
new particles, and secondly because theoreticians need encouragement.
On the encouragement side, how exactly were quarks and gluons not
unexpected in Geon theory? This has to do with the Gell-Mann matrices,
which comprise the algebra for chromodynamics. The entire Gell-Mann
algebra lives in the Clifford spacetime algebra 𝐶ℓ (1, 3), and so could have
been obtained from a geon theory, had anyone been using geometric al-
gebra instead of matrix algebra. Thus, so far, all known physics has
4I
use English spelling, but refuse to call chromodynamics “colour” symmetry, it is
nicer to think of color as a mere analogy to colour.
What was accomplished? 57
Experiment rules the roost in physics, a fact everyone knows except the
more zealous theoreticians. Zeal and ardour in searching are however
not a bad thing. The key is balance in diversity, not in homogeneity. It
is good to have all sorts of types of people doing physics and cooperating
and sharing results, even the crackpots. How many times did a genius
physicist “steal” an idea from a crackpot and claim greatness, perhaps
even without thinking they were borrowing the idea? I do not know,
perhaps never, but the chance greatness can spring from insanity is not
something to always shy away from, it is something to temper with some
appropriate moderation, and let a hundred flowers bloom (there, you see,
even Mao Zedong was not all despot).
But experiment rules the roost.
So when parity violation was observed in kaon decay by Cronin and
Fitch, few thought they were crackpots with a mad theory because this
was experiment. Although experiments can go astray (cold fusion, eu-
genics and other madness) when you gt the experiments right no one
will accuse of of being a crackpot, at least not for the reason of your ex-
perimental results. If you test a novel unproven drug on yourself then its
you who is the crackpot, not your experiment.
Subsequently 𝑇 (time symmetry) was found to be violated, for example
in neutral meson systems. Since CPT symmetry is thought to hold true,
this means CP must also be violated in some systems.
This section discusses some basic aspects of CPT symmetries for stu-
dents who are unfamiliar, it is not meant to be a detailed review. My
main motive is to give younger students some good intuitive feel for how
physics works, so to speak. To motivate the student I present this little
puzzle.
How can physics depend upon whether we mirror invert? Surely this
would mean given a ‘blind’ trial look through a mirror into our
universe we could tell which side of the mirror we were on, and would
CPT Symmetry 59
* * *
In quantum theory all force arises from particle interaction, even gravity.
There are no fields. This means a nice way to analyse what forces are
described by our physics is to consider elementary interactions. If we find
there are interactions that are not allowed, then we will get constraints
on the possible types of forces.
Once we have the known forces, then later on we can invent fictional
fields to calculate dynamics without needing to think about the elemen-
tary interactions. The elementary particle–particle interactions are
however always primary ontology, at least in all quantum mechanics we
know, including string theory.
A quick aside: (and since I write for students) when you read the fu-
ture press, do not get sucked in too much by the hype when the LHC or
the next generation of colliders has “discovered a new force.” All the pos-
sible elementary (single boson exchange) forces of nature are described by
Yang–Mills theory.1 It is like MMT in macroeconomics. One framework
can describe all. What a “new force” means is journalistic laziness for
discovery of a new type of boson. That is going to be a huge momentous
event when it happens, but it is not really a new type of force. It will
still be a Yang–Mills force. There is going to be a lot to hype when this
1Iwrite this with false confidence. We may eventually discover entirely new physics
that cannot ever be explained by a Yang–Mills theory. All I am saying, really, is that
I personally think the chances of that are none to zero. But that’s just a subjective
personal opinion. You can take that and Bayesianize it and deduce anything you like,
depending upon how much weight you give my opinion.
61
62 Scattering and Forces
I do not want extra space dimensions, ok. If you do, for some reason,
then fine, go off and study string theory. Please. If you like, think of this
as just a research effort to see how much we can get from 4 𝐷-spacetime.
Maximum parsimony.
Since the quantum theory of gravitons is unworkable, does this mean
gravity is not a quantum theory?
My answer is no! — and in fact GR is already a quantum theory be-
cause it admits wormhole topology on the Planck scale (at least). I mean,
how do you tell the cosmos to not create extremal wormholes? You can-
not. If one exists and it’s ends are minimal black holes, it cannot Hawking
evaporate, so it will be stable. Macroscopic wormholes are a no-go, but
Planck scale wormholes are fine. This provides ordinary old GR with a
definite quantum postulate: minimal wormholes exist. They are elemen-
tary particles of a sort.
There is precedent for such weird “possible but unobserved” physics. The
Dirac theory of electrons permits magnetic monopoles, and yet we have
never observed them. String theory permits all manner of exotic parti-
cles, but none of them are observed. String theory requires supersym-
metry, which is not observed. A lot of possible physical states are totally
allowed to exist but do not apparently exist.
Now I am not saying gravitons are fictional. My view is that pack-
ets of gravity waves are gravitons, and gravitons are nothing more. So
in the extreme conditions of black physics I would expect particle-like
gravitons: highly localized energetic gravity waves, soliton solutions in
general relativity. I would not see how nature can avoid them.
This means in black physics we will have to grapple with the mathe-
matical difficulties of having gravitons in the path integrals or scatter-
ing calculations. But that is probably going to be fine, because cutting
edge theory, like the Amplituhedron program, suggest perturbation the-
ory with Feynman diagrams is not the way to do honest quantum field
theory. If that turns out to be the case, then renormalizable calculations
can probably be accomplished in the future with gravitons included.
I’d beg some empathy here. These ideas are radically conservative
and highly unifying. Consider that gravitons are massless (gravity is
long range) so are gravity waves.
Consider (indulge the imagination) that in an alternative reality space
was filled with some kind of ‘luminiferous aether’ or phlogiston, respon-
sible for light waves. Although that turned out not to be the case, if it
had then electromagnetism would be a lot like gravity, only with pho-
tons being actual physical waves rather than topological particles. We
would not then be talking about the photoelectric effect and photons.
But, . . . small soliton-like waves of phlogiston might arise, and they’d be
photons, spin=1 massless “particles.” If we pushed physics to extremes
we would probably say these ‘photons’ can be generated. They’d radiate
like crazy, like gravitons would, but they’d exist, and we’d describe them
quantum mechanically by spin=1, if they were small enough to be treated
in superposition and entangled states.
Because photons are localized packets of energy, there is no aether
model for them. But for gravitons there is an aether, and it is just space-
time. Since spacetime is smooth on the macroscopic scale we see grav-
ity waves, not gravitons, and we have no need for gravitons at all. But
Maximum parsimony 65
we have not gone beyond the paradigm of quantum mechanics here, the
spin=2 “particle” is still a possible object that can be generated, precisely
by sufficiently localizing a gravity wave packet. How does one do that?
With great difficulty is the answer. Just as solitons in water are incred-
ible rarities, so would gravitational solitons. They are not stable, and
require carefully manufactured boundary conditions to get propagating.
Who is to say nature does not fashion such conditions all the time though?
I do not know. But I am pretty confident people who study gravity wave
solitons will tell you it is not easy to motivate them.
Because all the other fundamental particles in nature are stable and
appear soliton-like, they are probably not wave modes of any aether. They
are far more likely to be topological in character, not wave-like. But then
we need a decent theory to explain why the elementary particles behave
like waves in QM. To my knowledge, only geon theory has a reasonable
answer to this, which I have, and will, write about elsewhere. It is a long
topic. The short answer is that to get information-theoretic (epistemic)
probability amplitude waves, as in Schrödinger–de Broglie–Dirac theory,
you do not need a wave medium, you need a mechanism for a fundamental
geon to appear to be in more than one place at one time, and yet appear
in only one of those places whenever detected.
Although it might be sore on your eyes and ears to hear or read this,
the most conservative way I can think of getting such phenomenology is
with closed timelike curves. This is where geon theory begins. A beau-
tiful unifying thing here is that gravity furnishes these for us, they are
wormholes. They do not need to be large, they only need permit elemen-
tary particles to traverse them, and that’s entirely plausible given all
known theoretical physics. Remember monogamy of entanglement? Why
would nature restrict entanglement to pairs, for godssake? Can it be that
entanglement is formed by Einstein–Rosen bridges? That explains why
there are only two ends to an entangled state, and why exotic multipar-
tite entanglement can occur, as with GHZ states, but upon measurement
only pairs are found to be fully entangled.
So who the hell knows what is more real, the warping of spacetime or
the graviton?
All I know is that most gravity effects are enormous in scale, and so it
does little good to analyse most of these phenomena in terms of graviton
interactions. But does expedience dictate ontology? No. However, when
it comes down to aesthetics, I have a definite preference. I see elemen-
tary particles as topology obstructions in otherwise smooth spacetime. I
cannot model a graviton in the same way, whenever I model a graviton in
T4G theory, it is just a gravity wave, of one kind or another, and if these
waves get sufficiently localized they can propagate along closed timelike
curves through wormholes, and then they’ll have to be treated quantum
mechanically.
Does that mean the spin=2 graviton is a particle now, not a gravity
wave? I think not. It is still a gravity wave, but it can get highly local-
ized in special circumstances, and in black physics. It is still in some
sense a classical warp in spacetime, and never is otherwise. When I do
graviton Feynman or Amplithedron computations, I am most definitely
thinking in terms of the Huygens wavelet approximation in these cases.
I do not for one second believe there are tiny packets like gravitons in-
volved. The usual (non-black physics) gravitational scattering physics is
highly macro, and the graviton is a fiction used to do calculations.
I do not want to theoretically yoyo too much here, but I think it is
important to draw a sharp distinction now with photons. You can say
soft photon scattering is similar to graviton scattering, in that you can
treat the photon as a particle and do Feynman diagrams, but you have
no hope of localizing these photons, they are going to be practically un-
detectable. However, I can imagine getting lucky and detecting a soft
photon, and when I do it will be highly localized, and will jiggle some
electron — that’s how it gets detected, ultra-sensitively somehow. We do
not have such sensitive detectors, but this gedankenexperiment assumes
we’ve got some such detector from some advanced alien civilization. In
T4G theory the same is not going to be possible with gravitons. This is
a sharp prediction of geon theory in my view. The graviton can get local-
ized, but then will be high energy. Most gravitons are not, and so will
never be detected as localized quanta, because they are not topological,
they are geometrically smooth.
There is one “out” for geon theory here: if we can find a model where
68 Scattering and Forces
a star. But think about this for just a second. It is absurd. Any parti-
cle (even the massless) going near a star gets continually and smoothly
bent around the geodesic. There is no single graviton doing this. Yet the
correct angle deviation is obtained from the single graviton scattering
formula. What is going on here?
To my mind this is a strong hint that the graviton theory is a gross
quantum idealization, it is a pure 𝑆-matrix theory object. The 𝑆-matrix
recall, only cares about asymptotic states, so the incoming and outgoing
geodesics are those of flat space, no star, and you get the angle devia-
tion from just a single graviton. The path integral pretends to sum over
infinitely many interactions, but all that is not really physical in some
sense — in the sense you do not need them.
It is like gross overkill.
The imagery that springs to mind is that of Ayrton Senna driving the
particle. Senna was famous for pumping the accelerator wildly when
turning corners, in the old F1 cars this somehow gave him more corner-
ing control and made him fast. Fernando Alonso does something similar
but with steering. My imagination thinks of all the off-shell interactions
as wobbling the particle around as it goes by the star, with all the in-
finitely many wobbles doing nothing at the end of the day except pray to
the mathematical symmetry gods and give alms to the lord of symmetry.
How the hell do all these off-shell interactions know to conspire to cancel
each other out? Clearly it is a physical fiction that has mathematics that
just works out, because of the demands of diffeomorphism invariance you
write into your Lagrangian.
I think it is pretty obvious that a lot more happens as the particle
goes by the star than a single graviton exchange. I think that graviton
is a fiction. It is being exploited to tell a story that avoids all mention of
Einstein curvature. You can comb a cat in more than one way. You can
have gravity in more than one way.
I do not think it is wrong to do the scattering calculation with the on-
tree graviton. It has to work out, because it has to be possible to replace
the curved spacetime near the star with a fictional standing gravity wave
or such-like, as long as what the particle sees is similar geodesics the
angle of deviation will be the same. And it has to be possible to replace
a macroscopic gravity wave disturbance with a spin=2 gravity-wave type
of soliton that hits the particle just once. Same 𝑆-matrix asymptotics.
70 Scattering and Forces
* * *
End of story? No, not end of story, but the end of my tale of gravitons.
Scale invariance
I just lied, I do have three more things to say about gravity and gravitons.
One slightly fanciful, let’s be generous and say whimsical, way of
thinking about gravitons, which I do not think is totally wrong, is that
there is a graviton, there is exactly one graviton, and it has a radius of
about 15 billion light years. All sorts of interesting stuff happen internal
to this one graviton, which is called “life.”
A second thing concerns scale invariance. It does not surprise me (only
in retrospect of course) that the AdS/CFT correspondence works. The
CFT is a theory invariant to scale. When you do not care about scale then
the hydrodynamics approximation works well: one graviton scattering
becomes indistinguishable from warping of spacetime. The conformal
invariance says you cannot possibly tell the difference. The fact our world
Maximum parsimony 71
does not have perfect conformal invariance is why we can say meaningful
things about the difference between gravity waves and gravitons.
A third thing harks back to my earlier comments on the absurdity
of thinking gravity lensing by stars is a single graviton scattering (for
massless particles). For a massive particle it does take more than one
graviton interaction to get the angular deviation, but the basic idea is
the same. The problem is that all the infinite other interactions have to
cancel out precisely to get the correct agreement with GR. QFT is a way
to get precise cancellation because it says you have to add up the ampli-
tudes in a precise way, and you must impose all the correct symmetries,
if you order the sums differently in the perturbative approximation all
hell breaks loose and you can get an unfathomable mess. It is only when
either,
that the miracle cancellations occur. If you are a religious believer in QFT
then this is fine. For the believer, the miracles do occur because nature
does take all possible paths (or acts like it knows them all).
For my taste, this is a massive loss of parsimony. The GR spacetime
warping model suffices, and I think is more realistic in some sense. This
is the sense in which there really are no graviton quanta, and what we
have is the warping of spacetime, which can be thought of in the abstract
as one gigantic graviton the size of a solar system. The effect of space-
time curvature is thus a lot like a Yang–Mills interaction, but in a to-
tally different regime, a full hydrodynamics regime. Think of it this way:
what happens to your fluid water viscosity and compressibility when the
molecule of water is a metre in diameter and your boat or aeroplane is
about the same length? You do not have hydrodynamics, you’ve got bil-
liards. Clearly, at least to my mind, gravity lensing and planetary orbits,
and ourselves sitting on the Earth, are not the billiards regime. For us,
there are no gravitons. There is only Ricci curvature and gravity waves
(Weyl curvature).
* * *
72 Scattering and Forces
𝑝𝜇 = ( 𝑝0 , 𝑝 ) = ( 𝑝0 , 𝑝𝑖 ) , 𝑝𝜇 𝑝𝜇 ≡ 𝑝2 = −𝑚2 .
We do not need matrices. The Pauli and Dirac matrices are better
thought of as 3-space and 4-spacetime basis vectors respectively. For the
Pauli algebra,
𝜎𝑖2 = 1, 𝜎𝑖 𝜎 𝑗 = 𝜎𝑖 · 𝜎 𝑗 + 𝜎𝑖 ∧𝜎 𝑗 ,
𝝈 𝑖 · 𝝈 𝑗 = 0, 𝜎𝑖 ∧𝜎 𝑗 = −𝜎 𝑗 ∧𝜎𝑖 , 𝐼 ≡ 𝝈1 𝝈2 𝝈3 .
The symbols 𝐼 and 𝑖 here are respectively the space and the spacetime
pseudoscalar, for Minkowski geometry. Both have properties of a unit
imaginary except that 𝐼 commutes with both vectors and with bivectors,
while the spacetime pseudoscalar 𝑖 anticommutes with vectors but com-
mutes with bivectors. The reason being that when there is a term 𝑒𝑘 𝑒𝑘
in basis vectors upon a permutation, then we do not need an extra per-
mutation to swap these two, so to move the psuedoscalar across a single
74 Scattering and Forces
base vector only takes ( 𝑛 −1) permutatons. We illustrate here for a vector
𝑎 = 𝑎2 𝝈 2 :
𝑎 𝐼 = ( 𝑎2 𝝈2 )(𝝈1 𝝈2 𝝈3 )
= 𝝈2 (𝝈1 𝝈2 𝝈3 ) 𝑎2 , (here, the scalar 𝑎2
= −𝝈1 𝝈2 𝝈2 𝝈3 𝑎2
= +𝝈1 𝝈2 𝝈3 𝝈2 𝑎2
= (𝝈1 𝝈2 𝝈3 )𝝈2 𝑎2
𝑎
= 𝐼𝑎
Now try this again in spacetime algebra (STA), again for 𝑎 = 𝑎2𝛾2 :
𝑎𝑖 = ( 𝑎2𝛾2 )(𝛾0𝛾1𝛾2𝛾3 )
= −𝛾0𝛾2𝛾1𝛾2𝛾3 𝑎2
= +𝛾0𝛾1𝛾2𝛾2𝛾3 𝑎2
= −𝛾0𝛾1𝛾2𝛾3𝛾2 𝑎2
= −(𝛾0𝛾1𝛾2𝛾3 ) (𝛾2 𝑎2 )
= −𝑖𝑎
In any dimension the pseudoscalar will commute with even grade vec-
tors such as bivectors. A complex algebra structure, which requires a
commuting root of −1, is the even subalgebra of 𝐶ℓ (3).
By admitting vectors of all grades up to the vector space 𝑉 dimension
𝑛, for 𝑉 = ℝ3 the graded algebra is 𝐶ℓ (3), which has one scalar, 1, three
vectors, three bivectors and a pseudoscalar, so is a 𝑑 = 8 dimensional al-
gebra, and the Pauli algebra is just the even subalgebra of 𝐶ℓ (3), denoted
𝐶ℓ + (3). The Dirac algebra is the algebra of the unit scalar combined with
the vectors {1, 𝛾𝜇 }.
Formally, the Pauli algebra is the closed algebra generated by,
{1, 𝝈 𝑖 𝐼 }
where on√the left we have the 2-component Pauli spinors and the imagi-
nary 𝑖 = −1, while on the right we just have plain basis vectors and the
pseudoscalar 𝐼 = 𝝈1 𝝈2 𝝈3 and no need for imaginary numbers. You can
check for yourself as a homework exercise that with the unit scalar re-
placing the identity matrix, the above map is indeed a bijection with each
Pauli matrix 𝜎ˆ 𝑖 replaced by the corresponding Euclidean base vector 𝝈 𝑖 .
In case you forgot, the Pauli operators can be defined by the relations,
𝜎
ˆ 𝑖𝜎
ˆ 𝑗 = 𝛿𝑖 𝑗 + 𝑖𝜖 𝑖 𝑗𝑘 𝜎
ˆ 𝑘.
𝐴 ∗ 𝐵 ≡ h 𝐴𝐵i0
inside the brackets is the full geometric product of 𝐴 and 𝐵, which could
have terms of all 𝑛-grades in the full algebra, the bracket with subscript
𝑘, so h· · · i𝑘 , by definition projects out the 𝑘-grade part only, and the scalar
part is the 𝑘 = 0 grade.
The other useful product operator is the commutator product ×, de-
fined by
1
𝐴 × 𝐵 = ( 𝐴𝐵 − 𝐵𝐴) .
2
again, inside the parentheses are two full geometric products. We will
use the commutator product in our Lie algebra technology.
Bivectors in particular are the only geometric object we need to define
representations for all Lie groups. A bivecotr, say 𝐵 = 𝑎 ∧ 𝑏, where 𝑎 and
𝑏 are 1-vectors, is an oriented area element. It is shapeless, so you can
think of it as a parallelogram if you like, the parallelogram swept out
by moving 𝑏 along the vector 𝑎. But there is no need to think 𝑎 ∧ 𝑏 has
any particular shape, this is powerful, since it means infinitesimal bivec-
tors can represent arbitrary shape area elements in a geometric calculus.
Determinants will handle the area transforms.
Likewise, a trivector in a 𝑛 ≥ 3 dimensional vector space, say 𝑎∧𝑏∧𝑐, is
an oriented volume element, so a parallelepiped, but does not have to be
thought of as having a parallelipiped shape, though again one can sweep
the parallelogram defined by first two along the third vector to get an
76 Scattering and Forces
2·𝐹 = 𝐽
2∧ 𝐹 = 0
𝐹 = 𝐸 + 𝑖𝐵
𝐵,
𝑖 ≡ 𝛾0 𝛾1 𝛾2 𝛾3 , is the spacetime pseudoscalar.
1
𝑎·𝑏= ( 𝑎𝑏 − 𝑏𝑎)
2
1
𝑎 ∧ 𝑏 = ( 𝑎𝑏 + 𝑏𝑎)
2
But we can also define the inner product of a vector and any multivector
as the grade lowered part of the geometric product,
𝑎 · 𝐴𝑟 = h 𝑎𝐴𝑟 i | 𝑟−1|
any multivector even one of mixed grades. Similarly, the wedge or outer
product is the grade raised term,
𝑎 ∧ 𝐴𝑟 = h 𝑎𝐴𝑟 i | 𝑟+1|
Without too much mental effort this leads naturally to the inner and
outer products of any two pure grade multivectors,
𝐴𝑟 · 𝐵 𝑠 ≡ h 𝐴𝐵i | 𝑟−𝑠 |
𝐴𝑟 ∧ 𝐵 𝑠 ≡ h 𝐴𝐵i | 𝑟+𝑠 |
We have now pretty much covered all the different types of product opera-
tors in a full graded geometric algebra. This will give us enormous power
of dopes who still use tensors. Although in my books I never use all this
power, so you can wonder why I bother. I just think geometric algebra is
super cool. I’m that that guy who’d drive a Bugatti Chiron around just
to go grocery shopping and drop the kids at playschool (like, I wish).
If you ever need a dictionary translation back to matrix elements, the
matrix elements 𝑎𝑖 𝑗 , of a matrix acting on vectors 𝑎® always refer to a
chosen basis, {®𝑒𝑖 },
𝑎𝑖𝑘 = 𝑒®𝑖 · 𝑎
®𝑘 .
here each 𝑎® 𝑘 is an entire vector, the index is a label not a scalar compo-
nent. The traditional matrix of this matrix is the array where all the 𝑎® 𝑘
are the columns. Determinants are then,
det( 𝑎𝑖𝑘 ) = ( 𝑒 𝑛 ∧ . . . ∧ 𝑒 1 ) · ( 𝑎 1 ∧ . . . ∧ 𝑎 𝑛 )
= 𝐼˜ · ( 𝑎 1 ∧ . . . ∧ 𝑎 𝑛 ) .
where 𝐼˜ is the reverse of the pseudoscalar 𝐼 . I will try to not use matri-
ces, we do not need them. Occasionally to show the correspondence with
orthodox textbooks I may write one or two of them.
All multivectors in a geometric algebra 𝐶ℓ ( 𝑛) are our linear operators.
Consider any linear map, an endomorphism 𝑓 , on the vector space,
𝑓 : 𝑎 ∈ ℝ𝑛 −→ 𝑎 0 = 𝑓 ( 𝑎 ) ∈ ℝ𝑛 .
𝑓 (𝑎1 ∧ . . . ∧ 𝑎 𝑘 ) = 𝑓 (𝑎1) ∧ . . . ∧ 𝑓 (𝑎 𝑘 ) , 𝑘 ≤ 𝑛.
78 Scattering and Forces
In matrix language we use traces and determinants a lot, when our linear
operators are G 𝐴 multivectors like 𝑓 , we still have traces and determi-
nants, and they are very easy to compute,
det 𝑓 = 𝑓 ( 𝐼 ) 𝐼˜
tr 𝑓 = 𝜕𝑎 · 𝑓 ( 𝑎) .
𝑏 · 𝑓 ( 𝑎) = 𝑓 ( 𝑎) · 𝑏.
Pretty simple huh, so don’t you forget it! If you recall that the Dirac
bra-ket was just an inner product all along this should be easy for you
to digest. In QM terminology the adjoint, remember, is the Hermitian
transpose of the operator. It only takes one line, but it is tricky to show
the familiar adjoint product still holds for us,
𝐹𝐺 ( 𝑎) = 𝐺 𝐹 ( 𝑎)
Geometric algebra preliminaries 79
𝐹 −1 ( 𝐴) = 𝐼𝐹 ( 𝐼 −1 𝐴)(det 𝐹 ) −1
−1
𝐹 ( 𝐴) = 𝐼𝐹 ( 𝐼 −1 𝐴) (det 𝐹 ) −1 .
This is so slick, you should try recalling how long it took to find a for-
mula for a matrix inverse. These multivectors, the ones with non-zero
determinants, are isomorphic to matrices, so we’ve just done a whole lot
of traditional matrix algebra technology in a single page.
Our linear maps also have eigenvectors of course,
𝐹 ( 𝑒) = 𝜆 𝑒.
˜ − 𝑒1 𝑒2 𝜃/2
𝑓 ( 𝑎) = 𝑅𝑎𝑅, 𝑅=𝑒
then
𝑓 ( 𝑒1 ) = cos 𝜃 𝑒1 + sin 𝜃 𝑒2 , 𝑓 ( 𝑒2 ) = − sin 𝜃 𝑒1 + cos 𝜃 𝑒2
as you’d expect, since this is a rotation by the rotor 𝑅. Using the endomor-
phism extension of 𝑓 to act on the bivector 𝑒1 𝑒2 , we find an eigenblade,
specifically and eigen-2-blade or just “eigenbivector”;
= 𝑒1 ∧ 𝑒2 .
Or, for shorthand, 𝑓 ( 𝑒1 ∧ 𝑒2 ) = 𝑒1 ∧ 𝑒2 . So the rotation has an eigenbivector
with eigenvalue =1. We see this, and say: rotation in a plane leaves the
plane unchanged ( 𝑒1 ∧ 𝑒2 ), but objects living in the plane ( 𝑒1 itself say, or
𝑒2 ) get rotated. How come 𝑒1 and 𝑒2 individually change but 𝑒1 ∧ 𝑒2 does
not? It is because in G 𝐴 the bivectors are primitive oriented elements of
area, they have no residual shape or vector aspect. In G 𝐴 vector parts are
described by vector parts, surprise, surprise. Although bivectors can be
described by parallelograms, you could equally describe a bivector just
as well by an oriented disk. Rotate the disk in its own plane leaves it
unaltered.
Now what if we have a different 𝐹 , with a “complex” eigenvalue from
the characteristic polynomial roots, say,
𝑓 ( 𝑢 + 𝑖𝑣) = (𝛼 + 𝑖𝛽)( 𝑢 + 𝑖𝑣)
This is the general case because complex eigenvalues are generally as-
sociated with complex eigenvectors, the eigenvector might have all real
components (so 𝑣 = 0)® in a particular 𝑓 instance, but not in general. Now
to see how to shift to a geometric interpretation, first decompose into real
and imaginary parts,
𝑓 ( 𝑢) = 𝛼 𝑢 − 𝛽 𝑣, 𝑓 ( 𝑣) = 𝛽 𝑢 + 𝛼 𝑣
then find your eigen-bivector using the outermorphism,
𝑓 ( 𝑢 ∧ 𝑣) = (𝛼 𝑢 − 𝛽 𝑣) ∧ (𝛽 𝑢 + 𝛼 𝑣)
= (𝛼2 + 𝛽 2 ) ( 𝑢 ∧ 𝑣)
We thus see the eigen-bivector picks up the magnitude of the otherwise
complex eigenvalue. The phase of the complex eigenvalue specifies the
rotation of object in the 𝑢 ∧ 𝑣 plane. Overall we’ve decomposed the action
of 𝑓 into a dilation combined with a rotation. Well, that’s what your tra-
ditional linear algebra course taught you, only in 3 or so pages instead of
1 page.
Thus we’ve solved the problem of geometrically interpreting complex
eignvectors and complex eigenvalues. In our continuing theme, we again
find we do not need the complex numbers, except perhaps as an interme-
diate crutch.
TODO: Some more prep was needed here I think?
Spinor Helicity variables 81
Quantum Kinematics
The style for this chapter is to assume the reader has a background in
quantum mechanics and just wants to practise some problem solving or
grounded theory exploration. So I present minimal theory refreshers
when needed as Hints to selected problems. For GNU Free Document
contributors, each problem should be introduced with some motivation,
not just as a mathematical curiosity.
However, there is a running theme through all these notes in the orig-
inal version, which is the quest for a geometric understanding of quan-
tum mechanics through; (1) geometric algebra representations, and (b)
ER=EPR duality (wormholes are entanglement) which in the author’s
view provides us with a unification of QM with GR which is the reverse
of the orthodoxy, that is, through the geometric lens we will find general
relativity implies quantum mechanics and gravity is “already a quantum
theory.”
83
84 Quantum Kinematics
if you get a wave at any time, after a little time you will still have a wave.
Energy conservation. The wave in a free-field does not have anything
to interact with, so it goes nowhere except where it was first heading
for. This is why the famous “wave equations” can be derived from simply
writing down the law for energy conservation. You write down a Hamil-
tonian, then set it to the differential time evolution operator as Noether
told us to do.
What’s that you say? Who is this Noether? Noether’s results fol-
low from Lagrangian mechanics. The reason Noether’s principles hold
in quantum theory is because they are deep. You only need a least action
principle and some symmetries. Well, quantum theory has that.
The only real mystery here is twofold: (1) why waves? (2) Why the 𝑖?
Brief answers are:
(2) The mystery 𝑖 is concealing the fact that wave equation is for parti-
cles that have an orientation, so either a spin, or a preferred direc-
tion —like they could be extended objects, but extended objects can
rotate, so this all amounts to about the same thing; we need a way
to know how to rotate our measurement frame onto the orientation
frame of the elementary particle.
These two answers are naïvely in severe conflict. A particle, even if ex-
tended, like a string or brane or wormhole, can certainly rotate, so needs
some geometric factor like a pseudoscalar 𝑖 or a bivector 𝑖𝑒𝑘 , in it’s wave
equation. But fields do not rotate on the spot. So how can we use a wave?
The answer is given in the answer: the wave is not a real wave, it is de-
scribing propagation of information. Real information, but nonetheless
it is a theory of information. Why is QM a theory of information? Well,
this has nothing to do with weird It from Bit or Copenhagen nonsense,
in my opinion, quantum theory is about information because all physics
is about information, it’s just that quantum mechanics is a theory that
Probability and the Phase of the Wavefunction 85
I have always wanted to get deep into what the heck complex numbers
are doing in quantum mechanics as probability amplitudes? The thing
is, quantum mechanics looks superficially a lot like ordinary classical
statistical mechanics, but the role of superposition of the complex valued
probability amplitudes make all the difference in the world. People mar-
vel at the strange consequences and invent outlandish ideas like Many
Worlds Theory to try to explain these things, but I prefer to not worry
too much about the metaphysical consequences and worry instead about
why the amplitudes are so fundamental in the first place. So far I have
not been able to convince myself that Many Worlds Theory is the reason.
1 Classical
mechanics is not a theory of measurement. You can perform measure-
ments in CM, but CM is about real trajectories in phase space independent of any mea-
surement.
86 Quantum Kinematics
With all due respect to Roger Penrose and others, the role of complex
numbers in QM is vastly over-stated. The complex number field ℂ ap-
pears naturally in the real spacetime algebra of Clifford algebra. Since
fundamental particles are assumed to move around in spacetime there
should then be no mystery why complex numbers might appear. The field
ℂ is a subalgebra (the even subalgebra) of the Clifford algebra 𝐶ℓ 3 , and as
the algebra of rotors (generators of rotations) in the relativistic spacetime
algebra 𝐶ℓ 1, 3.
The rotors, of course, include the Lorentz rotations (boosts).
So why are quantum particles best represented by complex valued
wave-functions?
One answer has to be that we have historically just chosen a bad rep-
resentation. The wave-function is really a Clifford algebra rotor. The
algebraic dimension of the rotors is 2D, and the algebra is isomorphic
to the complex number algebra. In the Clifford (geometric) algebra 𝐶ℓ 3
the rotors are combinations of the scalars and bivectors, which square to
unity,
𝑅 𝑅˜ = 1.
* * *
I will now revert back to some standard QM to ease into some of the
mathematics. But I want the studnet in the back of their mind to be
always thinking of the wave function in the ℂ Hilbert space as really just
a dopey way of representing a normal spacetime spinor or rotor.
88 Quantum Kinematics
This next problem explores the relation between the real valued probabil-
ity density and the phase of the wavefunction, and we find the analogue
to the phase in classical mechanics. We will not gain a great mystical
insight into why our description of nature must use complex amplitudes
instead of straight-forward probabilities this way, but it is a good exercise
for developing some insight into what quantum mechanics is all about.
(b) (i) Use your continuity equation to identify the probability cur-
rent 𝑗 , (flow of probability) in terms of the wavefunction Ψ.
(ii) Then show that your result can be reduced to,
ℏ
𝜌= =m{Ψ∗ ∇Ψ} .
𝑚
Ψ = 𝜌 1/2 𝑒−𝑖𝑆/ℏ
(d) Derive an equation for the time derivative of the phase factor 𝑆,
and thus start exploring the physical meaning of the wavefunc-
tion phase 𝑆 by exploring the classical limit where ℏ → 0. A first
Probability and the Phase of the Wavefunction 89
𝑖ℏ 2 𝑖
= ∇ Ψ − 𝑉Ψ
2𝑚 ℏ
∗
𝜕Ψ 𝑖 ∗ 𝑖 𝑝2 ∗ ∗
and = 𝐻Ψ = Ψ + 𝑉Ψ
𝜕𝑡 ℏ ℏ 2𝑚
𝑖ℏ 2 ∗ 𝑖
=− ∇ Ψ + 𝑉 Ψ∗
2𝑚 ℏ
¤
Putting these expressions into 𝜌,
𝜕𝜌 𝑖ℏ ∗ 2 𝑖 𝑖ℏ 𝑖
∴ = Ψ ∇ Ψ − Ψ∗𝑉 Ψ − Ψ∇2 Ψ∗ + Ψ𝑉 Ψ∗
𝜕𝑡 2𝑚 ℏ 2𝑚 ℏ
and use the fact wavefunctions are just complex numbers, so they com-
mute with 𝑉 , so the terms in 𝑉 cancel, and we get,
𝜕𝜌 𝑖ℏ ∗ 2
= Ψ ∇ Ψ − Ψ∇2 Ψ∗
𝜕𝑡 2𝑚
From here you can probably finish up.
Solution for Problem 4.1.(a) Using the vector calculus result, we can
write,
Ψ∗ ∇2 Ψ = ∇ • Ψ∗ ∇Ψ − ∇Ψ∗ • ∇Ψ
and likewise for Ψ∇2 Ψ∗ , and noting ∇Ψ∗ • ∇Ψ = ∇Ψ • ∇Ψ∗ , we get,
𝜕𝜌 𝑖ℏ ∗ ∗
= ∇ Ψ ∇Ψ − ∇ Ψ∇Ψ
• •
𝜕𝑡 2𝑚
= −∇ • 𝑗
𝑖ℏ ∗
where 𝑗 ≡− Ψ ∇Ψ − Ψ∇Ψ∗ .
2𝑚
The continuity equation for particle probability density is thus,
𝜕𝜌
+ ∇ • 𝑗 = 0.
𝜕𝑡
That’s it. ,
Hints for Problem 4.1.(b) The solution to part (a) has done most of
the work, we have already identified the current 𝑗 in terms of Ψ. Since
Ψ is a complex valued function it has real and imaginary parts, and Ψ∗
is it’s complex conjugate. Then noting,
∗
Ψ∗ ∇Ψ = Ψ∇Ψ∗
and,
𝑧 − 𝑧∗ = 2𝑖 =m( 𝑧) , for any complex number 𝑧,
Probability and the Phase of the Wavefunction 91
Hints for Problem 4.1.(c) The polar form for Ψ can be taken as a def-
inition of the phase factor 𝑆,
Ψ = 𝜌 1/2 𝑒−𝑖𝑆/ℏ
we note firstly that 𝑆 has the dimensions of action (energy×time). So we
might expect a classical analogue for 𝑆 to be the classical action, which
is a functional integral of the particle Lagrangian. We want to see if this
is the case, basically. So this question is only a Lemma towards this goal.
Hint 3. But for this lemma we do not need the full continuity equation,
you should be able to squint your eyes and see that the imaginary part
of Ψ∗ ∇Ψ is all we need to obtain a relationship between 𝑗 and 𝑆.
Solution for Problem 4.1.(c) Start with Ψ = 𝜌 1/2 exp 𝑖𝑆/ℏ , and take
the gradient, using the product and chain rules for differentiation,
1 √ 𝑖
∇Ψ = √ ∇𝜌 + 𝜌 ∇ 𝑆 𝑒𝑖𝑆/ℏ
2 𝜌 ℏ
√
and since Ψ∗ = 𝜌 𝑒𝑖𝑆𝑡 ,
∇𝜌 𝑖
∴ Ψ∗ ∇Ψ = + 𝜌 ∇𝑆
2 ℏ
92 Quantum Kinematics
Hints for Problem 4.1.(d) The question hinted plugging Ψ(𝜌 , 𝑆) into
Schrödinger’s equation and then taking the semi-classical limit ℏ → 0.
So that’s what we will do.
and then using the Ψ(𝜌 , 𝑆) form, and the product and chain rules for
differentiation w.r.t. time 𝑡,
√
𝜕Ψ 1 𝑖 𝜌
= √ 𝜌¤ + 𝑆¤ 𝑒𝑖𝑆/ℏ
𝜕𝑡 2 𝜌 ℏ
√
√ 𝑖 𝜌
and ∇Ψ = ∇ 𝜌 + ∇ 𝑆 𝑒𝑖𝑆/ℏ
ℏ
2 2√ 𝑖 𝑖√ 2 1√ 2 𝑖𝑆/ℏ
∴ ∇ Ψ = ∇ 𝜌 + ∇𝜌∇ 𝑆 + 𝜌 ∇ 𝑆 − 2 𝜌(∇ 𝑆) 𝑒
ℏ ℏ ℏ
Hints for Problem 4.1.(d) Now just let ℏ → 0 in the above relation,
and you will see all we have left over classically is,
√
√ ¤ 𝜌 √
− 𝜌𝑆= (∇ 𝑆) 2 + 𝜌 𝑉
2𝑚
The Classical Action and Quantum Phase 93
1
∴ 𝑆¤ + (∇ 𝑆) 2 + 𝑉 = 0
2𝑚
which is precisely the Hamilton–Jacobi equation for a classical particle.
,
* * *
solutions. That ‘special thing’ is that the solutions minimize the action
between any two chosen points of the motion.
But note the Hamilton–Jacobi equation is a differential equation. So
it is a local dynamical differential equation, not global. How does this
arise form the global action minimization principle?
The way to turn the global action minimization principle into an in-
finitesimal calculus form is to first note that 𝑆 [𝑥 ( 𝑡)] can be viewed as a
function of the end points of the unique path which minimizes 𝑆,
𝑆 (𝑥 𝑖 , 𝑥 𝑓 , 𝑡 )
𝑆 = 𝑆 (𝑥 𝑓 , 𝑡 )
3. When you obtain the above expression for the propagator, and you
know the potential 𝑉 , you will still not know how to compute it as a
function of 𝑥 and 𝑡. The usual process is to guess a solution and show
that it solves Schrödinger’s equation. The best guess can be worked
out fairly simply by figuring out what a solution would be in the
classical limit as ℏ → 0. We find an action extremization principle,
Path Integrals and Phase 97
1. Waves Only. Abandon the idea particles are localized, and embrace
fully the wave function as reality. I reject this proposal because
every single experiment designed to detect systems described by a
wave function has found localized particles, not distributed energy
over all of space. We know waves can be localized, in the form of
Path Integrals and Phase 99
2. Fields Only. Particles are not real, non-localized fields are, but
whereas fields obey the laws of Schrödinger time evolution, when-
ever field energy is detected is coalesces into a localized quanta. I re-
ject this proposal because similarly to the waves-only picture it is far
too inelegant and post hoc. Why would fields magically coalesce into
lumps only when measured, but not at other unobserved times? It
is clear fields cannot be too lumpy in the unobserved times because
then the Feynman histories would not get sampled. Remember, we
are trying to take the path integral formalism seriously.
Some other notes: the Sum Over Histories formalism does not explic-
itly assume interference effects. Interference arises as a secondary result
of the sums of the complex phase factors. This is hard to identify as a lit-
eral wave interference. The phase factors are not particle probability
densities, they are just a mathematical component thereof. We can use
complex amplitudes or sinusoids to describe waves, but this is not how
the phase factors are motivated.
As we have seen, the phase factors are closely related to the classical
100 Quantum Kinematics
The main goal of early quantum theory was to unite QM with special
relativity (SR). The first obstacle was trying to find a Lorentz invariant
wave equation. Schrödinger actually found the Klein–Gordon equation
before the Schrödinger equation, but did not publish it, probably because
he knew or suspected it was not physical. We now know the Klein–Gordon
equation is kind of physical, but cannot describe the motion of a single
particle, rather, it can describe a massive spin-0 field. So it is a field equa-
tion, not a particle–wave equation. In the early years of QM physicists
had not realized quantum theory could be best expressed as a field the-
ory. So Schrödinger (unsuccessfully) and then later Dirac (successfully)
went in search of a relativistic wave equation using other tricks.
Probability Conservation 101
𝑖𝜕𝑡 Ψ = 𝐻 Ψ
becomes, −𝜕𝑡2 Ψ = ( 𝑝2 + 𝑚2 )Ψ. (Klein–Gordon)
Then the time derivative of the left-hand side must be identically zero.
Let’s see what it is?
ˆ ∞ ˆ +∞
𝜕𝑡 𝑑𝑥 𝜌( 𝑥, 𝑡) = 𝑑𝑥 𝜕𝑡 Ψ∗ Ψ + Ψ∗ 𝜕𝑡 Ψ
−∞ −∞
102 Quantum Kinematics
ˆ ∞
= 𝑑𝑥 (−𝑖𝜕𝑥2 Ψ∗ + Ψ∗ )Ψ + Ψ∗ ( 𝑖𝜕𝑥2 Ψ −
𝑖𝑉
𝑖𝑉
Ψ)
ˆ−∞
∞
= 𝑑𝑥 −𝑖𝜕𝑥2 Ψ∗ Ψ + 𝑖Ψ∗ 𝜕𝑥2 Ψ
h−∞ i∞
∗ ∗
= 𝑖 −𝜕𝑥 Ψ Ψ + Ψ 𝜕𝑥 Ψ
−∞
ˆ ∞
∗ ∗
−𝑖 𝑑𝑥 − 𝜕𝑥Ψ 𝜕𝑥 Ψ + Ψ
𝜕𝑥 𝜕𝑥 Ψ
−∞
h i∞
= 𝑖 −𝜕𝑥 Ψ∗ Ψ + Ψ∗ 𝜕𝑥 Ψ
−∞
= 0.
The boundary terms must of course vanish for physical reasons. You
could integrate by parts a second time and the boundary term would then
cancel identically, but it is already clear the whole integral vanishes. This
does not prove the probability density is unitary, it just shows that it is
constant. But that’s all we need for the present purposes. □
* * *
Note the caveat in the problem statement, “(unless you wish to toler-
ate a theory of physics where particles can arbitrarily vanish or appear).”
Well, maybe you already know that this is exactly what physicists have
indeed learned to tolerate! Quantum theory (more precisely, quantum
field theory, QFT) does not work relativistically unless we consider vir-
tual particles. The existence of virtual particles is basically the Feynman
diagram manifestation of quantum field theory. Feynman diagrams are a
particle picture, while the amplitudes of the field theory which the Feyn-
man diagrams are used for computing are what we get when the particles
are reinterpreted as “modes” of the fields.
The difference with the QFT picture and the Klein–Gordon wave-
equation problem is that in QFT the virtual particles or ghost field
modes must disappear when we integrate over all possible histories, and
we can tolerate this if we can make them field modes rather than real
particles. Whereas a Klein–Gordon particle that could appear or vanish
Probability Conservation 103
105
106 Kinematics and Amplituhedra
𝑆-matrix theory which was a failed attempt to do away with QFT (quan-
tum field theory) and just apply principles like the conservation laws plus
locality and unitarity to derive scattering amplitudes. The idea is that
we isolate a compact region of spacetime, send some stuff in, and observe
some stuff come out. Every time we repeat the experiment sending the
exact same stuff in, different stuff might come out, there is no consis-
tency in the scattering, we see only consistency in some abstract rules,
like momentum-energy conservation, angular momentum conservation
and unitarity. The thought occurs that maybe those abstract rules are
all there is to physical reality! This could be considered the essence of
quantum mechanics, though as we learn it is not the whole story, which
is why 𝑆-matrix theory fails. But for plenty of practical purposes this
idea of scattering processes being like a black box suffices to understand
the world form an elementary point of view.
What we know is that 𝑆-matrix theory is nevertheless very close to
being all there is no know about the world. The attempt to modernize
𝑆-matrix theory and tell the whole story about fundamental physics is
what the Amplituhedron program is all about.
The thing is, only tools one needs to do physics are the scattering am-
plitudes (or equivalently correlation functions). In this sense it is a darn
shame 𝑆-matrix theory does not work. Nature is telling us we need some-
thing like QFT or string theory to say a little bit more about the inner
workings of that abstract black box that is the scattering process, in order
to correctly calculate the scattering amplitudes.
Before we move on to describing the Amplituhedron program, it is
worth relating the 𝑆-matrix approach to the most recent advances in
theoretical physics and cosmology, namely the gauge/gravity dualities,
or AdS/CFT correspondence.
AdS/CFT, or gauge/gravity duality, tells us that in the fictional world of
anti-de Sitter spacetimes a consistent quantum field theory on the bound-
ary of the spacetime is dual to a pure gravity theory in the bulk with one
higher space dimension. Thus, if the bulk spacetime has a (3 + 1) signa-
ture then we would expect a 2 𝑑-quantum field theory to exist describing
the physics on the boundary. But we can add hidden space dimension by
either compactification (string theory) or decorating the space manifold
The S matrix origins 107
have to live on the boundary of our spacetime nor has access to the infor-
mation there (as we could in AdS) to do good physics. As far as fundamen-
tal particles are concerned, our position a few dozen metres distant from
a collider experiment may as well be at spatial infinity. So in everyday
terms, we are living on an effective boundary. We just have to deal with
the noise from nearby stuff that messes up the ideal approximation that
nothing can get in from our surrounding laboratory box here on Earth
to infect our collider experiments. We know how to do this, it is called
statistics.
Nevertheless, tell a pure theorist they have no exact theory and they
will be quite upset. You cannot stop them searching for an exact theory, it
is in their nature. The Amplituhedron idea is one way to conservatively
extend 𝑆-matrix theory so that we capture as much essential physics as
possible, in the spirit of Feynman, but without the painfulness of seem-
ingly endless Feynman diagram calculations.
When physicists look inside the black box of a scattering process they
pull out Feynman diagram methods. The Feynman method also has more
modern operator calculus formalism, which do the same thing only less
diagrammatically. It is of no use at all to argue about which method is
“more fundamental” because they are describing the exact same physics.
Nature does not care how we compute the amplitudes.
However, it is not so simple. To use the standard Feynman diagram
approach one assumes all manner of virtual particle process can occur,
one has to add terms in the path integral for all of them, even if they are
not real. The end result is a mess, and if you make no mistake you get
some approximation to some loop order for your scattering amplitude.
But amazingly it all tends to simplify with “miraculous” cancellations.
The reason why the Feynman diagram methods starts off as a horrible
mess is due to the Lagrangian treatment given. I have often considered
the Lagrangian formulation of QM to be the most natural and obvious
thing to do. It says nature explores all possible paths. However, because it
is used as a treatment of Hamiltonian time evolution it is not manifestly
Lorentz invariant. This means off-shell processes have to be included in
the perturbative expansion. When a particle is off-shell it is not propa-
The S matrix origins 109
Before you start to worry all your previous quantum field theory learning
gets thrown out the window, it is worth bearing in mind for the practic-
ing physicists employed say at CERN or Fermilab, QFT or Yang-Mills
fields theory is still the way to compute scattering amplitudes. It’s just
that over time the Amplituhedron on-shell methods are proving to be far
more efficient and gradually replacing a lot of the traditional computa-
tion methods.
It is also not hard to interpret every quantum particle theory (but not
any classical particle theory) as a field theory. In QM there is a non-
110 Kinematics and Amplituhedra
“If it turned out that some physical system could not be described by a
quantum field theory, it would be a sensation; if it turned out that the sys-
tem did not obey the rules of quantum mechanics and relativity, it would
be a cataclysm”
Weinberg is thus careful to note that he is not saying “the particles are
more fundamental than the fields,” — but while you can protest you can
protest too much. Weinberg did not know about the development of Am-
plituhedron methods at the time he wrote his QFT volumes, had he un-
derstood the amplituhedron he might well have capitulated to the “par-
ticles are more fundamental” point of view. Who knows. The Weinberg
that lives in my head certainly did.
We cannot, or morally should not, use both particle and field descriptions
at the same time. The great success of QFT is mainly in the gauge theory
of the fundamental forces. The basic way gauge theory works is that in
the field theory one (might) discover there is gauge redundancy, meaning
a gauge transformation can be applied to the field without changing the
physics, it should therefore not change the Lagrangian.
The S matrix origins 111
𝐴𝜇 ↦−→ 𝐴𝜇 + 𝑖𝜕𝜇 𝜃
then this retrieves an invariant Lagrangian. You can check for yourself.
In this case a symmetry in the fields (charged fermions) forced us to as-
sume nature has some other field (photons).
Note we do not really need the 𝑖 in the phase factor, that just expresses
the charged particle is a fermion (the 𝑖 is hiding a bivector in the proper
spacetime algebra). So a similar argument could be made for a charged
boson.
* * *
Speculative Ideas
This chapter collects some random thoughts that I know are probably all
wrong, they are the sorts of ideas you have when day-dreaming, maybe
one out of 100 turns out to be a good insight. I have by no means kept
all such ideas that spring to my mind, most are obviously flawed. The
remaining I think are crazy, absolutely crazy, but as we know the future
physics often looks absolutely crazy to the past, so I hope there are some
gems in here.
* * *
115
116 Speculative Ideas
This section is for main chapter contents that I either have not bothered
writing up or may never get to.
Haven’t had too many crazy ideas about the vacuum recently. With geon
theory I am biased towards thinking non-mystically, so I think vacuum
energy is just remnants of closed time-like curve traversals. This makes
it hard to compute a vacuum energy, but it makes it well-defined, since
there is an absolute flat Minkowski space to gauge here, whereas in or-
thodox QM there is no such thing since the Minkowski metric is not stable
to fluctuations in orthodox QM.
* * *
118 Speculative Ideas
as you can verify for yourself, all these square to +1, hence the positive
definitieness.
The other thing is that Schmeikal has colorspace defined only up to
scalar multiples. So the ideals generated say in the first colorspace,
121
122 BIBLIOGRAPHY
[11] J. Baez and J. Huerta, “The algebra of grand unified theories,” Bul-
letin of the American Mathematical Society, vol. 47, no. 3, pp. 483–
552, 2010.
[14] C. Furey, “Unified theory of ideals,” Physical Review D, vol. 86, no. 2,
p. 025024, 2012.
125
126 INDEX
Appendix A
https://fsf.org/
Preamble
127
128 GNU Free Documentation License
We have designed this License in order to use it for manuals for free
software, because free software needs free documentation: a free pro-
gram should come with manuals providing the same freedoms that the
software does. But this License is not limited to software manuals; it can
be used for any textual work, regardless of subject matter or whether it
is published as a printed book. We recommend this License principally
for works whose purpose is instruction or reference.
The “Cover Texts” are certain short passages of text that are listed,
as Front-Cover Texts or Back-Cover Texts, in the notice that says that
the Document is released under this License. A Front-Cover Text may
be at most 5 words, and a Back-Cover Text may be at most 25 words.
A “Transparent” copy of the Document means a machine-readable
copy, represented in a format whose specification is available to the gen-
eral public, that is suitable for revising the document straightforwardly
with generic text editors or (for images composed of pixels) generic paint
programs or (for drawings) some widely available drawing editor, and
that is suitable for input to text formatters or for automatic translation
to a variety of formats suitable for input to text formatters. A copy made
in an otherwise Transparent file format whose markup, or absence of
markup, has been arranged to thwart or discourage subsequent modifica-
tion by readers is not Transparent. An image format is not Transparent
if used for any substantial amount of text. A copy that is not “Transpar-
ent” is called “Opaque”.
Examples of suitable formats for Transparent copies include plain
ASCII without markup, Texinfo input format, LaTeX input format,
SGML or XML using a publicly available DTD, and standard-conforming
simple HTML, PostScript or PDF designed for human modification. Ex-
amples of transparent image formats include PNG, XCF and JPG.
Opaque formats include proprietary formats that can be read and edited
only by proprietary word processors, SGML or XML for which the DTD
and/or processing tools are not generally available, and the machine-
generated HTML, PostScript or PDF produced by some word processors
for output purposes only.
The “Title Page” means, for a printed book, the title page itself, plus
such following pages as are needed to hold, legibly, the material this Li-
cense requires to appear in the title page. For works in formats which
do not have any title page as such, “Title Page” means the text near the
most prominent appearance of the work’s title, preceding the beginning
of the body of the text.
The “publisher” means any person or entity that distributes copies
of the Document to the public.
A section “Entitled XYZ” means a named subunit of the Document
whose title either is precisely XYZ or contains XYZ in parentheses follow-
ing text that translates XYZ in another language. (Here XYZ stands for a
130 GNU Free Documentation License
2. VERBATIM COPYING
You may copy and distribute the Document in any medium, either com-
mercially or noncommercially, provided that this License, the copyright
notices, and the license notice saying this License applies to the Docu-
ment are reproduced in all copies, and that you add no other conditions
whatsoever to those of this License. You may not use technical measures
to obstruct or control the reading or further copying of the copies you
make or distribute. However, you may accept compensation in exchange
for copies. If you distribute a large enough number of copies you must
also follow the conditions in section 3.
You may also lend copies, under the same conditions stated above, and
you may publicly display copies.
3. COPYING IN QUANTITY
If you publish printed copies (or copies in media that commonly have
printed covers) of the Document, numbering more than 100, and the Doc-
ument’s license notice requires Cover Texts, you must enclose the copies
in covers that carry, clearly and legibly, all these Cover Texts: Front-
Cover Texts on the front cover, and Back-Cover Texts on the back cover.
Both covers must also clearly and legibly identify you as the publisher of
these copies. The front cover must present the full title with all words
of the title equally prominent and visible. You may add other material
on the covers in addition. Copying with changes limited to the covers, as
GNU Free Documentation License 131
long as they preserve the title of the Document and satisfy these condi-
tions, can be treated as verbatim copying in other respects.
If the required texts for either cover are too voluminous to fit legibly,
you should put the first ones listed (as many as fit reasonably) on the
actual cover, and continue the rest onto adjacent pages.
If you publish or distribute Opaque copies of the Document numbering
more than 100, you must either include a machine-readable Transpar-
ent copy along with each Opaque copy, or state in or with each Opaque
copy a computer-network location from which the general network-using
public has access to download using public-standard network protocols a
complete Transparent copy of the Document, free of added material. If
you use the latter option, you must take reasonably prudent steps, when
you begin distribution of Opaque copies in quantity, to ensure that this
Transparent copy will remain thus accessible at the stated location un-
til at least one year after the last time you distribute an Opaque copy
(directly or through your agents or retailers) of that edition to the public.
It is requested, but not required, that you contact the authors of the
Document well before redistributing any large number of copies, to give
them a chance to provide you with an updated version of the Document.
4. MODIFICATIONS
You may copy and distribute a Modified Version of the Document un-
der the conditions of sections 2 and 3 above, provided that you release the
Modified Version under precisely this License, with the Modified Version
filling the role of the Document, thus licensing distribution and modifica-
tion of the Modified Version to whoever possesses a copy of it. In addition,
you must do these things in the Modified Version:
A. Use in the Title Page (and on the covers, if any) a title distinct from
that of the Document, and from those of previous versions (which
should, if there were any, be listed in the History section of the
Document). You may use the same title as a previous version if the
original publisher of that version gives permission.
sion, together with at least five of the principal authors of the Doc-
ument (all of its principal authors, if it has fewer than five), unless
they release you from this requirement.
C. State on the Title page the name of the publisher of the Modified
Version, as the publisher.
I. Preserve the section Entitled “History”, Preserve its Title, and add
to it an item stating at least the title, year, new authors, and pub-
lisher of the Modified Version as given on the Title Page. If there is
no section Entitled “History” in the Document, create one stating
the title, year, authors, and publisher of the Document as given on
its Title Page, then add an item describing the Modified Version as
stated in the previous sentence.
5. COMBINING DOCUMENTS
134 GNU Free Documentation License
You may combine the Document with other documents released under
this License, under the terms defined in section 4 above for modified ver-
sions, provided that you include in the combination all of the Invariant
Sections of all of the original documents, unmodified, and list them all as
Invariant Sections of your combined work in its license notice, and that
you preserve all their Warranty Disclaimers.
The combined work need only contain one copy of this License, and
multiple identical Invariant Sections may be replaced with a single copy.
If there are multiple Invariant Sections with the same name but different
contents, make the title of each such section unique by adding at the end
of it, in parentheses, the name of the original author or publisher of that
section if known, or else a unique number. Make the same adjustment to
the section titles in the list of Invariant Sections in the license notice of
the combined work.
In the combination, you must combine any sections Entitled “History”
in the various original documents, forming one section Entitled “His-
tory”; likewise combine any sections Entitled “Acknowledgements”, and
any sections Entitled “Dedications”. You must delete all sections Entitled
“Endorsements”.
6. COLLECTIONS OF DOCUMENTS
You may make a collection consisting of the Document and other doc-
uments released under this License, and replace the individual copies of
this License in the various documents with a single copy that is included
in the collection, provided that you follow the rules of this License for
verbatim copying of each of the documents in all other respects.
You may extract a single document from such a collection, and dis-
tribute it individually under this License, provided you insert a copy of
this License into the extracted document, and follow this License in all
other respects regarding verbatim copying of that document.
8. TRANSLATION
9. TERMINATION
You may not copy, modify, sublicense, or distribute the Document ex-
cept as expressly provided under this License. Any attempt otherwise to
copy, modify, sublicense, or distribute it is void, and will automatically
terminate your rights under this License.
136 GNU Free Documentation License
However, if you cease all violation of this License, then your license
from a particular copyright holder is reinstated (a) provisionally, unless
and until the copyright holder explicitly and finally terminates your li-
cense, and (b) permanently, if the copyright holder fails to notify you of
the violation by some reasonable means prior to 60 days after the cessa-
tion.
Moreover, your license from a particular copyright holder is reinstated
permanently if the copyright holder notifies you of the violation by some
reasonable means, this is the first time you have received notice of vio-
lation of this License (for any work) from that copyright holder, and you
cure the violation prior to 30 days after your receipt of the notice.
Termination of your rights under this section does not terminate the
licenses of parties who have received copies or rights from you under
this License. If your rights have been terminated and not permanently
reinstated, receipt of a copy of some or all of the same material does not
give you any rights to use it.
11. RELICENSING
GNU Free Documentation License 137
If you have Invariant Sections without Cover Texts, or some other com-
bination of the three, merge those two alternatives to suit the situation.
If your document contains non-trivial examples of program code, we
recommend releasing these examples in parallel under your choice of free
software license, such as the GNU General Public License, to permit
their use in free software.