Qo) UT Rom al
Modern
Mathematical
Ae a
and Differential
Geometry
, — ‘This book provides an introduction to the major mathematical structures used in physics
today. It covers the concepts and techniques needed for topics such as group theory, Lie
algebras, topology, Hilbert spaces and differential geometry. Lmportant theories of physics
such as classical and quantum mechanics, thermodynamics, and special and general rela-
tivity are also developed in detail, and presented in the appropriate mathematical language.
The book is suitable for advanced undergraduate and beginning graduate students im
mathematical and theoretical physics. It includes numerous exercises and worked examples
to test the reader's understanding of the various concepts, as well as extending the themes
covered in the main text. The only prerequisites are clementary calculus and linear algebra.
No prior knowledge of group theory, abstract vector spaces or topology is required,
PerrrSZeKkeres received his Ph.D. from King’s College London in 1964, in the area
of general relativity, He subsequently held rescarch and teaching positions at Cornell
University, King's College and the University of Adelaide, where he stayed from 1971
till bis recent retirement. Currently he is 2 visiting research fellow at that mstitution. He is
well known internationally for his research in general relativity and cosmology, and has an
excellent reputation for his teaching and lecturing. ae
pew
S2é
PUBLISHED BY THE PRESS SYNDICATE OF THE UNIVERSITY OF CAMBRIDGE
The Pitt Building, Trumpington Street, Cambridge. United Kingdom
CAMBRIDGE UNIVERSITY PRESS
The Edinburgh Building, Cambndge CB2 2RU, UK
40 West 20th Street, New York, NY 10011-4211, USA
47 Willkarnstown Road, Port Melbourne, VIC 3207, Australia
Ruz de Alarcon 13, 28014 Madnd, Spain.
Dock House, The Waterfront, Cape Town 8001, South Affica
btlpy/www cambridge org
© P Sackeres 2004
“This book is in copyright. Subject to statutory exception
and to the provisions of relevant collective heensing agreements,
no reproduction of any part may take place without
the writen permission of Cambridge University Press
First published 2004
Printed in the United Kingdom at the University Press, Cambridge
Typeface Times New Roman 10/13 pt and Frutiger — Svsrem WIEXZe [rH]
A caratogue record for thus book ts available from the Brinsh Library
Library of Congress Cataloguing in Publication data
Szckeres, Peter, 1940-
‘A course in modern mathematical physics. groups. Hilbert space, and differential geometry / Peter Szekere's
poem
Includes bibliographical references and index.
ISBN 0521 629607 — ISBN 0521 53645 6 (pb.)
| Mathematical physics Title
QC20 S965 2004
530 15 -e22
2004045675
ISBN 0521 82960 7 hardback
ISBN 0 521 53645 6 paperback *
The publisher has used its best endeavours to ensure that the URLs for external websites referred to 1n this book
sare Correct and active at the time of poing to press However, the publisher has no responsibility for the websttes
and cen make no yuerantee that site will remain five or that the content is of will remain appropriate A Course in Modern
Mathematical Physics
Groups, Hilbert Space and Differential Geometry
Peter Szekeres
Formerly of University of Adelaide
CAMBRIDGE
INIVERSITY PRESS Contents
Preface
Acknowledgements
1 Sets and structures
1
1.2
13
14
Ls
16
17
Sets and logic
Subsets, unions and intersections of sets
Cartesian products and relations
Mappings
Infinite sets
Structures
Category theory
2 Groups
Elements of group theory
Transformation and permutation groups
Matrix groups
Homomorphisms and isomorphisms
Normal subgroups and factor groups
Group actions
Symmetry groups
3. Vector spaces
a1
32
a3
34
3.5
3.6
37
Rings and fields
Vector spaces
Vector space homomorphisms
Vector subspaces and quotient spaces
Bases of'a vector space
Summation convention and transformation of bases
Dual spaces
4 Linear operators and matrices
4l
42
Eigenspaces and characteristic equations
Jordan canonical form
page ix
xiti
27
27
30
35
45
49
$2
59
59
60
63
72
BL
88
98
99
107 Contents
0
4.3 Linear ordinary differential equations
44 Introduction to group representation theory
Inner product spaces
5.1 Real inner product spaces
52 Complex inner product spaces
5.3 Representations of finite groups
Algebras
6.1 Algebras and ideals
62 Complex numbers and complex structures
63 Quaternions and Clifford algebras
64 — Grassmann algebras
65 Lie algebras and Lie groups
Tensors
7.1 Free vector spaces and tensor spaces
7.2. Multilinear maps and tensors
7.3. Basis representation of tensors
7.4 — Operations on tensors
Exterior algebra
8.1 Vectors and r-forms
8.2 Basis representation of r-vectors
83 Exterior product
84° Interior product
85° Oriented vector spaces
8.6 The Hodge dual
Special relativity
9.1 Minkowski space-time
9.2 Relativistic kinematics
9.3 Particle dynamics
94 — Electrodynamics
9.5 Conservation faws and energy. stress tensors
‘Topology
10.1 Euclidean topology
10.2 General topological spaces
10.3 Metric spaces
10.4 Induced topologies
10.5 Hausdorff spaces
10.6 Compact spaces
116
120
126
126
133
141
149
149
152
157
160
166
178
178
186
193
198
204
204
206
208
213
215
220
228
228
235
239
244
251
255
255
257
264
265
269
271 Contents
14
16
10.7 Connected spaces
10.8 Topological groups
10.9 Topological vector spaces
Measure theory and integration
11.1 Measurable spaces and functions
11.2. Measure spaces
11.3. Lebesgue mtegration
Distributions
12.1 Test functions and distributions
12.2 Operations on distributions
12.3. Fourier transforms
12.4 Green’s functions
Hilbert spaces :
13.1 Definitions and examples
13.2. Expansion theorems
13.3 Linear functionals
13.4 Bounded linear operators
13.5. Spectral theory
13.6 Unbounded operators
Quantum mechanics
14.1 Basic concepts
14.2 Quantum dynamics
14.3. Symmetry transformations
14.4 Quantum statistical mechanics
Differential geometry
15.1 Differcntiable manifolds
15.2 Differentiable maps and curves
15.3. Tangent, cotangent and tensor spaces
15.4 Tangent map and submanifolds
15.5 Commutators, flows and Lie derivatives
15.6 Distributions and Frobenius theorem
Differentiable forms
16.1 Differential forms and exterior derivative
16.2 Propertics of exterior derivative
16.3. Frobenius theorem: dual form
16.4 Thermodynamics
16.5 Classical mechanics
223
2716
279
287
287
292
301
308
309
314
320
323
330
330
335
341
344
351
357
366
379
387
397
410
41
4is
4i7
426
432
447
451
454
457
464
vil viii
Contents
7
18
19
Integration on manifolds
17.1 Partitions of unity
17.2 Integration of n-forms
17.3. Stokes’ theorem
17.4 Homology and cohomology
17.5. The Poincaré lemma
Connections and curvature
18.1 Linear connections and geodesics
182 Covariant derivative of tensor fields
183. Curvature and torsion
18.4 Pseudo-Riemannian manifolds
18.5 Equation of geodesic deviation
18.6. The Riemann tensor and its symmetries
18.7 Cartan formalism
188 General relativity
18.9 Cosmology
18.10 Variation principles in space-time
Lie groups and Lie algebras
19.1 Lie groups
19.2 The exponential map
19.3. Lie subgroups
19.4 Lie groups of transformations
19.5 Groups of isometries
Bibliography
Index
481
482
484
486
493
500
506
506
510
512
516
522
524
527
534
548
553
559
559.
564
569
572
578
587
589 Preface
Afier some twenty years of teaching different topics in the Department of Mathematical
Physics at the University of Adelaide I conceived the rather foolhardy project of putting
all my undergraduate notes together in one single volume under the ttle Mathematical
Physics. This undertaking turned out to be considerably more ambitious than I had originally
expected, and it was not until my recent retirement that | found the time to complete it
Over the years I have sometimes found myself in the midst of a vigorous and at times
quite acrimonious debate on the difference between theoretical and mathematical physics.
‘This book is symptomatic of the difference. 1 believe that mathematical physicists put the
mathematics first, while for theoretical physicists it is the physics which is uppermost. The
latter scck out those areas of mathematics for the use they may be put to, while the former
have a more unified view of the two disciplines don’t want to say one ts better than the
other — it is simply a different outlook. In the big scheme of things both have their place
but, as this book no doubt demonstrates. my personal preference 1s to view mathematical
physics as a branch of mathematics.
The classical texts on mathematical physics which | was originally brought up on, such
as Morse and Feshbach [7], Courant and Hilbert [1], and Jeffreys and Jeffreys [6] are es-
sentially books on differential equations and linear algebra. The flavour of the present book
is quite different. It follows much more the lines of Choquet-Bruhat, de -Morette and
Dillard-Bleick [14] and Geroch [3]. in which mathematical structures rather than mathemat-
ical analysis is the main thrust. Of these two books, the former is possibly a little daunting as
an introductory undergraduate text, while Geroch’s book, written in the author's inimitably
delightful lecturing style, has occasional tendencies to overabstraction. | resolved therefore
to write a book which covers the material of these texts, assumes no more mathematical
knowledge than clementary calculus and linear algebra, and demonstrates clearly how theo-
ries of modern physics fit into various mathematical structures. How well | have succeeded
must be feft to the reader to judge
At times | have been caught by surprise at the natural development of ideas in this book.
For example, how is it that quantum mechanics appears before classical mechanics? The
reason 18 certainly not on historical grounds. In the natural organization of mathematical
ideas, algebraic structures appear before gcometrical or topological structures, and linear
structures are evidently simpler than nom-tinear. From the point of view of mathemaucal
simplicity quantum mechanics, being a purely linear theory in a quasi-alebraic space
(Hilbert space), is more elementary than classical mechanics, which can be expressed in
ix Preface
terms of non-linear dynamical systems in differential zcometry. Yet, there is something
of a paradox here, for as Niels Bohr remarked: ‘Anyone who is not shocked by quantum
mechanics does not understand it”, Quantum mechanics is not a difficult theory to express
mathematically, but it is almost impossible to make epistomological sense of it. | will not
even attempt to answer these sorts of questions, and the reader must look elsewhere for a
discussion of quantum measurement theory [5]
Every book has its limitations. At some point the author must call it a day, and the
omissions in this book may prove’a disappointment to some readers, Some of them are
a disappointment to me. Those wanting to go further might explore the theory of fibre
bundles and gauge theories [2, 8, 13], as the stage is perfectly set for this subject by the end
of the book. To many, the biggest omission may be the lack of any discussion of quantum
field theory. This, however, 1s an area that seems to have an entirely different flavour to the
rest of physics as its mathematics is difficult if nigh on impossible to make rigorous. Even
quantum mechanics has a ‘classical’ flavour by comparison. It is such a huge subject that I
felt daunted to even begin it. The reader can only be dirceted to a number of suitable books
to introduce them to this field [10-14].
Structure of the book
This book is essentially in two parts, modern algebra and geometry (including topology).
The carly chapters begin with set theory, group thcory and vector spaces, then move to more
advanced topics such as Lie algebras, tensors and exterior algebra. Occasionally ideas fron)
group representation theory are discussed. If calculus appears in these chapters it 1s of an
elementary kind. At the end of this algebraic part of the book, there is included a chapter
‘on special relativity (Chapter 9), as it seems a nice example of much of the algebra that has
gone before while introducing some notions from topology and calculus to be developed in
the remaining chapters. I have treated it as a kind of crossroads: Minkowski space acts as 5
link between algebraic and geometric structures, while at the same time it is the first place
where physics and mathematics are seen to interact in a significant way.
In the second part of the book, we discuss structures that are essentially geometrical
in character, but generally have an algebraic component as well. Beginning with topology
(Chapter 10), structures are created that combine both algebra and the concept of continuity,
The first of these is Hilbert space (Chapter 13), which is followed by a chapter on quantum,
mechanics. Chapters on measure theory (Chapter 11) and distribution theory (Chapter 12)
precede these two, The final chapters (15-19) deal with differential geometry and examples
‘of physical theories using: manifold theory as their setting — thermodynamics, classical
mechanics, general relativity and cosmology. A flow diagram showing roughly how the
chapters interlink is given below.
Exercises and problems are interspersed throughout the text. The exercises are not de-
signed to be difficult — their aim is either to test the reader's understanding of a concept
just defined or to complete a proof needing one or two more steps. The problems at ends
of sections are more challenging. Frequently they are in many parts, taking up a thread Preface
oh
el
of thought and running with it. This way most closely resembles true research, and is my
preferred way of presenting problems rather than the short one-liners often found in text
books. Throughout the book, newly defined concepts are written in bold type. If a con-
cept is written in italics, it has been introduced in name only and has yet to be defined
properly.
References
{1 R. Courant and D. Hilbert. Methoels of Mathematical Physics, vols | and 2. New York
Interscience, 1953.
[2] T. Frankel. The Geometry of Physics. New York, Cambridge University Press, 1997.
[3] R. Geroch. Mathematical Physics. Chicago, The University of Chicago Press, 1985.
[4] J. Glimm and A. Jaffe. Quantum Physics: A Functional Integral Point of View. New
York, Springer-Verlag, 1981.
[5] J. M. Jauch. Foundations of Quantum Mechanics. Reading, Mass., Addison-Wesley.
1968
[6] H. J. Jeffreys and B. S. Jeffreys. Methods of Mathematical Physics. Cambridge,
Cambridge University Press, 1946.
[7] P.M. Morse and H. Feshbach. Methods of Theoretical Physics, vols 1 and 2. New York,
McGraw-Hill, 1953
[8] C. Nash and S. Sen. Topology and Geometry for Physicists. London, Academic Press,
1983.
[9] P. Ramond. Field Theory: A Modern Primer. Reading, Mass., Benjamin/Cummings,
1981.
[10] L. H. Ryder, Quantum Field Theory. Cambridge, Cambridge University Press, 1985.
[11] S.S. Schweber. An Introduction to Relativisttc Quantum Field Theory. New York,
Harper and Row, 1961.
xi xii
Preface
U2] R. F Streater and A. $. Wi
W.A Benjamin, 1964,
U3] A. Trautman. Fibre bundles associated with space-
Physics, 1:29-62, 1970,
[14] C de Witt-Morette, Y. Choquet-Bruhat and M. Dillard-Bleick. Analysis, Manifolds
and Physics. Amsterdam, North-Holland, 1977
‘ghtman. PCT. Spin and Statistics, and All That. New York, Acknowledgements
‘There are an cnormous number of people | would like to express my gratitude to, but 1
will single out just a few of the most significant. Firstly, my father George Szekeres, who
introduced me at an early age to the wonderful world of mathematics and has continued
to challenge me throughout my life with his doubts and criticisms of the way physics
(particularly quantum theory) is structured. My Ph.D. supervisor Felix Pirani was the first to
give mean inkling of the importance of differential geometry in mathematical physics, while
others who had an enormous influence on my education and outlook were Roger Penrose,
Bob Geroch, Brandon Carter, Andrzej Trautman, Ray McLenaghan, George Ellis, Bert
Green, Angas Hurst, Sue Scott, David Wiltshire, David Hartley, Paul Davies, Robin Tucker,
Alan Carey, and Michael Fastwood, Finally, my wife Angela has not only been an endless
source of encouragement and support, but often applied her much valued eritical faculties
to my manner of expression. T would also like to pay a special tribute to Patrick Fitzhenry
for his invaluable assistance in preparing diagrams and guiding me through some of the
nightmare that is today’s computer technology.
xiii Sets and structures
The object of mathematical physics is to describe the physical world in purely mathemat-
ical terms. Although it had its origins in the science of ancient Greece, with the work of
Archimedes, Buclid and Aristotle, it was not until the discoveries of Galileo and Newton that
mathematical physics as we know it today had its true beginnings. Newton's discovery of
the calculus and its application to physics was undoubtedly the defining moment. This was
built upon by generations of brilliant mathematicians such as Euler, Lagrange, Hamilton
and Gauss, who essentially formulated physical law in terms of differential equations, With
the advent of new and unintuitive theories such as relativity and quantum mechanies in the
twentieth century, the reliance on mathematics moved to increasingly recondite areas such
as abstract algebra, topology, functional analysis and differential geometry. Even classical
areas such as the mechanics of Lagrange and Hamilton, as well as classical thermody-
namics, can be lifted almost directly mto the language of moder differential geometry.
‘Today, the emphasis is often more structural than analytical, and it is commonly believed
that finding the right mathematical structure is the most important aspect of any physical
theory. Analysis, of the consequences of theorics, still has a part to play in mathematical
physics — indeed, most research is of this nature — but it is possibly less Fundamental in the
total overview of the subject.
When we consider the significant achievements of mathematical physics, one cannot help
but wonder why the workings of the universe are expressable at all by rigid mathematical
‘Jaws’. Furthermore, how is it that purely human constructs, in the form of deep and subtle
mathematical structures refined over centuries of thought, have any relevance at all? The
nineteenth century view of a clockwork universe regulated deterministically by differential
equations seems now tu have been banished for ever, both through the fundamental appear
ance of probabilities i quantum mechanics and the indeterminism associated with chaotic
systems. These two aspects of physical law. the deterministic and indeterministic, seem to
interplay in some astonishing ways, the impact of which has yet to be fully appreciated. [tis
this interplay, however, that almost certainly gives our world its richness and variety. Some
of these questions and challenges may be fundamentally unanswerable, but the fact remains
that mathematics seems to be the correct path to understanding the physical world.
The aim of this book is to present the basic mathematical structures used in our subject,
and to express some of the most important theories of physics in their appropriate mathe-
matical setting, It isa book designed chiefly for students of physics who have the need for a
more rigorous mathematical education. A basic knowledge of calculus and linear algebra,
including matrix theory, is assumed throughout, but little else. While different students will
1 1.41
Sets and structures
of course come to this book with different levels of mathematical sophistication, the reader
should be able to determine exactly what they can skip and where they must take pause
Mathematicians, for example, may be interested only in the later chapters, where various
theories of physics are expressed in mathematical terms. These theories will not, however,
be developed at great length, and their consequences will only be dealt with by way of a
few examples.
‘The most fundamental notion in mathematics is that ofa set, or ‘collection of objects’.
‘The subject of this chapter is ser sheory — the branch of mathematics devoted to the study
of sets as abstract objects in their own right. It turns out that every mathematical structure
consists of a collection of sets together with some defining relations. Furthermore, as we
shall see in Section 1,3, such relations are themselves defined in terms of sets. It is thus a
commonly adopted viewpoint that all of mathematics reduces essentially to statements in
set theory, and this is the motivation for starting with a chapter on such a basic topic.
‘The idea of sets as collections of olyects has anon-rigorous, or “naive” quality, although it
is the form in which most students are introduced to the subject [1-4]. Early in the twentieth
century, it was discovered by Bertrand Russell that there are inherent self-contradictions
and paradoxes in overly simple versions of set theory. Although of concern to logicians and
those mathematicians demanding a totally rigorous basis to their subject, these paradoxes
usually involve inordinately large self-referential sets — not the sort of constructs likely to
occur in physical contexts. Thus, while special models of set theory have been designed
to avoid contradictions, they: generally have somewhat artificial attributes and naive set
theory should suffice for our purposes. The reader's attention should be drawn, however,
to the remarks at the end of Section 1.5 concerning the possible relevance of fundamental
problems of set theory to physics. These problems, while not of overwhelming concern,
may at least provide some food for thought.
While a basic familiarity with set theory will be assumed throughout this book, it never:
theless scems worthwhile to go over the fundamentals, if only for the sake of completeness
and to establish a few conventions. Many physicists do not have a good grounding in set
theory, and should find this chapter a useful exercise in developing the kind of rigorous
thinking needed for mathematical physics. For mathematicians this is all bread and butter,
and if you fee! the material of this chapter is well-worn ground, please feel free to pass on
quickly.
Sets and logic
There are essentially two ways in which we can think of a set S. Firstly, it can be regarded
as a collection of mathematical objects a, 6, ... , called constants, written
S=(ab,
‘The constants a, b, ... may themselves be sets and, indeed, some formulations of set theory
require them to be sets. Physicists in general prefer to avoid this formal nicety, and find it
much more natural to allow for ‘atomic’ objects, as it is hard to think of quantities such as
temperature or velocity as being ‘sets’. However, to think of sets as consisting of lists of 1.1. Sets and logic
objects is only suitable for finite or at most countably infinite sets. [f we try putting the real
numbers into a list we encounter the Cantor diagonalization problem — see Theorems 1.4
and 1.5 of Section 1.5.
‘The second approach to set theory is much more general in character. Let P(x) be a
logical proposition involving a variable x. Any such proposition symbolically defines a set
S= {x | P(x),
which can be thought of as symbolically representing the collection of all x for which the
proposition P(x) is true. We will not attempt a full definition of the concept of logical
proposition here — this is the business of formal logic and ts only of peripheral interest to
theoretical physicists — but some comments are in order. ntially, logical propositions are
statements made up from an alphabet of symbols, some of which are termed constants and
some of which are called variables, together with logical connectives such as not, and, or
and implies, to be manipulated according to rulcs of standard logic. Instead of *P imphes
O° we frequently use the words “if P then Q” or the symbolic representation P => Q. The
statement ‘P if and only if Q", or *P iff Q’, symbolically written P Q. 15a shorthand
for
(P — Q) and (QO > P),
and signifies logical equivalence of the propositions P and Q. The two quantifiers V and
4, said for all and there exists, respectively, make their appearance in the following way: if
P(x) is @ proposition involving a variable v, then
vx(PQx)) and 3x(POr))
are propositions.
Mathematical theories such as set theory, group theory, etc. waditionally involve the
introduction of some new symbols with which to generate further logical propositions.
The theory must be complemented by a collection of logical propositions called axioms for
the theory — statements that are taken to be automatically true in the theory. All other true
statements should in principle follow by the rules of logic.
Set theory involves the introduction of the new phrase is a set and new symbols
{... | ...} and © defined by:
(Setl) If S is any constant or variable then *S is a set’ is a logical proposition.
(Set2) If P(x) is a logical proposition involving a variable x then (x | P(x)} 182 set.
(Se13) If S is a set and a is any constant or variable then @ € S is a logical proposition,
for which we say belongs to S or a is a member of S, or simply a is in S. The
negative of this proposition is denoted a ¢ S—said a is not in S.
These statements say nothing about whether the various propositions are true or false — they
merely assert what are “grammatically correct” propositions in set theory. They merely tell
us how the new symbols and phrases are to be used in a grammatically correct fashion. The
main axiom of set theory is: if P(x) is any logical proposition depending on a variable x,
3 Sets and structures
then for any constant or variable @
ae {x| P(x)} & Pla).
y mathematical theory uses the equality symbol = to express the identity of math-
ematical objects in the theory. In some cases the concept of mathematical identity needs a
separate definition. For example equality of sets 4 — B is defined through the axiom of
extensionality:
Two sets A and B are equal if and only if they contain the same members. Expressed
symbolically,
A=B @ Yaae A eae B).
A finite set A = (a), a2, ..., a4} is equivalent to
A= {x|(x =a)) or(x =a) or... or (x =a,)}-
A set consisting of just one element a is called a singleton and should be written as {a} to
distinguish it from the clement a which belongs to it: (a) = [x |x =a).
As remarked above, sets can be members of other sets. A set whose elements are all sets
themselves will often be called a collection or family of sets. Such collections are often
denoted by script letters such as A, 24, etc. Frequently a family of sets has its members
indexed by another set /,, called the indexing set, and is written
U={U, lie TN,
For a finite family we usually take the indexing, set to be the first » natural mumbers,
1 ={1.2,...,n}. Strictly speaking, this set must also be given an axiomatic definition
such as Peano’ axioms. We refer the interested reader to texts such as [4] for a discussion
of these matters. .
Although the finer details of logic have been omitted here, essentially all concepts of set
theory can be constructed from these basics. The implication is that all of mathematics can
be built out of an alphabet for constants and variables, parentheses (. . . ) logical connectives
and quantifiers together with the rules of propositional logic, and the symbols {... | ..-]
and €. Since mathematical physies is an attempt to express physics in purely mathematical
language, we have the somewhat astonishing implication that all of physics should also
be reducible to these simple terms. Eugene Wigner has expressed wonderment at this idea
in a famous paper entitled The unreasonable effectiveness of mathematics in the natural
sciences [5].
‘The presentation of set theory given here should suflice for all practical purposes, but it
1s not without logical difficulties. The most famous is Russell's paradex: consider the set of
all sets which are not members of themselves. According to the above rules this set can be
written R = (AA ¢ A}. Is R a member of itself? This question does not appear to have
an answer. For, if R € R then by definition R ¢ R, which is a contradiction. On the other
hand, if & ¢ R then it satisfies the criterion required for membership of R; that is, R € R. 1.2
1.2 Subsets, unions and intersections of sets
To avoid such vicious arguments, logicians have been forced to reformulate the axioms of
set theory in a very careful way. The most frequently used system is the axiomatic scheme
of Zermelo and Fraenkel — sce, for example, |2] or the Appendix of [6]. We will adopt the
‘naive’ position and simply assume that the sets dealt with in this book do not exhibit the
self-contradictions of Russell's monster.
Subsets, unions and intersections of sets
A set T is said to be a subset of S, or 7 is contained in S, if every member of 7 belongs
to S, Symbolically, this is written 7 CS,
TCS iff aceToaeS.
We may also say S is a superset of 7 and write S > 7. Of particular importance is the
empty set (J, to which no object belongs,
* Va (a € 9),
The empty set is assumed to be a subset of any set whatsoever,
VS@ S).
This is the default position, consistent with the fact that a © ) —> @ © S. since there are no
a such that @ € # and the left-hand side of the implication is never true. We have here an
example of the logical dictum that ‘a false statement implies the truth of any statement’,
‘A common criterion for showing the equality of two sets, 7 = S, is to show that T CS
and S © T. The proof follows from the axiom of extensionality:
T=Se5 eT eacS)
(ae TS ac Sjand(acS act)
<> (7 Syand (SCT),
Exercive: Show that the empty set is unique; Le. sf" 18. an empty set then = 4.
The collection ofall subsets of a set 5 forms a set in its own right, called the power set
of S, denoted 2°.
Example 1.1 If S isa finite set consisting of n elements, then 2° consists of one empty
set # having no elements, 77 singleton sets having just one member, (') sets having two
clements, ete. Hence the total number of sets belonging to 2° is, by the binomial theorem,
vf) G)eo()eeen
This motivates the symbolic representation of the power set,
Unions and intersections
‘The union of two sets Sand T, denoted SU T, is defined as
SUT ={xlx cSorx eT). Sets and structures
The intersection of two sets Sand 7, denoted $9 T, is defined as
SOT =(x|x © Sandx <7)
‘Two sets S and 7 are called disjoint if no clement belongs simultaneously to both Sets,
SOT =. The difference of two sets S and 7 1s defined as
S-T = |x © Sandx ¢ 7).
Exercise: WS and T are disjoint, show that §— T = $.
‘The union of an arbitrary (possibly infinite) family of sets .A is defined as the set of all
elements x that belong to some member of the family,
UA = te [AS such that ( € Ajand (x € $)).
Similarly we define the intersection of S to be the set of all elements that belong to every
set of the collection,
A= tive S forall Se A)
When A consists of a family of sets S; indexed by a set 7, the union and intersection are
frequently written
Utsi and O51
er rel
Problems
Problem 1.1 Show the distributive laws
ACBUC)=(AN BANE), AU(BNC)=(AU BN (AUC)
Problem 1.2. If B = (B,|/ <2) is.any family of sets, show that
4nU8=Ulangien AUP B=(i4u Bier)
Problem 13 Let B be any set. Show that (49 B)UC = 4 (BUC) iPand only if cA
Problem 1.4 Show that
A-BUQ)=(A- BN(A-O, A-(BNO =(4~ BUA -C)
Problem 15 If 8 = (8,{¢ € 1) is any family of sets, show that
4-U8=Ul4 Bien
Problem 1.6 If and F are any sets, prove the identities
2nara2t aug co
Problem 1.7 Show that if C is any family of sets then
(2 =29 Uarcaue 13
1.3. Cartesian products and relations
Cartesian products and relations
Ordered pairs and cartesian products
As it stands, there is no concept of order in a set consisting of two clements, since {a, 6) =
{h, a}. Frequently we wish to refer to an ordered pair (a, 6). Essentially this is a set of two
elements (a. 5} where we specify the order in which the two clements are to be writien
‘A purely set-theoretical way of expressing this idea is to adjoin the clement @ that is to
be regarded as the “first’ member. An ordered pair (a, b) can thus be thought of as a set
consisting of {a, b} together with the element a singled out as bemg the first,
(a,b) = {la, 5}, a} ay
While this looksa little artificial at first, it docs demonstrate how the concept of ‘order* can be
defined in purely set-theoretical terms. Thankfully, we only give this definition for illustrative
purposes - there is essentially no need to refer again to the formal representation (1.1).
‘rom the definition (1+1) show that (a,b) = (a’, b') tffa = a’ and b = bt
Exercis
Similarly, an ordered n-tuple (a), a2, .... @,.) isasetin which the order of the elements
must be specified. This can be defined inductively as
a a CO
Exercise: Write out the ordered triple (a,b, €) a8 a set
The (cartesian) product of two sets, S x T, is the set of all ordered pairs (s, 1) where s
belongs to S and ¢ belongs to 7,
Sx T=((,1)\s © Sandre T}
The product of n sets is defined as
Sy XS. x 0+ XS, = US1, 52, ---, Su) $1 © S182 © Sr, 00 Sn © So}
If the 7 sets are equal, S; = S = --- = S, = S, then their product is denoted 5".
Exercise: Show that Sx T = #iff S=Wor T=.
Relations
Any subset of S” is called an n-ary relation on a set S. For example,
unary relation = L-ary relation = subset of S
binary relation = 2-ary relation = subset of S? = S x S
ternary relation = 3-ary relation = subset of S' = § x S x S, ete.
We will focus atiention on binary relations as these are by far the most important. If
RCS S isa binary relation on S, it is common to use the notation ab in place of
(a,b) eR. Sets and structures
‘Some commonly used terms describing relations are the following:
R is said to be a reflexive relation if @ Ra for alla € S.
R is called symmetric if «Rb => bRa for alla, hb € S.
Ris transitive if (aRb and bRe) => aRe for alla, b,c € S.
Example 1.2 Let be the set of all real numbers. The usual ordering of real numbers is a
relation on I, denoted v < y, which is both reflexive and transitive but not symmetric. The
relation of strict ordering x <_y is transitive, but is neither reflexive nor symmettic. Similar
statements apply for the ordering on subsets of 2, such as the integers or rational numbers
‘The notation x < y is invariably used for this relation in place of the rather odd-looking
(x, ») © < where a =h.
The ordering = on real numbers has the further special property of being a total order,
by which it is meant that for every pair of real numbers x and , we have either ¥ < y or
yen
Example 15 The power set 2° of a set S is partially ordered by the relation of set in-
clusion C,
UCU forallU ES,
UCVad¥ CW => UCH,
UG Vand V CU => U=¥.
Unlike the ordering of real numbers, this ordering is of in general a total order.
A set S together with a partial order < is called a partially ordered set or more briefly
a poset. This is an example of'a structured set, The words “together with’ used here are a
rather casual type of mathspeak commonly used to describe a set with an imposed structure.
Technically more correct is the definition of a poset as an ordered pair,
where <© S x S satisfies the axioms of a partial order. The concept of a poset could
be totally reduced to its set-theoretical elements by writing ordered pans (s, ¢) as sets of
the form [{s, t}. s}, etc., but this uninstructive task would only serve to demonstrate how
simple mathematical concepts can be made totally obscure by overzealous use of abstract
definitions. 1.4
10
Sets and structures
Problems
Problem 1.8 Show the following identities:
(AUB) x P=(A x P)U(B x P).
(ANB) x (PQ) = (A x PYM x P),
(A— B)x P=(4 x P)—(B* P)
Problem 1.9 If A= (4, |¢ © JJ and B = (B, | j ¢ J} are any two families of sets then
UAxUs8= U 4x8,
icles
NAxNB= 1 4x4,
tehuer
Problem 110 Show that both the following. two relations:
(a,b) <(x,») tf a a Yvis a
subset of Xx ¥ such that for every x © X there is a unique y © Y for which (x, ») ¢g
By unique we mean
(ey) € gand (x, y') eg => y
Mappings are also called functions or maps. It is most common to write y= g(x) for
(. 9) € g. Whenever x) it is said that x is mapped to y, written y sx >».
Inelementary mathematics it is common to referto the subset y C ¥ x ¥ as representing
the graph of the function y. Our definition essentially identifies a function with its graph
‘The set X’is called the domain of the mapping g, and the subset y(X) CY defined by
AX) = ty EV ly = gfx), x CX}
is called its range.
Let U be any subset of ¥. The inverse image of U is defined to be the set of all points
of X that are mapped by g into U, denoted
o'W)= tr € Xe) € UI).
This concept makes sense even when the inverse map y~' does not exist. ‘The notation
@ '(U) isto be regarded as one entire symbol for the inverse image set, and should not be
broken into component parts. 1.4 Mappings
E> E be the standard sine function on the real numbers I. The
(0, +b, 427, +37, ...}, while the inverse image of 2 1s
Example 1.6 Let s
inverse image of O is sin '(0)
the empty set, sin '(2) = 0.
An n-ary function from X to ¥ is a function g : X"” > Y. In this case we write y
YO), X2.-.-, Xn) for (O44, x2, 4). ») € gy and say that y hast arguments in the set S,
although strictly speaking it has just one argument from the product set 4° = X x xX.
Itis possible to generalize this concept even further and consider maps whose domain
is.a product of possibly different sets,
gi Xx Xp x xX > ¥.
Important maps of this type are the projection maps
pr 2X x Xp x Xy > X
defined by
PH, = 1x2. Kn) Me
Ify : ¥—> ¥ and y : ¥ > Z, the composition map yy © y : X -» Z is defined by
v oe) = We).
‘Composition of maps satisfies the associative law:
ac(hey)=(@owey
where o: Z -> HW’, since for any x c X
20 (Fo PX) = ACHGEN) = (@o HNL) = (eo We we).
Hence, there is no ambiguity in writing « © yy © for the composition of three maps.
Surjective, injective and bijective maps
A mapping y : X —> ¥ is said to be surjective or a surjection if its range is all of 7. More
simply, we say ¢ is a mapping of X onto Y if y(X) = ¥. It is said to be one-to-one or
injective, or an injection, if for every y © Y there is a unique x © X such that » = g(x);
that is,
(x) = G(X") => x =
A map ¢ that is injective and surjective, or equivalently one-to-one and onto, is called
bijective or a bijection. In this and only this case can one define the inverse map g! :
Y > X having the property
Vx EX.
g glx)
‘Two sets ¥ and ¥ are said to be in one-to-one correspondence with each other if there
exists a bijection y 2 X > ¥
Exercise: Show that ify : X > ¥ ts a biyection, then so is y', and that p(y '(x)) =x, Vr © X.
an} Sets and structures
A bijective map y : X > X from X onto itself is called a transformation of X. The
most trivial transformation of all 1s the identity map id, defined by
idyx)= x, Ve eX.
Note that this map can also be described as having a ‘diagonal graph’,
idy = {@,x)|x CXPO XXX.
Frercise: Show that for any map y | X > Y, id) oy =p oid.
When g ; X —+ Y is a bijection with inverse g~', then we can write
dy
etop=idy yo
If both @ and yy are bijections then so is ¥ oy, and its mverse is given by
Wogy'=e oy!
since
gloyopopay™
If U is any subset of X and y : X — Y is any map having domain X, then we define
the restriction of y to U as the map y|,, : U -> Y by g|,.0¢) = g(a) for all x ¢ U. The
restriction of the identity map
cidyog=y lop =idy.
% = ids, :UsXx
is referred to as the inclusion map for the subset U. The restriction of an arbitrary map y
to U is then its composition with the inclusion map,
vp =e 0%
Example 1.7 {f U 1s a subset of X, define a function yx: X¥ > {0,1}, called the
characteristic function of U,by -
0 ifx¢U,
rn {} ifreu.
Any function y : X > {0, 1} 1s evidently the characteristic function of the subset UC X
consisting of those points that are mapped to the value 1,
P=xXo where U=_y (tl),
‘Thus the power set 2" and the set ofall maps y : X — {0, 1} are in one-to-one correspon-
dence.
Example 1.8 Vet R be an equivalence relation on a set X. Define the canonical map
@ +X — X/R from X onto the factor space by
lx) = Erle, Vx eX.
It is easy to verify that this map is onto 15
1.5. Infinite sets
More generally, any map y : X > Y defines an equivalence relation R on X by a Rb iff
a) = ob). The equivalence classes defined by R are precisely the inverse images of the
singleton subsets of ¥,
X/R=lp ‘WDIY ET),
andthe map y : ¥ > X/R defined by y(y) = w '({y'}) is one-to-one, for if ¥r(y) = WO")
then y = y’ — pick any element x € (vy) = WO") and we must have g(x)
Infinite sets
A set S is said to be finite if there is a natural number #1 such that S is in one-to-one
correspondence with the set WV = {1, 2,3, ..., n} consisting of the first 2: natural numbers,
We call the cardinality of the set S, written 2 = Card(s
Example 1.9 For any two sets § and T the set of all maps y : S > T will be denoted
by 7°. Justification for this notation 1s provided by the fact that if S and 7° are both fi-
nite and s = Card($), ¢ = Card(7) then Card( 7S) = #*. In Example 1.7 it was shown that
for any set S, the power set 2° is in one-to-one correspondence with the set of charae~
teristic functions on (1, 2)>. As shown in Example 1.1, for a finite set S both sets have
cardinality 2°.
A set is said to be infinite if it is not finite. The concept of infinity is intuitively quite
difficult to grasp, but the mathematician Georg Cantor (1845-1918) showed that infinite sets
could be dealt with ina completely rigorous manner. He even succeeded in defining different
‘orders of infinity’ having a ‘ransfinite arithmetic that extended the ordinary arithmetic of
the natural numbers,
Countable sets
The lowest order of infinity is that belonging to the natural numbers. Any set S that is in
one-to-one Correspondence with the set of natural numbers I) = [1, 2, 3, ...] is said to be
countably infinite, or simply countable. The elements of S can then be displayed as a
sequence, 5, $3, s3.... on setting s; = f-'(i).
Example 1.10 ‘The set of all integers Z = (0, £1, +2,...} is countable, for the map
Sf: Z—> NS defined by ((0) = 1 and f(r) = 2n, fn) = 2n +1 forall n > Ois clearly
a bijection,
SO=L. (U)=2W EY =3, fA=4, f(-B=5,..-
Theorem 1.1 Every subset of a countable set is cither finite or countable.
Proof: Let S be a countable set and f: $ > Na bijection, such that f(s}) = 1. f(s2) =
2,... Suppose S' 1s an infinite subset of S. Let s{ be the first member of the sequence
S1,82,.++ that belongs to S’. Set s} to be the next member, ete. The map f” > S’ > 1
% 14
Sets and structures
(81; 41) (s1, to) (81, t3) . .
x oj
(s2, 1) (2, te) (sz, ts) a
fo
(83, t2) ° °
a
x :
. ° .
Figure 1.1 Product of two countable sets 1s COUntable
defined by
isa bijection from S10 N .
Theorem 1.2 he cartesian product ofan parr of countable sets is coumable
Corollary 1.35 The rational numbers Q form a countable set.
Proof: A rational number isa fraction 12/m where m isa natural number (positive integer)
and 7 48 an integer having no common factor with m. The rationals are therefore in one-to-
phe correspondence with a subset of the product set Z x By Example 1.10 and Theorem
12.2 x Nts a countable set. Hence the rational numbers @ are countable, a
In the set of real numbers ordered by the usual <
thatx
US such that £() € S, forall eI While correct for finite and countably infinite families of sets, the status of this axiom is much less clear for uncountable families. Cohen in fact showed that the axiom of choice was 1.6 1.6 Structures an independent axiom and was independent of the continuum hypothesis. It thus appears that there are a variety of alternative set theories with differing axiom schemes, and the real numbers have different properties in these alternative theories. Even though the real numbers are at the heart of most physical theories, no truly challenging problem for mathematical physics has arisen from these results, While the axiom of choice is certainly useful, its availability is probably not critical in physical applications. When used, it is often invoked ina slightly different form: Theorem 1.6 (Zorn’slemma) Let | P, <] hea partially ordered set (poset) with the prop- erty that every totally ordered subset is bounded above. Then P has a maximal element. Some words of explanation are in order here. Recall that a subset Q is totally ondered if forevery pair of elementsx, y’ < Qeitherx < yor y® has a higher cardinality than that of the real numbers by using a Cantor diagonal argument to show it cannot be put in one-to-one correspondence with R. Problem 1.14 If /': (0, 1] > Risa non-decreasing function such that /(0) = 0, f(1) = 1, show that the places at which f 1s not continuous form 2 countable subset of [0. 1] Structures Physical theories have two aspects, the static and the dynamic. The former refers to the general background in which the theory is set. For example, special relativity takes place in Minkowski space while quantum mechanics is set in Ulilbert space. These mathematical structures are, to use J. A. Wheeler's term, the ‘arena’ in which a physical system evolves; they are of two basic kinds, algebraic and geometric. In very broad terms, an algebraic structure is a set of binary relations imposed on a set, and ‘algebra’ consists of those results that can be achieved by formal manipulations using the rules of the given relations. By contrast, a geometric structure is postulated as a set of 7 18 Sets and structures relations on the power set of a set. The objects in a geometric structure can in some sense be * visusilized’ as opposed to being formally manipulated. Although mathematicians frequently divide themselves into “algebraists’ and *geometers’, these two kinds of structure interrelate inall kinds of interesting ways, and the distinction is generally difficult to maintain. Algebraic structures A (binary) law of composition on a set S is a binary map pi 8xSoS. For any pair of elements a, b © S there thus exit new element (a,b) € S called their product. The product is often simply denoted by ab, while at other times symbols such as a-b, ach, a+b. ax bh. ab, [a,b], etc. may be used, depending on the context. Most algebraic structures consist of a set $ together with one or more law's of composition defined on S. Sometimes more than one set ts involved and the law of composition may take a form such as p : Sx T —> S.A typical example is the case of a vector space, where there are two sets involved consisting of vectors and scalars respectively, and the law of composition is scalar multiplication (see Chapter 3). In principle we could allow laws of composition that are n-ary maps (#1 > 2), but such laws can always be thought of as families of binary maps. For example, a ternary map @ = S? —> S is equivalent to an indexed family of binary maps {@, |@ © S} where ¢, ; S? —> Sis defined by $,(b, c) = $(a, b,c). A law of composition is said to be commutative if ab = ba. This is ahvays assumed 10 be true for a composition denoted by the symbol |; that is. @ + b = b +a. The law of composition is associative if a(be) = (ah)c. This is true, for example, of matrix multipli cation or functional composition f 6 (go hy = (f © g) © h, butis not true of vector product a = b in ordinary three-dimensional vector calculus, ax (b xc) =(a.eyb —(abje F(a x b) xe. Example 1.12 A semigroup is a set S with an associative law of composition defined on it. It ts said to have an identity element if there exists an element e € S such that ea =ae=a, Wac. Semigroups are one of the simplest possible examples of an algebraic structure. The theory of semigroups is not particularly rich, and there is little written on their general theory, but particular examples have proved interesting (1) The positive integers NV form a commutative semigroup under the operation of addi- tion. Tf the number 0 is adjomed to this set it becomes a semigroup with identity e = 0, denoted I. (2)Amap > Sofaset S into itself is frequently called a diserete dynamical system ‘The successive iterates of the function f, namely F = {fi f*,.... f= fo"). form a commutative semigroup with functional iteration as the law of composition. If we inchide the identity map andset f = idy, the semigroup is called the evolution semigroup generated by the function /. denoted E ;. 1.6 Structures The map @ : 1) > Ey defined by (1) = f" preserves semigroup products, on +m)= f"r". Such a product-preserving map between two semigroups is called a homomorphism, If the homomorphism 1s a one-to-one map itis called a semigroup isomorphism. Two seru- groups that have an isomorphism between them are called isomorphic; to all intents and purposes they have the same semigroup structure. The map ¢ defined above need not be an i, — {2}, the real numbers excluding the number isomorphism. For example on the set § 2, define the function f : S—> S by 2x3 ¥=2 Simple algebra reveals that /(/(2)) = x, so that ? = ids. In this case E, is isomorphic with the residue class of integers modulo 2, defined in Example 1.3. {= (3) Allof mathematics can be expressed as a semigroup. For example, set theory is made up of finite strings of symbols such as{... | ...}. and, not, €,V, ete. anda countable collection of symbols for variables and constants, which may be denoted ¥), 2, ... Given two strings 7 and o2 made up of these symbols, it is possible to construct a new string 6,2, formed by concatenating the strings. The set of all possible such strings 1s. semigroup, where “product” 1s defined as string concatenation. Of course only some strings are logically meaningful, and are said to be well-formed. The rules for a well-formed string are straightforward to asare the rules for ‘universally valid statements’ and the rules of inference. Gédel’s famous incompleteness theorem states that if we include statements of ordinary arithmetic in the migroup then there are propositions P such that neither P nor its negation, not P, can be reached from the axioms by any sequence of logically allowable operations. In a sense, the truth of such statements is unknowable. Whether this remarkable theorem has any bearing ‘on theoretical physics has still to be determined. Geometric structures In its broadest terms, a geometric structure defines certain classes of subsets of Sas in some sense ‘acceptable’, together with rules concerning their intersections and unions. Alternatively, we can think of a geometric structure G on a set S as consisting of one or more subsets of 2°, satisfying certain properties. In this section we briefly discuss two examples: Euclidean geometry and topology. Example 1.13 Euctidean geometry concerns points (singletons), straight lines, triangles, circles, ete., all of which are subsets of the plane. There isa ‘visual’ quality of these concepts, even though they are idealizations of the ‘physical’ concepts of points and lines that must have size or thickness to be visible. The original formulation of plane geometry as set out in Book | of Euclid’: Elements would hardly pass muster by today’s criteria as a tigorous axiomatic system. For cxample, there is considerable confusion between definitions and undefined terms. Historically, however, it is the first systematic approach to an area of mathematics that turns out to be both axiomatic and interesting. 19 20 Sets and structures The undefined terms are point, line segment, line, angle, circle and relations such as incidence on, endpoint, length and congruence. Euclid’s five postulates are: 1. Every pair of points are on a unique line segment for which they are end points. 2. Every line segment can be extended to a unique line. 3. For every point 4 and positive number r there exists a unique circle having A as its centre and radius r, such that the line connecting every other point on the circle to 4 has length ». 4. All right angles are equal to one another. 5. Playfatr’s axiom: given any line £ and a point A not on €, there exists a unique line through A that does not intersect ( — said to be parallel to £ ‘The undefined terms can be defined as subsets of some basic set known as the Euclidean plane. Points are singletons, line segments and lines are subsets subject to Axioms | and 2, while the relation incidence on is interpreted as the relation of set-membership ©. An angle would be defined as a set (A, €;, £] consisting of a point and two lines on which it is incident. Postulates 1-3 and 5 seem fairly straightforward, but what are we to make of Postulate 4? Such inadequacies were tidied up by Hilbert in 1921. ‘The least ‘ebvious’ of Fuclid’s axioms is Postulate 5, whichis not mam/festly independent of the other axioms. The challenge posed by thts axiom was met in the nineteenth century by the mathematicians Bolyai (1802-1860), Lobachevsky (1793-1856), Gauss (1777-1855) and Riemann (1826-1866). With their work arose the concept of non-Euclidean geometry, which was eventually to be of cructal importance in Einstein's theory of gravitation known general relativity. see Chapter 18, Although often regarded as a product of pure thought, Euclidean geometry was in fact an attempt to classify logically the geometrical relations in the world around us. It can be regarded as one of the earliest exercises in mathematical physics. Einstein's general theory of relativity carried on this ancient tradition of unifying geometry and physics, a tradition that lives on today in other forms such as gauge theories and string theory. The discovery of analytic geometry by Rene Descartes (1596-1650) converted Euclidean geometry into algebraic language. The cartesian method is simply to define the Euclidean plane as R* = E x with a distance function d : R° = IR? E given by the Pythagorean formula (x,y). (a, 1) = Mr — P+ = vy (1.2) This theorem is central to the analytic version of Euclidean geometry ~ it underpms the whole Euclidean edifice. The generalization of Euclidean geometry to a space of arbitrary dimensions ” is immediate, by setting ‘The ramifications of Pythagoras’ theorem have revolutionized twentieth century physics in many ways, For example, Minkowski discovered that Einsfein’s special theory of relativity could be represented by a four-dimensional pserido-Enclidean geometry where time ts 1.6 Structures interpreted as the fourth dimension and a minus sign is introduced into Pythagoras” law. When gravitation ts present, Einstein proposed that Minkowski’s geometry must be ‘curved’, the pseudo-Fuclidean structure holding only focally’ at each point. A coniplex vector space having a natural generalization of the Pythagorean structure is known as a Hilbert space and forms the basis of quantum mechanics (see Chapters 13 and 14). It is remarkable to think that the two pillars of twentieth century physics, relativity and quantum theory, both have their basis in mathematical structures based on a theorem formulated by an eccentric mathematician over two and a half thousand years ago. Example 1.14 \n Chapter 10 we will meet the concept of a topology on a set S, defined as a subset © of 2° whose elements (subsets of S) are called open sets. To qualify as a topology, the open sets must satisfy the following properties: 1. ‘The empty set and the whole space are open sets, 6 € O und S € ©. 2. Ue Cand V cOthenUCY CO. 3. If Lis any subset of © then JU € O. The second axiom says that the intersection of any pair of open sets, and therefore of any finite collection of open sets, is open. The third axiom says that an arbitrary, possibly infinite, union of open sets is open. According to our criterion, a topology is clearly a geometrical structure on S. ‘The basic view presented here is that the key feature distinguishing an algebraic structure from a geometric structure on a set S is algebraic structure = a map S x S-> $= asubset of S*, while geometric structure = a subset of 2°. This may look to be a clean distinction, but it is only intended as a guide, for in reality many structures exhibit both algebraic and geometric aspects. For example, Euclidean geometry as originally expressed in terms of relations between subsets of the plane such as points, lines and circles is the geometric or “visual” approach. On the other hand, cartestan geometry is the algebraic or analytic approach to plane geometry, in which points are represented as elements of 2’. In the latter approach we have two basic maps: the difference — R? definedas(x,¥) (u,v) = (8 — wv, y — v), and the distance map E 2 defined by Eq. (1.2). The emphasis on maps places this method much more definitely in the algebraic camp, but the two representations of Euclidean geometry are essentially interchangeable and may indeed be used simultaneously to best understand a problem in plane geometry Dynamical systems The evolution of a system with respect to its algebraic/geometric background invokes what is commonly known as ‘laws of physics’. In most cases, particularly when describing 21 22 Sets and structures a continuous evolution, these laws are expressed m the form of differential equations. Providing they have a well-posed initial value problem, such equations generally give rise toa unique evolution for the system, wherein lies the predictive power of physics. However. exact solutions of differential equations are only available in some very specific cases, and it is frequently necessary to resort to numerical methods designed for digital computers with the time parameter appearing in discrete packets, Discrete time models can also serve as a useful technique for formulating “toy models’ exhibiting features similar to those of a continuum theory, which may be too difficult to prove analytically. ‘There 1s an even more fundamental reason for considering discretely evolving systems We have good reason to believe that on time scales less than the Planck rime. given by the continuum fabric of space-time is probably invalid and a quantum theory of gravity becomes operative. It is highly likely that differential equations have little or no physical relevance at or below the Planck scale. As already discussed in Example 1.12. a discrete dynamical systeni 1s a set S together withamap f S— § Themap f : S— Siscalled a discrete dynamical structure on § The complexities generated by such a simple structure on a single set $ can be enormous A well-known example 1s the Jogéstic map f : [0 1| > [0. 1] defined by fey=Crl—~) where 0< 6 <4 and used to model population growth with limited resources or predator-prey systems in ecology. Successive iterates give rise to the phenomena of chaos and strange attructors limiting sets having a Cantor-like structure, The details of this and other maps such as the Hénon map [8], f = R° — 1° defined by f= W +l ax", bx) can be found in several books on non-linear phenomena, such as [9] Discrete dynamical structures are often described on the set of states on a given set S. where a sfate on S1s.a function ¢ > S—> (0. 1}. As each state is the characteristic function of some subset of § (see Example 1.7), the set of states on S can be identified with 2° A discrete dynamical structure on the set of all states on $ 1s called a cellular automaton on S. Any discrete dynamical system (S,_f) induces a cellular automaton (2°, f* : 2° — 2°), bysetung f°: 1+ Go f foranystate@ ; S—> {0.1}, Thiscanbe pictured in the following way, Every state ¢ on S attaches a | or 0 to every point p on S. Assign to p the new value 4(f(p)). which is the value 0 or | assigned by the original state ¢ to the mapped pont { (p) This process is sometimes called a pullback — it caries state values ‘backwards’ rather than forwards. We will frequently meet this idea that a mapping operates on functions, states in this case. in the opposite direction to the mapping. Not all dynamicat structures defined on 2°. however, can be obtained in the way just described. For example, if S has elements. then the number of dynamical systems on Sis n". However, the number of discrete dynamical structures on 2° is the much larger 17 1.7 Category theory number (2”)” = 2". Even for small initial sets this number 1s huge; for example, for n= Ajtis2 ~ 2 x 10! while forslightly larger » it easily surpasses all numbers normally encountered in physics. One of the most intriguing cellular automata is Conway's game of life, which exhibits complex behaviour such as the existence of stable structures with the capacity for self-reproducibility, all from three simple rules (sce [9, 10). Graphical versions for personal computers are readily available for experimentation. Category theory Mathematical structures generally fall into ‘categories’, such as sets, semigroups, groups, vector spaces. topological spaces, differential manifolds, ete. The mathematical theory devoted to this categorizing process can have enormous benefits in the hands of skilled practioners of this abstract art. We will not be making extensive use of category theory, but in this section we provide a flavour of the subject. Those who find the subject too obscure for their taste are urged to move quitkly on, as little will be Jost in understanding the rest of this book. A category consists of (Cat) Acclass © whose clements are called objects. Note the use of the word ‘class’ rather than ‘set” here. This is necessary since the objects to be considered are generally themselves sets and the collection of all possible sets with a given type of structure is too vast to be considered as a set without getting into difficulties such as those presented by Russell's paradox discussed in Section 1.1. (Cat2) For each pair of objects A, B of © there is a set Mor(4, B) whose clements are called morphisms from A to B, usually denoted 4 *> B (Cat3) For any pair of morphisms 4 -®> B, Bs C there is a morphism 4 ““% C, called the composition of @ and such that 1. Composition is assoctative: for any three morphisms A B, B > C.¢ 4 D, (eo Wed =polWed). 2. Fot cach object 4 there is a morphism 4 ‘4 A called the identity morphism on A, such that for any morphism 4 *> B we have bou=b and for any morphism C + 4 we have wow=y. Example 1.15 The simplest example of a category ts the category of sets, in which the objects are all possible sets, while morphisms are mappings from a set 4 to a set B. In this case the set Mor(A, B) consists of all possible mappings from 4 to B. Composi- tion of morphisms is simply composition of mappings, while the identity morphism on 23 24 Sets and structures an object 4 is the identity map id, on A. Properties (Catl) and (Cat2) were shown in ‘Section 1.4. Exervise, Show that the class of all semigroups, Example 1.12, forms a category, where morphisms are defined as semigroup homomorphisms. ‘The following are some other important examples of categories of structures to appear in later chapters: Objects Morphisms Refer to Groups Homomorphisms Chapter 2 ‘Vector spaces Linear maps Chapter 3 Algebras Algebra homomorphisms Chapter 6 Topological spaces Continuous maps Chapter 10 Differential manifolds Differentiable maps Chapter 15 Lie groups Lie group homomorphisms Chapter 19 ‘Two important types of morphisms are defined as follows. A morphism 4S B is called a monomorphism if for any object X and morphisms X “> 4 and X “> Awe have that goa=yoou' => a=c' The morphism ¢ 1s called an epimorphism if for any object X and morphisms B > ¥ and Bay Bog=Bicy => p=p' “These requirements are often depicted in the form of commutative diagrams. For example, @ is a monomorphism if the morphism a 1s uniquely defined by the diagram shown in Fig, 1.3. The word ‘commutative’ here means that chasing arrows results in composition of morphisins, vv = (yea). x - A — = B ° Figute 1.3 Monomorphism y 1.7 Category theory A _ ba Y Figure 1.4 Epimorphism ¢ On the other hand, ¢ is an epimorphism if the morphism # is uniquely defined in the commutative diagram shown on Fig. 1.4. In the case of the category of sets a morphism 4 ++ B is a monomorphism if and only fit is a one-to-one mapping. Proof \ Mfg: A — B isone-to-one then for any pair of mapsa : X > A anda’ : X > 4, (ee) = gla’) => a(x) = a(x) for all x € X. This is simply another way of stating the monomorphism property y o @ = goo’ =a =a 2. Conversely, suppose ¢ is a monomorphism. Since X isan arbitrary set, inthe definition of the monomorphism property, we may choose it to be a singleton X = {x}. For any pair of points a,a’ © A define the maps a, a’ : X > A by setting a(x) = a and «'(x) = a’. Then ya) = g(a’) => goats) = goa (x) => pou=¢geo’ = a=a =) a=a(%)= a(x) =a". Hence y is one-to-one. . Itis left as 2 problem to show that in the category of sets a morphism isan epimorphism if and only if it is surjective. A morphism 4 * B is called an isomorphism if there exists amorphism B + A such that gcop=ts and yoy’ =u In the catepory of sets 2 mapping is an isomorphism if and only if it is bijective, that is, it is both an epimorphism and a monomorphism. There ean, however, be a trap for the unwary here. While every isomorphism is readily shown to be both a monomorphism and an epimorphism, the converse is not always true. A classic case is the category of Hausdorff topological spaces in which there exist continuous maps that are epimorphisms and monomorphisms but are not invertible. The interested reader is referred to [11] for further development of this subject. 25 26 Sets and structures Problems Problem 1.15 Show that in the category of sets a morphism 1s an epimorphism if'and only if it onto (surjective) Problem 1.16 Show that every isomorphism is both @ monomorphism and an epimorphism References [1] T. Apostol. Mathematical Analysis. Reading, Mass., Addison-Wesley, 1957. [2] K. Devlin. The Jov of Sets. New York, Springer-Verlag, 1979. [3] N. B. Haaser and J. A. Sullivan. Real Analysis. New York, Van Nostrand Reinhold. Company, 1971 [4] PR. Halmos. Natve Set Theory. New York, Springer-Verlag, 1960, [5] E. Wigner. The unreasonable effectiveness of mathematics in the natural sciences, Communications in Pure and Applicd Mathematics, 13:1-14, 1960. [6] J. Kelley. General Topology. New York, D. Van Nostrand Company, 1955. [7] PJ. Cohen. Set Theory and the Contintaum Hypothesis. New York, W. A. Benjamin, 1966. {8] M. Hénon. A two-dimensional map with a strange attractor. Communications in Math- ematical Physics, 50:69-77, 1976. [9] M. Schroeder. Fractals, Chaos, Power Laws. New York, W. H. Freeman and Company, 1991. [10] W. Poundstone. The Recursive Universe. Oxford, Oxford University Press, 1987. [11] R. Geroch. Mathematical Physics. Chicago, The University of Chicago Press, 1985 2 21 Groups The cornerstane of modern algebra is the concept of a group, Groups are one of the simplest algebraic structures to possess a rich and interesting theory, and they are found embedded in almost all algebraic structures that occur in mathematics [1-3]. Furthermore, they are important for our understanding of some fundamental notions in mathematical physic: particularly those relating to symmetries [4] The concept of a group hgs its origins in the work of Evanste Galots (1811-1832) and Niels Henrik Abel (1802-1829) on the solution of algebraic equations by radicals. The latter mathematician is honoured with the name of a special class of groups, known as abelian, which satisfy the commutative law. In more recent times, Emmy Noether (1888— 1935) discovered that every group of symmetries of a set of equations arising from an action principle gives nse to conserved quantities. For example, energy, momentum and angular momentum arise from the symmetries of time translations, spatial translations and rotations, respectively. In elementary particle physics there are further conservation laws related to cxotic groups such as SU/(3), and their understanding has led to the discovery of new particles. This chapter presents the fundamental ideas of group theory and some examples of how they arise in physical contexts. Elements of group theory A group is a set G together with # law of composition that assigns 1 any pair of ele- ments gh © G an element gh © G, called their product, satisfying the following three conditions (Gpl) ‘The associative law holds: e(hk) = (gh)k, forall g, h, k € G. (Gp2) There exists an identity element ¢ ¢ G, such that eg=ge-g forallgeG. (Gp3) Each element g € G has an inverse g- | ¢ G such that 1 g'g=eg' =e. ‘More concisely, a group isa semigroup with identity in which every element has an inverse. Sometimes the fact that the product of two clements is another clement of G is worth noting as a separate condition, called the closure property. ‘This is particularly relevant 27 28 Groups when G is a subset of a larger set with a law of composition defined. In such cases it 1s always necessary to verify that G is closed with respect to this faw of composition: that 1s, for every pair g, fi ¢ G, their product gh G. Examples will soon clarify this point. Condition (Gp1) means that alf parentheses in products may be omitted. For example, a((be)d) = a(b(cd)) = (ab)(cd) = (ab)e)d. It is a tedious but straightforward matter to show that all possible ways of bracketing a product of any number of elements are equal. ‘There is therefore no ambiguity in omitting all parentheses in expressions such as abcd. However, it is generally important to specify the order in which the elements appear in a product. The identity element ¢ is easily shown to be unique. For, if e’ 1s a second identity such that eg = ge’ =g forall g © G then. setting g = e, we have by (Gp2). Exercise: By a similar argument, show thal every g € G has a unigue inverse g~! =e Exercise: Show that (gh)! = hr 'g! A group G is called abelian if the law of composition is commutative, gh=he forallg.heG. ‘The notation gh for the product of two clements is the default notation. Other possibilities area b,a x ba +h, a0 b, etc. When the law of composition is written as an addition g +A, we will always assume that the commutative law holds, ¢ +h = h + g. In this case the identity element is usually written as 0, so that (Gp2) reads g + 0=0 +g =g. The inverse is then written —g, with (Gp3) reading g + (2) = 0 or, more simply, g — ¢ =0 Again, the associative law means we never have to worry about parentheses in expressions such sat b+e4o + fi. A subgroup H of a group G isa subset that 1s a group in its own right A subset 7 G is thus subgroup iit contains the identity element of G and is closed under the operations of taking products and inverses: (a) h.k CH => hk © H (closure withrespect to taking products); (b) the identity e © H; (c) hE H = h-' CH (closure with respect to taking inverses). ILis not necessary to verify the associative law since H automatically inherits this property from the larger group G. Every group has two trivial subgroups {e] and G, consisting of the identity alone and the whole group respectively. Example 2.1 The integers Z. with addition as the law of composition forma group, called the additive group of integers. Strictly speaking one should write this group as (Z, +), but the kaw of composition is implied by the word ‘additive’, The identity element is the integer 0, and the inverse of any integer m is =n. ‘The even integers {0, +2,-4,...) form a subgroup of the additive group of integers. Example 2.2. The real numbers 2. form a group with addition x 4 yas the law of com- position, called the additive group of reals. Again the identity is 0 and the inverse of. rhe additive group of integers is clearly a subgroup of 2. The rational numbers is are 2.1. Elements of group theory closed with respect to addition and also form a subgroup of the additive reals I, since the number 0) is rational and if p/g is a rational number then so is —p/q Example 2.3 The non-zero real numbers t = Ik — {0} forma group called the mudtiplca- tive group of reals. In this case the product 1s taken to be ordinary muluphieation xy, the identity is the number | and the inverse of x sx! = 1 /x. The number 0 must be excluded since it has no inverse. Exercise Show that the non-zero rational numbers Q form a mulnplicative subgroup of Ie Exercise: Show that the complex numbers C form a group with respect to addition, and C = 1S a group with respect to multiplication of complex numbers — (0) Exercise. Which of the following sets form a group with respect to addition: (i) the rational numbers, Gi) the trrational numbers, (111) the complex numbers of modulus 1? Which of them is a group with respect to multiplication? A group G consisting of only a finite number of elements is known as a finite group ‘The number of elements in Gis called its order, denoted |G Example 2.4 Let k be any natural number and %, = (10). [J.... . [A — 1} the integers modulo k, defined in Example 1.3, with addition modulo & as the law of composition fa] + [hb] = [a +4], 2, is called the additive group of integers modulo k. Itis a finite group of order &, written |Z;| =k. There is little ambiguity im writing the elements of Z; as 0, 1,..., 4 — 1 and [a + BJ is often replaced by the notation @ +b mod k. Exercise: Show that the definition of addition modulo & is independent of the choice of representatives from the residue classes [al] and [6] Fxample 2.5 \f a group G has an element a such that its powers fa. a’, a’, ...) run through all of ts elements, then G 1s said to be a eyelic group and a is called a generator of the group. If G is a finite cyclic group and a is a generator, then there exists a positive integer m such that a” = e. If m is the lowest such integer then every element g © G can be uniquely written g = a’ where | 2 be any prime number. The non-zero integers modulo p form a group of order p — I with respect to multiplication modulo p, [alfb] = [ab] = ab mod p. denoted G,,. The identity is obviously the residue class [1], but in order to prove the existence of inverses one needs the following result from number theory: if p and g are relatively 29 2.2 30 Groups prime numbers then there exist integers k and m such that kp + mg = 1. Since p isa prime number, if [g] # [0] then g is relatively prime to p and for some k and m [rq] = 1 — Fellpy = 11). Hence [¢] has an inverse {q]~! = [in]. For finite groups of small order the law of composition may be displayed in the form of ‘a multiplication table, where the (i. /)th entry specifies the product of the #th element and the /th element. For example, here is the multiplication table of G;: Gili 2 3 4 5 6 ryt 2 30 40 5 6 2)2 4 6 41 3 5 3|/3 6 2 5 1 4 4/4 1 5 2 6 3 5/5 3 1 6 4 2 6]6 5 4 38 2 4 Transformation and permutation groups All groups in the above examples are abelian. The most common examples of non- commutative groups are found in a class called transformation groups. We recall from Section 1.4 that @ wansformation of a set X is a map g: X —> X that 1s one-to-one and onto. The map g then has an inverse g~! = X — X such that g!og=gog! = idy Let the product of two transformations g and h be defined as their functional composition gh=goh, (gh\x) = go hex) = g(h(x)). The set of all transformations of X’ forms a group, denoted Transf{X): Closure: if g and h are transformations of X then so is gh; Associative law: (gh) = (feyh: Identity: e = idy < TransftX), Inverse: if g is a transformation of X then so is ¢ ! Closure follows from the fact that the composition of two transformations (invertible maps) results in another invertible map, since (f © g)-! = g-! © f~'. The associative law holds automatically for composition of maps, while the identity and inverse are trivial. By a transformation group of X is meant any subgroup of Transf{.%). 1£X isa finite set of cardinality» then the transformations of ¥ are called permutations of the elements of X. The group of permutations of Y = {1,2,.....n} is called the symmetric group of order 7, denoted S,. Any subgroup of S, is called a permutation group, A permutation z on » clements can be represented by the permutation symbol Po2..48 Cr 2.2. Transformation and permutation groups where a, = (1), a2 = 7(2), etc. The same permutation can also be written as n=(" bb .. a “1&2 ©n, where by, b3....,b, are the numbers 1. 2..... inanarbitrary orderand cy = (by), 6: = (bi). (by). For example, the permutation zr that interchanges the elements 2 and 4 from a four-element set can be written in several ways, re(l 234 2314123) “\i43 2) a3 1 2) > 21 4 3) In terms of permutation symbols, if oa(h 2) ang gp (40 Boe Ge Nay dg ee. Gy, Ab) bp oe. By then their product is the permutation 0 = x oo, Se u(l 2” TON iy by oe Bn Note that this product involves first performing the permutation « followed by 7, which is opposite to the order in which they are written; conventions can vary on this pomt. Since the product is a functional composition, the associative law is guaranteed. The identity permutation is i £ owes if A, = 7 (; @. ") while the inverse of any permutation is given by wi-(! 2. 2) fa @ an “Naya... a) XL 2 2 on ‘The symmetric group S, is a finite group of order n!, the total number of ways n objects may be permuted, It 1s not abelian in general. For example, in Ss G32)Gta)-G 1 det “G1 2) G1 )G3 )-G3 03 )-6 31) A more compact notation for permutations is the cyclic notation. Begin with any element to be permuted, say a). Let a2 be the result of applying the permutation z to a), and let ay be the result of applying it to @2, etc. Eventually the first element a, must reappear, say 88 Gat = ay, This defines a eyele, written (a; a> ...4;,). Hm =n, then 7 is said to be a cyclic permutation. If'm ... By) of successive images of by under 7. Continue while 31 32 Groups until all the elements 1.2... are exhausted. The permutation 7° may be written as the product of its cycles; for example, | 1234567 721 a ) = (1476)(25)3) Note that it does not matter which element of a cycle is chosen as the first member, so that (1476) = (76 14) and (25) = (52). | Cycles of length | such as (3) merely signify that the permutation z Icaves the element 3 unchanged. Nothing is lost if we totally ignore such |-cycles from the notation, writing (1.476)(25)(3) = (147625) ‘The order in which cycles that have no common elements is written 1s also immaterial, (1476)(25) = (25)(1476). Products of permutations are easily carried out by following the effect of the cycles on cach element in succession, taken in order from right to left. For example, (1375421 23467146) = (1654)(23\(7) follows from 1 > 4 > 6,6 > 1> 25,5 +4,456+7- Let. Exercise: Express each permutation on (1.2. 3} in cyclic notation and write out the 6 + 6 multipli- cation table for Sy Cycles of length 2 are called interchanges. Every cycle can be written as a product of interchanges, (Ay G23 +++ Gy) = (42 43 Mats 4). (1 Ay Keb 1) and since every permutation zr 1s a product of cycles, it 1s in tum a product of interchanges ‘The representation of a permutation as a product of interchanges is not in general unique, but the number of interchanges needed is either always odd or always even. To prove this, consider the homogeneous polynomial flew 825-206 4) = [] or 3) = (er = My 85) 6 = Xn RZ — 83) + Ot = Xn) If any pair of variables x, and x, are interchanged then the factor (x, — x) changes sign and the factor (x, — x;) is interchanged with (x, — x) forall k 1. j. When k <0 You might also like
