I0

ur/
Is
II
II

I%,
/I/I
I.-ft
l i
AN INTRODUCTION

/%'-U;I II /I
11.41
" // 11
it

i
i/I,//I/i11
IF

I/III/I'lV/I/l/ II I/I
/
1'lv(l I
l , /I \
I/
SW v/I rII.W/ rr lbli

OSHE CARMELI
 AN IIYTRODUCTiON

m Upbwr$

MOSHE CARMELI
Ben Gurion University, Israel

SHIMON MALIN
Colgate University, USA

World Scientific
Singapore -New Jersey London Hong Kong
Published by
World Scientific Publishing Co. Pte. Ltd.
P O Box 128, Fairer Road, Singapore 912805
USA office: Suite 1B, 1060 Main Street, River Edge, NJ 07661
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library.

THEORY OF SPINORS: AN INTRODUCTION

Copyright m 2000 by World Scientific Publishing Co. Pte. Ltd.
All rights reserved. This book, or parts thereof may not be reproduced in any form or by any means,
electronic or mechanical, including photocopying, recording or any information storage and retrieval
system now known or to be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright
Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to
photocopy is not required from the publisher.

ISBN 981-02-4261-1

This book is printed on acid-free paper.

TO OUR GRANDCHILDREN

Nadav, Guy, Daniel, Shelly

and

Michaela
Preface
This is a textbook intended for advanced undergraduate and graduate stu-
dents in physics and mathematics, as well as a reference for researchers.
The book is based on lectures given during the years at the Ben Gurion
University, Israel. Spinors are used extensively in physics; it is widely, ac-
cepted that they are more fundamental than tensors and the easy way to
see this fact is the results obtained in general relativity theory by using
spinors, results that could not have been obtained by using tensor methods
only. The book is written for the general physicist and not only to the
workers in general relativity, even though the latter will find it most useful
since it includes all what is needed in that theory.
But the foundations of the concept of spinors are groups; spinors appear
as representations of groups. In this text we give a wide exposition to the
relationship between the spinors and the representations of the groups. As
is well known, both the spinors and the representations are widely used in
the theory of elementary particles.
After presenting the origin of spinors from representation theory we,
nevertheless, apply the theory of spinors to general relativity theory, and a
part of the book is devoted to curved spacetime applications.
In the first four chapters we present the group-theoretical foundations
of the concept of two-component spinors. Chapter 1 starts with an intro-
duction to group theory emphasizing the rotation group. This followed
by discussing representation theory in Chapter 2, including a brief out-
line of the infinite-dimensional case. Chapters 3 and 4 discuss in detail
the Lorentz and the SL(2,C) groups. Here we give an extensive discussion
on how two-component spinors emerge from the finite-dimensional repre-
sentations of the group SL(2,C). Chapter 4 also includes the derivation
of infinite-dimensional spinors as a generalization to the two-component
spinors.
In Chapters 5 and 6 we apply the two-component spinors to a variety of

vii
vii'

problems in curved spacetime. In Chapter 5 we discuss the Maxwell, Dirac

and Pauli spinors. Also given in this chapter the passage to the curved
spacetime of spinors. The gravitational field spinors are subsequently dis-
cussed in detail in Chapter 6. Here we derive the curvature spinor and give
the spinors equivalent to the Riemann, Weyl, Ricci and Einstein tensors.
In Chapter 7 we present the gauge field spinors and discuss their geo-
metrical properties. As is well known, gauge fields are extremely important
nowadays. The Euclidean gauge field spinors are finally discussed in Chap-
ter 8.
All chapters of the book start with the ordinary physical material before
introducing the spinors of that subject. Thus, for instance, the chapters
dealing with the Lorentz group and gravitation start with detailed discus-
sion of the theories of special relativity and general relativity.
It is a pleasure to thank our wifes Elisheva and Tova for creating the
necessary atmosphere and for their patience while writing this book. We are
grateful to the many students who attended the courses in spinors during
the years for their suggestions which led to a better presentation of the
material in the book. We also want to thank Silvia Behar for her help with
the Index of the book. Finally, we want to thank Julia Goldbaum for the
excellent job of typing the book, prepairing the Index, and for the many
suggestions for improvements.

Moshe Carmeli
Beer Sheva, Israel

Shimon Malin
Hamilton, N. Y.
U.S.A.
Contents
Preface Vii
1 Introduction to Group Theory 1
1.1 Review of Group Theory ... . . . . . . . . . . . . . . . . . 1

1.1.1 Group and Subgroup . . . . . . . . . . . . . . .

. . . 1

1.1.2 Normal Subgroup and Factor Group . . . . . . . . . 3

1.1.3 Isomorphism and Homomorphism . . . . . . . . .. 3
1.2 The Pure Rotation Group SO(3) . . . . . . . . . . . . . . . 4
1.2.1 The Euler Angles . 4
1.3 The Special Unitary Group SU(2) . . . . . . . . . . . . . . 6
1.3.1 Homomorphism between the Groups SO(3) and SU(2) 6
1.4 Invariant Integrals over Groups . . . . . . . . . . . . . . .
. 8
1.4.1 Invariant Integral over the Group SO(3) . . . . . . . 8
1.4.2 Invariant Integral over the Group SU(2) . . . . . .. 9
1.5 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.6 References for Further Reading .. . . . . . . . . . . . . . . 10

2 Representation Theory 11
2.1 Some Basic Concepts . . ... .. . . . . . . . . . . . . . . . 11
2.1.1 Linear Operators . . . . . . . . . . . . . . . . . . . . 11
2.1.2 Finite-Dimensional Representations . . . . . . . . . 12
2.1.3 Unitary Representations . . . . . . . . . . . . . . . . 13
2.2 Representations of SO(3) and SU(2) . . . . .. 13
2.2.1 Weyl's Method ........... ... .....
... . .. ... .
2.2.2 Infinitesimal Generators .. .
.. . . . . . .

. .
.

.
13
14
..
2.2.3 Basic Infinitisimal Operators . . . . . . . . . . . 16
2.2.4 Canonical Basis . ... . ... .. . . . . . . . . . . . 16
2.2.5 Unitary Matrices Corresponding to Rotations . . . . 17
2.3 Matrix Elements of Representations .. .. . . . . . . . . . 18
2.3.1 The Spinor Representation of the Group SU(2) . . . 19

ix
x CONTENTS

2.3.2 Matrix Elements of Representations . . . . . . . . . 20

2.3.3 Properties of (u) . . . . . . . . . . . . . . . . . 21
2.4 Differential Operators of Rotations . . . . . . . . . . . . . . 22
2.4.1 Representation of SO(3) in Space of Functions . . . 23
2.4.2 The Differential Operators . . . . . ... .
. . . . . . 24
2.4.3 Angular Momentum Operators . . . . . . . . . .. . 26
2.5 Infinite-Dimensional Representations . . . . . . . . . . . . . 27
2.5.1 Banach Space . .. .. ..
. . . . . . . . . . . . . . . 27
2.5.2 Hilbert Space . . . . . . . . . . . . . . . . . . . . . . 29
2.5.3 Operators in a Banach Space . . . . . . . . . . . . . 30
2.5.4 General Definition of a Representation . . . . . . . . 30
2.5.5 Continuous Representations ..
. . . . . . . .. . . . 31
2.5.6 Unitary Representations . . . . . . . . . . . . . . . . 31
2.6 References for Further Reading . . . . . . . . . . . . . . . . 32

3 The Lorentz and SL(2,C) Groups 35

3.1 Elements of Special Relativity . . . . . . . . . . . . . . . . . 35
3.1.1 Postulates of Special Relativity . . . . . . . . . . . . 35
3.1.2 The Galilean Transformation . . . . . . . . . . . . . 37
3.1.3 The Lorentz Transformation . . . . . . . . . . . . . 38
3.1.4 Derivation of the Lorentz Transformation . . . .. . 38
3.1.5 The Cosmological Transformation . . .. . . . . . . 41
3.2 The Lorentz Group . . . . . . . . . . . . . . . . . . . . . . . 44
3.2.1 Orthochronous Lorentz Transformation . . . . . . . 46
3.2.2 Subgroups of the Lorentz Group . . . . . . . . . . . 46
3.3 The Infinitesimal Approach ... . . . . . . . . . . . . . . .
47
Infinitesimal Lorentz Matrices . . . . . . . . . . . . .
3.3.1 47
3.3.2 Infinitesimal Operators . . .
. ..
. . . . . . . . . . 49
3.3.3 Determination of the Representation by its Infinites-
imal Operators . . . . . . . . . . . . . . . . . . .
. . 51
3.3.4 Conclusions . . ..
. . . . . . . . . . . . . . . . . . . 52
3.3.5 Unitarity Conditions . . . . . . . . . . . . . . . . . . 53
3.4 The Group SL(2,C) and the Lorentz Group . . . . . . . . . 54
3.4.1 The Group SL(2,C) . ... .... ..... .. ... 54
3.4.2 Homomorphism of the Group SL(2,C) on the Lorentz
Group L . . . . .... .. . ... ... .
. . . . . . . 56
3.4.3 Kernel of Homomorphism . . ..
. . . . . . . . . . . 58
3.4.4 Subgroups of the Group SL(2,C) . . . . . . . . ... 59
3.4.5 Connection with Lobachevskian Motions . . . . . . . 60
3.5 Problems ... . . . . .... ..
. . . . . . . . . . . . . . . 61
CONTENTS xi

3.6 References for Further Reading . . . . . . . . . . . .. . . . 62

4 Two-Component Spinors 65
4.1 Spinor Representation of SL(2,C) . . . . . . .. . ... .. . 65
4.1.1 The Space of Polynomials . . . . . . .. .. ... . . 65
4.1.2 Realization of the Spinor Representation . ... . . . 66
4.1.3Two-Component Spinors . . . . . . . . . . . . . . . . 68
4.1.4Examples . . . .... ... . .. .. . ... ...
. . 71
4.2 Operators of the Spinor Representation . . . . . . . . . . . 73
4.2.1 One-Parameter Subgroups . . . . . . . . . . . . . . . 73
4.2.2 Infinitesimal Operators . . . . . . . ... . .. . . . 74
4.2.3 Matrix Elements of the Spinor Operator D (g) .. . 75
4.2.4 Further Properties of Spinor Representations . . . . 77
4.3 Infinite-Dimensional Spinors . . ..
. . . . . .. . ... . . . 77
4.3.1 Principal Series of Representations . . . . . . . . . . 77
4.3.2 Infinite-Dimensional Spinors . . . . . . . . ... ... 79
4.4 Problems . . . . . . . .... ..
. . . . . . . . . . .. . . . 81
4.5 References for Further Reading . . . . . . . . . . .. . . . . 81

5 Maxwell, Dirac and Pauli Spinors 83

5.1 Maxwell's Theory . . . . . . . . .. .. ..
. . . . . . . . . . 83
5.1.1 Maxwell's Equations in Curved Spacetime . ..
. . . 86
5.2 Spinors in Curved Spacetime . . . . . .. ... . .. ...
. 88
5.2.1 Correspondence between Spinors and Tensors . . . . 88
5.2.2 Raising and Lowering Spinor Indices .. .. ..
. . . 89
5.2.3 Properties of the o Matrices . . .. .. . ... .
. . . 89
5.2.4 The Metric 9AB'CD' and the Minkowskian Metric 77 90
5.2.5 Hermitian Spinors .. . . . ... . .. .. . . . . . . 91
5.3 Covariant Derivative of a Spinor . . . . . . . . .. ..

. . . . 92
5.3.1 Spinor Affine Connections . . . . . . . . .. .. . . . 92
5.3.2 Spin Covariant Derivative . . . . .. ..
. . . . . . . 93
5.3.3 A Useful Formula ... . ... . . . . . . . .. . . . . 95
5.4 The Electromagnetic Field Spinors . . . . . . .. .. . . . . 95
5.4.1 The Electromagnetic Potential Spinor . . . . . . . . 95
5.4.2 The Electromagnetic Field Spinor . . . . .. . . . . 96
5.4.3 Decomposition of the Electromagnetic Spinor . . . . 96
5.4.4 Intrinsic Spin Structure . . . . . . . . . . . . . . . . 97
5.4.5 Pauli, Dirac and Maxwell Equations . .. .. . . . . 98
5.5 Problems . ..
. . . . . ... .
. . . . . . . ... .. . . . . 100
5.6 References for Further Reading . . . . . . . . . . . . . . . . 107
xii CONTENTS

6 The Gravitational Field Spinors 109

6.1 Elements of General Relativity . . . . . . . . . . . . . . . . 109
6.1.1 Riemannian Geometry . . . . . . . . . . . . . . . . . 110
6.1.2 Principle of Equivalence . . . . . . . . . . . . . . . . 118
6.1.3 Principle of General Covariance . . . . . . . . . . . . 119
6.1.4 Gravitational Field Equations . . . . . .
. . . . . . . 120
6.1.5 The Schwarzschild Solution . . . . . . . . . . . . . 123
6.1.6 Experimental Tests of General Relativity . . . . . . 127
6.1.7 Equations of Motion . . . . . . . . . . . . . . . . . . 133
6.1.8 Decomposition of the Riemann Tensor . . . . . . . 142
6.2 The Curvature Spinor . . . . . . . . . . . . . .. . . . . . . 143
6.2.1 Spinorial Ricci Identity . . . . . . . . . . . . . . . . 144
6.2.2 Symmetry of the Curvature Spinor . . . . . .. . . . 145
6.3 Relation to the Riemann Tensor . . . .. . . .. . . . . . . 146
6.3.1 Bianchi Identities ... .
. . . . . . . . . . . . . . . . 147
6.4 The Gravitational Field Spinors . . ..
. . . . . . . . . . . . 148
6.4.1 Decomposition of the Riemann Tensor . . . . . . . . 148
6.4.2 The Gravitational Spinor . . . . . . . . . . . . . . . 150
6.4.3 The Ricci Spinor . . . . . . . . . . . . .
. . . . . . . 152
6.4.4 The Weyl Spinor . . . . . . . . . . . . .
. . . . . . . 153
6.4.5 The Bianchi Identities . . . . . . . . .
. . . . . . . . 157
6.5 Problems . . . .. . . . .
. . . .
. . . . . . .
. . . . . . . . 157
6.6 References for Further Reading . . . . . . . . . . . .. . . . 163

7 The Gauge Field Spinors 167

7.1 The Yang-Mills Theory . . . . . . . . . . . . .. . . . . 167
7.1.1 Gauge Invariance . .. ..
. . . . . . . . . . . . . . . 167
7.1.2 Isotopic Spin . . . ..
. . . . . . . . . . . . . . . . . 168
7.1.3 Conservation of Isotopic Spin and Invariance . . . . 168
7.1.4 Isotopic Spin and Gauge Fields . . . .. . . . . . . . 169
7.1.5 Isotopic Gauge Transformation . . . . . ... . . . . 169
7.1.6 Field Equations . . . . . . . . . . . . . . . . . . . . . 171
7.2 Gauge Potential and Field Strength . . . .. . . . . . . . . 172
7.2.1 The Yang-Mills Spinor ..
. . . . . . . . . .. . . . . 172
7.2.2 Energy-Momentum Spinor . . . . . . . . . . . . . . . 174
7.2.3 SU(2) Spinors . ... . ..
. . . . . . .. . . . . . . . . 176
7.2.4 Spinor Indices . . . . .. .. .. .. ... . .. ..
. . 176
7.3 The Geometry of Gauge Fields . . . .. ..
. . . . . . . . . 177
7.3.1 Four-Index Tensor . . . . . . . . . . . . .. ... . . 177
7.3.2 Spinor Formulation . . . . . . . . . . . .. .. . . . . 179
CONTENTS xiii

7.3.3Comparison with the Gravitational Field . . . . . . 180

7.3.4Ricci and Einstein Spinors . . . . . . .
. . . . . . . . 182
7.4 References for Further Reading . . . . . . . . .
. . . . . . . 183

8 The Euclidean Gauge Field Spinors 185

8.1 Euclidean Spacetime . . . .. . . . . . . . .. . . . . . . . . 185
8.1.1 The Euclidean Dirac Equation . . . .. .. . . . . . 186
8.1.2 Algebra of the Matrices s 188
8.2 The Euclidean Gauge Field Spinors . . . . .. . . . . . . . . 190
8.2.1 0(4) Two-Component Spinors . . .. . . . . . . . . 190
8.2.2 Self-Dual and Anti-Self-Dual Fields . .. . . . . . . . 193
8.3 Problems . . . . . . . . . . . . . . . .
. . . .. . . . . . . . 194
8.4 References for Further Reading . . . . . . .. . . . . . . . . 194

Index 197
Chapter 1
Introduction to Group
Theory
In this chapter a brief discussion on group theory is given. This includes the
concept of group and subgroup, normal subgroup and factor group. Isomor-
phism and homomorphism are subsequently discussed. This thep followed
by introducing the rotation group and the group SU(2), the aggregate of
unitary matrices of order two and determinant unity. A homomorphism
between the pure rotation group and the group SU(2) is subsequently es-
tablished. The chapter is concluded with presenting invariant integrals over
the groups.

1.1 Review of Group Theory

In this section the fundamental concepts of group theory are briefly pre-
sented. For details the reader is refered to the books of Pontrjagin, van der
Waerden and others suggested at the end of the chapter.

1.1.1 Group and Subgroup

A non-empty set G of elements a, b, c,..., such as numbers, mappings,
transformations, is called a group if the following axioms are satisfied:
(1) There exists an operation in the set G which associates to each two
elements a and b of G a third element c of G. This operation is called

1
2 CHAPTER 1. INTRODUCTION TO GROUP THEORY

multiplication, and the element c is called the product of a and b, denoted

by c = ab;
(2) The multiplication is associative, namely, if a, b and c are elements
of G, then (ab) c = a (bc);
(3) The set G contains a right identity, namely, there exists an element
e such that ae = a for each element a of G; and
(4) For each element a of G there exists a right inverse element, denoted
by a-1, such that as-1 = e.
If the set G is finite, then the group G is called finite and the number of
elements of G is called its order. Otherwise, the group G is called infinite. If
the product of any two elements a and b of G is commutative, namely, ab =
ba, the group is called abelian. In abelian groups the multiplication notation
ab is replaced by an addition notation a + b, and the group operation is
called addition. The identity is called zero and denoted by 0, and the inverse
of a is called the negative of a and denoted by -a.
Since the product of group elements is associative, one writes for (ab) c =
a (bc) simply abc and for (a + b) + c = a + (b + c) just a + b + c. The same
holds for products of any number of elements. One can easily show (see
Problem 1.1) that a right identity e is also a left identity, namely, ea = a,
for any element a of G.
Likewise, a right inverse a-1 of a is also a left inverse, a-la = e. Hence
the inverse of a-1 is simply a. Moreover, it follows that both the identity
and the inverse are unique. This allows the use of the notation of the
notation of algebra such as a1+1 = a"'a, with al = a, for any natural
number m. Negative powers of a are introduced by a-' = = (a- 1)m, ao = e.
Hence aPa9 = aP+q, and (aP)4 = aP4, where p and q are integers.
An example of a group is the set of all nonzero rational numbers, if the
rule of combination is ordinary multiplication. The identity is the number
1

Another example of a group whose elements are not numbers is the

aggregate of rotations of a plane or of space about a fixed point. Two
rotations a and b are combined by performing the rotations successively. If
b is carried out first and then a, the same result, i.e. the same final position
of all points of the space, may also be obtained by a single rotation, denoted
by ab. The group of rotations in space is an example of non-abelian group
since it is not immaterial whether one performs first the rotation a and then
b, or first b and then a. The identity of the rotation group is the identical
transformation that leaves every point in its original position. The inverse
of a rotation is the rotation in the opposite sense which cancels the first
one.
1.1. REVIEW OF GROUP THEORY 3

A set H of elements of a group G is called a subgroup of G if it is a group

with the same law of multiplication which operates in G. A necessary and
sufficient condition for a subset H of a group C to be a subgroup is that
if H contains two elements a and b it must also contain the element ab-1
(see Problem 1.2).

1.1.2 Normal Subgroup and Factor Group

Let G be a group and H a subgroup, and let a and b be two elements of
G. One calles a and b equivalent, a oc b, if ab-1 is an element of H. The
group G is thus devided into classes of equivalent elements each called a
right coset of H relative to G. It follows that if A is a right coset of H and
a is an element of A then A = Ha. Moreover, every set of the form Hb is
a right coset and the subgroup H itself is one of the cosets. One can also
introduce left cosets of H, written in the form aH. They are obtained from
an equivalence relation such that a oc b if a-1b belongs to H.
A subgroup N of a group G is called an invariant or normal subgroup of
G if for every element n of N and a of G the element a-1na belongs to N.
It follows that a necessary and sufficient condition for right and left cosets
of a subgroup N to coincide is that N be a normal subgroup. Every group
has at least two normal subgroups, the subgroup which includes only the
identity, and the subgroup which coincides with the group itself. A group
which has no normal subgroup except for these two subgroups is called
simple.
If N is a normal subgroup of a group G and A and B are two cosets of
N, A = Na, B = Nb, then AB is also a coset of N. The multiplication of
cosets thus defined satisfies the group axioms, and the set of all cosets is
called the factor group of G by the normal subgroup N and is denoted by
GIN.

1.1.3 Isomorphism and Homomorphism

A mapping f of a group G on another group G' is called isomorphism if
it (1) is one-to-one; and (2) preserves the multiplication. G and G' are
then called isomorphic. The inverse f-1 of an isomorphism f is itself an
isomorphism. An isomorphism of a group onto itself is called automorphism.
The aggregate of all automorphisms of a group forms a group.
A mapping f of a group G on another group G' is called homomorphism
if it preserves the operation of multiplication. The set N of all elements of
G which go over into the identity of G' under the homomorphism is called
4 CHAPTER 1. INTRODUCTION TO GROUP THEORY

the kernel of the homomorphism. If the kernel coincides with the identity
of G then the homomorphism is an isomorphism. It follows that N is a
normal of G, and G' is isomorphic to GIN. The isomorphism between G'
and GIN is called the natural isomorphism.
The mapping f of a group G on GIN defined by associating with each
element a of G the element f (a) = A of GIN containing a is a homomor-
phism, called the natural homomorphism of a group on its factor group. If f
is a homomorphism of a group G on another group G' and H is a (normal)
subgroup of G, then f (H) is a (normal) subgroup of C'. If f is a homo-
morphism of a group G on another group G', and g is a homomorphism of
G' on a third group G", then the mapping gf is a homomorphism of G on
G".
One finally notes that if f is a homomorphism of a group G on part
of another group G' then the set of all elements of G' which are images of
elements of G forms a subgroup of G'. Also, if f -1 (H') is the set of all
elements of G which go into H' C G' under the homomorphism f, and if
H' is a (normal) subgroup of the group G', then f -1 (H') is also a (normal)
subgroup of the group G.

1.2 The Pure Rotation Group SO(3)

A linear transformation g of the variables x1, x2, and x3i which leaves the
form xi + x2 + x3 invariant, is called a three-dimensional rotation. The
aggregate of all such linear transformations g forms a continuous group,
which is isomorphic to the set of all real orthogonal (namely, ggt = 1,
where gt is the transposed of g) 3-dimensional matrices and is known as
the three-dimensional rotation group. One can easily show that the deter-
minant of every orthogonal matrix is equal to either +1, in which case the
transformation describes pure rotation, or to -1, in which case it describes
a rotation-reflection. The aggregate of all pure rotations forms a group,
which is a subgroup of the 3-dimensional rotation group, and is known as
the pure rotation group. We will be concerned with the 3-dimensional pure
rotation group. This group is denoted by us by SO(3). (For more details,
see in the sequel.)

1.2.1 The Euler Angles

Let g be an element of the group SO(3), i.e., a 3-dimensional orthogonal
matrix with determinant unity. It is well known that one then can express
each such element in terms of a set of three parameters. An example of
1.2. THE PURE ROTATION GROUP SO(3) 5

such parameters is that of the familiar Euler angles, which are defined as
the three successive angles of rotation describing the transformation from
a given Cartesian coordinate system to another one by means of three
successive rotations performed in a specific sequence.
The sequence will be started by rotating the original system of axes X
by an angle 01 clockwise about the z axis. The new coordinate system will
be denoted by E. One then has

E= g (41) X, (1.1)

where the orthogonal matrix g (01) is given by

cos4i -sinol 0
g (01) = sin 01 cos 01 0 (1.2)
0 0 1

We use the notation according to which X = (x, y, z) = (xl, x2i x3),

- 17, (), and X' = (x', y', z) _ (xI x2, xs), = W , 71, S .
In the second stage the intermediate axes E are rotated about its axis
clockwise by an angle 0 to another intermediate set which is denoted 8',
thus one has
"' = g (0) X', (1.3)
where the orthogonal matrix g (0) is given by

1 0 0
9(0) = 0 cos 0 - sin B (1.4)
0 sine cos 0

The ' axis is called the line of nodes. Finally the E' axes are rotated
clockwise by an angle (02 about the ' axis to produce the desired X' system
of axes,
X' = g (02) ,=, (1.5)
where the orthogonal matrix g (02) is now given by

cos (02 - sin 02 0

g (02) = sin 02 cos 02 0
0 0 1

The matrix of the complete transformation

X' = gX, (1.7)

6 CHAPTER 1. INTRODUCTION TO GROUP THEORY

is given, therefore, by the product of the matrices g = g (02) g (0) g (0l).

Hence it is given by

COS 12 cos 01 - cos 2 sin 01 sin 02 sin 6

-cos9sinclsin(b2 -cos0cos0lsin(62
g= sin 02 cos 0 1 - sin 02 sin 01 - cos 02 sin 0 . (1.8)
+ cos 0 sin 01 cos 02 +cos 0 cos 01 cos 02
sin9sin1 sin0cos01 cosh

The angles 01, 9, 02 are independent parameters, fully determining the

rotation g. They are known as the Euler angles. By their definition, one
- - -
has 0 < q5l < 27r, 0 < 0 < 7r, and 0 < 02 < 27r for the intervals of the angles
01, 9, and 02.

1.3 The Special Unitary Group SU(2)

Rotations can also be described by unitary matrices of order two and de-
terminant unity. The aggregate of all such matrices provides a group which
is usually denoted by SU(2). The relation between the groups SO(3) and
SU(2) can be established as follows.

1.3.1 Homomorphism between the Groups SO(3) and

SU(2)
Let xl and xk, with k, l = 1, 2, 3, denote the coordinates of two Cartesian
frames related by the transformation

xk = gklxl,
where gkl are elements of the matrix g ESO(3), and repeated indices means
summation from 1 to 3. With each coordinate system xk one associates a
2 x 2 Hermitian matrix P defined by

P=x 0, z x+iy
X - x iy z ,
(1.10)

where ak are the familiar Pauli spin matrices,

-(1 0)' a2
-(0
al 0 1 0 2

-i 0)' a3 1 0
-1 (1.11)
1.3. THE SPECIAL UNITARY GROUP SU(2) 7

In terms of the matrix P one requires that the coordinates transform

according to the formula
P' = uPut, (1.12)

where u is an element of the group SU(2),

P'=xjr1, (1.13)

and ut is the Hermitian conjugate of the matrix u. The relations between

u of SU(2) and g of SO(3) are given by

grs = 1 Tr (UruJsut) (1.14)

u=T(1+QrQsgrs)/2(1+Tl g)1/2, (1.15)

where Tr stands for trace.
Accordingly, to each rotation g of the group SO(3) there correspond,
by Eq. (1.15), two matrices ::Fu of the group SU(2) and, conversely, to
each unitary matrix u of the group SU(2) there corresponds, by Eq. (1.14),
some rotation g of the group SO(3). It thus follows that the group SU(2)
is homomorphic to the group SO(3) (see Section 1.1). For example, the
unitary matrices corresponding to the rotations g (01), g (0), and g (02)
given by Eqs. (1.2), (1.4) and (1.6) are easily found, using Eq. (1.15).
They are given by
et0i/2 0
u(01) _ T ( e-L0i/2 ) (1.16a)
0

0 0
cos 2 isin 2
U (0) = T 0 0
(1.16b)
i sin - cos -
2 2
and
etim2/2 0
u(0z)=T-( 0 e-i02/2 (1.16c)

A general rotation g, described by the matrix (1.8), will then correspond

to the unitary matrix u = U (02) U (0) u (01), and is thus given by

u=
cos eisin 22 (1.17)
e-i(m2+m1)/2
isin 2 e-=(02-01)/2 cos
8 CHAPTER 1. INTRODUCTION TO GROUP THEORY

1.4 Invariant Integrals over Groups

A function y = f (g) is said to be defined over the group G if to each
element g of G there corresponds a number y. If the group is taken to be
the rotation group SO(3) and one uses the Euler angles as parameters then
f (g), where g c SO(3), becomes simply a function of the angles 01, 0, 02,
i.e.
f (9) = f (01, 0, 02) . (1.18)

The function f then satisfies

f (01 + 27r, 8, -02) = f (01, 0, 02), (1.19a)

f (01, 0, 02 + 27') = f (01, 0, 02) . (1.19b)

1.4.1 Invariant Integral over the Group SO(3)

The integral f f (g) dg is then called invariant integral of the function f (g)
over the group SO(3) if it satisfies

jf(g9o)dg= ff(gog)dg = ff(g)dg (1.20)

for any go E SO(3), and

f f (9-1)d9= f f(9)dg. (1.21)

The expression dg is called a measure. When the Euler angles are used
to parametrize the elements g of the group SO(3), one can write dg in terms
of the angles 01, 0, (02 as

dg = sin 9d01d0dc2. (1.22)

82
One then can easily verify that it satisfies

f dg = 1. (1.23)

The integration limits extend over the whole domain of definitions of the
variables, i.e., 0 < c51 < 27r, 0 < 0 < 1r, and 0 < 02 < 27r.
1.4. INVARIANT INTEGRALS OVER GROUPS 9

1.4.2 Invariant Integral over the Group SU(2)

The concepts of functions defined over the group SO(3) and invariant inte-
grals defined over the rotation group SO(3) can easily be extended to the
special unitary group SU(2). Again, a function f (u) defined over the group
SU(2) can be considered as a function of the angles 01, 9, 02, i.e.,

f (u) = f (01, 9, 02) (1.24)

if the Euler angles are used for parametrization. The analogous periodicity
conditions to those of Eq. (1.19) for functions defined over SO(3) will now
be
f (01 + 47r, 9, 02) = f (01, 0, 02), (1.25a)

f (01, 0, 02+47r) = f (01, 0, 02), (1.25b)

f (01 + 2ir, 9, 02 + 27r) = f (01i 9, 02) . (1.25c)

The invariant integral over the group SU(2) then satisfies

f f (uuo) du = Jf(uou)du=Jf(u)du (1.26)

for any u E SU(2), and

uo
f f (U-1) du =j f (u) du. (1.27)

The measure du can then be expressed in terms of the Euler angles as

r2 sin 9d71 dOdO2.

du = -1767- (1.28)

It can be shown that it satisfies

du = 1. (1.29)
J
The integration limits here will be: 0 < 1 < 47r, 0 < 0 < ir, and 0 < 02 <
27r.
In the next chapter the theory of representations of groups is given and
applied to the rotation group.
10 CHAPTER 1. INTRODUCTION TO GROUP THEORY

1.5 Problems
1.1. Show that a right identity e is also a left identity, namely, ea = a, for
any element a of a group G. Show also that a right inverse a-1 of a is also
a left inverse, namely, a-1a = e.
Solution: The solution is left for the reader.
1.2. Show that a necessary and sufficient condition for a subset H of a
group G to be a subgroup is that if H contains two elements a and b it
must also contain the element ab-1.
Solution: The solution is left for the reader.

1.6 References for ]Further Reading

M. Carmeli and S. Malin, Representations of the Rotation and Lorentz
Groups (Marcel Dekker, New York and Basel, 1976).
C. Chevalley, Theory of Lie Groups (Princeton University Press, New Jer-
sey, 1962). (Section 1.1)
L.P. Eisenhart, Continuous Groups of Transformations (Dover Publica-
tions, Inc., New York, 1961). (Section 1.1)
H. Goldstein, Classical Mechanics (Addison-Wesley Publishing Co., Read-
ing, Mass., 1965). (Section 1.3)
M.A. Naimark, Linear Representations of the Lorentz Group (Pergamon
Press, New York, 1964). (Sections 1.3, 1.4)
L. Pontrjagin, Topological Groups (Princeton University Press, Princeton,
New Jersey, 1946). (Section 1.1)
A. Salam, The formalism of Lie groups, Lecture Notes, 1960.
B.L. van der Waerden, Modern Algebra (Fredric Ungar Publishing Co., New
York, 1953). (Section 1.1)
A. Well, Actualites Sci. Ind., No. 869 (1938); L'integration dans les groups
topologiques et ces applications (Hermann et Cie., Paris, 1940). (Section
1.4)
Chapter 2
Representation Theory
In the last chapter the important concept of groups was discussed. In this
chapter the theory of representations of groups is given and applied to the
rotation group and the group SU(2). The spinor representation of the group
SU(2), along with the matrix elements, are then given. This subsequently
followed by finding the differential operators of the rotations. The more
complicated theory of infinite-dimensional representations is briefly given
in the last section of the chapter.

2.1 Some Basic Concepts

In this section the fundamentals of the theory of finite-dimensional repre-
sentations are given. For more details the reader is reffered to the books
of Naimark, of Gelfand, Graev, and Vilenkin, and of others given in the
suggested references at the end of the chapter. The more complicated the-
ory of infinite-dimensional representations is also given in the last section
of the chapter.

2.1.1 Linear Operators

Let S be a linear space and x a vector in it. A function A (x) is called an
operator in S if for any vector x of S there corresponds a vector y = A (x) of
S. An operator A in S is then called linear if A (x + y) = A (x) + A (y) and
A (ax) = aA (x), for any x, y of S and a complex number a. Addition of two
operators A and B is defined in the space S by (A + B) x = Ax+Bx for all
vectors x of S. Similarly, multiplication by a number a and multiplication

11
12 CHAPTER 2. REPRESENTATION THEORY

of operators A and B in the space S are defined by (aA) x = a (Ax) and

(AB) x = A (Bx). If, furthermore, A and B are linear operators in S then
A + B, aA, and AB are also linear operators in S.
Linear operators in a finite-dimensional space S can be represented as
matrices by introducing a basis, e1, ..., e, in S. Accordingly, if A is a linear
operator in the space S, then Aek can be written as a linear combination
of el, ..., en, or,
n
Aek=EA3k (k = 1,..., n). (2.1)

j-1
A,k are the elements of the matrix representing the operator A relative
to the basis el, ..., en. One can show that the operator A is completely
determined by its matrix Att. Furthermore, the operations of addition,
multiplication by a number, and multiplication of operators correspond to
the same operations of their matrices relative to a fixed basis.

2.1.2 Finite-Dimensional Representations

Let G be a group and g an arbitrary element of G. A correspondence
g --+ D (g) of each element g of the group G to a linear operator D (g) in a
finite-dimensional space S is called a representation if: (1) D(gl)D(92) =
D (9192) and (2) D (e) is the unit element in S, where e is the identity
element of G. The space S is called the space of representation and its
dimension is called the dimension of the representation. (For the more
general definition of a representation see Subsection 2.5.4.)
Two finite-dimensional representations g -k D1 (g) and g -+ D2 (g) of
the group C in two spaces Sl and S2 having the same dimensions, respec-
tively, are called equivalent if bases in the spaces Si and S2 can be chosen
so that the matrices of the operators D1 (g) and D2 (g) are identical. A
subspace S' of the space S is called invariant with respect to the represen-
tation g --+ D (g) if for every vector x of S' one finds that D (g) x is also
a vector in S' for all elements g of the group G. If there are no invariant
subspaces in the space S with respect to the representation g -+ D (g),
except for the trivial cases of the null subspace and the whole space, the
representation is then called irreducible.
A representation g -+ D (g) of a group C is called continuous if D (g) is a
continuous operator function on the group G. (An operator function D (g)
is called continuous on a group G if the elements of the matrix of D (g),
relative to a fixed basis, are continuous functions on G. This definition
of continuity of D (g) does not depend on the choice of the basis since
2.2. REPRESENTATIONS OF SO(3) AND SU(2) 13

the matrix elements relative to another basis are linear combinations, with
constant coefficients, of the matrix elements relative to the original basis.)
Only continuous representations will be considered here.

2.1.3 Unitary Representations

A linear space is called Euclidean if from each two vectors x and y of it
one can define a function, called the scalar product of x and y, denoted by
(x, y), which satisfies:
(1) (x, x)>0, (x, x) = 0 if and only if x = 0;
(2) (y, x) _ (x, y);
(3) (ax, y) = a (x, y);
(4) (xl + x2, y) = (x1, y) + (x2, y)
One can show that a scalar product can be introduced in every finite-
dimensional space. (The infinite-dimensional case is discussed in the ap-
pendix at the end of the chapter.)
An operator D in a finite-dimensional Euclidean space E is called uni-
tary if it preserves the scalar product, namely, (Dx, Dy) = (x, y) for all
x, y of the space E. A representation g -* D (g) is called unitary if all its
operators D (g) are unitary.
In the following we find the irreducible representations of the three-
dimensional pure rotation group. This is done by Weyl's method which
makes use of the homomorphism of the special unitary group of order two
onto the rotation group. The representations are expressed in terms of the
angle of rotation in a specified direction and the spherical angles of the
direction of the rotation.

2.2 Representations of SO(3) and SU(2)

We have seen that the unimodular unitary group of order two, SU(2), is
homomorphic to the pure rotation group SO(3) such that to every rotation
g of SO(3) there correspond two matrices +u and -u of SU(2) and, con-
versely, to every element u of SU(2) there corresponds some rotation g of
SO(3).

2.2.1 Weyl's Method

It thus follows that the description of the representations (see Section 2.1) of
the group SO(3) is equivalent to that of the group SU(2); a representation
g --+ D (g) of the group SO(3) is single- or double-valued according to
14 CHAPTER 2. REPRESENTATION THEORY

whether or not D (u) is equal to D (-u). The use of the group SU(2) for
finding the representations of the group SO(3) was originally suggested by
H. Weyl and has been wildly adopted when the Euler angles are used to
parametrize the groups. The advantage of Weyl's method is in giving the
double-valued representations along with the proper representations. The
double-valued representations are important in physical problems dealing
with spin-like properties of particles whose spins are half integers.
We point out that, by using Weyl's method, one can obtain a general
invariant result that is a function of the element u c SU(2), valid for any
parametrization one uses to describe the rotation. To find the representa-
tions of the group SO(3) in terms of a specific set of parameters, one has
merely to express u in terms of these parameters, as is the case when the
Euler angles are adopted.
In addition, by having the results as functions over the group SU(2),
certain relations will be obtained which are invariant under change of the
parameters. As an example, the orthogonality relations between the matrix
elements of the irreducible representation can be written in the form of an
invariant integral over the group SU(2). Hence the relations are valid for
any parametrization.

2.2.2 Infinitesimal Generators

An orthogonal matrix describing a rotation with an angle 0 about some
direction
n = (sin 0 cos o, sin 0sin 4, cos B) (2.2)
is given by

9rs = brs cos,' + nrns (1 - cos 1') - Erstnt sin 0, (2.3)

where r, s and t take the values from 1 to 3. Rotations gi (0), 92 (0) and
g3 (0) around Ox1i Ox2 and OX3 axes are then obtained from Eq. (2.3) by
putting the proper values for the polar angles 0 and 0. These matrices are
given by
1 0 0
g1 (vi) = 0 cos,0 - sin,0 (2.4a)
0 sin -0 cos,0

cos 0 0 sin o
92 (') = 0 1 0 (2.4b)
-since 0 cos 0
2.2. REPRESENTATIONS OF SO(3) AND SU(2) 15

coso -sine 0
93 (z/,) = sin,0 cos 0 (2.4c)
0 0 1

Infinitesimal Matrices
The infinitesimal matrices gr, corresponding to rotations about the axis
Oxr are defined by
9r= IL d,0(&) J .0=o
Ldgr
and satisfy the commutation relations

[9r, 9s] = Erst9ti

where [gr, gs] = grg3 - gage

The matrices gr's are given by
0 0 0
91 = 0 0 -1 , (2.7a)
0 1 0

0 0 1

92 = 0 0 0 , (2.7b)
-1 0 0

0 -1 0
93 = 1 0 0 (2.7c)
0 0 0

and one has the relation

gr (,0) = exp (mgr)

Let us denote a representation of the group SO(3) in an n-dimensional

Euclidean space R by g D (g)) and, for convenience, we denote

Ar (W)
^ D (9r) ('Y) . (2.9)

Ar (v') are called the basic one-parameter groups of the given representation
and define one-parameter groups of operators that satisfy Ar (ibi) Ar ('02) =
Ar (01 + 02); they are differentiable functions of 0 and may be expanded
as Ar (0) = exp (OAr), where Ar is defined by Eq. (2.10)
16 CHAPTER 2. REPRESENTATION THEORY

2.2.3 Basic Infinitisimal Operators

The basic infinitesimal operators of the representation are then obtained by
Ar dA, (V))1
lL JJ
(2.10)
,p-o

A representation of the group SO(3) is uniquely determined by its basic

infinitesimal operators Ar. The determination of all the finite-dimensional
representations of the group SO(3) is based on the fact that the operators Ar
satisfy the same commutation relations that exist among the infinitesimal
matrices gr:
[Ar, As] = ErstAt. (2.11)
The operators Ar are skew-Hermitian, At = -Ar, since, without loss of
generality, every finite-dimensional representation of SO(3) can be consid-
ered to be unitary. (An operator B in a finite-dimensional Euclidean space
E is called adjoint to the operator A in the same space if (Ax, y) = (x, By)
for all x, y of E. The adjoint of an operator A is usually denoted by At. It
can be shown that for any linear operator A there exists one and only one
adjoint operator At, and that the adjoint operator to At is A. An operator
A is called Hermitian if At = A. An operator A can be shown to be unitary
if and only if AtA = 1.)

2.2.4 Canonical Basis

Defining the new operators

iAl A2, (2.12a)

L3 = iA3, (2.12b)

one then finds for the commutation relations of the infinitesimal generators
L+, L_, and L3 the following:

[Lw, L3] = L:F, (2.13a)

[L+, L-] = 2L3, (2.13b)

L+ = L-, L3 = L3. (2.13c)

The problem of determining the representation is then reduced to the

determination of the operators LT- and L3 satisfying the conditions (2.13).
2.2. REPRESENTATIONS OF SO(3) AND SU(2) 17

This problem is solved by the following: every finite-dimensional represen-

tation of the group SO(3) is uniquely determined by a non-negative integer
or half-integer j, the weight of the representation.
The space of the representation corresponding to such a number j has
the dimension 2j + 1; the operators L:F and L3 of the representation are
given relative to its canonical basis f_j, f_,+1, ..., f; by
1/2
Lfm = [(7 + m) (.7 m + 1)] fmf1, (2.14a)

L3 fm = mfm (2.14b)
where m=-j,-j+1, ...,j.
It also follows that for each j there corresponds an irreducible repre-
sentation of SO(3). If the operators Lv and L3 of a representation of
SO(3) in a (2j + 1)-dimensional space are given relative to some basis
f_., f_j}1, ..., f3 , then by Eqs. (2.14) that representation is irreducible.

2.2.5 Unitary Matrices Corresponding to Rotations

We now find the unitary matrix u corresponding to the rotation g, of Eq.
(2.3). The matrices u and g are related by Eqs. (1.14) and (1.15). A direct
calculation then gives:
w
cos + i sin cos B i sin sin 9e'O
u=+ 2 (2.15)
2
i sin 2
sin Oe-io cos 2 i sin 2 cos B

This is the unitary matrix u E SU(2) corresponding to a rotation with an

angle 0 around the direction n specified by the spherical angles B and 0.
The corresponding matrix, when the Euler angles are employed, was given
in Eq. (1.17). It will be noted that

u (-V, 9, 0) = u-1 (V, 0, 0) . (2.16)

The unitary matrices u1 (i)), u2 (V)) and u3 (0) corresponding to the

rotations g1 (v/i), 92 (0) and g3 (0) around the axes of coordinates Ox1i
OX2 and Ox3i can be obtained from Eq. (2.15) by putting the appropriate
values for the angles 0 and 0. They are:

cos i sin
u1() (2.17a)
i sin cos 2
2
18 CHAPTER 2. REPRESENTATION THEORY

cos - - sin 2
U2 (Y) _ (2.17b)
sin 2 cos 2

el1,/2 0
1/
(2.17c)
U3 (0) = T 1 0 e-'i'/2 )
Using these matrices, the operators A, (0) of the group SU(2) will be de-
termined in the next chapter.
The infinitesimal matrices u,., corresponding to rotations around Ox,.,
are given by
u,. _ [dui.('b)1 (2.18)
dpi

and explicitly,

ul l i
_ 1r0 i
0
u2=:F 2
1 0
1
0
1
u32 (0 1 i 0
-i
. (2.19)
These are related to the Pauli matrices, Eq. (1.11), by
z r
ur=F2o (2.20)

2.3 Matrix Elements of Representations

A matrix u of the group SU(2) can be considered as that of a linear trans-
formation of the space of all pairs of complex numbers (61, e2):
2
61P
r, upg6q (p=1,2). (2.21)
q=1

A representation of the group SU(2) can be obtained if one considers several

pairs (i, C2) , ..., (Ck, k) and forms all products ...Ckk, letting p1, ..., Pk
take the values 1, 2, independently. Under the transformation (2.21), this
product transforms like
2
CC
upigi...UpkgkSgl...Qq (2.22)
qi ,....qk=1
2.3. MATRIX ELEMENTS OF REPRESENTATIONS 19

The product P' may be considered as a vector in the linear space Rk

of all 2k complex numbers The linear transformation D(k) (u) of
the space Rk is then given by
2
61Pi...Pk = v'Pk9k691...gk
(2.23)
E uP191
9i.... ,9k=1

2.3.1 The Spinor Representation of the Group SU(2)

The correspondence u -i D(k) (u) is a representation of the group SU(2),
not irreducible in general, since the subspace Sk of Rk of all symmetrical
vectors is invariant with respect to all the operators D(k) (u). The cor-
respondence u -p D(k) (u) is irreducible, however, in the subspace Sk. We
denote this representation by Zk. It is called the spinor representation of
the group SU(2) and is of weight k/2.
An equivalent realization of the representation Zk is obtained if one
identifies the subspace Sk with the (k + 1)-dimensional space of homoge-
neous polynomials p (zi, z2) of degree k in the two complex variables z1 and
z2 and sets up a one-to-one correspondence between of Sk and p (z1, z2)
in the form
2
Cpl...pkzp
P (z1, z2) = E S 1 "'zPk (2.24)

P1,...,pk=1
The operator D(k) (u) for this new realization in the space of polynomials
Sk is then given by
D(k) (u) P (z1, z2) = p (zi, Z2'), (2.25a)

where
2

zq upgzp (9 = 1, 2). (2.25b)

P=1

Introducing now a new variable z = zl/z2, the polynomial p (zl, z2) can
then be written as z2 p (z), where p (z) is a polynomial in the variable z
of degree not exceeding k. The operators D(c) (u) of the representation zk
are, accordingly, given by

D(k) (u) P (z) = (u12z + u22)k P l u12z

( u1 1z + U21 (2.26)
+ u22
This equation gives, in particular, the operators A, (ii) = D (u1.(7i)) when
the matrices ur (0), Eqs. (2.17), are used. (For the determination of the
20 CHAPTER 2. REPRESENTATION THEORY

operators A, (0), one needs u,. (0) only for small values of 'i. The signs in
Eqs. (2.17) are determined by the conditions limu,. (V)) = 1 when 1i - 0;
hence the + sign must be used.)

2.3.2 Matrix Elements of Representations

It follows that every irreducible finite-dimensional representation of the
group SU(2) is uniquely determined by some non-negative integer or half-
integer j = k/2, the weight of the representation. Conversely, for any non-
negative integer or half-integer j, there exists an irreducible representation
of the group SU(2) of weight j. A representation of weight j can be realized
as the spinor representation Zk, where k = 2j; and every finite-dimensional
irreducible representation of the group SU(2) is equivalent to one of the
representations Zk.
The functions

fm (z) _
(-z)' (2.27)
M)!11/2'
[(j - m)! (j +
where m = -j, -j + 1, ..., j then form a canonical basis for the represen-
tation zk in the space Sk. Using Eq. (2.26), one finds
j
Dlkl (u) fn (z) _ D1mn (u) fm (z) , (2.28)
m=-j
where Dan (u) are the matrix elements of the operator D (u) of the ir-
reducible representation of weight j relative to the canonical basis, which
corresponds to an arbitrary rotation g. Its explicit expression is

(-1)2j-m-n
(j - m)! (j +m)! 1/2
D nn (u) _
[ (j -n)! (j +n)! I
.) +n ) "a11"
712-m-au21 j-+L-a m+n+a
(2.29)
a (j-m-a u22 ,

where a runs from a = max (0, -m, -n) to a = min (j - m, j - n), and
m m!
(2.30)
n (m-n)!n!
In Eq. (2.29) the indices m and n take the values -j, -j + 1, ..., j and
j = 0, 1/2, 1, 3/2, 2, ... .
2.3. MATRIX ELEMENTS OF REPRESENTATIONS 21

It will be noted that (-u) = (-1)22 Dj.n (u). Thus the represen-
tation is single-valued for integer j and double-valued for half-integer j. In
the sequel the matrix u of Eq. (2.15) will be taken with the + sign.
To find the matrix elements (2.29) in terms of the variables Vi, 0 and 0
we simply substitute for up, their expressions as functions of these variables
as given by Eq. (2.15). (One can easily find the expression of Di in terms
of Euler's angles.) One obtains

(-1)2j-m-n (? - m)! (j + m)! 1/2

Lmn (Y'' 0, o) _
1 (j - n)! (j + n)! I
Y' m-n/ Y' \m}n
x (isin sin Be-im (cos 7(21 - i sin 2 cos B I S (j, m, n; x) .
2
(2 . 31)
Here we have used the notation
2n`-1
S (j, m, n; x) = (j - n) ! (j + n)

xua!(j - n - )!(j ( m 1) a)!(a+m+n)!'

(2 32)
.

where x is defined by
x = 1 - 2 sine sine B. (2.33)

2 m, n; x) is equal to the Jacobi

It will be noted that the function S (j,
polynomial p, 13(x) when s= =j (I m+ n l+ j m- n I ), a =1 m- n 1,
2
and1i=Im+n1.

2.3.3 Properties of D;,,n (u)

Finally we discuss the properties of the matrices Line (u).
One first notices that the matrices Di (u) are unitary. The correspon-
dence u -> D! (u) is a representation of the group SU(2). Therefore one
has

Dmn (u1u2) _ Jymn' (u1) Un'n (u2) . (2.34)

n'=-,7
Furthermore, one has

D' (u-i) = [D3 (u)] -1 = [D' (u)] 1 , (2.35a)

22 CHAPTER 2. REPRESENTATION THEORY

or
(u)-1
Dmn = V. (u) . (2.35b)

Denoting now by -y the unitary matrix

^= ( e-'0/2 0
(2.36)
0 i-0/2 )
where 0 is a real number. If one applies now the representation formula
(2.26), where p (z) is taken as the basis functions fm (z) of Eq. (2.27), one
obtains

e""Ofm
D7fm (z) (z). (2.37)
(7-m) ((j+M)I
Hence the matrix Di (-y) is diagonal, and Din,,, (-y) = e2no. Furthermore,
one easily finds that

Damn (_YU) = (U), (2.38a)

D'mn (u'Y) = ei"DYmn (u) . (2.38b)

We conclude this section by giving the orthogonality relation that the

matrices Di satisfy:

f 'ftini (u) Vm2n2 (U)du= (2i1 +

1)-1
thhbM1m26nln2. (2.39)

Relations similar to (2.39) are valid for any compact group. See, for exam-
ple, the book of Pontrjagin.

2.4 Differential Operators of Rotations

We are now in a position to find the differential operators corresponding to
infinitesimal rotations about the coordinate axis, namely, the operators A1i
A2 and A3 and, consequently, the operators LT and L3. These operators
are well known in the literature when the Euler angles are employed. We
here derive these operators in terms of the variables 0, 0 and 0.
2.4. DIFFERENTIAL OPERATORS OF ROTATIONS 23

2.4.1 Representation of SO(3) in Space of Functions

Let g -+ D (g) be an irreducible representation of weight j of the group
SO(3) and let Dmn = Dmn be its matrix elements. We consider these
elements as functions of the rotation g, Dm.n = Dmn (g). Since g -+ D (g)
is a representation, one has

D (99) = D (g) D (g') . (2.40)

In terms of matrix elements, the last relation is

i
Dmn (99) _ q Dmq (9) Dqn (9') , (2.41)
q=-i
where Dmn (99') are the matrix elements of the operators D (gg').
Define now a transformation U such that

U (9') Dmn (9) = Dmn (99') . (2.42)

Comparing Eqs. (2.41) and (2.42) we obtain

j
U (9) Dmn (9) _ q Dqn (9) Dmq (g) . (2.43)
4=-j
Furthermore, one can show that

U(9)U(g")=U(9g"). (2.44)

It thus follows that the transformation U (g') realizes a representation of

the group SO(3) in the space of 2j + 1 functions of the mth row of the
matrix D (g) [compare Eq. (2.28)], and that the matrix elements of U (g')
are Dqn (9').
The representation g' -+ U (g') in the space of functions Dmq (g), q =
-j, -j + 1, ..., j, is irreducible, and the Dmq (g) form a canonical basis in
this space. Hence the operators L:F and L3 of this representation satisfy
the relation (2.14), i.e.,

L Dm. (g) _ [(7 n + 1) (9 n)] 1 /2 D n,nl (9), (2.45a)

L3Dmn (9) = (9) . (2.45b)

24 CHAPTER 2. REPRESENTATION THEORY

2.4.2 The Differential Operators

To find the operators A, we take g' as the rotation through some angle
a around the axis Ox, and expand the relation (2.42) in powers of a.
Expansion of Dmn (gg'), which we denote by Dmn 8, ) , gives

Dmn ( ; & ) = Dmn (b, 0, 0)

i,&

BDmn d I) BD,,,, dB aDmn d

+a + (2.46)
I a da + aB da + 8o da
-1 =o

To determine the infinitesimal operators A, we have to determine the func-

tions

dal-o [di , and I

da]
._
(2.47)
J ` o

for each rotation.

Now the 3x3 matrix of the rotation g is a function of the angles V), 9
and 0 which, by Eq. (2.3), has the form

cos I/i sin 2 0 cos 0 sin q5 sin 9 cos 9 cos q5

+ sin' 0 cost , x (1 - cos V)) x(1-cosO)
x (1 - cos7l,) -cos0sinV) +sin0sinq5 sin 1'
sin20sin0cos cos V) sin0cos0sin0
x (1 -cos'O) +sin' 0sin2.0 x (1 - COs,') . (2.48)
+cos0sin0 x (1 - cos'O) -sin9cos0sinVi
sin 9 cos 9 cos 0 sin 9 cos 0 sin 0 cos 'O
x (1 - cos O) x (1 - cos V)) + cos' 8 (1 - cos V))
-sinOsin0sinii + sine cos 0 sin z/' l
The matrix of rotation gg' is given by some angles Vii, 8 and t which depend
on the rotation angle a and which are equal to 0, 9 and 0 when a = 0. It
will also be noted that expansion of the matrix gg' in a power series in a
gives

[ddfl g +.. .
99 = 9 (0, 9, 0)+a 18th + 80 [da] + aS [da] ]
a=0 &=0 a=0
(2.49)
2.4. DIFFERENTIAL OPERATORS OF ROTATIONS 25

To find the infinitesimal operator Al we identify g' with the rotation

with angle a around Ox1 given by
1 0 0
91 (a) = 0 cos a - sin a (2.50)
0 sin a cos a

Therefore
1 0 0 0 0 0
g1 (a) = 0 1 0 +a 0 0 -1 +. (2.51)
0 0 1 0 1 0

As a consequence we obtain for the product of g with gi:

0 913 -912
991 = 9 (I, 0, 0) + a 0 923 -922 + (2.52)
0 933 -932

On the other hand, ggl is given by Eq. (2.49) when g1 = g'. Comparing
these two expressions for gg1, we obtain equations from which the three
expressions given in (2.47) can be determined for the case of rotation about
Ox1. We obtain

2 sin 0 cos 0 sin (_sin8sin I da] + cos 0 cos I _J

L Cg=0 L a=0/
2
- cos (1 - sine B cost (2.53a)
da I = 0,
Q=0
2
2sin0sinosin (sin8cosf da +cos0sinf d;J
L a=0 L a=0)
2
- cos . (1 -sin 2 9 sine ) I ddip
o
a=o

= sin 0 (cos 0 sin 0 sin - cos 0 cos , (2.53b)

/
2 2
2cos0sin d_+cos 2 sin0 Ld_J
0(=0 a=o
2
26 CHAPTER 2. REPRESENTATION THEORY

= cos O sin 0 sin +cos 0 cos 2. (2.53c)

(One actually obtains nine equations;2only three of them are independent.

Equations (2.53) are obtained by equating the diagonal element of the ma-
trices (2.49) and (2.52).)

2.4.3 Angular Momentum Operators

The solution of Eqs. (2.53) can easily be shown to be

= cos 0 sin 9, (2.54a)

dcx
Q=o

di 1 1
= Cs in m + cos 2 cos B cos 1J , (2.54b)
d 2
L J=o

= 2 cosecB (cos B cos q5 - cot 2 sin J . (2.54c)

[da]
a=0
Using Eqs. (2.54) in Eq. (2.46), we find the operator Al corresponding to
the rotation around Oxl:

Al = cos o sin B ai + 2 (sin 0 + cot 2

cos 8 cos q5 50-

+2 cosecB (cosOcos o - cot 2 sin of a . (2.55a)

The operators A2 and A3 are found in a similar way:

a
A2 = sin o sin B Ccos o - cot cos B sin o)50-
a -2
2
+2 cosecB (cos O sin m + cot cos o) (2.55b)

a 2
A3 = cos9 - 2 cot sin BAe (2.55c)
2
Using the last three equations in Eq. (2.12) one obtains for the operators
L+, L_ and L3:
a
L = iei4' in 0 _ + 2 Ti + cot 2
cos B)
IS a70-
11 ]
2.5. INFINITE-DIMENSIONAL REPRESENTATIONS 27

+iet`o [cosec9 (cos 0 f i cot 2 (2.56a)

/ 1,
L3 = i (cos 050 - 2 cot 2 sin BYO (2.56b)
5-0 .

The operators derived above were expressed in terms of the angle of

rotation V and the spherical angles of direction of rotation 0 and 0. One
can use, however, the Euler angles and obtain the standard expressions of
angular momentum operators given in the books of Naimark and Wigner.
In the following the Lorentz group is introduced along with its infinitesi-
mal matrices and basic infinitesimal operators. The commutation relations
that these matrices and operators satisfy are also given. This is the in-
finitesimal approach to finding the representation of the Lorentz group.
Each representation is shown to be completely determined by a pair of
numbers.

2.5 Infinite-Dimensional Representations

In this section a brief review of the theory of infinite-dimensional repre-
sentations is given. For more details the reader is referred to the books of
Naimark and of Gelfand et al.

2.5.1 Banach Space

A linear space is called Euclidean if in it a function (x, y), called the scalar
product of x and y, is defined and satisfies the following:
(1) (x, x) > 0, (x, x) = 0 if and only if x = 0;
(2) (y, x) = (x, y);
(3) (ax, y) = a (x, y);
(4) (x1 + x2, y) = (x1,11) + (x2, y),
for any number a.

Normed Space
A linear space R is said to be norrned if a function, denoted by I x I, is
defined in it, which satisfies:
(1) Ix I>0, Ixl=O if and onlyifx=0;
(2) ax1=IaIIxI for any number a andanyxER;
(3) Ix+yI<Ix1+IyI foranyx,yER.
Such a function I x I is called a norm.
28 CHAPTER 2. REPRESENTATION THEORY

An example of a normed space is the aggregate C of all complex numbers

x. The norm of a complex number is taken as its modulus. Another
example is provided by the aggregate of all sequences x = {1i e2, ...} of
complex numbers S1, b2, ... for which the series I i 12 + I e2 12 +
converges. The operations in the space are defined as ax = {aEj, ae2,...}
and x + y = {S1 +771, e2 +7)2, ...} for x = {e1i 62, ...} and y = {7)1i 7)2, ...}.
The norm is defined as I x 1= {161 12 + 162 12 + } 1'2. This space is
sometimes denoted by 12.
Let R be a Euclidean space, not necessarily finite-dimensional. Then,
in R, a norm can be defined by

I X I= (x x). (2.57)

The axioms for a norm will be satisfied; in fact, the first two axioms are
trivially satisfied. To prove the triangle inequality one needs

I(x,y)I<-IxIIyI, (2.58)

which is the well-known Cauchy-Buniakovsky inequality. From the above

inequality one has

I x+y I2= (x+y,x+y) _ (x,x)+(x,y)+(x,y)+(y,y)

<Ixi2+2Ix1IyI+Iyi2=(IxI+IyD)2, (2.59)
and consequently
Ix+yISIxI+I I. (2.60)

A sequence of elements xn of a normed space R is called convergent in

norm to the element x of R if I x - xn I-+ 0 as n -+ oc. A sequence xn
of R is called fundamental if it satisfies the Cauchy condition (i.e. if for
every e > 0 there exists a number N = N (e) such that I xn - x,n. I< e for
n, m > N.) A space R is called complete if every fundamental sequence in
R converges in norm to some element x of R. A complete normed space is
called a Banach space.
Examples of complete normed spaces are the space C of all complex
numbers and the space 12, both mentioned above. An example of a non-
complete normed space is the set of all sequences x = {l;l, b, } in which
only a finite number of n is non-zero, all other operations of the space are
the same as those of the space 12.
Let S be an arbitrary set in a Banach space R. The set S obtained from
S by adding to it all the limits in norm of sequences of elements xn of S is
2.5. INFINITE-DIMENSIONAL REPRESENTATIONS 29

called the closure of the set S. A set S is called dense in R if S = R. A set

S is called closed if S = S. A closed subspace of a Banach space is itself a
Banach space.
A series x1 + X2 + of elements xn of R is called convergent and the
element x of R is called the sum of the series if x1 + x2 + + xn -4 x
as n - oo in the sense of the norm in R. A series x1 + x2 + is called
absolutely convergent if the series I xl i + I X2 I + of real numbers is
convergent. In a Banach space every absolutely convergent series converges.
This follows from the inequality

I xn+ 1 + . + xn+p ICI xn+1 (2.61)

and the fact that the space is complete.

2.5.2 Hilbert Space

A Euclidean space R, complete with respect to the norm I X I= (x x), is
called a Hilbert space.

Examples of Hilbert Spaces

The space 12 discussed in Subsection 2.5.1 is a Hilbert space if the scalar
product is defined by (x, y) = >2 ekrlk for x = {6, 2, } and y = {rll, 772, }.
Another example of a Hilbert space is the aggregate of all functions f (x),
measurable in a fixed interval (a, b) and satisfying the conditions f b I

f (x) 12 dx < oo, if the operations of addition and multiplication by a

number are defined in the usual way, and the scalar product is defined by
lb f(x)
(fl, f2) = fl (x) dx.

This Hilbert space is sometimes denoted by L2 (a, b). In the same way the
Hilbert space L2(SU(2)) is defined as the aggregate of all functions f (u)
satisfying f I f (u) I2 du < oo, where the scalar product is defined by

(fl, f2) = f fl (u) f2 (u) du.

A linear functional f (x) in a linear space R is a numerical function

satisfying f (ax) = of (x) and f (x + y) = f (x) + f (y) for any number
a and x and y of R. A linear functional in a normed space R is called
bounded if there exists a constant c > 0 such that I f (x) I < c I x I for all
30 CHAPTER 2. REPRESENTATION THEORY

x of R. The smallest number c > 0 satisfying this condition is called the

norm of the functional and is denoted by I f I. Thus I f (x) 1:51 f 11 X I.
The bounded linear functionals in R form a normed linear space where
the sum and product are defined by (fl + f2) (x) = fi (x) + f2 (x) and
(a f) (x) = Z if (x). This space is called the conjugate to the space R. It is
a complete space.
In a Hilbert space R every bounded linear functional f (x) is represented
in the form f (x) = (x, y), where y belongs to R and I f 1=1 y 1. Hence the
space R', conjugate to a Hilbert space R, may be identified with R itself,
R'=R.
Two Hilbert spaces R1 and R2 are called isometric if there exists a
linear operator U mapping RI onto R2 and preserving the scalar product,
(Ux, Uy) = (x, y) for all x, y of R1. The operator U itself is then called
isometric and it satisfies I Ux 1=1 x I for all x of R1.

2.5.3 Operators in a Banach Space

A linear operator A in a Banach space R is called bounded if there exists a
constant c > 0 such that I Ax 1< c I x I for all x of R. The smallest number
c satisfying this condition is called the norm of the bounded operator A
and is denoted by I A I. Hence I Ax I<I A II x I. If A and B are bounded
operators, then also the operators aA, A + B, and BA are bounded and
satisfy

IaAI=IaIIAI; IA+BI<IAI+IBI; IAB I<_IAIIBI. (2.62)

It then follows that every bounded linear operator A is continuous. Fur-
thermore, if two bounded operators A and B coincide on a set S which is
dense in a space R, then they coincide on the whole of R.

2.5.4 General Definition of a Representation

A mapping g -+ D (g) of a group G on a Banach space R is called a
representation if to every element g of G there corresponds a bounded linear
operator D (g) in R such that D (e) = 1 and D (9192) = D (91) D (92). A
representation g --+ D (g) in a Banach space R is called irreducible if R
contains no closed subspace (other than the null one and R itself) which is
invariant with respect to all operators D (g). This definition coincides with
that of irreducibility for the finite-dimensional one. This is so since every
finite-dimensional subspace is closed.
2.5. INFINITE-DIMENSIONAL REPRESENTATIONS 31

Let a linear space be denoted by R and let its conjugate space be denoted
by R', Then for every element x of R there exists a functional f of R' such
that f (x) =1 x I and I f 1= 1. Hence if f (x) = 0 for all f of R' then
x = 0. Furthermore, if M is a closed subspace of a Banach space R and xo
is a vector in R not belonging to M, then there exists a functional f of R'
satisfying f (xo) 54 0, and f (x) = 0 for al l x of M.
Since the conjugate space R' is a normed space, one can therefore con-
sider the linear bounded functional Fn it. Such functionals are ob-
tained, for example, if we put Fx (f) = f (x) for a fixed element x of R since
Fx (f) is a bounded linear functional in R'. A space R is called reflexive if
the functionals FF (f), for all x of R, exhaust all the bounded linear func-
tionals in R'. In other words if every bounded linear functional F (f) in R'
is given by F (f) = f (x) for some x of R. (Throughout our discussion we
consider only representations in reflexive Banach spaces.)

2.5.5 Continuous Representations

function ofapoint
in an m-dimensional space with values in R. A vector function x (t) is called
continuous in a set D in m-dimensional space if for every functional f of
the conjugate space R' the numerical function f [x (t)] is continuous in D.
A bounded linear operator function A (t) in R is called continuous in D if
for every x of R and f of R' the numerical function f (a (t) x) is continuous
in D. For example if G is a group of matrices, then it may be regarded as
a subset of m-dimensional space for a sufficiently large m. Hence one may
speak of a vector-function x (g) or an operator function A (g) as continuous
in the group G.
A representation g -+ D (g) of a group of matrices is called continu-
ous if D (g) is a continuous operator function. [Throughout the text the
term representation stands for continuous representation (unless otherwise
stated)]. It then follows that if x (t) and A (t) are vector and operator func-
tions, respectively, which are continuous in a closed bounded set D, then
the numerical functions I x (t) I and I A (t) I are bounded in that set.

2.5.6 Unitary Representations

The concept of a unitary representation in finite-dimensional spaces dis-
cussed in Subsection 2.1.3 can be generalized to infinite dimensions as fol-
lows.
Let U be an isometric operator mapping R onto itself, then U is called a
32 CHAPTER 2. REPRESENTATION THEORY

unitary operator in R. A representation g -+ D (g) of a group G in a space

R is called unitary if R is a Hilbert space and D (g) is a unitary operator
for all g of G. A representation g --> D (g) in a Hilbert space R is unitary if

(D (g) x, D (g) y) = (x, y) (2.63)

for all g of G and x, y of R.

Now let A be a bounded operator in a Hilbert space R. An operator At
is called adjoint to A if (Ax, y) = (x, Aty) for all x, y of R. One can show
that Att = A, (&A)t = aAt, (A+B)t = At + Bt, (AB)t = BtAt, and
At 1=1 A 1. An operator U is unitary if and only if UtU = UUt = 1. The
operator A-1 is called inverse to A if AA-1 = A-1A = 1. Hence a unitary
operator satisfies Ut = U-1. An operator A is called Hermitian if At = A.
A Hermitian operator P is called a projection operator if p2 = P.
Finally, let R1, R2, be closed, mutually orthogonal, subspaces of a
Hilbert space R. The aggregate of all sums x = x1 + X2 + .. of con-
vergent series of elements xk E Rk is called the orthogonal sum of the
Hilbert spaces R1, R2, , and is denoted by R , It follows
that R1 R2 is a closed subspace of R. If En is a projection op-
erator in R onto R1R2 . . . Rn, then Enx = xl + + xn for any
vector x = X1 + X2 + of R, where Xk E Rk.
The bounded linear operator A in a space R is called the orthogo-
nal sum of the operators Ak in Rk, denoted by Al A2 , if R =
R1R2 and Ax =
A unitary representation g --+ D (g) of a group G in a Hilbert space R is
called the orthogonal sum of the representations g --> D(k) (g) in the closed
subspaces Rk if D (g) = D(1) (g) + D(2) (g) + for all g of G.
In the next chapter the Lorentz group and the group SL(2,C) are dis-
cussed in details.

2.6 References for Further Reading

M. Carmeli, Representations of the three-dimensional rotation group in
terms of direction and angle of rotation, J. Math. Phys. 9, 1987-1992
(1968). (Sections 2.2, 2.4)
M. Carmeli and S. Malin, Representations of the Rotation and Lorentz
Groups (Marcel Dekker, New York and Basel, 1976).
I.M. Gelfand, M.I. Graev and I.I. Pyatetskii-Shapiro, Representation The-
ory and Automorphic Ftnctions (W.B. Saunders Co., Philadelphia, (1969).
2.6. REFERENCES FOR FURTHER READING 33

(Section 2.5)
I.M. Gelfand, M.I. Graev and N.Ya. Vilenkin, Generalized Functions, Vol.5:
Integral Geometry and Representation Theory (Academic Press, New York,
1966) (Sections 2.1, 2.5).
I.M. Gelfand, R.A. Minlos and Z.Ya. Shapiro, Representations of the Ro-
tation and Lorentz Groups and their Applications (Pergamon Press, New
York, 1963). (Section 2.1)
I.M. Gelfand and Z.Ya. Shapiro, Representations of the group of rotations
in three-dimensional space and their applications, Usp. Mat. Nauk 7, 3-117
(1952); Amer. Math. Soc. Transl. Ser.2., 2, 207-316 (1956). (Section 2.1)
H.E. Moses, Irreducible representations of the rotation group in terms of
the axis and angle of rotation, Ann. Phys. (N. Y) 37, 224-226 (1966); 42,
343-346 (1967). (Section 2.2)
H.E. Moses, Irreducible representations of the rotation group in terms of
Euler's theorem, Nuovo Cimento 40A, 1120-1138 (1965). (Section 2.2)
M.A. Naimark, Linear Representations of the Lorentz Group (Pergamon
Press, New York, 1964). (Sections 2.1, 2.4, 2.5)
L.S. Pontrjagin, Topological Groups (Princeton University Press, Princeton,
New Jersey, 1946). (Section 2.3)
E.P. Wigner, Group Theory and its Application to the Quantum Mechanics
of Atomic Spectra (Academic Press, New York, 1964). (Section 2.4)
Chapter 3
The Lorentz and SL(2,C)
Groups
In this chapter the elements of the theory of special relativity are given
and the relationship between the Lorentz group and the group SL(2,C)
is discussed extensively. We first discuss the Lorentz group. The group
SL(2,C), the aggregate of all 2 x 2 complex matrices with determinant unity,
is consequently introduced. It is shown that SL(2,C) is homomorphic to
the homogeneous, orthochronous, Lorentz group. A direct correspondence
between elements of the matrices of the two groups is given explicitly. The
representations of these groups are given in the next chapter.

3.1 Elements of Special Relativity

In this section the fundamentals of Einstein's special relativity theory are
given. It is within this theory that the I;orentz group originates.

3.1.1 Postulates of Special Relativity

In the following we give the basic principles of the special theory of relativ-
ity. These principles are needed to describe the electromagnetic field and
other physical phenomena, and they constitute their spacetime symmetry
background.
The special theory of relativity was developed by Einstein in 1905 in or-
der to overcome and correct certain basic concepts that were in use at that

35
36 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

time, such as asymmetries in relative motion of bodies. Examples of rela-

tive motion in electrodynamics, and the unsuccessful attempt to detect the
motion of the Earth by the experiment of Michelson and Morley, suggested
that the phenomena of electrodynamics and mechanics do not depend on
the Newtonian notion of absolute rest. Rather, the laws of electrodynam-
ics should be valid in all frames of references in which the equations of
mechanics are valid.
Einstein raised the above observation to the status of a postulate and
called it the principle of relativity. He also introduced another postulate
(which is only apparently inconsistent with the former one) according to
which light always propagates in empty space with a constant velocity c
which is independent of the motion of the emitting body and the measuring
instrument.
The above two postulates were shown by Einstein to be enough for the
development of a consistent theory of electrodynamics of moving charges
which is based on Maxwell's original theory that was assumed to be valid in
stationary systems only. The theory did not require an "absolute stationary
space."
To describe the electromagnetic field, or any other classical field, one
needs a system of coordinates in terms of which the fields are described.
Such a coordinate system will include three spatial coordinates to which
we add the time coordinate. The three spatial coordinates will be denoted
by 'xk, where lower case Latin indices k = 1, 2, 3, and the time coordinate
by x = ct, where c is the speed of light in vacuum. The four coordinates
will collectively be denoted by xa, where Greek indices take the values
a=0,1,2,3.

Inertial Coordinate System

A system of coordinates in which the law of inertia holds is called an inertial
coordinate system. Hence Newton's laws of mechanics are valid only in
inertial coordinate systems.
If K is an inertial coordinate system, then every other coordinate system
K' is also an inertial system if it is in uniform motion with respect to K.
Hence if, relative to K, K' is a uniformly moving coordinate system then the
physical laws can be expressed with respect to K' exactly as with respect
to K.
One of the most important physical consequences of the special relativity
theory is the existence of a maximum signal speed in nature, which coincides
with the velocity of light In empty space. It is therefore natural to define
3.1. ELEMENTS OF SPECIAL RELATIVITY 37

the same time at separate points by means of light signals. This then raises
the problem of defining simultaneity.

Simultaneity
The definition of simultaneity is made as follows. If light requires the same
time to pass across a path A - M as for a path B --> M, where M is the
middle of the distance AB, then we say that the light signals at A and B
started simultaneously if the observer at M sees the two light signals at the
same time.
Will the two events, which occur simultaneously in one system, also be
simultaneous in another system moving with a velocity v with respect to
the first one? The answer is negative; events which are simultaneous in one
coordinate system are not necessarily simultaneous in others. It follows
that every inertial system has its own particular time.

3.1.2 The Galilean Transformation

As was mentioned above, inertial coordinate systems are those which are
in uniform, rectilinear, translational motions with respect to each other.
Accordingly, inertial systems of coordinates differ from each other by or-
thogonal rotations, accompanied by translations of the origins of the sys-
tems, and by motion with uniform velocities. One can, furthermore, add
the translation of the time coordinate thus enabling an arbitrary choice of
the origin of time t = 0.
Counting the number of parameters which each system of coordinates
has with respect to any other, we find that there are ten.
A transformation between inertial coordinate systems which has ten pa-
rameters, as described above, is called a Galilean transformation. The ag-
gregate of all Galilean transformations provides a group, called the Galilean
group, which has ten parameters.
One can choose two inertial systems of coordinates so that their corre-
sponding axes are parallel and coincide at t = 0. If v is the velocity of one
inertial coordinate system with respect to the other, the Galilean transfor-
mation can then be reduced into a simple transformation as follows:

x'=x-vxt, (3.1a)

y = y - vyt, (3.1b)

z' = z - vzt, (3.1c)

38 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

where v, vy and vz are the components of the velocity v along the x axis,
y axis, and z axis, respectively. Of course the Newtonian laws of classical
mechanics are invariant under the full ten-parameter Galilean group of
trannformations, and we have what can be called a Galilean invariance.

3.1.3 The Lorentz Transformation

According to the Galilean transformation which relates the spatial coordi-
nates and the time between inertial systems in prerelativity, the postulates
of the constancy of the speed of light and of the principle of relativity (which
applies, in particular, to the propagation of light and hence its constant ve-
locity is independent of the choice of the inertial system) are mutually
incompatible, even though both are experimentally valid.
The special theory of relativity resolves this impasse as follows.
The above two postulates will be compatible with each other if a new
transformation relating the spatial coordinates and times of different iner-
tial systems replaces the Galilean transformation. The new transformation,
of course, follows to be the Lorentz transformation. This, subsequently, re-
quires certain behavior of the moving measuring rods and clocks.
The principle of relativity may, thus, alternatively be restated as fol-
lows: The laws of physics should be covariant (or invariant) under the
Lorentz transformations relating different inertial coordinate systems. This
Lorentz invariance is in accordance with the Michelson-Morley null experi-
ment which showed that on the moving Earth light spreads with the same
speeds in all directions.
Consequently, the behavior of light is not incompatible with the princi-
ple of relativity. The incompatibility is only apparent.

3.1.4 Derivation of the Lorentz Transformation

We now derive the Lorentz transformation connecting the two coordinate
systems K and K' when they have the same orientations and their origins
coincide at t = 0, but K' moves along the coordinate x with a velocity v.
The directions perpendicular to the motion are obviously left unaffected
by the transformation. Hence

x 2 = x2, xis = x3 (3.2)

3.1. ELEMENTS OF SPECIAL RELATIVITY 39

and only the x0 and xl coordinates require changes when transforming from
one system to the other. One will therefore have the form

Ao A 1 0 0
A- Al o All 0 0
(3 . 3)
0 0 1 0
0 0 0 1

for the matrix of the Lorentz transformation in our particular case.

The "orthogonality" condition of the Lorentz transformation yields
?1CDACAADB = riAB, (3.4)
where the indices A, B, C, D = 0, 1, and lioo = -7111 = 1, 7701 = 7710 = 0.
The above formula gives three relations connecting the four elements of the
matrix (3.3):
(A00)2 - (A10)2 = 1,
(A1)2-(A'1)2=-1,
A0oA01 - AloAll = 0.
The solution of these equations can therefore be determined up to an
arbitrary parameter. One then finds that
A00 = cosh &, A1 =sinh1/i,
(3.6)
A10 = sinh Vi, All = cosh Vi.
is such an appropriate solution. With these values for the four elements,
we obtain
cosh ik sinh ' 0 0
A sinh,0 cosh V 0 0
(3.7)
0 0 1 0
0 0 0 1

for the matrix (3.3) of the Lorentz transformation.

The parameter -0 is related to the relative velocity v between the two
inertial coordinate systems K and K'. The relationship between them is
found by determining the motion of the origin of the coordinate system K
as seen from K', for instance. This motion is determined by putting xl = 0
in the Lorentz transformation, and using Eq. (3.7). This gives
xi0 = x cosh
(3.8)
x'1 = x sinh /i.
40 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

We therefore obtain
xi1x'
x70_c
6
t' = = tanh o, (3.9)

where the parameter Q is defined by

_ v
Q (3.10)
c.

Accordingly we obtain from Eq. (3.9)

1
cos h ,= -/32' (3 . 11 a )
1

-Q
sink V' = (3.11b)
1-a2
Using these results in Eq. (3.7) then yields

( 1
1 - Qz
-Q
1 _'32 oof
-Q 1
A= l - Q2
0 0 ( 3 . 12 )
1 - #2
0 0 1 0
0 0 0 1

for the matrix of the Lorentz transformation. We also obtain

0 0
I 1102 1-Q2
Q
A-1= 1
0 0 (3.13)
1 - QZ 1 --)32
I 0 0 1 0
0 0 0 1

for the inverse matrix describing the inverse Lorentz transformation.

The Lorentz transformation along the x axis is therefore given by

ct - ox
ct = (3.14a)

X' 1 - fl2 , (3.14b)

3.1. ELEMENTS OF SPECIAL RELATIVITY 41

y =y, Z=z. (3.14c)

We also obtain
ct' +,3x'
ct = 1-,Q2 , (3.15a)

x'+/3ct'
x ( 3 . 15b )

y=y, z=z. (3.15c)

for the inverse transformation from the coordinates x' back to x.

Equations (3.15) show that the inverse transformation differs from Eqs.
(3.14) only by a change in the sign of v. This result is obvious since the
coordinate system K is moving relative to the system K' with the velocity
-V.
A Lorentz transformation involving the time coordinate x and one or
more spatial coordinates xk, such as that derived above, is often called
a boost. A Lorentz transformation which keeps the time coordinate un-
changed is, of course, just an ordinary three-dimensional rotation of the
spatial coordinates.

3.1.5 The Cosmological Transformation

Universe Expansion versus Light Propagation
The Lorentz transformation has an analog in cosmology. (For full details
see Carmeli's book appearing in the References at the end of this chapter.)
Under the assumption that gravitation is negligible and thus Hubble's con-
stant is constant in cosmic time, there is an analogy between the propaga-
tion of light, x = ct, and the expansion of the Universe, x = rv, where 'r is
Hubble's time, a constant which is also the age of the Universe under the
above assumption, and c is the speed of light in vacuum.
Thus one can express the expansion of the Universe, assuming that it is
homogeneous and isotropic, in terms of the null vector (v, x, y, z) satisfying

T2v2 - (x2 +y 2 + z2) = 0, (3.16)

where v is the receding velocity of the galaxies. Equation (3.16), in the 4-

dimensional flat space of the Cartesian 3-space and the velocity, is similar
to
c2t2 - (x2 + y2 + z2) = 0, (3.17)
42 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

describing the null propagation of light in Minkowskian spacetime. We

assume, furthermore, that a relationship of the form (3.16) is valid at all
cosmic times, just as the assumption that (3.17) holds in all inertial systems
moving at any speed.
Accordingly, at a cosmic time t' at which the coordinates and velocity
are labeled with primes, we have
,r2,Ui2 _ (x/2 + 2 + z'2) = 0,
y (3.18)

with the same T, just as for light emitted from a source with velocity v with
respect to the first one,
c2t'2 - (x12 + 12 + z'2) = 0. (3.19)

As a result, we have a 4-dimensional space with zero curvature of v, x, y,

z just as the Minkowskian spacetime of t, x, y, z.
We now assume that at two cosmic times t and t' we have
(x/2 12 (x2
7' 2'U 2 - + + z'2) = T2v2 - + y2 + z2) , (3.20)

in analogy to the special relativistic formula

c2t'2 - (X/2 +
y 2 + z'2) = c2t2 - (x2 + y2 + z2) . (3.21)

The question is then what is the transformation between v', x', y', z'
and v, x, y, z that satisfies the invariance formula (3.20).
The transformation can be derived like deriving the Lorentz transfor-
mation by writing, in the two-dimensional case,

T2ti2-xi2=T2v2-x2 (3.22)

whose solution is
Tv' = Tv cosh it i - x sinh i,b,
(3 . 23)
x' = x cosh ik - Tv sinh,i.
At x' = 0 we obtain
tanhO=
x
Tv
= t
T
(3.24)

and therefore
t/T
sinh w = t2 '
3.25a)
1- T2
3.1. ELEMENTS OF SPECIAL RELATIVITY 43

1
cosh,/, = (3.25b)

which lead to the transformation

V - xt/T2
V = (3.26a)

,
(3.26b)

1-2
T
t2

Y' =y, z'=z. (3.26c)

The cosmic time appearing in the transformations (3.26) is measured back-

ward with respect to us. Thus the present time is zero, and that at the big
bang is T.
The transformation (3.26) is called the cosmological transformation and
it relates physical quantities at different cosmic times. It is in complete
analogy to the Lorentz transformation which relates physical quantities at
different velocities.

Interpretation of the Cosmological Transformation

Equations (3.26) give the transformed values of x and v as measured in the
system K' with a relative cosmic time t with respect to K. The roles of the
time and the velocity are exchanged as compared to special relativity. This
fits our needs in cosmology where one measures distances and velocities
at different cosmic times in the past. The parameter t/T replaces v/c of
special relativity.
It should be emphasized that the transformation (3.26) is not a trivial
exchange of v/c, appearing in the Lorentz transformation, and t/T here.
For example, the redshift z = v/c at low velocities, but is certainly not
equal to t/T for small t/-r.

The Galaxy Cone

The invariant equation (16), describing the distribution of galaxies in the
Universe at any cosmic time, has a very simple geometrical interpretation.
It enables one to present the locations of galaxies as a cone in the dual space
of distance and velocity. One then has a galaxy cone, similar to the familiar
44 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

light cone in special relativity. The symmetry axis of the cone coincides
with the x axis which extends from -7-c to +7-c.

3.2 The Lorentz Group

Consider two inertial coordinate systems K and K' whose origins coincide
at time t = 0, and the events with respect to which are denoted by t, x, y,
z and t', x', y', z', respectively. A light pulse emitted from the origin of K
will be spread spherically with the speed c, according to the equation
x2 +Y 2 +Z 2 = c2t2. (3.27)
Invariance of the speed of light tells us that an observer in K' will also see
the light propagating from his origin spherically according to the equation
x'2+y2+ z'2=C2t'2 (3.28)

From Eqs. (3.27) and (3.28) one then obtains

c2t'2 _ (x'2+y2+z'2) =c2t2 _ (x2+y2+ z2) (3.29)

or
77vx4x'v

= 77vxxv, (3.30)
where x and x'" are defined by
x = (ct, x, y, z) , x" = (ct', x', y', z') , (3.31)

and the symbol 77v,v (and later on 77v) is the flat-space metric, given by the
matrix
+1 0 0 0
0 -1 0 0
7 (3.32)
0 0 -1 0
0 0 0 -1
In the above equations, and throughout the following, repeated indices
indicate the use of the summation convention.
We will seek a linear transformation of the form
x'"` = Avxv (3.33)
between the times and spatial coordinates of the two inertial systems K
and K'. Using matrix notation, Eqs. (3.30) and (3.33) can then be written
in the form
x't77x' = xt?7x (3.34)
3.2. THE LORENTZ GROUP 45

and
X'= Ax, (3.35)
respectively. Here x and x' are the one-column matrices
xo x"
xl x'i
x= x2 ,x= I x,2 (3.36)
I
x3 xi3

xt and x't are the transposed matrices to the matrices x and x', respectively,
and A is the 4 x 4 matrix whose elements are AA,,.
Using now Eq. (3.35) in Eq. (3.34) then gives

xtAt??Ax = xtrlx, (3.37)

from which we obtain the condition

AMA = n (3.38)

that the 4 x 4 matrix A of the transformation has to satisfy. The transforma-

tion (3.33) is the Lorentz transformation. Equation (3.38) is a generalization
of the familiar relation
RtIR = I, (3.39)
which the 3 x 3 orthogonal matrix R, describing ordinary rotations of
the spatial coordinates alone, satisfies. The essential difference between
the two cases is in the replacement of the unit matrix I in the ordinary
three-dimensional rotations by the matrix rl in the four-dimensional Lorentz
transformations.
Equation (3.38) shows that

(detA)2 = 1, (3.40)

and accordingly the determinant of every Lorentz transformation A is equal

to either +1,
detA = +1, (3.41)
in which case the transformation is called proper, or to -1,
detA = -1, (3.42)

where the transformation is called improper. This is similar to the case of

the rotation group discussed in the previous chapter.
46 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

3.2.1 Orthochronous Lorentz Transformation

From Eq. (3.12), when written with indices, one finds

f]zAaAvp = flap, (3.43)

and taking a =,8 = 0, one obtains

(A)2 - (A 0)2 - (A20)2 - (A'0)2 = 1. (3.44)

Therefore (A0)2 > 1, and consequently we have either

A0 >_ +1, (3.45)

in which case the transformation is called orthochronous, or

A0 < -1. (3.46)

The aggregate of all orthochronous Lorentz transformations provides a sub-

group of the Lorentz group. The aggregate of all proper, orthochronous,
Lorentz transformations also provides a group which is a subgroup of the
Lorentz group.
In the following we will be concerned with the group of all proper,
orthochronous, Lorentz transformations. This group will be denoted by L.

3.2.2 Subgroups of the Lorentz Group

To conclude this section we give a brief discussion of the groups which can
be obtained from the Lorentz transformations.
The Lorentz transformations form a group called the (homogeneous)
Lorentz group. It is a subgroup of the inhomogeneous Lorentz group, also
known as the Poincare group. The latter group is formed from the inho-
mogeneous Lorentz transformations

x' = A" x' + xo , (3.47)

where x0 describes translations.

The Lorentz group possesses four disconnected parts which arise as fol-
lows.
(1) L+: detA = +1, A0 > +1. This part contains the identity element
of the group. The aggregate of all proper, orthochronous, Lorentz trsans-
formations provides a group, which is a subgroup of the Lorentz group. It
3.3. THE INFINITESIMAL APPROACH 47

is called the proper, orthochronous, Lorentz group.

(2) L1-: detA = -1, Ao > +1. This part contains a space inversion
element S which describes a reflection relative to the three spatial axes:
xio = x
x/1 = -xl
2 = -x2' (3.48)
x/3
= -x3
(3) L-1: detA = -1, Ao < -1. This part contains a time reversal element
T which describes a reflection relative to the time axis:

(3.49)

(4) 'L+: detA = +1, Ao < -1. This part contains the element ST.
As was mentioned before, from the above four parts of the Lorentz group
one obtaines the subgroup L1 = L+ U LT (the union of L+ and L? ), called
the orthochronous Lorentz group. Likewise, the subgroup L+ = L+ UL' ,
called the proper Lorentz group, is obtained.
Finally, we notice that every improper Lorentz transformation can be
written in the form
A = SAP, (3.50)
where S is a space-inversion element and AP is a proper Lorentz transfor-
mation.

3.3 The Infinitesimal Approach

3.3.1 Infinitesimal Lorentz Matrices
Rotations a1 (7&), a2 (sb), a3 (0) and Lorentz transformations (boosts) bl (,b),
b2 (0), b3 (0), around and along Oxl, Ox2i Ox3 can then be written explic-
itly. These matrices are given by
1 0 0 0
_ 0 1 0 0
al 0 0 cos 0 - sin 0
0 0 sin 0 cos 0
48 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

1 0 0 0
0 cos 0 0 sin 0
a2 (3.51b)
0 0 1 0
0 -sin o 0 cos 0
1 0 0 0
0 cos4i -sin0 0
(3.51c)
a3 (0) = 0 cos 0 0
sin V
0 0 0 1

and
coshb sinh o 0 0
sinh 0 cosh 0 0 0
bl (0) - 0 0 1 0
(3.52a)
0 0 0 1

cosh 0 0 sinh 7/i 0

b2
_ 0 1 0 0
(3.52b)
sinh -0 0 cosh 0 0
0 0 0 1

cosh ,0 0 0 sinh,i
0 1 0 0
bs (3.52c)
0 0 1 0
sinh 0 0 0 cosh

Infinitesimal Matrices
As in the case of the rotation group, the infinitesimal matrices a, and b, of
the group L are defined by

a, = [dar(IP)] b,, =
db J =o PW ,0=0
(3.53)

The a, and b, are related to a,. (0) and b,. (0) by

a,. (&) = exp (V a-) , bT (',) = exp ('i/)b,), (3.54)

and are given by

0 0 0 0
_ 0 0 0 0
(3.55a)
al 0 0 0 -1
0 0 1 0
 We and 3.3.
Here

denote 3.3.2
One
Ei.,k

denote THE
dimensional
is can

an
the

space easily
Ar
B usual see
(b)
arbitrary

= (see
Infinitesimal
that a3 a2
D b3 b2 bl
INFINITESIMAL

linear = = = = =
(ar [bi, these
[ai, [ai,
Levi-Civita
Chapter
b9]
(V,)) b3] a.]
, 2) = 0 10 0
= = 0 0 00 0 000 0 0 0 1 -1 00 0
Br by Operators matrices
symbol
g 0 0 00 0 0 0 1 0 0 0 0 0 0 -1 0 0 00 0 APPROACH

(0) -+ representation

Eijkbk.
0 00 1 0 000 0 0 0 0
Eijkak,
-Ei,,kak, 0 00 0 00 10
= D of satisfy
defined
D (g)
'
the the
(br by
and
E123
(b)) group
. for =
L 1.
in
commutation

an

convenience

(3.58) we infinite- (3.57c) (3.57b) (3.57a) relations (3.56c) (3.56b) (3.56a) (3.55c) (3.55b) 49
50 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

Ar (0) and Br (14') are continuous functions of 0 and are called basic one-
parameter groups of operators for the given representation. They satisfy
the relations (no summation on r)
Ar (01) Ar (02) = Ar (01 +02), (3.59a)

Br (VI 1) Br (02) = Br (ykl + 02) . (3.59b)

Ar (0) = 1, Br (0) = 1. (3.59c)
If the representation is finite-dimensional then the operators Ar (0i) and
Br (0) are differentiable functions of 0. If the representation is infinite-
dimensional, however, these operators might be non-differentiable.

Basic Infinitesimal Operators

The basic infinitesimal operators of the one-parameter groups Ar (0) and
Br (0) are then defined by
Ar_ [dAr(b)] Br[dBr()] (3.60)
d 0=o
if t he representation is finite-dimensional. Ar (;i) and Br (0) might then
be expanded in terms of Ar and Br as
Ar ('W) = exp (OA,-), Br (Y') = exp (OBr) . (3.61)
If the representation g -+ D (g) is infinite-dimensional, however, the op-
erator functions Ar (0) and Br (0) might be non-differentiable, but there
may still exist a vector x for which Ar (0) x and Br (0) x are differentiable
vector-functions.
In general, let A (t) be a continuous one-parameter group of operators in
a Banach space R (see Chapter 2), and denote by X (A) the set of all vectors
x E R for which the limit of (A (t) x - x) It, when t -40, exists in the sense
of the norm in R. Obviously the set X (A) contains the vector x = 0.
Define now the operator A for all x E X (A) by Ax = lim [(A (t) x - x) /t]
at the limit t -+ 0. The domain of definition, X (A), of the operator A is a
subspace of R, and A is linear, i.e.,
A(,\ 1x1 + )2x2) = )'1 Ax1 + ))2Ax2 for x1i X2 E X (A). (3.62)
Such an operator A is called the infinitesimal operator of the one-parameter
group A (t). If A (t) = D (a (t)) is the group of operators of the represen-
tation g -+ D (g), corresponding to a one-parameter subgroup a (t) of the
group L, the corresponding operator A is then called the infinitesimal op-
erator of the representation g - D (g).
3.3. THE INFINITESIMAL APPROACH 51

3.3.3 Determination of the Representation by its In-

finitesimal Operators
A representation g -+ D (g) of the group L is completely determined by
its infinitesimal operators Ai and Bi, i = 1, 2, 3. The determination of the
irreducible representations of the group L is based on the fact that the basic
infinitesimal operators of a representation satisfy the same commutation
relations that exist among the infinitesimal matrices a, and b,., namely:

(Ai, Aa] = Ei,;kAk, (3.63a)

[Bi, B,] _ -EijkAk, (3.63b)

[Ai, B3] = Ei,kBk. (3.63c)

Defining now the new infinitesimal operators

L:F = iAj A2i L3 = iA3, (3.64a)

FT=iB1B2i F3=iB3, (3.64b)

one then finds that they satisfy the following commutation relations:

[LT, L3] = [L+, L-] = 2L3, (3.65a)

[Fm, F3] = ::FL:F, [F+, F-] = -2L3, (3.65b)

[Lf, Ff] = 0, [L3, F3] = 0, (3.65c)

[L, F3] = ::FFf, [Ft, L3] = ::FF, (3.65d)

[L, FF) = 2F3. (3.65e)

The problem of determining a representation then reduces to the deter-

mination of L, L3, Ff, F3 satisfying the conditions (3.65).
Now, since the three-dimensional pure rotation group SO(3) is a sub-
group of the proper, orthochronous Lorentz group L, obviously every rep-
resentation of the group L is also a representation of the group SO(3). In
fact, any infinite-dimensional representation of the group L, when regarded
as a representation of the group SO(3), is highly reducible; it is equivalent
to a direct sum of an infinite number of irreducible representations. The
space of representation R of any irreducible representation of the group L
is, therefore, a closed direct sum of subspaces Mi, where Mi is the (2j + 1)-
dimensional space in which the irreducible representation of weight j of the
group SO(3) is realized.
52 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

Following the standard convention, one chooses the 2j + 1 normalized

eigenvectors of the operator L3 as the canonical basis for the subspace Mi.
Let these base vectors be denoted as fj,,, where m = -j, -j + 1, ..., j, the
superscript j indicates the subspace to which f L belongs, and the subscript
is the eigenvalue of the operator L3. The superscript in f4 specifies the sub-
space uniquely since each irreducible representation of SO(3) is contained
at most once in any given irreducible representation of the group L.

3.3.4 Conclusions
A detailed investigation of the commutation relations (3.65) in terms of the
canonical basis f j,, then leads to the following conclusions:
(a) Each irreducible representation of the group L is characterized by
a pair of numbers (jo, c), where jo is integral or half-integral, and c is a
complex number.
(b) The space R (jo, c) of any given irreducible infinite-dimensional rep-
resentation of the group L is characterized by integer or half-integer jo such
that
R (jo, c) = M' Mj0+1 ... (3.66)
The whole space R (jo, c) is apanned, therefore, by the set of base vectors
f;,,, where j = jo, jo + 1, jo + 2, ..., and m = -j, -j + 1, ..., j. If the given
irreducible representation is finite-dimensional then the direct sum of the
subspaces M's terminates after a finite number of terms.
(c) A given reppresentation is finite-dimensional if and only if
c2 = (jo + n)2, (3.67)

for some natural number n.

(d) The irreducible representation corresponding to a given pair (jo, c)
is, with a suitable choice of basis f,;, in the space of representation, given
by the formulas

Lff'n = [(j m+ 1)(j+m)]1,2fm1' (3.68a)

L3 f,7,, = ,,n (3.68b)

F'tf = [(j+m)(j+m-1)]1/2C;fmt1
1)11/2
- [(j m) (j m + Aj fmf 1
(3.68c)
f [(j m + 1) (j f m + 2)] 1/2 Cj+lfm 1
F'3fm = [(j - m) (j+ m)]1/2 C; f;,,-1 - mA1fm (3.68d)
-[(j+m+1)(j-m+1)11/2Cj+lfn 1.
3.3. THE INFINITESIMAL APPROACH 53

Here we have used the symbols A2 and C3 which are defined by

icjo
A; (3.69a)
j(j+1),
and

C2
_ i(j2-X0)1/2(72-C2)1/2 (3.69b).
j (4j2 - 1)1/2
(e) To each pair of numbers (jo, c), where jo is integral or half-integral
and c is complex, there corresponds a representation g -+ D (g) of the group
L, whose infinitesimal operators are given by Eqs. (3.68).
Equations (3.68), for the unitary representations case and under certain
assumptions, were first obtained by Gelfand (see the book of Naimark);
they later on were rederived by Harish-Chandra, and by I.M. Gelfand and
A.M. laglom.

3.3.5 Unitarity Conditions

A representation of a group G in a space R is called unitary if R is a Hilbert
space (see Chapter 2) and D (g) is a unitary operator for all g E G. This
implies that (D (g) x, D (g) y) = (x, y) for all g E G and all x, y E R, where
(x, y) denotes the scalar product in R. (For the physical significance of non-
unitary representations see Barut et al..) If the representation g -+ D (g)
of the group L is unitary, then Eqs. (3.68) satisfy certain conditions which
are summarized below.

Adjoint Operator
An operator B is called an adjoint to the operator A if (Ax, y) _ (x, By) for
all x, y E R. Let A be an infinitesimal operator of a unitary representation
g --4D (g) of the group L. Then A (t) = D (A (t)) is a unitary operator and
therefore its adjoint satisfies:
[A (t)]' = A (-t). (3.70)

Accordingly one has

(A (t) f, g) = (f, A (-t) g). (3.71)

Differentiating both sides of this equation with respect to t we obtain for

t = 0,
(Af, g) = - (f, Ag) . (3.72)
54 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

Using this relation one then easily finds that

(L+f, g) = (f, L_g), (3.73a)

(L3f, 9) = (f, L39) , (3.73b)

(F+ f, g) = (f, F_ g) , (3.73c)

(F3f, 9) = (f, F39) . (3.73d)

A systematic use of Eqs (3.73) in (3.68) then leads to the following.

If the irreducible representation g -+ D (g) of the group L is unitary then
the pair (jo, c) characterizing it satisfies either: (a) c is purely imaginary
and jo is an arbitrary non-negative integral or half-integral number; or (b)
c is a real number in the intervals 0 <1 c 1< 1 and jo = 0.
The representations corresponding to case (a) are called the principal
series of representations and those corresponding to case (b) are called the
complementary series.

3.4 The Group SL(2,C) and the Lorentz Group

We now introduce the group of all 2 x 2 complex matrices with determi-
nant unity, the group SL(2,C), and establish a homomorphism between the
group SL(2,C) and the proper, orthochronous, Lorentz group (see Subsec-
tion 3.2.2). Subgroups of the group SL(2,C) are then discussed, and the
connection with the Lobachevskian motion is pointed out.

3.4.1 The Group SL(2,C)

In what follows we establish the fact that elements of the proper, or-
thochronous, homogeneous Lorentz group L discussed in the last section
can be described by means of elements of SL(2,C), the group of all 2 x 2
complex matrices
(3.74)

with
det g = ad - be = 1. (3.75)
In the natural topology of matrices the group SL(2,C) is simply connected.
The relation between the two groups can be established as follows.
3.4. THE GROUP SL(2,C) AND THE LORENTZ GROUP 55

One associates with each four-vector x a Hermitian matrix

xo + x3 x1 + ix2
Q= x1-ix2 xo-x3 (3.76)

In this way one defines a one-to-one linear correspondence between all four-
vectors and all 2 x 2 Hermitian matrices. Equation (3.76) can also be written
as
Q = X'u-, (3.77)

where ak, k = 1, 2, 3, are the three Pauli matrices and ao is the 2 x 2 unit
matrix:
Ol al (0 11 i (1
Qo
1

O 1 0 J'
y2
-i 0'
0 3
0
0
-1
1

(3.78)
.

It is often also very convenient to parametrize the elements g of the group

SL(2,C) by
9 = 9Q, (3.79)
where go, gk, k = 1, 2, 3, are complex numbers.
Corresponding to every element g of the group SL(2,C) consider the
following transformation in the space of the Hermitian matrices Q:

Q' = 9Q9 t, (3.80)

where gt is the Hermitian conjugate of g, and Q' = x,,,ci . The corre-

sponding operation in the Minkowskian space of four-vectors is a linear
transformation
x" =Ap (g) xp, (3.81a)

or, in matrix notation,

i = A (g) x, (3.81b)

where the transformation matrix A can be expressed in terms of the matrix

g of the group SL(2,C). The transformation (3.81) preserves the scalar
product since
(xio)2 _ (XI 1)2 (X12) 2
- - (x')2 = det Q'

= det Q = (x0)2 - (x1)2 - (x2)2 - (x3)2 (3.82)

56 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

3.4.2 Homomorphism of the Group SL(2,C) on the

Lorentz Group L
The matrix elements AP can be expressed in terms of the corresponding
matrix g of the group SL(2,C). Using the properties of the Pauli spin ma-
trices, and using Eq. (3.80) and Eqs. (3.81), one has

WrQ')
x'O = vx p = 1 T1 (o o') XP = I Tr
2 2

(a"gcpgt) xp.
= 2Tr (a,9Q9t) = 2Tr (3.83)

Comparing this result with Eq. (3.81a) one obtains

Aap = 1 Tr (Q9ap9t) , (3.84a)

2

where gt is the Hermitian conjugate of the matrix g, and Tr stands for

trace.
The explicit expression of the transformation A (g) in terms of the pa-
rameters go and gk of the matrix g is as follows (Problem 3.1):
3

Ao=IgoI2+EI9kI2, (3.84b)
k=1

A i6klmgtgm, (3.84c)
0 = 90'9k + 9o9k -
Ak = 909k + 909k + jEklm9t9m (3.84d)
3

Ask = ak I go I2 - I g9 I2 + 9k91 + 9k91 - i6klm (-go9m - 9o9m)

s=1
(3.84e)
where the e symbols are fixed by E123 = EO123 = +1. In particular, one
notices the useful relation

Tr A(9)=ITrgI2=4I go 12. (3.85)

One also notices that because the group SL(2,C) is connected, and the
mapping into the homogeneous Lorentz group is a continuous homomor-
phism, the image of the group SL(2,C) must be a subgroup of the proper
orthochronous, Lorentz group L.
3.4. THE GROUP SL (2, C) AND THE LORENTZ GROUP 57

Equations (3.84) show that to an arbitrary matrix g of SL(2,C) there

corresponds a 4 x 4 matrix A. We now show that the matrix A belongs to
the proper, orthochronous, Lorentz group L.
First, from Eq. (3.82), one sees the quadratic form (xo)2 - (x1)2 -
-
(x2)2 (x3)2 is invariant under the transformation A, and therefore the
matrix A is an element of the homogeneous Lorentz group. As a conse-
quence, det A = 1. But for the special case for which g is the 2 x 2 unit
matrix, the corresponding A is the identity transformation, and hence det
A = 1. Since det A is a continuous function of the four variables a, b, c, d
of the matrix g of the group SL(2,C), and since the domain of variation
of these four variables is simply connected, a discontinuous jump from det
A = +1 to det A = -1 is excluded.
Consequently, det A = +1 for all values of a, b, c, d, subject to the
restriction (3.75). Hence A belongs to the proper Lorentz group. Finally,
from Eqs. (3.84) one sees that A0 cannot be negative. Accordingly, A is
orthochronous. Consequently, A is an element of the proper, orthochronous,
Lorentz group L.
Suppose now that an element A of the group L is given. Let us try to
invert the relations (3.84). If Tr A # 0 we obtain (Problem 3.2):
3

g = gou0 + > gkck

k=1
3
=D-1 Tr AQO + > (A 0 + Aok - ieokPA P) Qk (3.86)
k=1
where
D2 = 4 - Tr A2 + (Tr A)2 - iepv A aAp (3.87)
The sign of the denominator D is undetermined. Since the smallest sub-
group of the group L that contains all elements with Tr A # 0 is A itself,
the image of the group SL(2,C) is the whole of the group L.
It is possible to find the elements g of SL(2,C) which go into L in the
case Tr A = 0, also. If
3
E (Ao)296

0, (3.88)
k=1
the matrix A then describes a rotation with an angle Ic, and one has
3

90 = 0, 9 = I: 9kak, 92 = -e, (3.89a)

k=1
58 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

A (g) A (g) = A (g2) = A (-e) = I, (3.89b)

where e is the 2 x 2 unit matrix. The 3 x 3 matrix Mtk = btkA +Atk
is symmetric and possesses the three eigenvalues 0, Ao - 1, Ao + 1. The
eigenvalue 0 belongs to the normalized eigenvector vo defined by
3 -1/2
r
(vo)k = L (At0)2]
t_1
ll
A 0. (3.90)

If we denote the normalized real eigenvector, corresponding to the eigen-

value Ao + 1, by v1, we can then express the components 9k in terms of
the vectors v1 and v1 x vo, as follows:
[(1 1/2 (1 1/2 1
9k =
2
(Ao - 1)) (v1 x vo)k + i
2
(Ao + 1) ) vk . (3.91)

Again the matrix g is determined only up to a sign. The remaining case,

for which Tr A = Ako = 0, k = 1, 2, 3, is contained in Eq. (3.91) as the
limit Ao = 1. One obtains (Problem 3.3):

9k = ivj. (3.92)

In this fashion one reaches the conclusion that there exists a two-to-one
mapping between all the elements of the group SL(2,C) and all the elements
A of the proper, orthochronous, Lorentz group L such that to each element
A of the group L there correspond two elements g of the group SL(2,C),
and to each element g of the group SL(2,C) there corresponds an element
A of the group L. The mapping preserves the group multiplication and
constitutes, therefore, a homomorphism of the group SL(2,C) on the group
L. As a result of this, the description of the representations of the group L
is equivalent to that of the group SL(2,C); a representation g -+ D (g) of
L is single- or double-valued according to whether or not D (g) is equal to
D (-g) or not.

3.4.3 Kernel of Homomorphism

The sign ambiguity of g = g (A) means, in particular, that the unit matrix I
of the group L is the image of both central elements e of the group SL(2,C),
where e = e, and e is the 2 x 2 unit matrix. (Group elements are called
central if they commute with all group elements. Central elements form the
center of the group.) Hence we have established an isomorphism between
3.4. THE GROUP SL(2,C) AND THE LORENTZ GROUP 59

the proper, orthochronous, Lorentz group L and the group SL(2,C)/Z2,

where Z2 denotes the center of the group SL(2,C) consisting of the elements
e+.

3.4.4 Subgroups of the Group SL(2,C)

The group SL(2,C) possesses some important subgroups, some of which
play crucial roles in further investigations. These subgroups correspond
to subgroups of the proper, orthochronous, homogeneous, Lorentz group
L as well. Since the group SL(2,C) is more natural to handle than the
group L, one prefers to deal with the group SL(2,C)xT4, where T4 is the
translational group. The group SL(2,C)xT4 is sometimes called the in-
homogeneous SL(2,C) group. (See Subsection 3.2.2 for the inhomogeneous
Lorentz group.)
The group SU(2) has already been mentioned as a subgroup of SL(2,C).
It consists, of course, of those elements u satisfying ut = u-1. A possible
parametrization of the group SU(2) is as follows:
U = u.ao" (3.93)

with the condition

3
up+juk=1. (3.94)
k=1
Here uo and uk, k = 1, 2, 3, are real numbers.
Another subgroup of SL(2,C) is the group SU(1,1). It consists of those
elements v of SL(2,C) satisfying the condition vta3v = a3. A possible
parametrization is as follows:
V = voao + vial + v2a2 + iv30r3, (3.95)

with the condition

vo-v2-v2+v3=1. (3.96)
Here the numbers vo and vk, k = 1, 2, 3, are real.
A third subgroup of SL(2,C) is the group SL(2,R). It consists of elements
a of the group SL(2,C) satisfying ata2a = a2. They can be presented as
a = aoao + alas + ia2a2 + a3o . (3.97)

Here the numbers ao and ak, k = 1, 2, 3, are real and satisfy the condition
2
ao - ai2 + a2 - a3 = 1. (3.98)
60 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

The matrix a is a real 2 x 2 matrix.

By the rotation exp (iiro'/4) = 2-1/2 (a +ioA) in the x2 - x3 plane,
we can map the group SU(1,1) on the group SL(2,R):

a=exp (Z 41)v exp (i 41) (3.99)

from which one infers that ao = v, al = vi, a2 = V3, and a3 = -v2. This
one-to-one mapping of the two groups SU(1,1) and SL(2,R) onto each other
is sometimes called the standard isomorphism.
Finally, the group of triangular matrices
e-+O/2
pe-i0/2 eio/2 (3.100)
0 /
where is complex and 0 < .0 < 4ir, with the group multiplication law
(01, Al) x (02412) = (01 + 02 (47r), p1 + e'Ol A2), is isomorphic to the
group of Euclidean motions on the Riemannian plane of functions zl/2.

The corresponding subgroup of L is isomorphic to the group of motions in

the complex z-plane itself. The notation of this subgroup is U(1)xT2 if
one means the subgroup of L, and U(1)'xT2 if one means the subgroup of
SL(2,C).

3.4.5 Connection with Lobachevskian Motions

We have seen that each complex, unimodular, two-dimensional matrix g
induces a Lorentz transformation in the Minkowskian space according to
Q' = gQgt, where Q is given by Eq. (3.76). These Lorentz transformations
map the surfaces
2 2-
x-x1-x2-x3-c
2 2
(3.101)

into themselves, since they preserve the corresponding quadratic form.

There are three types of such surfaces. These are either sheet of a two-
sheeted hyperboloid when c > 0, a single-sheeted hyperboloid when c < 0,
and either the positive or the negative cone when c = 0. [If, instead of
considering points x in Minkowskian space, we deal with Hermitian matrices
Q, the surfaces would be the following three types of manifolds in the space
of Hermitian matrices: all positive definite (or negative definite) Hermitian
matrices with fixed determinant c > 0; all Hermitian matrices with fixed
determinant c < 0; and all Hermitian matrices Q > 0 (or Q < 0), that is
matrices Q whose corresponding Hermitian form takes on nonnegative (or
3.5. PROBLEMS 61

nonpositive) values, with determinant zero. See, for example, Gelfand et

al.]
The Lorentz transformations induce transformations that are called
motions of these surfaces. In this way to each complex unimodular two-
dimensional matrix g there corresponds a motion on each of the surfaces
above. One can show that a given motion corresponds to two matrices gl
and 92 if and only if gi = 92.
The upper sheet of a two-sheeted hyperboloid together with the motions
defined in this way is one model of Lobachevskian space. This means that
the group of complex two-dimensional unimodular matrices is locally iso-
morphic to the group of Lobachevskian motions. In addition to Lobachevskian
space, there exist two related spaces with groups of motions locally iso-
morphic to the same group of matrices. Models of these spaces are the
single-sheeted hyperboloid and the positive cone.
We conclude this brief discussion on the Lobachevskian space by point-
ing out that the group of motions on each of these surfaces is transitive, that
is every point of the space can be transformed by some motion to any other
point. Let us prove this assertion for the upper sheet of the two-sheeted
hyperboloid xp - xi - x2 - x2 = 1, as the proof for the other surfaces is
similar. Using Eq. (3.76), then the points on our surface correspond to
positive definite unimodular Hermitian matrices. Since every such matrix
can be written in the form Q = ggt = gegt, where g is a complex unimod-
ular matrix and e is the 2 x 2 unit matrix. This proves that there exists a
motion transforming the fixed unit matrix into Q.
In the next chapter the theory of two-component spinors is developed
in detail.

3.5 Problems
3.1 Use Eq. (3.84a) to prove Eqs. (3.84b)-(3.84e).
Solution: The solution is left for the reader.
3.2 Prove Eq. (3.86).
Solution: The solution is left for the reader. (See the book of Riihl.)
3.3 Prove Eq. (3.92).
Solution: The solution is left for the reader. (See Riihl.)
3.4 Prove that the kernel of the homomorphism of the group SL(2,C) onto
the group L coincides with the center of the group SL(2,C). (Notice that
62 CHAPTER 3. THE LORENTZ AND SL(2,C) GROUPS

by definition the element g is in the kernel of the homomorphism if for all

Hermitian matrices Q one has Q = gQgt.)
Solution: The solution is left for the reader.
3.5 Use Eq. (3.84a) to show that if a b, c and d are the four elements of the
matrix g of the group SL(2,C), with ad - be = 1, then the corresponding
matrix A of the proper, orthochronous, homogeneous, Lorentz group is
given by
(aa + bb) R (ab + cd) s (ab - cd) (a a - bb)

+2
2 (cc + dd) +2 (cc - dd)
2
t (ac + bd) R (ad + bc) s (ad + bc) R (ac - bd)
A= s (ac + bd) s (ad + bc) R (ad - bc) s (ac - bd)
(aa + bb) R (ab - cd) s (ab + cd) (aa - bb)
2 (cc+ dd)
-2 2
-2 (cc - dd)
where R ands denote real and imaginary parts. Show that the same matrix
can also be obtained directly from either Eq. (3.80) or Eqs. (3.84b)-(3.84e).
Solution: The solution is left for the reader.

3.6 References for Further Reading

A.O. Barut, H. Kleinert and S. Malin, The "anomalous Zitterbewegung"
of composite particles, Nuovo Cimento 58A, 835-847 (1968). (Section 3.3)
A.O. Barut and S. Malin, Position operators and localizability of quantum
systems described by finite- and infinite-dimensional wave equations, Revs.
Mod. Phys. 40, 632-651 (1968). (Section 3.3)
D. Bohm, The Special Theory of Relativity (Benjamin, New York, 1965).
(Section 3.1)
M. Born, Einstein's Theory of Relativity (Dover, New York, 1962). (Section
3.1)
M. Carmeli, Group Theory and General Relativity (McGraw-Hill, New York,
1977).
M. Carmeli, Cosmological Special Relativity: The Large-Scale Structure of
Space, Time and Velocity (World Scientific, Singapore, 1997). (Section 3.1)

M. Carmeli and S. Malin, Finite- and infinite-dimensional representations

3.6. REFERENCES FOR FURTHER READING 63

of the Lorentz group, Fortschritte der Physik 21, 397-425 (1973). (Section
3.3)
M. Carmeli and S. Malin, Representations of the Rotation and Lorentz
Groups (Marcel Dekker, New York and Basel, 1976).
A. Einstein, Zur Elektrodynamik bewegter Korper, Ann. Physik 17, 891-
921 (1905); English translation, On the electrodynamics of moving bodies,
in: The Principle of Relativity (Dover, New York, 1923). (Section 3.1)
A.P. French, Special Relativity (W.W. Norton, New York and London,
1968). (Section 3.1)
I.M. Gelfand, M.I. Gravy and N.Ya. Vilenkin, Integral Geometry and Rep-
resentation Theory (Academic Press, New York, 1966). (Section 3.4)
I.M. Gelfand and A.M. Iaglom, General relativistic invariant equations and
finite-dimensional representations of the Lorentz group, Zh. Eksp. Theor.
Fiz. 18, 703 (1948). (Section 3.3)
Harish-Chandra, Infinite irreducible representations of the Lorentz group,
Proc. Roy. Soc. (London) A 189, 372 (1947); On relativistic wave equa-
tions, Phys. Rev. 71, 793 (1947). (Section 3.3)
A.I. Miller, Albert Einstein's Special Theory of Relativity (Addison-Wesley,
Reading, Massachusets, 1981). (Section 3.1)
M.A. Naimark, Linear Representations of the Lorentz Group (Pergamon
Press, New York, 1964). (Section 3.3)
W. Riihl, The Lorentz Group and Harmonic Analysis (W.A. Benjamin,
New York, 1970). (Sections 3.2, 3.4)
Chapter 4

Two-Component Spinors
Spinors were invented by Elie Cartan without reference to the theory of
representations of groups. The natural way to study spinors, however, is
through the representation theory of the groups SU(2) and SL(2,C); two-
component spinors occur in the spinor representation of the group SL(2,C).
The latter is an irreducible, finite-dimensional, nonunitary representation.
It can be shown that any finite-dimensional representation of the group
SL(2,C) is equivalent to the spinor representation. In this chapter we study
the above-mentioned topics. Infinite-dimensional spinors are subsequently
given.

4.1 Spinor Representation of SL(2,C)

We now construct the spinor representation which contains all the finite-
dimensional irreducible representations of the group SL(2,C). A generaliza-
tion to infinite dimensions is given in the last section.

4.1.1 The Space of Polynomials

Let us denote by P,,,, the aggregate of all polynomials with complex coef-
ficients p (z, z) of the two variables z and z, where z is a complex variable
and x is its complex conjugate. We assume that the polynomial p (z, z)
is of degree not exceeding m in the variable z and not exceeding n in the
variable z, where m and n are two fixed nonnegative integers. Hence we

65
66 CHAPTER 4. TWO-COMPONENT SPINORS

have
m,n
p(z,z)= Eprszrz'=P00+p10 +p01 + +p,nnz'n!n- (4.1)
r,s=O

The space P,,,n is therefore determined by the two integers m and n.

The space P. may be considered a linear vector space, the components
of the "vectors" being the coefficients pr with r = 0, 1, ..., m and s =
0, 1,...,n. For each value of s there are m + 1 values for r, and for each
value of r there are n + 1 values for s. Hence the dimension of the space
Pmn is (m+ 1) (n+ 1). The operation of addition of two polynomials in
Pmn is defined, as usual, by

P (Z"5) +P" (z+ z) _ >2Prszrzs + >prszrzs = prszrz9, (4.2)

where pr, = p;, + p','9 .

The operation of product by a number is also defined, as usual, by

ap (z, x) = a E pr, zr.z = p;, zrz', (4.3)

where p;, = apr,. We obviously have

ap(z,z)+bp(z,x) = (a+b)p(z,z), (4.4)

ap'(z,a)+ap"(z,z)=a(p'+p")(z,z). (4.5)
Thus Pmn is a linear space. The space Pn will be used in the following as
the space of representation for the group SL(2,C).

4.1.2 Realization of the Spinor Representation

Let us now denote an element of the),ad_bc=1i
group SL(2,C) by

g= I a d

where a, b, c and d are four complex numbers. We then define the operator
D (g) in the space Pnn by
3) n
D (g) p (z, z) = (bz + d)m (bz + P (w, w) , (4.7)

where w is the image of z under the Mobius transformation

az+c (4.8)
T" bz+d'
4.1. SPINOR REPRESENTATION OF SL (2, C) 67

Accordingly, to each element g of the group SL(2,C) there corresponds an

operator D (g) defined in the linear space P,, , .
We now verify that the correspondence g --> D (g) is a linear represen-
tation of the group SL(2,C). To this end we have to show that

D (91) D (92) P (z, z) = D (9192) P (4.9)

D (I) = 1, (4.10)

for arbitrary elements g1 and 92 of the group SL(2,C). In the above formulas
I denotes the unity element of SL(2,C) and 1 denotes the unit operator in
the space P,,,,,,,.
Let now the elements g1 and 92 of the group SL(2,C) be denoted by

91 = C a1 b1 ) 92 = a2 d2
b2 I (4.11)
Cl di c2

and hence their product is given by

ala2 + b1c2 aib2 + bide
9=99 = 1 z
c1a2 +d1c2 c1b2 +d1d2
(4.12)

Accordingly we have, using the representation formula (4.7),

D(92)P(z,z) = (b2z+d2)m(b2z+22)"' P(w2, '12), (4.13)

where w2 is the image of z under the Mobius transformation associated

with the matrix 92,
a2z + C2
W2 b2z + d2. (4.14)

Applying now the operator D (gi) on both sides of Eq. (4.13), we obtain

D (91) D (92) P (Z,'5)

_ (biz + di)"` (biz + d1)n (b2w1 + d2)"` (b2w1 +d2)n P
(v,
v) , (4.15)
where w1 is the image of z under the Mobius transformation corresponding
to the element g1 of SL(2,C),
a1z + c1
W1 = b1z+d1, (4.16)

and v is obtained from w2 by replacing the variable z by w1i

a2W1 + C2
(4.17)
b2w1 + d2 .
68 CHAPTER 4. TWO-COMPONENT SPINORS

A simple calculation then gives the following for the products of the
terms with equal powers m and n in Eq. (4.15):
(blz+di)(b2wi+d2)=(aib2+bid2)z+(cib2+did2)=bz+d. (4.18)
Accordingly we obtain

D (gi) D (92) p (z, z) _ (bz + d)m (bx + )n p (V, v) (4.19)

for Eq. (4.15).

The variable v of Eq. (4.17) may be calculated using Eqs.(4.16) and
(4.12), giving

(aia2 + bjc2)z + (cia2 + dice) _ az + c

v = (4.20)
(alb2 + bide) z + (clb2 + d1d2) bz + d

In the above formulas a, b, c and d are the elements of the matrix g = 8192
given by Eq. (4.12). Accordingly we obtain for Eq. (4.19), using Eq. (4.20),
the following:

D(9i)D(92)p(z,z) = D(9192)p(z,x), (4.21)

thus proving Eq. (4.9). The proof of Eq. (4.10) is immediate since
D (I) p (z, z) = p (z, z) by the representation formula (4.7).
The correspondence g -+ D (g) is thus a finite-dimensional representa-
tion of the group SL(2,C) since it is being realized in the finite-dimensional
space P,,,,,. It is known as the spinor representation and is usually denoted
by D( i). Its dimension is, of course, equal to (m + 1) (n + 1).
,

4.1.3 Two-Component Spinors

In order to introduce the two-component spinors we realize the spinor rep-
resentation D(-i,!) discussed above in a somewhat different form.
To this end let us consider all systems of numbers

OA1...A,,,Xi --x;,1 (4.22)

which are symmetric in their indices Ai . . . A,,, and Xi X;, taking the
values 0, 1 and 0', 1', respectively. Such numbers may be considered as
the components of vectors of a linear space. Let us denote such a space by
Because of the symmetry of the indices in (4.22), we actually have
only m + 1 independent indices Ai . . A,,, and n + 1 independent indices
4.1. SPINOR REPRESENTATION OF SL(2,C) 69

Xi X,,. These are 0,..., 0, 0; 0,..., 0, 1;...; 1, ..., 1, 1, for instance,

for Al A,,,._jA,,,.. Hence the dimension of the space Pn is equal to
(m+1)(n+1).
We may relate the two spaces Pmn and Ann by one-to-one mapping by
associating to each number (4.22) of the polynomial

P (z, z) OA1...A-X,'...Xnz (4.23)

This polynomial is of degree not larger than m in the variable z and not
larger than n in the variable z. Hence the polynomial (4.23) belongs to the
space P,nn.
On the other hand every polynomial

P (z, z) = EPrszrz' (4.24)

of the space P,nn may be written in the form given by Eq. (4.23) by relating
the coefficients of z'V of the two polynomials (4.23) and (4.24). We then
obtain

OA1...A-Xi...X;, = Prs (4.25)

along with the conditions

(4.26)

In Eq. (4.25)
M _ m!
. (4 . 27)
n
(m - n)!n!
A second form of the spinor representation is obtained if we apply the
operator D (g) on the polynomial (4.23). We then obtain

D (g) p (z, z) = D (g)

,+B-. Y,'+ +Y

"' (9-10z - Bl+...+B,,, w-Yl+...+yn

_ (91o z + 9o) + 9 - o0) OB, ... B m yl. y* w
(4.28)
where w is given by Eq. (4.8) and use has been made of the notation
911 = a, 910 = b, go' = c, and go = d. Hence we obtain

/
D(g)p(z,z 1z

0 0 m-Bl -...-B.. / 1'- 1. Y" +Y,

70 CHAPTER 4. TWO-COMPONENT SPINORS

(4.29)

and therefore
-
D (9) P (z, z) = YgoB' ...9oB'"9o'," ... 9o'
Y
Q'Bi ...B,,,YI...Y,

+...+91BI ...91BTM91Yi

= rr
A ,X' B[,,YY'
Bi B,,.- Yi
9AI ...9Am 9Xi
- Y'0Bi...B,,.Y1...Y
"X,

XzA`+...+A-. X,'+...+X; .
(4.30)

Accordingly we may finally write the following for the spinor represen-
tation:
0iA1...q,,.Xi XAl +...+A,,, -Xi+...+Xn
D (9) p (z, z) ;.
z z (4.31)

where
0q1...q
nXi...Xn =
B1 ...9A-B_
gAl ...9X;,-Y
gxi Y' OB1...B-yl...yn (4.32)
B,Y'

is the transformed 0 under the group SL(2,C).

The quantities 0A1 are called two-component spinors, and
are complex numbers. The indices Al ... A, take the values 0, 1 and are
called unprimed (or undotted) indices, whereas Xi X;, take the values
0', 1' and are called primed (or dotted) indices. Similarly to tensors, every
spinor has an order. The spinor 0 defined above, for instance, is of order
m in its unprimed indices and n in its primed indices. Equation (4.32)
shows that these two kinds of indices transform under elements of the group
SL(2,C) and its complex conjugate, respectively. The summation over the
indices can also be made to run over 0, 1 instead of over 1, 0 by relabeling
the matrix g of SL(2,C) so that g11 = a, g1 = b, got = c, and g o = d,
where a, b, c and d are defined by Eq. (4.6).
Finally we notice that although two-component spinors were introduced
above as numbers, they can actually be made functions of spacetime when
applied in physics. This is again similar to tensors. The essential difference
between tensors and spinors is their association with groups. While tensors
are associated with the Lorentz group, spinors are associated with the group
SL(2,C), which is the covering group of the Lorentz group. As a result of
4.1. SPINOR REPRESENTATION OF SL(2,C) 71

this fact, spinors can be used to describe particles with spins .1 , 2,... in
addition to those with spins 0, 1, 2,..., whereas tensors can describe only
the latter kind of particles. As a consequence, spinors are considered to
be more fundamental than tensors from both the mathematical and the
physical points of view.

4.1.4 Examples
1. The spinor representation D(z''-z).
The representation DO,') corresponds to m = 2j1 = 1 and n =
2j2 = 1. The space of representation has accordingly the dimension of
(2j1 + 1) (2j2 + 1) _ (m + 1) (n + 1) = 4. The space P,nn is the aggregate
of all polynomials of the form

p (z, z) = poo +ploz +polz+pllzz. (1)

When the operator D (g) is applied to the above polynomial, we obtain ;

D (g) p (z, z) = (bz + d) ('+d) (poo + plow + pol'J + pllww)

= poo (bz + d) (bz + d) + plo (az + c) (bz + )

+pol (bz + d)( az + c) +pll (az + c) (az + -6), (2)

by the representation formulas (4.7) and (4.8).
Using now the correspondence between the spaces P,,, and P,,, we
then find the following, using Eq. (4.25), for the relationship between the
polynomial coefficients p,., and the components of the corresponding spinor:

poo = ooa, plo =.010', pol = t01', P11 = o11', (3)

Hence we have a spinor with two indices, one is unprimed and one is primed.
The polynomial p (z, z) is therefore defined in the space P,,,, by

p(z,z) _>,OAX'ZAZX = 0oo'+01o'z+0o1'z+011'zz. (4)

Using now Eqs. (2) and (3) we obtain

D(9)p(z,z) = 40oo, (910Z +g0) (9i z+90 )

01
+010' (911z+go1) (919 + 9o )
72 CHAPTER 4. TWO-COMPONENT SPINORS

+0oi' (gi0z + g0) (911 z + 901 )

+oii' (911z+go1) (91 +9o1) (5)

which may also be written in the form

D (9) p (z, z) = Ooo' (91 9h, za + ... + go 90 + '

+Oil, (911911 zz+...+g01901 ) (6)

Hence we have

D (9) p (90 909'000' + ... + 9o19ol .011) + .. .

+ (91090 0oo' + ... + 911911 X11) zz

_ > 9oB 90Y W BY' + ... + E g1 B 9lY, O,BY' z (7)

or

D (9) p 9A 9X' OBY' ZAZX

A,X' B,Y'
EA ,01
ZAZX , (8)
A,X'
where
PAX' _ 9A 9x' OBY' (9)
B,Y'
In the above formulas A, B = 1, 0 and X', Y' = 1', 0' (or, alternatively, 0,1
and 0',1').
2. The spinor corresponding to the space of representation with dimension
m=2.71=2andn=2.72=1.
The polynomials of the space P21 are given by
p(z,z) =poo + plot + poll + pllzz + p2oz2 + p21z2z (1)
By Eq. (4.25) the spinor corresponding to the coefficients p,., is given by
OABX', namely, of order 2 in its unprimed indices and of order 1 in its
primed indices. One then easily finds that
P00 = Oooo', plo = 20olo' = 24'lo0',
pol = 0001', pll = 20011' = 2101', (2)

P20 = 0110', P21 = 0111'-

4.2. OPERATORS OF THE SPINOR REPRESENTATION 73

Hence we finally have for the polynomial p (z, x), in terms of the spinor
OABX' the following:

p (z, z) = 0000, + (0oio' + .01oo') z + 00011z

+ (0oi1' + 010 1,) zz + 0iio'z20iii'z2

_ cABX'Z`9+BZX'

(3)

4.2 Operators of the Spinor Representation

We now find the infinitesimal operators L:F, L3, K:F, K3 of the spinor
representation for the group SL(2,C) discussed in the last section.

4.2.1 One-Parameter Subgroups

First we find the one-parameter subgroups of the group SL(2,C) corre-
sponding to the one-parameter subgroups ak (v') and bk (0) of the proper,
orthochronous, homogeneous Lorentz group L. These can easily be derived
using Eqs. (3.86). One finds for these one-parameter groups:

al
() _ cos 2 i sin 1 /' _ cos ,27t - sin
( i sin cos 22 )' a2 (w) sin 2 cos 2
2
0
a3 0 (4.33a)
eti+'/2 e- ii'/2

_ cosh sinh cosh 2 i sinh

b1 (,0) = ( sinh 2
cosh 2
) ' b2 (') - ( -i sinh 2 cosh
2
b3 0 a-+,/2 (4.33b)
Ce
In terms of the infinitesimal matrices ak and bk, where k = 1, 2, 3, of
the group SL(2,C), they can be written as

ak (V)) = e"a', bk (?,b) = eOb', (4.34)

where as,, and bk are given in terms of the three Pauli matrices (3.78) by
ak = iak/2 and bk = ak/2.
74 CHAPTER 4. TWO-COMPONENT SPINORS

4.2.2 Infinitesimal Operators

Using now Eqs. (4.7) and (4.8) one can find the operators Ak (vi) and
Bk (0), where k = 1, 2, 3. For example,

Aj (i,b) p (z, z) = (isinz+cos)m (_isin-1'+cos)Th

(cosl'z+isinV) cosz-isin'!'
2 2 2 2
xp
V)
i sin Z z +cos
b' ?b
-isin Z z +cos 2 J t , (4.35a)

A2(i,b)p(x,z)= (_sinz+cos.)m (_sin-+cos)'

cos z+ sin cos z+ sin 2
xp 2 2 (4.35b)
-sin 2z+cos2 2' -sin 2z+cos 2
e_imio/2eino/2p (eJo Z' cioz) . (4.35c)
A3 ('') p (z, z) =
In the same way one finds the operators Bk (0), k = 1, 2, 3. Differen-
tiating both sides of these equations with respect to the parameter b, and
putting 0 = 0, one obtains:

Aip=12(1-xz)z2(1-zz) +2(mxnz)A (4.36a)

Azp =
112

(1 zz)
az
+ 2 (1 + -j2) 8 - 2 (mz + nz)] p' (4.36b)

A3p = L z z - iz izn - 2 (m - n)] p, (4.36c)

Bip = 2 (1 - z2) 8z 1 (1
- zz) 5i + 2 (mz + nz)1 p, (4.36d)

B2p = L-2 (1 + z2) a + 2 (1 + z2) + 2 (rnz - nz)1 p, (4.36e)

B3p =
Lz 8z +
z
is 2
(m + n)] p' (4.36f)
4.2. OPERATORS OF THE SPINOR REPRESENTATION 75

The operators L, L3, K, and K3 can now be found, using Eqs. (4.36):

L+p _
az
- x2 + n71 P, (4.37a)

L_p = Lz2 az + a - mz] p, (4.37b)

Lap = [_zL +'i + (m - n)J p, (4.37c)

2
K+p = i aza - iz2 a + inzJ p, (4.37d)

K- p = I -iz2 az + i a + imz1 p, (4.37e)

K3P = [iz_ + 2z - 2 (m + n)] p, (4.37f)

4.2.3 Matrix Elements of the Spinor Operator D (g)

To conclude this section we find the matrices,of the spinor operators D (g).
For more details see the book of Riihl.
Consider the complex two-vectors = W, 0, which transform as

' = fig, (4.38)

under application of a matrix g of the group SL(2,C). Let us construct a

linear vector space of complex polynomials by defining p p e2)
which are homogeneous in 1 and 2 of degree 2J, where 2J is an arbitrary
nonnegative integer. In this space, having the dimension (2J + 1), we define
the transformation D (g) for any g of SL(2,C) by

D (g) p p p (fig). (4.39)

These transformations provide a (2J + 1)-dimensional representation of the

group SL(2,C).
In order to relate the transformation (4.39) to a more familiar notation
we expand the polynomial p into powers of 1 and 2:
J
(1)J+M (e2)J-M
P XMNM (4.40)
M=-J
76 CHAPTER 4. TWO-COMPONENT SPINORS

Here NM j are normalized constants defined by

1/2
NM, (2J)!
(4.41)
(J+M)!(J-M)!
and XM are expansion coefficients. In terms of XM the transformation D (g)
can be expressed as

(D (9) XJ)M = j DMM' (9) XM,. (4.42)

M'=-J
From Eqs. (4.39), (4.40), and (4.42) one finds

((J+M)!(J-M)!11/2 J+M' J-M'

DMM'(9)_ J+M-n
L(J+M')!(J-M')!J ( n

n J+M-n 921
X 911912
J+M'-n 922
n-M-M' (4.43)

where the sum over n extends over all integers for which neither of the
binomial coefficients vanish.
If we restrict the group SL(2,C) to the subgroup SU(2), we obtain
the matrix DMM, (u) defining the unitary irreducible representation of the
group SU(2). Each element DMM, (u) is a homogeneous polynomial of de-
gree 2J in the matrix elements of u, and the coefficients of the polynomial
are real. A substitution uij -+ gj2, with i, j = 1, 2, leads us back to the
matrix elements DMM, (g). If we substitute

ul1 = a + if3,
U12=7'+iS," (4.44)
U21 = -'Y + id,
u22=a-i13,
with real a, ,3, 'y, 5, such that

a2+R2+'y2+52=1, (4.45)

the substitution uij --+ gij can be interpreted as an extension of the real
parameters a, /.3, 'y, b into the complex domain. This is called a complexi-
fication of the group SU(2).
4.3. INFINITE-DIMENSIONAL SPINORS 77

4.2.4 Further Properties of Spinor Representations

We conclude this section by the following statements.
(1) The spinor representations of the group SL(2,C) are irreducible.
(2) Every finite-dimensional irreducible representation of SL(2, C) is
equivalent to some spinor representation. The pair of numbers jo and c, ap-
pearing in the representation formula, is then related to the pair of numbers
m and n of the spinor representation by

jo = I m -n (4.46a)
2

_ [sign (m - n)] [z (m + n) + 1] ; mn (4.46b)

c- [I(m+n)+1]; m=n
(3) The spinor representations are all nonunitary.
The proofs of (1), (2), and (3) are left for the reader (Problems 4.4-4.6).

4.3 Infinite-Dimensional Spinors

In this section the method used in Section 4.1 to obtain symmetrical spinors
and their transformation law from finite-dimensional representations of the
group SL(2,C) is extended to infinite-dimensional representations.

4.3.1 Principal Series of Representations

As we have seen in Section 4.1, two-component spinors are associated with
finite-dimensional representations of the group SL(2,C) when realized in
the space of polynomials. Spinors appear (up to factorial terms) as the
coefficients of the polynomials of the space in which the representation
is realized. Their transformation law then provides another form for the
representation.
The group SL(2,C), however, has also infinite-dimensional representa-
tions the most notable of which is the principal series of representations
(see Subsection 3.3.5). In this section we define an infinite set of numbers
which can be associated with the principal series of representations in a
way which is very similar, but as a generalization, to that two-component
spinors appear in describing the finite-dimensional representations. The
transformation law of these numbers, at the same time, defines another
form of the principal series of representations of SL(2,C). Just as in the
spinor case, these numbers become functions of spacetime when applied in
physics.
78 CHAPTER 4. TWO-COMPONENT SPINORS

The principal series of representations of SL(2,C) is an irreducible uni-

tary representation, which can be realized in several ways according to the
space of realization. For our purpose, we employ that particular realization
of it by means of the special unitary group SU(2).

The Hilbert Space L2s(SU(2))

We denote by L2,(SU(2)) the set of all functions 0 (u), where u is an element
of SU(2), which are measurable and satisfy the conditions

0' (?'u) = e"" 0 (u) , (4.47)

J 10 (u)12 du < oo, (4.48)

where ry ESU(2) is given by

CeI
'Y = 0 (4.49)
eY'k )
L23(SU(2)) provides a Hilbert space where the scalar product is defined by

(01, 02) = J 01 (u) 2 (u) du. (4.50)

The principal series of representations is then given by the formula

V (9) 0 (u) = a (u9) (4.51)

a (ug) 0 (u9)

where
g=C 911 912 )
921 922 J

is an element of the group SL(2,C) and a (g) is a function given by

a (9) = 922 1922 (4.52)

Here p is a real number and 2s is an integer.

4.3. INFINITE-DIMENSIONAL SPINORS 79

4.3.2 Infinite-Dimensional Spinors

The Hilbert Space 12'
Consider now all possible systems of numbers (P,,,, where m = _j _j +
1, , j and j =j s 1, 1 s I +1, 1 s 1 +2, , with the condition
00 3
12< oo.
(2j + 1) (4.53)
j=191 'm=-.7

The aggregate of all such systems ,,, forms a Hilbert space, which we
denote by 12', where the scalar product is defined by
00
(2j + 1) (4.54)
j=19I m=-j
for any two vectors O,,, and Vi;;,, of 12'.
With each vector O',,, E 12', we associate the function
00 j

0 (u) = E (27 + 1) L OmDm (u) , (4.55)

j=191 Tn=-7

where Dm (u) is the matrix element Ds,,, (u) of the irreducible representa-
tion of SU(2). Since
(7u) = e"I D;,, (u) , (4.56)
the function given by Eq. (4.55) belongs to the space L2'(SU(2)).
On the other hand, every function in L2'(SU(2)) can be written in the
form (4.55), since the D;,, (u) provide a complete orthogonal set,

f D;,, (u) (u) du = 1 5jj'5mm,. (4.57)

2j+1
The two spaces L2'(SU(2))
2 and 12'
2 are, in fact, isometric where the transi-
tion from one space to the other can be made by means of the generalized
Fourier transform
r
(u) D;,, (u) du. (4.58)

Similarly to spinors, which appear as coefficients in the polynomials of

the space of representation, we see that the numbers appear as coef-
ficients in the expansion given by Eq. (4.55) of the functions 0 (u) of the
80 CHAPTER 4. TWO-COMPONENT SPINORS

space L23(SU(2)). By means of the mapping (4.58), the operator V (g) of

the representation (4.51) may also be regarded as an operator in the space
123, whose explicit expression we find below. This expression also defines
another form of the principal series of representations.
Applying the operator V (g) to the function 0 (u) as given by Eq. (4.55),
we obtain

V (g)0(u) _ (2j (u9), (4.59)

a (ug)

or

V(g)0(u)=L,(2j+1)E01E(2j'+1)
7 m it

(4.60)
m'

where
a (ug)
(g; s, P) = (4.61)
V mm '
a (ug) Din (vg) Dm/ (u) du.
Accordingly, we obtain

V (g) 0 (U) (2j + 1) 0m D', (u) (4.62)

'm

where, using Eq. (14), we have

01i' 00

_ E (2j + 1) Vmm' (g; s, p) m. (4.63)

i=131 m=-3

Thus, the operator V (g) of the principal series of representations of

SL(2,C) in the space 122, is the linear transformation determined by Eq.
(4.63) describing the law of transformation of the quantitieswhere
7 = s s 1 +1, 1 s 1 +2, ... and m = -j, -j + 1, ... J. Here,Wit,
V, (g; s, p)
are functions of g ESL(2,C) and of p and s, where p is a real number and
2s is an integer.
In the next two chapters we apply the two-component spinors to curved
spacetime topics.
4.4. PROBLEMS 81

4.4 Problems
4.L Show that the matrix elements DMM, (g) of Eq. (4.43) satisfy the
properties:
DMM'(9) = DMM' (g),
DMM, (9t) = DMM' (g),
(-1)M-M,
DMM' ((g-1)t) = DjM,-M' (g).

Solution: The solution is left for the reader.

4.2 Show that the parameter 0 of Eqs. (4.33b) is related to the relative
speed v of the Lorentz transformation by
,0 =cosh-1 (1 - 2/,2)-1/2]

Solution: The solution is left for the reader.

4.3 Find the operators Bk (vi) for the spinor representation of the group
SL(2,C) and from them prove Eqs. (4.36d)-(4.36f).
Solution: The solution is left for the reader.
4.4 Show that the spinor representations of the group SL(2,C) are irre-
ducible.
Solution: The solution is left for the reader.
4.5 Prove Eqs. (4.46).
Solution: The solution is left for the reader.
4.6 Prove that the spinor representations are all nonunitary. [Thus the
group SL(2,C) does not contain finite-dimensional unitary representations.]

Solution: The solution is left for the reader.

4.5 References for Further Reading

H. Boener, Representations of Groups (North-Holland, Amsterdam, The
Netherlands, 1963). (Sections 4.1-4.3)
82 CHAPTER 4. TWO-COMPONENT SPINORS

R. Brauer and H. Weyl, Spinors in n dimensions, American J. Mathematics

57, 425 (1935). (Sections 4.1-4.3)
M. Carmeli, Infinite-dimensional representations of the Lorentz group, J.
Math. Phys. 11, 1917-1918 (1970). (Section 4.3)
M. Carmeli, Classical Fields: General Relativity and Gauge Theory (John
Wiley, 1982).
M. Carmeli and S. Malin, Finite- and infinite-dimensional representations
of the Lorentz group, Fortschr. der Phys. 21, 397-425 (1973). (Sections
4.1, 4.3)
E. Cartan, The Theory of Spinors (The M.I.T. Press, Cambridge, Mas-
sachusetts, 1966). (Sections 4.1-4.3)
C. Chevalley, The Algebraic Theory of Spinors (Columbia University Press,
New York, 1954). (Sections 4.1-4.3)
C. Chevalley, Theory of Lie Groups (Princeton University Press, Princeton,
New Jersey, 1962). (Sections 4.1-4.3)
P.A.M. Dirac, Spinors in Hilbert Space (Plenum Press, New York, 1974).
(Sections 4.1-4.3)
I.M. Gelfand, M. 1. Graev and N.Ya. Vilenkin, Generalized Functions, Vol.5:
Integral Geometry and Representation Theory (Academic Press, New York,
1966) (Sections 4.1, 4.3).
M.A. Naimark, Linear Representations of the Lorentz Group (Pergamon
Press, New York, 1964). (Sections 4.1, 4.3)
W. Riihl, The Lorentz Group and Harmonic Analysis (Benjamin, New York,
1970). (Section 4.2)
B.L. van der Waerden, Nachr. Wiss. Gottingen, Math.-Physik 100 (1929).
(Sections 4.1-4.3)
H. Weyl, The Classical Groups (Princeton University Press, Princeton, New
Jersey, 1964). (Sections 4.1-4.3)
Chapter, 5

Maxwell, Dirac and Pauli

Spinors
In the last chapter we derived the two-component spinors and their trans-
formation law under the group SL(2,C) from the theory of representations.
We now apply the two-component spinors to curved spacetime. Hence these
quantities will be functions of spacetime. In this chapter and the follow-
ing chapters two-component spinors will be applied to the electromagnetic,
Dirac, neutrino and gravitational fields. The first three fields can be in-
troduced in flat (Minkowskian) spacetime, but not the gravitational field.
Our presentation, however, will all be done in curved spacetime since that
needs no further effort. It should be very easy to go from the curved to the
Minkowskian spacetime if it is necessary. Hence our discussion will start
with spinors in curved spacetime. We start the chapter with a brief review
of Maxwell's theory.

5.1 Maxwell's Theory

In this section we give a brief account on the Maxwell equations for elec-
trodynamics.
The Lagrangian density for the electromagnetic field is given by

L
16-7rfapfap- Ij`Aa+Le. (5.1)

83
84 CHAPTER 5. MAXWELL, DIRAC AND PAULI SPINORS

The field f" is related to the potential A,,, by

f" = aA", (5.2)
and ja is the four-current density. Le is the Lagrangian density of the
charged particles.
Maxwell's equations are then given by
47rj,
(5.3)
C

.9-yf, + ap fya + a. f, = 0. (5.4)

Equation (5.3) is the field equation obtained from the Lagrangian density
(5.1), whereas Eq. (5.4) is a consequence of Eq. (5.2).
The electric field E and the magnetic field H are related to the electro-
magnetic field tensor f" by the following identification:
E _ (E., Ey, E.) = (El, E2, E3), (5.5a)

H = (Hr, Hy, Hz) = (Hi, H2, H3), (5.5b)

where E; and Hi, with i = 1, 2, 3, are given by
1
Ei = ffo, HH = 1 fijkf7k. (5.6)

Here 6ijk is the three-dimensional totally skew-symmetric tensor with values

+1 and -1, depending upon whether ijk is an even or an odd permutation
of 123, and zero otherwise.
The electromagnetic field tensors f and f " may then be written
explicitly as follows:
0 -Ex -Ey -Ez
Ey 0 Hz -Hy (5.7a)
f" -
Ey -Hz 0 H.
E. Hy -H. 0

0 Ex Ey Ez
f" - -E, 0 HZ -Hy (5.7b)
- Ey - HZ 0 H.
-EE Hy -H. 0
In terms of the dual * f to the electromagnetic field tensor fap, given
by
* fop = 2 Eap"f
5.1. MAXWELL'S THEORY 85

the Maxwell equation (5.4) may also be written in the alternative form
a*faA
0. (5 . 9)
axp
We then have for the dual the following explicit expressions:
0 -Hx -Hy -HZ
fuv _ Hx 0 -Ez Ey
(5.10a)
Hy E. 0 - Ex
H. -Ey E. 0

0 H. Hy H.
* f _ -Hx 0 -E. Ey
(5.10b)
-Hy E. 0 -Ex
-Hz -Ey Ex 0
To obtain the Maxwell equations in the usual notation, we have merely
to make the following identifications:
A = (A, A-) _ (0, A), (5.11a)

A = (Ao, Am) = -A), (5.11b)

and
7" _ (7,7"`) = (cp,j), (5.12a)

7 = (?,7m) = (cp, -j) (5.12b)

In the above equations is the scalar potential, A is the vector potential,
p is the charge density, and j is the vector current density.
A straightforward calculation, using Eq. (5.2), gives

E _ -0 O - c , (5 13)
.

H=OxA. (5.14)
Equations (5.3) and (5.4), on the other hand, give
(5.15)

VE=-c W, (5.16)
O.H=0, (5.17)
1 aE 4ir ,
O x H = at + j. (5.18)
86 CHAPTER 5. MAXWELL, DIRAC AND PAULI SPINORS

5.1.1 Maxwell's Equations in Curved Spacetime

It is easily seen that generalizations of these equations to curved spacetime
are achieved by the following equations:

OV fv = _jA (5.19)
c

f, = V A - vAv, (5.20)
for Eqs. (5.3) and (5.2), whereas

va [Eaprya (-g)-112 f.01 = 0, (5.21)

generalizes Eq. (5.4).

It will be noted that Eqs. (5.20) and (5.21) are identical to Eqs. (5.2)
and (5.4), respectively. In Eq. (5.21) &010 is the totally skew-symmetric
tensor density of weight +1 with values +1 and -1, depending on whether
af3ry6 is an even or an odd permutation of 0123 and zero otherwise. In
the above equations the symbol Da means covariant differentiation (see
Chapter 6).
The Lagrangian density (5.1) can be extended, in the presence of grav-
itation, as follows:

fapf`R_ CV---g7A.+Ce, (5.22)

1-
where now we use the curved spacetime metric tensor to raise the indices
of the Maxwell tensor,

fa/3f-O = g-9,3V fapfj . (5.23)

The Lagrangian density (5.22) then leads to the field equation

a as ax 0. (5.24)
axp a (aAa/axp) - aA. =
Using the Lagrangian density (5.22) in the field equation (5.24) gives
a.e 1
(5.25)

ac
= _ 1 vr--g fao
9 - " 9 P V f
, - V = _ 1 (5.26)
a (aA,, /axp) 4-7r
- v - - g
4-7r
5.1. MAXWELL'S THEORY 87

Accordingly, using Eq. (5.24) we obtain the following equation:

1 a (rfap) 47r
(5.27)
Vr-g axo =c
or
fap = 47rja
Op (5.28)
c

Equations (5.27) and (5.28) are Maxwell equations in the presence of grav-
itation and are a generalization of their flat-space counterpart, Eq. (5.3).
It remains to generalize the Maxwell equations (5.4) into curved space-
time. This may be achieved by replacing the partial derivatives appearing
in those equations by covariant derivatives. The result is

O7fap + V afp7 + Opf7a = 0, (5.29)

or, in its alternative form,

Op *fao = 0. (5.30)

One can easily show that the above equations are identical to Eqs. (5.4)
for the flat-space case.

Continuity Equation
The equation of continuity is obtained in the electromagnetic theory from
the Maxwell equations (5.3). We obtain, because of the antisymmetry of
the Maxwell field f", the following:
a,p
= o. (5.31)
ax

In the presence of gravitation, on the other hand, the Maxwell equations

(5.27) yield the following equation of continuity in curved spacetime:

('V__9ja) = 0. (5.32)
axa
Again, it can easily be shown that the latter formula may also be written
in the equivalent form
Vaja = 0. (5.33)
88 CHAPTER 5. MAXWELL, DIRAC AND PAULI SPINORS

5.2 Spinors in Curved Spacetime

Two-component spinors are introduced in curved spacetime at each point in
a "tangent" two-dimensional complex space. We then associate with every
tensor a spinor. The opposite is not correct; not all spinors correspond to
tensors. This is a consequence of the fact that tensors are associated with
the Lorentz group, whereas spinors are associated with its covering group
SL(2,C), and the correspondence between the two groups is a homomor-
phism rather than an isomorphism. Another way of looking at this is that
the group SL(2,C) yields all representations, including those with the half-
integral spins, whereas the Lorentz group yields only the representations
with integral spins.

5.2.1 Correspondence between Spinors and Tensors

The correspondence between spinors and tensors is achieved by means of
mixed quantities that were first introduced by Infeld and van der Waerden.
These are four 2 x 2 Hermitian matrices, denoted by QAB,. Here Greek-
letter indices are the usual spacetime indices of tensors taking the values
0, 1, 2, 3, whereas Roman capital indices are the spinor indices taking the
values 0, 1. The primed indices refer to the complex conjugate and take
the values 0', 1.
The hermiticity of the matrices QAB' means, using spinor notation, that
! _
CAB' -SBA' = SB'A'
(5.34)

The matrices CAB, are generalizations of the unit matrix and the three Pauli
matrices. They are functions of spacetime. There is no need to calculate
them explicitly, however, when spinors are used in general relativity theory.
The relationship between the matrices o and the geometrical metric
tensor g.,,, is as follows:

gvaAB'aCD' = EACEB'D'.
(5.35)

Here CAC and EB'D', along with EAC and OW to be used in the sequel, are
the skew-symmetric Levi-Civita metric spinors. They are defined by the
matrix
_0 . (5.36)
/
The raising and lowering of spinor indices is accomplished by means of the
above metric spinors as follows.
5.2. SPINORS IN CURVED SPACETIME 89

5.2.2 Raising and Lowering Spinor Indices

The role of the Levi-Civita metric spinors in raising and lowering spinor
indices is similar to that of the geometrical metric tensors in raising and
lowering spacetime indices. There is a difference, however, since the metric
spinors are antisymmetric. We will use the convention according to which
CA = EAB6B, 77A,
= CA'B' 77B'i (5.37)

and
cA = CBEBA, 77A' = 77B/EB'A', (5.38)
for arbitrary spinors C and 77. The above formulas give, for instance, CO = Ct
and C1 = -eo. We have, moreover,
CA77A
= CA77BEBA = -CAEAB77B = -CB 77B. (5.39)

Hence contraction with spinor indices should be done according to the con-
vention given by Eqs. (5.37) and (5.38).
We finally notice that the Levi-Civita spinor satisfies the simple identity

EABECD+ EACEDB + EADEBC = 0- (5.40)

The above identity may easily be proved by taking the different values of
the indices A, B, C, and D.

5.2.3 Properties of the a Matrices

In addition to relation (5.35), which the Hermitian matrices Q satisfy, they
also satisfy the following formulas:

orA14
B'avAB' = 9v (5.41 a)

or
aAB'01V B (5.41b)
= 5L,
which is equivalent to Eq. (5.41a).
The spinor equivalent of a tensor is a quantity that has an unprimed
and a primed spinor index for each spacetime tensor index. The spinor
equivalent of the tensor Tap, for instance, is given by

TAB'CD' = QAB'aCD'T-p (5.42)

90 CHAPTER 5. MAXWELL, DIRAC AND PAULI SPINORS

The above formula may be reversed. We then obtain the tensor correspond-
ing to the spinor TAB'CD'. We obtain
AB'allCD'
TAB'cD' = Q
AB'QvCD' a P
Q aAB'CCD'TaP

= b bQTap = T." (5.43)

by Eqs. (5.41).
The spinor equivalent of the geometrical metric tensor g,,,, is given by

9AB'CD' = QAB'aCD'9a3 = EACeB'D', (5.44x)

9AB'CD' = QAAB'Q ACD'gap = EACEB'D' (5.44b)
by Eq. (5.35). The above spinors are, in fact, the usual flat spacetime
metric tensors, but are now having the form
0 1
0
9AB'CD' -1 0
= 9AB'CD' = (5.45a)
0 -1
0
1 0

rather than that of the Minkowskian metric tensor reap.

The indices of the matrix (5.45a) are arranged in such a way that the
first pair, AB' = 00, 01', 10', 11', denotes the rows, whereas the second pair
CD', taking the same values, denotes the columns. We also have
1 0

9ABD = QaAB'0
CD'
= SAB' = jCjD' _
0 1
(5.45b)
0

which is equivalent to Eq. (5.45a).

5.2.4 The Metric 9AB'CD' and the Minkowskian Metric

77uw

The relationship between the flat spacetime metric gAB'cD' and the ordi-
nary Minkowskian metric tensor 7 7,,, is obtained if we take for the matrices
QAB' the Pauli matrices and the unit matrix, divided by Accordingly
we may take
0 1 1 _ 1 (0 11
QAB' 0 )1' C1
QAB'
1 72 1
0
5.2. SPINORS IN CURVED SPACETIME 91

AB'_
2 1( 0 il ag _ 1 (1 -0 1 (5.46a)
01
-i 0 JJ ' AB 0 1J'
and, raising the indices AB', we obtain

Q
OAB' __
f(0 1a
1 1 0 1AB'
2
0
-1
-1
0

i)
2AB' =
f -i
1 0 3AB'

The tensor equivalent to the spinor 9AB'CD' is then given by

_ 1
( -1 0 /
(5.46b)

AB' CD' (5.47)

77v = U 47 U 1/ 9AB'CD' _

which is, of course, the Minkowskian metric tensor.

5.2.5 Hermitian Spinors

When taking complex conjugate of a spinor, unprimed indices become
primed, and vice versa. The complex conjugate of the spinor SAB', for
instance, is given by
SAB' = SA' B . (5.48)
When a tensor is real, its spinor equivalent is Hermitian. Suppose, for
instance, that Va is a real vector and its spinor equivalent is VAB'. Then
we have

VB'A=VBA' =aBA'Va - B'AVa=aAB'Va=VAB', (5.49

by Eq. (5.34).
If the vector Va is null, namely, VaVG = gapV'V,3 = 0, then its cor-
responding spinor can be given as a product of an unprimed spinor with a
primed spinor,
VAB' = aAQB (5.50)
If, moreover, the vector Va is real, then 3B' is a multiple Of -&B',

VAB' = aAaB'. (5.51)

Any direction along the light cone, therefore, corresponds uniquely to a

one-index spinor ray, namely, to a set of spinors proportional to a given
spinor.
92 CHAPTER 5. MAXWELL, DIRAC AND PAULI SPINORS

5.3 Covariant Derivative of a Spinor

The covariant derivative of a spinor A, denoted by V1,CA, is given by

VA= aA
ax
- B
(5.52)

Here rAp is the spinor afjine connection. When taking the covariant deriva-
tive of the complex conjugate of the spinor eA, we have

V A' = aX (5 . 53)

Analogous equations hold for the spinors A and -CA ,

VA= axN'
+ r BC B . (5 . 54)

aC A/ A' B'
D A' ax,
+ r B'N, C (5 . 55)

Generalizations of the above formulas to spinors with more than one

index are done similarly as for tensors. Thus we have for a spinor with two
indices,
a AB'
+ FCiicB' + C''ec
B' ,
(5.56)
axp
for instance.

5.3.1 Spinor Affine Connections

The spinor affine connections introduced above are fixed by the requirement
that the covariant derivatives of the matrices a and the Levi-Civita spinors
all vanish,
VQAB, = 0, (5.57)

V.6AB = 0, V1116AB = 0, (5.58a)

pafA'B' = 0, VaCA'B' = 0. (5.58b)

The vanishing of the covariant derivative of 6AB, for instance, implies

V aCAB = x- raecB - rBaEAC = 0. (5.59)

5.3. COVARIANT DERIVATIVE OF A SPINOR 93

Hence we obtain
C
rCgaECB = rBaECA, (5.60a)
or
FBA. = rABa, (5.60b)
where rABa = rgaECA-
In the sequel we use the covariant derivative operator DAB defined by

VAB' = Q ( Voo' Vol' 1

ae V Vio, V11,
. (5.61)

Of course the two components Voo and V11, are real, whereas Vol, and
Vio' are complex, each being the complex conjugate to the other one,
Vlo' = Do,l = Dol- The operator DAB' is often denoted as follows:

DAB, _ (D 5 / . (5.62)

In flat spacetime, and when the matrices a are presented as in Eqs.

(5.46), the operator DAB' has the simple presentation
1 at+a. ax+2ay
VAB' _
f ay-2ay
C at - a. J (5.63))

In the above formula it has been assumed that the coordinates are Carte-
sian, x = t, x1 = x, x2 = y, and x3 = z, with the speed of light c = 1.

5.3.2 Spin Covariant Derivative

As was shown in Chapter 4, two-component spinors are obtained as (com-
plex) numbers in the representation formula of the group SL(2,C). When
applied in physics they become functions of spacetime and subsequently
one needs to define the (coordinate) covariant derivative as was done pre-
viously. Such derivatives have a vectorial form V or a spin form VA$'.
The question arises as to whether one can define spin covariant derivatives
of the form VA and VA' which are more basic than the previous ones. In
the following such a derivative is presented.
To this end one considers two-component spinors as functions of some
complex variables. More definitely it will be assumed that they are func-
tions of two complex variables, z and z1, and their complex conjugates.
This is equivalent to four real variables which might or might not have any
94 CHAPTER 5. MAXWELL, DIRAC AND PAULI SPINORS

relationship to spacetime coordinates. The two complex variables will be

denoted collectively by zA with A = 0, 1.
The spin covariant derivative may then be defined by

VA(B = a(Bq - rAB(C, (5.64a)

D A ( B' = az - r AB' ( C', (5 . 64b)

D A' (B = -A
A, - rA'B (C, (5 . 64c)

DA'(B' = a A, - rA'B'(C,, 5.64d

for the spinors (B and 8,. Similar formulas for the spin covariant deriva-
tives of the spinors (B arid B :
B
DA (B = azA + rAC(c, (5.65a)

a(B
V A(
B'
= azA
+rB' -c1
AC'( , (5 . 65b)
_ B
V A'( B = A, + r A'c( , (5 . 65c)

B' a(B B' C'

D A' ( _ 1A, + I' A,c ( , (5 . 65d)

It will be noted that Eqs. (5.64c) and (5.65c) are the complex conjugates
of Eqs. (5.64b) and (5.65b), respectively. The quantities IF are called spin
affine connections.
We finally relate the spin covariant derivative to the ordinary coordinate
covariant derivatives. This is done by expressing the ordinary covariant
derivative as a product of the newly defined ones,
DAB' = VAVB', (5.66)
where VAB' = O',AB,OA, and thus one has
O'aAB DAB, = v';B'VADB,. (5.67)
Since DAB' is Hermitian, VA is real. In terms of the complex variables zo
and zi, one then has

CAB -
- ( VzoVy V '-' 1
5.68
( V=I O= Ozi Oz, J )
5.4. THE ELECTROMAGNETIC FIELD SPINORS 95

5.3.3 A Useful Formula

Finally it is worthwhile mentioning that any spinor with two indices eAB
satisfies the relation

eAB - eBA = SC CAB, (5.69)

where ECDECD. Equation (5.69) is a consequence of the identity

(5.40) and is obtained from it by multiplying it by ff. Formulas similar
to Eq. (5.69) are valid for spinors having more than two indices. Thus we
have, for instance,

2S[ABICD = SABCD - SBACD = EABSF CD, (5.70)

for an arbitrary spinor SABCD.

In the next section we apply the theory presented above to the electro-
magnetic field.

5.4 The Electromagnetic Field Spinors

We may now apply the theory presented in the last section to the electro-
magnetic and gravitational fields. Accordingly, all the field variables will be
presented in spinorial form. This includes the electromagnetic field poten-
tial and tensor, the gravitational field curvature tensor, the Weyl conformal
tensor, and the Ricci and Einstein tensors. This section is devoted to the
electromagnetic field.

5.4.1 The Electromagnetic Potential Spinor

The spinor equivalent to the electromagnetic potential A is the spinor
ACD' given by
ACD' = aCD'A, . (5.71)

Since the vector A is real, the spinor ACD' is Hermitian, namely, ACD' =
AD'C. Thus the components A00, and All, are real, whereas A01, and Alo,
are complex conjugate to each other,

Alo, = Aol, = Ao,l. (5.72)

96 CHAPTER 5. MAXWELL, DIRAC AND PAULI SPINORS

5.4.2 The Electromagnetic Field Spinor

The spinor equivalent to the electromagnetic field tensor f is given by

fAB'CD' = C AB'UU
CD'fl+1 (5.73)

Since f1,, is skew-symmetric and real, the spinor (5.73) is antisymmetric un-
der the exchange of the two pairs of indices AB' and CD' and is Hermitian
with respect to these indices. Accordingly we have

fAB'CD' = -JCD'AB', (5.74)

fAB'CD' = fB'AD'C (5.75)

Because of the antisymmetry property of the electromagnetic field, we
may decompose its spinor equivalent (5.73). To this end we present Eq.
(5.73) as follows:

fAB'CD' = 2 (fAB'CD' - fCB'AD') + (fCB'AD' - fCD'AB') (5.76)

2
Here the first expression in parentheses is skew-symmetric in the indices A
and C, while the second expression is skew-symmetric in the indices B' and
D'. According to Eq. (5.70) we then have

fAB'CD' - fCB'AD' = EACfGBC D" (5.77)

fCB'AD' - fCD'AB' = EB'D'fCG'A (5.78)

Hence Eq. (5.76) may now be written in the form

fAB'CD' = 2 (EACfCBCD' + EB'D'fCG'A (5.79)

5.4.3 Decomposition of the Electromagnetic Spinor

Let us now denote
1 C,
PAB = 2 fAC'B (5.80)

We then have
1 C'1 1 '0' ,y,
OBA = 2fBC'A = -2fA BC' = 2fAC'B = WAB (5.81)
5.4. THE ELECTROMAGNETIC FIELD SPINORS 97

Hence OAB is a symmetric spinor. Taking now the complex conjugate of

OAB we obtain
1 1 C
c = 2fA'CB' = 2fCA' B'+ 1 c (5.82)
OA'B' = AB = 2 fAC'B
where use has been made of the hermiticity property of the spinor fAB'CD'
As a result, Eq. (5.79) may now be written in the form
fAB'CD' = EACcB'D' +cACEB'D', (5.83)

where use has been made of Eqs. (5.81) and (5.82).

We thus see that the electromagnetic field tensor f,, is equivalent to
the symmetric spinor OAB. The six real components of f are presented
by the three complex components of AB. These are Ooo, Ooi = 'io, and
oil. The spinor OAB will be referred to by us as the electromagnetic field
spinor, or simply the Maxwell spinor. In the sequel use will be made of the
notation
00 =000, 01=(001=010, 02=Oii- (5.84)
We finally find the spinor equivalent to the dual to the electromagnetic
field tensor. If f, is the electromagnetic field tensor and *fop is its dual,
(5.85)
*fap = 2
V, gEapti6fry6,

where e,,p.ya is the Levi-Civita tensor density of weight W = -1, then the
spinor equivalent to the dual is given by
*fAB'CD' = i tbACEB'D') . (5.86)

5.4.4 Intrinsic Spin Structure

We now apply the commutator (VNVM - V MV N) to an arbitrary spinor
l;Q. A simple calculation then yields
(VNVM - VMVN) eQ = -'P QMNCP, (5.87)
where

QMN = rQM,N - rQN,M + rQMrBN - rQNrBM (5.88)

In Eq. (5.88) a comma followed by a capital letter means a partial deriva-

tive, fm = of/azM (see Subsection 5.3.2). In the same way we obtain

(V NV M - VMVN) CQ = PMNCP. (5.89)

98 CHAPTER 5. MAXWELL, DIRAC AND PAULI SPINORS

The above formulas are analogous to those for defining the Riemann cur-
vature tensor (see next chapter).
From Eq. (5.87) we obtain
I (5.90)
V NV M - VMVN) CQ = 4?PQMNeP.
On the other hand, by lowering the free index Q in Eq. (5.89), we have

(ONV M - OM ON) eQ = d)QPMNeP (5.91)

Comparing the last two formulas we find that

4DPQMN = '1'QPMN, (5.92)

thus the spinor 4) is symmetric with its first two indices. It is also obvious,
by its derivation, that 4? is skew-symmetric with respect to its last two
indices,
'PQMN = -'PPQNM. (5.93)
Using now Eq. (5.70), one thus has

'DPQMN = "DPQ EMN, (5.94)

where 4i PQ = 4?QP = 4?PQAA

The spinor 4)pQ thus has the same symmetry properties as the electro-
magnetic field spinor cpQ (see Subsection 5.4.3).

5.4.5 Pauli, Dirac and Maxwell Equations

1. The neutrino equation
The neutrino is described by a two-component Pauli spinor, 77A. The neu-
trino field equation is then given by

VAB'I7A = 0. (5.95)

2. The Dirac equation

The Dirac equation for a spin-1 particle is given by
ry(VA-ieA4 )0=m?,b,
in units in which h = c = 1. Here 0 is a four-component spinor and ry
are four 4 x 4 matrices that satisfy certain commutation relations. The
5.4. THE ELECTROMAGNETIC FIELD SPINORS 99

4-component spinor,0 is composed of two 2-component spinors aA and RA

as follows:

The Dirac equation can also be described by coupled two equations with
the two 2-component spinors aA and QA (see Problem 5.12):

(OCD' ieACD') aC ='rnQD', (5.96a)

(VCD' + ieACD')13C = aD', (5.96b)

where ACD' = QCD,AA, with A being the electromagnetic potential, and

m and e are the mass and the charge of the particle.
3. The Maxwell equations
These are given by
47r
VV f = (5.97a)
c
VV *f AV
= 0, (5.97b)
where *f is the dual to f, and c is the speed of light in vacuum.
The spinor version of the above equations is

OCD' f
AB'CD' = 4*7r,AB' (5.98a)
c

VCD'
*fAB'CD' _ (5.98b)
where f and *f are given by
fAB'CD' = tACEB'D' (5.99a)
* fAB'CD' = 5ACEB'D') (5.99b)
Defining now the spinor f+ _ ``f + i *f, we then obtain for the Maxwell
equations
OCD'
f+AB'CD' = 4?r,AB' (5.100)
c
with fgB'CD' = 20ACEB'D'.
In the next chapter we discuss the gravitational field dynamical vari-
ables, starting with a review of general relativity theory.
100 CHAPTER 5. MAXWELL, DIRAC AND PAULI SPINORS

5.5 Problems
5.1 Decompose the commutator of the covariant differentiation operators
in spinor form.
Solution: The spinor equivalent to the commutator (0"V - is
given by
(VCD'VAB' - DAB'VCD') (1)

By adding and subtracting identical terms we obtain

VCD'VAB' -DAB'VCD'
= (VCD'VAB' - VCB'VAD') + (VCB'VAD' - DAB'V CD') (2)
The first bracket on the right-hand side is antisymmetric in the indices D'
and B', whereas the second bracket is antisymmetric in the indices C and
A. Hence we can use Eq. (5.69). We then obtain

VCD'VAB' - DAB'OCD' = ED'B'O(AC) + ECAD(B'D'), (3)

where use has been made of the notation

V(AC) = 2 (VAE'VC + OCE'DA ') , (4)

V(B'D') = (VEB'VD' +DED V DI) . (5)

5.2 Show that the Levi-Civita contravariant tensor density E"aA of weight
W = +1 may be presented in the form
E"a/j = 2 111-QCB',,"AD' (QCD'O'AB'
- QCD'QAB') (1)

Solution: We start with the identity

EMN'PQ'AB'CD' _ - (E AMEB'Q'ECP6D'N' - EAPEB'N'ECMED'Q' (2)

where EMN'PQ'AB'CD' is the spinor equivalent to the Levi-Civita tensor

density E"`p. Identity (2) may be verified by checking the components of
Eq. (2) for the various values of its indices. The indices MN', PQ', AB',
CD' take the values 00', 01', 10', 11', and eMN'PQ'AB'CD' takes the values
+1 and -1, depending upon whether MN', PQ', AB', CD' is an even or
an odd permutation of 00', 01', 10', 11', and zero otherwise.
5.5. PROBLEMS 101

Using now the relation

Eor
va/MN'PQ'AB'CD'

av aa aP = a6 MN'PQ'AB'CD' (3)

where
a = deta B' = f = -i g, (4)
Eq. (3) then yields, using Eq. (2),
EvaPaMN' 01 PQ' aAB' or CD'
v a P

= 2 /- (E AMEB Q'ECPED'N' - EAP,B'N'ECM,D'Q' I . (5)

Multiplying now the latter formula by a MN'aPQ'aAB'aCD' then yields

Eprry3
= i V -aMN'"P
J Q ' (a
fMQ'a6PN'
- aoMQ a7PN ) (6)

Equation (6) is identical to Eq. (1) if we raise the indices MN' and
PQ' of aMN, and aPQ, and lower the same indices of the bracket without
causing any change.
5.3 Show that the spinor equivalent to the tensor

Ea/3
- V - `! 9aP9PEpQv =
=
9p9vaEpaaf3

Fyv =
1 (1)

where Epa,v and Epaaf are the Levi-Civita covariant and contravariant ten-
sor densities of weights W = -1 and +1, respectively, is given by

CEP CH, = i (UEUCC VH UF, - UG"E6F''H') (2)

Solution: Using the representation for the Levi-Civita tensor density given
by Eq. (1) of the previous problem, we obtain

E,ayb = S- alCB' as D' (aCD'aAB, - aAB'aC13

D' (3)

The spinor equivalent of Eq. (3) is

EF'GH' CB' 6AD' 6EF'.CH' 6EF'6--"%
EIJ'KL' = ZbIJ' KL' (CD' AB' - AB' CD' (4)

which may also be written as

EEF'GH' - i (XEF'XCH'
IJ'KL IL KJ' - bEF'tSGH')
KJ IL (5)
102 CHAPTER 5. MAXWELL, DIRAC AND PAULI SPINORS

In the above formulas use has been made of the notation

"CD' = LC LD' . (6)

5.4 Show that the spinors equivalent to the tensor c'O and to the tensor
Lryb are related by
AB'CD' - AB'CD'
CEF'GH' - ti"EH'GF' (1)

Solution: The tensor L7a is given by

Lp = L Lp - Lu L. (2)
Hence its spinor equivalent is given by
LAB'CD' =LAB' LCD' LAB' LCD'
E F' G H' EF' G H' - G H' EF' (3)

where
ISEF' = LELF' . (4)
Comparing Eq. (3) with Eq. (5) of the previous problem for the spinor
equivalent of the tensor ep, we obtain Eq. (1).
5.5 Write the spinor affine connections in terms of the ordinary tensor affine
connections and the matrices a and their derivatives.
Solution: We use the fact that the covariant derivative of the matrices a"
vanish. We then obtain
Qv v v Cp v
QAB' = a,LaAB' + F PaAB' - I'AaC"B'
_- D'
B'aAVD'
V = 0, (1)

where alf = af/ax. Multiplying now the above equation by aEF' and
summing over the index v, we obtain

a'EF' aaA"B' + O I ,PaAB' - rA B' - I`B'1 A = 0. (2)

Contracting now the indices F' and B' in Eq. (2), yields
BB' v
awyA B' + av
EB'rjAp
v P AB' - 2q,
E = 0,
av (3)

and therefore
rAp = aEB' (r1PaAB' + a O'AB') . (4)
2
Likewise we obtain, by contracting indices A and E in Eq. (2),
rB' = 2av F"
(rPaAB' + a'AB') (5)
5.5. PROBLEMS 103

5.6 Write the ordinary tensor affine connections (see Chapter 6) in terms
of the spinor connections and the matrices Qa and their derivatives.
Solution: From the vanishing of the covariant derivatives of the matrices
oA B' we obtain
AB'-r)3 B'
+ rC01a
B'
+ rB,c C' = 0. (1)
l1JFy01
o0
Multiplying this equation by cAB' and summing up over the indices AB'
then yields
AB'
4I = TAP

B'aFL T + I'ClAacr B' aAB

P
+ is BUA,B. (2)

In Eq. (2) use has been made of the fact that the matrices Qa are Hermitian.
The first term on the right-hand side of the above equation is real since
B BA'
aAB'al.aaB/ = A'BaA0 a = BA = OrAB'alOraB (3)
Hence we finally obtain

4rPa = rCUaB aAB' + complex conjugate. (4)

5.7 Find the relationship between the spinor affine conections I A and the
spin affine connections IBC.
Solution: The solution is left for the reader.
5.8 Show that if fap is the electromagnetic field tensor and * fap is its dual,
then the spinor equivalent to the dual is given by Eq. (5.86).
Solution: The spinor equivalent to the dual * fap is given by

*fAB'CD' = aAB'QCD' *fap

2QAB'QCD'V a, = u (1)
Changing now the spacetime tensorial indices in Eq. (1) by the spinorial
indices, we get
1
KL'MN'
*fAB'CD' = 2EAB'CD'fKL'MN' (2)
Using now Eqs. (5) and (6) of Problem 5.3 in the above formula, we
obtain
*fAB'CD' = 2 (fAD'CB' - fCB'AD') = ifAD'CB' (3)
104 CHAPTER 5. MAXWELL, DIRAC AND PAULI SPINORS

Hence we obtain, using (5.83),

*fAB'CD' = i tACED'B')
or

*fAB'CD' = i OACEB'D'), (4)

where use has been made of the fact that OAB is symmetric and eB'D' is
skew-symmetric.
5.9 Find the spinor equivalent to the tensor

fv = fv +i*fv. (1)

Solution: Using Eqs. (5.83) and (5.85) we obtain

JAB'CD' = 2cACEB' D' (2)

5.10 Find the expression of the spinor equivalent to_the energy-momentum

tensor of the electromagnetic field.
Solution: The energy-momentum tensor of the electromagnetic field is
given by Eq. (62) of Appendix A, Chapter 6:

T = 41r
(9wfafp - (1)

Using spinor notation, the above expression may subsequently be written

in the form
EF GH EF
TAB'CD' = 4EACEB'D'fEF'GH' f - fAB'EF' fCD (2)
471r (1
The two quadratic terms in f in Eq. (2) may be calculated. We then obtain

fEF'CH'fEF'GH'
= 2 (.pEG EG+ (3)

fAB'EF' fCD EF' _ -20ACcB'D' + eACgB'F' cDF' + OAEOC EB'D' (4)

Denoting now the spinor equivalent to the energy-momentum tensor
TAB'CD' by T,,,.n, where

Tmn = TA+C,B'+D', (5)

5.5. PROBLEMS 105

with m, n = 0, 1, 2, and using the notation given by Eq. (5.84) for the
electromagnetic field spinor cAB we then obtain from Eq. (2) the following
simple formula:
Tmn = 0-Wn (6)
ZTr
Here m, n = 0, 1, 2.
5.11 Show that the energy-momentum tensor for the neutrino field is given
by
Tv = Z(QX,?AV ?1x + av X'r7AV X,

-UAX/rIX,VV? A - ULI ' !X'vr1A)- (1)

Solution: The solution is left for the reader.

5.12 The "standard" form of the Dirac equation is given by
-y(V-ieA)V)=mi, (1)

in units in which h = c = 1, where V) is a four-component spinor, y are

four 4 x 4 matrices, and m and e are the mass and charge of the particle.
Show that ip can be expressed in terms of the two 2-component spinors aA
and 3A, as
ao

-0 (2)
Q

Finally, show that Eq. (1) yields the two equations (5.96). Find the matri-
ces y" and write them in terms of the three Pauli matrices and the 2 x 2
unit matrix.
Solution: The solution is left for the reader.
5.13 "Ghost neutrinos" in general relativity are solutions of the Einstein-
Dirac equations with a neutrino current but with zero energy-momentum.
Solve the Einstein field equations for a static, plane-symmetric spacetime
generated by neutrinos. [See T.M. Davis and J.R. Ray, Phys. Rev. D 9,
331-333 (1974); also C. Collinson and P. Morris, Nuovo Cimento 16B, 273
(1973).]
Solution: The Einstein-Dirac equations are given by
1 8irG ( )
Gv = Rv - 29R = c4
TV, 1
106 CHAPTER 5. MAXWELL, DIRAC AND PAULI SPINORS

Ac
Tv =- 4
OGtyoG;' - t v-rp + i,t7,4b;" -0 -Y"0) (2)
where satisfies the zero-mass Dirac equation

-Yav);. = 0, (3)

and the semicolon denotes covariant differentiation. Since for zero-mass

Dirac particles the trace of the energy-momentum tensor vanishes, the
scalar curvature R also vanishes, and the Einstein equations reduce to
87rG
Rv = C4 T11,.. (4)

Spacetimes with plane symmetry along the x axis can be given by

ds2 = e2u (dt2 - dxz) - e2 (dy2 - dxz) , (5)

where u and v are functions of x and t only, and the speed of light c is
taken as unity. For the static case u, v, Vi, etc. depend on only x.
The exact solution to the Einstein-Dirac equations is given by

u=ln(kx+1)-114, v=ln(kx+1)12, (6)

and
P,, = (kx+ 1)-3/8 ybao, (7)
where
1

f1 '
'Yao=a (8)

fi
and k and a are arbitrary constants. The neutrino current density,
J = i,'t,YI,O, (9)

is consequently given by

J" = 4 1 a 12 (kx+ l)-3Y4 (bo f b1 (10)

Equations (6) and (7) represent an exact solution to the Einstein-Dirac

equations for a static, plane-symmetric spacetime. Its most interesting
property is that the neutrino energy-momentum tensor vanishes, whereas
the neutrino field and current density do not vanish.
5.6. REFERENCES FOR FURTHER READING 107

5.14 Show that the solution to the Einstein-Dirac equations given in Prob-
lem 5.13 can be generalized to represent "non-ghost" neutrinos,

u=-4In (ax+b)+cx+d, (1)

1
v= 2ln(ax+b), (2)

s
= e i" io (3)

where a, b are constants, and s, q and 0 are arbitrary real numbers. All
components of T are zero except Too and T11, which are equal.
Solution: The solution is left for the reader. [See K.R. Rechenick and J.M.
Cohen, Phys. Rev. D 19, 1635-1640 (1979).]
5.15 Generalize the solutions given in Problems 5.13 and 5.14 to the case
in which an electromagnetic field is present. Show that in the presence
of an electromagnetic field even the time-dependent Dirac field has ghost
solutions, but the solutions become ghost-free in the presence of charged
matter. Show also that the time-independent Dirac field has a ghost-free
solution if the neutrinos are considered to possess some mass.
Solution: The solution is left for the reader. [See K.D. Krori, T. Chaud-
hury and R. Bhattacharjee, Phys. Rev. D 25, 1492-1498 (1982).]

5.6 References for Further Reading

W.L. Bade and J. Jehle, An introduction to spinors, Revs. Mod. Phys. 25,
714-728 (1953). (Section 5.2)
J.D. Bjorken and D. Drell, Relativistic Quantum Mechanics (McGraw-Hill
Book Co., New York, 1964). (Section 5.4)
M. Carmeli, Group Theory and General Relativity (McGraw-Hill, New York,
1977).
M. Carmeli, Classical Fields: General Relativity and Gauge Theory (John
Wiley, 1982).
M. Carmeli, E. Leibowitz and N. Nissani, Gravitation: SL(2,C) Gauge The-
ory and Conservation Laws (World Scientific, Singapore, 1990). (Sections
108 CHAPTER 5. MAXWELL, DIRAC AND PAULI SPINORS

5.2-5.4)
E.M. Corson, Introduction to Tensors, Spinors and Relativistic Wave Equa-
tions (Hafner Publishing Co., New York, 1953). (Section 5.4)
P.A.M. Dirac, Proc. Roy. Soc. A 117, 610 (1928). (Sections 5.2-5.5)
M. Fierz, Uber die Relativistische Theorie kraftefreier Teilchen mit beliebi-
gen Spin, Helv. Phys. Acta 12, 3 (1938). (Sections 5.2-5.5)
L. Infeld and B.L. van der Waerden, The wave equation of the electron in
the general relativity theory, Sb. preuss. Akad. Wiss., Phys.-mat. K1. 380
(1933). (Section 5.2)
J.D. Jackson, Classical Electrodynamics (John Wiley, New York, 1975).
(Section 5.1)
T. Kahan (Editor), Theory of Groups in Classical and Quantum Theory,
Vol. I (Oliver and Boyd, London, 1965). (Sections 5.2-5.5)
W. Pauli, Zur Quantenmechanik des magnetischen Electrons, Z. Physik 43,
601 (1927). (Sections 5.2-5.5)
R. Penrose, A spinor approach to general relativity, Ann. Phys. (N. Y.) 10,
171-201 (1960). (Sections 5.2-5.4)
0. Veblen and J. von Neumann, Geometry of Complex Dynamics (The In-
stitute for Advanced Study, Princeton, New Jersey, 1958). (Section 5.2)
B.L. van der Waerden, Die gruppentheoretische Methode in der Quanten-
mechanik (Springer, Berlin, 1932). (Sections 5.2-5.5)
Chapter 6

The Gravitational Field

Spinors

In this chapter the gravitational field spinors are presented. These include
the curvature spinor and the spinors equivalent to the Riemann, Weyl,
Ricci and Einstein tensors. A decomposition of the Riemann spinor into
its irreducible components is also given. We start the chapter with a brief
review of the essentials of general relativity theory.

6.1 Elements of General Relativity

In this section a brief review of general relativity theory is given. Only
the essentials of the theory that is needed for the chapter are given. We
begin the discussion with a brief review of Riemannian geometry, followed
by a description of the physical foundations of general relativity. These are
the principles of equivalence and of general covariance. The gravitational
field equations are then derived in a tensorial form. The Schwarzschild
solution of Einstein's field equations is derived. Experimental verification
of general relativity is subsequently given. The section is then concluded
with a review of the problem of motion in the gravitational field.

109
110 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

6.1.1 Riemannian Geometry

Transformation of Coordinates
Any four independent variables x, where Greek letters take the values 0,
1, 2, 3, may be considered as the coordinates of a four-dimensional space
V4. Each set of values of x defines a point of V4. Let there be another set
of coordinates x' related to the first set x" by
x'1` = f, (x") , (6.1)

where fu are four independent real functions of x". A necessary and suffi-
cient condition that fu be independent is that their Jacobian,
8 U o 3

af
ax"
afo
8x3

does not vanish identically. Equation (6.1) defines a transformation of

coordinates in the space V4. When the Jacobian is different from zero, one
can also write x1` in terms of x" as

x = g1` (x'") . (6.3)

A direction at a point P in the space V4 is determined by the differential

dx. The same direction is determined in another set of coordinates x"` by
the differential dx'. The two differentials are related, using Eq. (6.1), by

dx" = ax" dx" = ofv dx". (6.4)

Here the Einstein summation convention is used, according to which re-

peated Greek indices are summed over the values 0, 1, 2, 3.

Contravariant Vectors
Let two sets of functions VA and V" be related by

V, = ax'IA
V"
ax" '

similar to the way the differentials dx' and dx'` are related. V'2 and V'2
are then called the components of a contravariant vector in the coordinate
6.1. ELEMENTS OF GENERAL RELATIVITY 111

systems x" and x''`, respectively. Hence any four functions of the x's in
one coordinate system can be taken as the components of a contravariant
vector whose components in any other coordinate system are given by Eq.
(6.5).
A contravariant vector determines a direction at each point of the space
V4. Let V' be the components of a contravariant vector and let dx' be a
displacement in the direction of V'`. Then dx/V = . . . = dx3/V3. This
set of equations admits three independent f k (x'`) = ck, where k = 0, 1, 2,
and the c's are arbitrary constants and the matrix afk/ac is of rank
three. The functions fk are solutions of the partial differential equation
V"afk/ax" = 0. Hence using the transformation laws (6.1) and (6.3)
one obtains V'k = 0 for k = 0, 1, 2, and V'3 # 0. Hence a system of
coordinates can be chosen in terms of which all components but one of a
given contravariant vector are equal to zero.

Invariants. Covariant Vectors

Two functions f (x) and f' (x') define an invariant if they are reducible to
each other by a coordinate transformation.
Let f be a function of the coordinates. Then
af a f ax" tr r)

Two sets of functions Vo and V.' are called the components of a covariant
vector in the systems x and x', respectively, if they are related by the
transformation law of the form (6.6),

VJ = X7'A V (6.7)

For example, if f is a scalar function, then a f /ax'` is a covariant vector.

It is called the gradient of f. The product V'`W, is an invariant if V is a
contravariant vector and W is a covariant vector. Conversely, if the quantity
V'2W, is an invariant and either V'' or W are arbitrary vectors, then the
other set is a vector.

Tensors

Tensors of any order are defined by generalizing Eqs. (6.5) and (6.7). Thus
the equation
ax''' ax'1m axo' ax^
"1' "^ 5xP1 axPm ax-'"1 ax'"^ 1 ..an
112 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

defines a mixed tensor of order m + n, contravariant of the mth order

and covariant of the nth order. If the Kronecker delta function is taken
as the components of a mixed tensor of the second order in one set of
coordinates, for example, then it defines the components of a tensor in any
set of coordinates. An invariant is a tensor of zero order and a vector is a
tensor of order one.
When the relative position of two indices, either contravariant or co-
variant, is immaterial, the tensor is called symmetric with respect to these
indices. When the relative position of two indices of a tensor is interchanged
and the tensor obtained differs only in sign from the original one, the ten-
sor is called skew-symmetric with respect to these indices. The process by
means of which from a mixed tensor of order r one obtains a tensor of order
r - 2 is called contraction.

Metric Tensor
Let g,, be the components of the metric tensor, i.e., a symmetric covariant
tensor, which is a function of coordinates, and let g = det g,,,,. The quantity
g"', denoting the cofactor of g,,, divided by g, is a symmetric contravariant
tensor and satisfies
9p9vp = by (6.9)

The element of length is defined by means of a quadratic differential form

ds2 = g dxdx". By means of the tensors g,,,, and g" one can lower or
raise tensor indices:
T'` up = 9 '`Tovpe (6.10a)

TaP7 = 9apT)Ah. (6.10b)

Certain other quantities transform according to the law

T :. = jN ax' ax p
... Tp::: (6.11)
axp ax
Here J is the Jacobian determinant I arc'/ax'p 1. The superscript N is the
power to which J is raised. Tom'.'.' is called a tensor density of weight N. For
example, if g' denotes det g' then g' = J2g, where g = det g,,. Hence
one has for the four-dimensional elements in two coordinate systems the
equality:
V, gd = -g7d4x'. (6.12)
6.1. ELEMENTS OF GENERAL RELATIVITY 113

Christoffel Symbols
From the two tensors gp" and gp" one can define the two functions

agp. a9oa agQ

r ape = 2 1

axa + axp axa

6 . 13)
C

rA = g, arapa (6.14)
They are symmetric in p and a, and are called the Christofel symbols of
the first and second kind, respectively.
Both kinds of Christoffel symbols are not components of tensors. By
starting with the differential transformation law for g,, it is not too diffi-
cult to show that rapo transforms according to the following relation (see
Problem 6.10):

axp ax'r ax5 axp a2x7

r"pa = ax,p ax,a rap + gpry ax," ax,pax,a (6.15)
axp"

Making use of the transformation law for gap then leads to the transfor-
mation law of as

axi5 axp ax" a ax' a2xa

(6.16)
p" xa ax'p ax'" r/ + ax ax,px,"

agJAV
(6.17)
r ap 2gax
" a
This equation can be rewritten in terms of the determinant g of g,,. The
rule for expansion of a determinant leads to the formula

(6.18)
a9"

where AP" is the cofactor of the element gp". From the law for obtaining
the inverse of a determinant, and from the definition of gp", Eq. (6.18)
may be written as
ag
= 99p", (6.19)
agp"

and consequently
dg = 99p"dg, _ -99p"dgp". (6.20)
114 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

Hence we have
aa9 = 99V aa9pv = -99Vaa9V (6.21)

The use of Eq. (6.21) enables us to write Eq. (6.17) in the form

rap
1
2g
ag 1 a- (6.22)
axa axa

Covariant Differentiation
We have seen that the derivatives of an invariant are the components of a
covariant vector. This is the only case for a general system of coordinates in
which the derivative of a tensor is a tensor. However, there are expressions
involving first derivatives which are components of a tensor. To see this we
proceed as follows.
Let V and V'V be a contravariant vector in two coordinate systems x
and x'. Then
(6.23)

Differentiating this equation with respect to xa and using Eq. (6.16) gives
(see Problem 6.11):
l+ 'P 'V aXjA

axa = (ax'V + V'ra axa ax'P -

vprpa. (6.24)

Hence if we define a covariant derivative of V by

vaV = aaV + rpaVP, (6.25)

Eq. (6.24) can be written as

vy'Pax'V ax
vaV = (6.26)
axa ax'P
Therefore vaV is a mixed tensor of second order.
In the same way one shows that the covariant derivative of a covariant
vector V. is given by:

vaV = aaV - rjPAaVP, (6.27)

From the above equation one has for the curl of a vector V:

Vpva - vavp = apva - a9Vp. (6.28)

6.1. ELEMENTS OF GENERAL RELATIVITY 115

Hence a necessary and sufficient condition that the first covariant derivative
of a covariant vector be symmetric is that the vector be a gradient.
It is easily seen, using the law of covariant differentiation of tensors (see
Problem 6.12), that
Vpgl" = 0, (6.29a)
Vpgv = 0, (6.29b)

VP 5'1" = 0. (6.29c)
Other properties of covariant differentiation can be established (see Problem
6.13).

Riemann and Ricci Tensors

If we differentiate covariantly the tensor VaV,,, given by Eq. (6.27), we
obtain
(V7V0 - V cV-,) V. = R' ap7V6, (6.30)

where R6 apy is called the Riernann tensor and is given by

R6 ap7 = apra7 - a7rap + ra7r' - rapr7. (6.31)

A generalization of Eq. (6.30) to an arbitrary tensor can be made (see

Problem 6.14).
One can show that in order that there can exist a coordinate system in
which the first covariant derivatives reduce to ordinary ones at every point
in space, it is necessary and sufficient that the Riemann tensor be zero and
that the coordinates be those in which the metric is constant.
One notices that the Riemann tensor satisfies

Rap-y6 = -Rpa76 = -Rap67 = Ry6ap, (6.32a)

Rap-y6 + Ray6p + Ra6p7 = 0. (6.32b)

Moreover, counting the number of components, one finds that in a four-
dimensional space the Riemann tensor has 20 components.
From the Riemann tensor one can define the Ricci tensor and the Ricci
scalar by

R1111 = Ra a , = /- (v Jr) a - (In rprQa (6.33)

R=R"JAI (6.34)
116 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

respectively. Here a comma denotes partial differentiation, f , = 8" f. The

Einstein tensor is then defined by
1 (6.35)
GV = Rv - 9v R.
2

The last important tensor constructed from the Riemann tensor is the
Weyl conformal tensor (see Section 6.1.8):

C,, ,, = Rpov - 1 (9PRvo - 9PvRo - 9oRvp + 9ovRp)

1
- 6 (9pv9a - 9P9vo) R. (6.36)

It has the special property that

Cp Pv
0. (6 . 37)

Furthermore, if the Weyl tensor vanishes everywhere, then the metric

is said to be conformally flat. (Two spaces V and V are called conformal
spaces if their metric tensors g,, and gL,,, are related by g = ,OgA,,, where
)3 is a function of the coordinates.) That is, there exists a mapping such that
g can be diagonalized, with 3 (x) appearing in the diagonal positions,
and where 0 (x) is some function. This follows from the fact that the Weyl
tensor can be expressed entirely Y iin terms of the density
Y 9v = (-9) 9v
and its inverse, and is equal to the Riemann tensor formed by replacing g,,v
by 9v, Rap-,b (9v) = C,, (g,).
Consequently, the vanishing of the Weyl tensor implies the vanishing of
Rp.yj which in turn implies that there exists a mapping such that
g,,v is everywhere diagonal, with 1 appearing along the diagonal. Only g
is arbitrary and (-g)114 appears along the diagonal of g,,,,.

Geodesics
The differential equations of the curves of extremal length are called geodesic
equations. To find their equations we seek.the relations which must be sat-
isfied to give a stationary value to the integral f ds. Hence we have to find
the solution of the variational problem

5 1 L ds = 0, (6.38)
6.1. ELEMENTS OF GENERAL RELATIVITY 117

where the Lagrangian L is given by

dx dx") 1 /2
(6.39)
L - Cg" ds ds
Accordingly we have

b f L ds = f [ax 6x + 6 dd 1 ds. (6.40)

a (dx/ds)
The second term of the integrand may be written as the two terms
d
6x] (6.41)
Ws [a (dx/ds) - ds [a (dx/ds)) Sx.
On integration, the first of these expressions contributes nothing since the
variations are assumed to vanish at the end points of the curve.
As expected, the equation obtained is the usual Lagrange equation:
d aL aL
0. (6.42)
ds a (dx/ds) 8x

A simple calculation then gives, using the Lagrangian given by Eq. (6.39),

d2x dx dxP
+ r-p 0. (6.43)
ds2 ds ds =

Bianchi Identities
A study of Eq. (6.16) shows that it is always possible to choose a coordinate
system in which all the Christoffel symbols vanish at a point. For, suppose
the Christoffel symbols do not vanish at a point A. One can then carry out
the coordinate transformation

x'a = xa - xA + 21'P--'y (A) xp - x A) (x7 - x A) , (6.44)

where the subscript A indicates to the value at the point A. By Eq. (6.16)
one finds that the Christoffel symbols in the new coordinate system vanish
at the point A.
A coordinate system for which the Christoffel symbols vanish at a point
is called geodesic. (It is also possible to transform away the Christoffel
symbols along a given curve.)
118 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

If we choose a geodesic coordinate system at a point A, then at A one

has
aparary - a7arap. (6.45)
Consequently, at the point A one has:

O,Rapry+OryRPa,p+VpR`.ry,,=0. (6.46)

Since the terms of this equation are components of a tensor, this equation
holds for any coordinate system and at each point. Hence Eq. (6.46) is an
identity throughout the space. It is known as the Bianchz identities.
Multiplication of Eq. (6.46) by gapb gives

gbA (v"iv + V7Rry5vp + OpRryary, ) = 0. (6.47)

Using the symmetry properties of the Riemann tensor, the last equation
becomes:
0 CRry
- Z5 R 1 = 0. (6.48)
V contracte/Jd

Equation (6.48) is called the Bianchi identity.

After having developed the mathematical tools to describe general rel-
ativity theory, we now turn to the physical foundations of the theory.

6.1.2 Principle of Equivalence

Null Experiments. Eotvos Experiment
One of the most interesting null experiments in physics is due to Eotvos, first
performed in 1890 and recently repeated by Dicke. The experiment showed,
in great precision, that all bodies fall with the same acceleration. The
roots of the experiment go back to Newton and Galileo, who demonstrated
experimentally that the gravitational acceleration of a body is independent
of its composition.
The importance of the Eotvos experiment is in the fact that the null
result of the experiment is a necessary condition for the theory of general
relativity to be valid.
Eotvos employed a static torsion balance, balancing a component of the
Earth's gravitational pull on the weight against the centrifugal force field
of the Earth acting on the weight. He employed a horizontal torsion beam,
40 cm long, suspended by a fine wire. From the ends of the torsion beam
were suspended two masses of different compositions, one lower than the
6.1. ELEMENTS OF GENERAL RELATIVITY 119

other. A lack of exact proportionality between the inertial and gravitational

masses of the two bodies would then lead to a torque tending to rotate the
balance. There appears to be no need for the one mass to be suspended
lower than the other.
The experement of Eotvos showed, with an accuracy of a few parts in
109, that inertial and gravitational masses are equal.
In the experiment performed by Dicke, the gravitational acceleration
toward the Sun of small gold and aluminium weights were compared and
found to be equal with an accuracy of about one part in 1011. Hence
the necessary condition to be satisfied for the validity of general relativity
theory seems to be rather satisfactory met.
The question therefore arises as to what extent is this experiment also
a sufficient condition to be satisfied in order that general relativity theory
be valid.
It has been emphasized by Dicke that gold and aluminium differ from
each other rather greatly in several important aspects. First, the neutron
to proton ratio is quite different in the two elements, varying from 1.08
in aluminium to 1.50 in gold. Second, the electrons in aluminium move
with nonrelativistic velocities, but in gold the k-shell electrons have a 15
per cent increase in their masses as a result of their relativistic velocities.
Third, the electromagnetic negative contribution to the binding energy of
the nucleus varies as z2 and represents i per cent of the total mass of a
gold atom, whereas it is negligible in aluminium. Fourth, the virtual pair
field and other fields would be expected to be different in the two atoms.
We thus conclude that the physical aspects of gold and aluminium differ
substantially, and consequently the equality of their accelerations represents
an important condition to be satisfied by any theory of gravitation.
Since the accuracy of the E6tv6s experiment is great, the question arises
as to whether it implies that the equivalence principle is very nearly valid.
This is true in a limited sense; certain aspects of the equivalence principle
are not supported in the slightest by the Eotvos experiment.

6.1.3 Principle of General Covariance

We have seen in the preceeding subsection that a gravitational field can be
considered locally equivalent to an accelerated frame. This implies that the
special theory of relativity (see Section 3.1) cannot be valid in an extended
region where gravitational fields are present. A curved spacetime is needed
and all laws of nature should be covariant under the most general coordinate
transformations.
120 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

The original formulation of general relativity by Einstein was based on

two principles: (1) the principle of equivalence (discussed in detail in the
last subsection); and (2) the principle of general covariance.
The principle of general covariance is often stated in one of the following
forms, which are not exactly equivalent:
(1) All coordinate systems are equally good for stating the laws of
physics, and they should be treated on the same footing.
(2) The equations of physics should have tensorial forms.
(3) The equations of physics should have the same form in all coordinate
systems.
According to the principle of general covariance, the coordinates become
nothing more than a bookkeeping system to label the events. The principle
is a valuable guide to deducing correct equations.
It has been pointed out that any spacetime physical law can be written
in a covariant form and hence the principle of general covariance has no
necessary physical consequences, and Einstein concurred with this view.
In spite of Einstein's acceptance of this objection, it appears that the
principle of general covariance was introduced by Einstein as a generaliza-
tion of the principle of special relativity and he often referred to it as the
principle of general relativity. In fact the principle of equivalence (which
necessarily leads to the introduction of a curved spacetime), plus the as-
sumption of general covariance, is most of what is needed to generate Ein-
stein's theory of general relativity. They lead directly to the idea that
gravitation can be explained by means of Riemannian geometry. This is
done in the next subsection.

6.1.4 Gravitational Field Equations

We have seen in Subsection 6.1.1 that the Riemannian geometry is charac-
terized by a geometrical metric, i.e., a symmetric tensor g,,, from which one
can construct other quantities. Classical general relativity theory identifies
this tensor as the gravitational potential. Hence in general relativity there
are ten components to the gravitational potential, as compared with the
single potential function in the Newtonian theory of gravitation.

Einstein's Field Equations

In trying to arrive at the desired gravitational field equations that the metric
tensor has to satisfy, we are guided by the requirement that, in an appro-
priate limit, the theory should be reduced to the Newtonian gravitational
6.1. ELEMENTS OF GENERAL RELATIVITY 121

theory. In the latter theory, the gravitational potential .0 is determined by

the Poisson equation:
V2 = 41rGp, (6.49)
where G (= 6.67 x 10-8 cm3 gm-1 sec-z) is the Newton gravitational
constant and p is the mass density of matter. Hence g,,,, should satisfy
second order partial differential equations. The equations should then be
related to the energy-momentum tensor T linearly. Such equations are
1
R - 1 nTiv, (6.50)

where is is some constant to be determined. In cosmology theory, one

sometimes adds an additional term, Ag,,,,, to the left-hand side of Eq. (6.50).
The constant A is known as a cosmological constant.
But the contracted Bianchi identities, Eq. (6.48), show that the covari-
ant divergence of the left-hand side of Eq. (6.50) vanishes. Hence

V,T11 =o, (6.51)

which expresses the covariant conservation of energy and momentum. The

constant n can be determined by going to the limit of weak gravitational
field (see Problem 6.20). Its value is c = 87rG/c4. The constant rc is known
as Einstein's gravitational constant.

Deduction of Einstein's Equations from Variational Principle

We start with the action integral

I = f / (Lc - 2KLF) d4x, (6.52)

and demand its variation to be zero. Here La and LF are the Lagrangians
for the gravitational and other fields, respectively. We take LG = R, where
R is the Ricci scalar, R = RIA g".
The first part of the integral (6.52) gives

(/gv) d4x. (6.53)

8 f /Rd4x = f V1_-ggvbRA,d4x + f
To find JR we note that in a geodesic coordinate system one has

6R = Va VV (ora) . (6.54)
122 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

But the latter is a tensorial equation. Hence it is valid in all coordinate

systems. Consequently, the first integral on the right-hand side of Eq.
(6.53) can be written as

f 9V f / V a (g- 3I' - g``,arAp d4x, (6.55)

and hence (by Problem 6.16) is equal to

1 (gV br;hv
- 9a 5r13)1
d4x. (6.56)

This integral, however, vanishes since by Gauss' theorem it is equal to a

surface integral which is equal to zero in consequence of the vanishing of
the variations at the boundary.
The second integral on the right-hand side of Eq. (6.53) gives, by Eq.
(6.21),

f RV6 (v g`) d4x = f (iw - 29V R bg"d4x. (6.57)

The second part of the integral (6.52) leads to (see Problem 6.21)

6 f 1_-gLFd4x = -2 f gT,,&AL d4x, (6.58)

where TV is the energy-momentum tensor and is given by

TV
-2 (a(/ LF) 9(y'LF)l (6.59)

a9 V / ,a - 9gV J
and a comma denotes partial differentiation, f, = as f . Combining Eqs.
(6.52), (6.57) and (6.58) then leads to the field equations (6.50):
1
RV - 911V R = nTV. (6.60)
2

The Electromagnetic Energy-Momentum Tensor

The energy-momentum tensor T,V for the electromagnetic field is obtained
from the general expression (6.59) with the field Lagrangian LF given by
the first part of the Lagrangian density (5.22) of Chapter 5,. namely,

LF =- 1 pV fapfV
9 (6.61)
16ir9
6.1. ELEMENTS OF GENERAL RELATIVITY 123

It can easily be shown to be given by

TPv = 4 (gpTfc/3i-,6 - fpafaX) . (6.62)

If we calculate the trace of the energy-momentum tensor (6.62) we find

that it vanishes,
T=TpP=g'Tpa=0. (6.63)
Using now R = -rT then leads to the vanishing of the Ricci scalar curva-
ture, R = 0. We therefore obtain
R = KT,,,, (6.64)

for the Einstein field equations in the presence of an electromagnetic field.

In Eq. (6.64) the energy-momentum tensor T,, is given by Eq. (6.62).
The Einstein field equations (6.64) and the Maxwell equations (5.28)
(Chapter 5) constitute the coupled Einstein-Maxwell field equations.

6.1.5 The Schwarzschild Solution

In spite of the nonlinearity of the Einstein field equations, there are numer-
ous exact solutions to these equations. Moreover, there are other solutions
which are not exact but approximate. Exact solutions are usually obtained
using special methods.
The simplest of all exact solutions to Einstein's field equations is that
of Schwartzschild. The solution is spherically symmetric and static. Such a
field can be produced by a spherically symmetric distribution and motion
of matter. It follows that the requirement of spherical symmetry alone is
sufficient to yield a static solution.
The spherical symmetry of the metric means that the expression for the
interval ds = (g,,,,dxdx')1/2 must be the same for all points located at the
same distance from the center. In flat space their distance is equal to the
radius vector, and the metric is given by (c is taken as equal to 1):
ds2 = dt2 - dr2 -r 2 (d02 + sin 2 OdO2) . (6.65)

In a non-Euclidean space, such as the Riemannian one we have in the

presence of a gravitational field, there is no quantity which has all the
properties of the flat space radius vector, such as that it is equal both to
the distance from the center and to the length of the circumference divided
by 2ir. Therefore, the choice of a radius vector is here arbitrary.
124 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

When a mass with spherical symmetry is introduced at the origin, the

flat space line element (6.65) must be modified but in a way that retains
spherical symmetry. The most general spherically symmetric expression for
ds2 is

ds2 = a (r, t) dt2 + b (r, t) dr2 + c (r, t) drdt + d (r, t) (d02 + sin 2 0d02).
(6.66)
Because of the arbitrariness in the choice of the coordinate system in gen-
eral relativity theory, we can perform a coordinate transformation which
does not destroy the spherical symmetry of ds2. Hence we can choose new
coordinates r' and t' given by some functions r' = r' (r, t) and t' = t' (r, t).
Making use of these transformations, we can choose the new coordinates
so that the coefficient c (r, t) of the mixed term drdt vanishes and the coeffi-
cient d (r, t) of the angular part to be -r'2, in the metric (6.66). The latter
condition implies that the radius vector is now defined in such a way that
the circumference of a circle whose center is at the origin of the coordinates
is equal to 27rr. It is convenient to express the functions a (r, t) and b (r, t)
in exponential forms, e" and -ea, respectively, where v and A are functions
of the new coordinates r' and V. Consequently, the line element (6.66) will
have the form

ds2 = e"dt2 - eadr2 - r2 (d02 + sin 2 0d2) , (6.67)

where, for brevity, we have dropped the primes from the new coordinates
r' and t', and the speed of light c is taken as equal to 1.
We now denote the coordinates t, r, 0, 0 by x, x1, x2, x3, respectively.
Hence the components of the covariant metric tensor are given by:
fe" 0 0 0
0 -ea 0 0
(6 . 68a)
g " 0 0 -r2 0
0 0 0 -r2 sine B

whereas those of the contravariant metric tensor are:

e-" 0 0 0
0 -e-A 0 0
9 (6.68b)
0 0 -r-2 0
0 0 0 -r-2 sin-2 0
To find out the differential equations that the functions v and A have
to satisfy, according to Einstein's field equations, we first need to calculate
6.1. ELEMENTS OF GENERAL RELATIVITY 125

the Christoffel symbols associated with the metric (6.68). The nonvanishing
components are:

Ioo=2 I'io=2, I'll=2 (6.69a)

1
V/
Ioo=2-e v- F0 =2,
A
1 I'11= 2
1
A/
( 6.69b )

r22 = -re-X, 133 = -rsin2 Be-a, r2= T (6.69c)

133 = - sin 0 cos 0, I'13 = 1, I 23 = cot 0, (6.69d)

r
where dots and primes denote differentiation with respect to t and r, re-
spectively.
With these Christoffel symbols, we compute the following expressions
for the nonvanishing components of the Einstein tensor:

coo = _e- X a 2 - + r2 = rcTo , (6.70a)

r

Go' icT0 (6.70b)

r

r + r-) + -r= kT1

G11 = -e-a v/ 1
(6.70c)

C2 2=-1e-X v 'A'
2 2 r 2 )
2
z
e-" =IcT22, (6.70d)
+2 L

G33 = G22 = xT3 3. (6.70e)

All other components vanish identically.

The gravitational field equations can now be integrated exactly for the
spherical symmetric field in vacuum, i.e., outside the masses producing the
field. Setting Eqs. (6.70) equal to zero leads to the independent equations:

(6.71a)
126 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

CA 1\\ + 0, (6.71b)
r r2 r2
A=0. (6.71c)
From Eq. (6.71a) and (6.71b) we find v' + A' = 0, so that v + A = f (t),
where f (t) is a function of t only. If we perform now the coordinate trans-
formation x0 = h (x'o) xk = x/k then goo = h2goo. Such a transformation
amounts to adding to the function v an arbitrary function of time, while
leaving unaffected the other components of the metric. Hence we can choose
the function h so that v + A = 0. Consequently, we see, by Eq. (6.71c),
that both v and A are time-independent. In other words the spherically
symmetric gravitational field in vacuum is automatically static.
Equation (6.71b) can now be integrated. It gives:

e-X=eL= 1- K
r
(6.72)

where K is an integration constant. We see that for r -- oo, a-a = e" =

1, i.e., far from the gravitational bodies, the metric reduces to that of
the flat space (6.65). The constant K can easily be determined from the
requirement that Newton's law of motion be obtained at large distances
from the central mass. From the geodesic equation it follows that the
radial acceleration of a small test mass at rest with respect to the central
mass is (see Problem 6.20):
1 1 K) K
-roo = --2 1 - r J r2
K
(6.73)
2r2
Comparing this expression with the Newtonian value -Gm/r2 gives K =
2Gm, where m is the central mass and G is the Newton constant.
The constant 2Gm, or 2Gm/c2 in units where c is not taken as equal
to 1, is often called the Schwarzschild radius of the mass m. For example,
the Schwarzschild radius for the Sun is 2.95 km, that for the Earth is 8.9
mm, and that for an electron is 13.5 x 10-56 cm.
We therefore obtain for the spherically symmetric metric the form:
1 - 2Gm/r 0 0 0
0 - (1 - 2Gm/r)-i 0 0
9AV =
0 0 -r 2 0
. (6.74)

0 0 0 -r2 sin2 B
It is known as the Schwarzschild solution and describes the most general
spherically symmetric solution of the Einstein field equations in a region
6.1. ELEMENTS OF GENERAL RELATIVITY 127

of space where the energy-momentum tensor T"' vanishes. Although g,,,,

goes to the flat space metric when r goes to infinity, it was not necessary
to require this asymptotic behavior to obtain the solution.
It is worth mentioning that all spherically symmetric solutions of the
Einstein field equations in vacuum which satisfy the boundary conditions at
infinity mentioned above are equivalent to the Schwarzschild field, i.e., their
time-dependence can be eliminated by a suitable coordinate transformation.
This result is due to Birkhoff.
Finally, it is convenient to introduce Cartesian coordinates by means of
the coordinate transformation
x1 =rsin0cos
x2=rsin0sin (6.75)
x3 = r cos 0.

In terms of these coordinates, the Schwarzschild metric (6.74) will then

have the form
2Gm
=1-

90r = 0, (6.76)

2Gm/r xrx9
9rs = -ars - 1 - 2Gm/r r2

6.1.6 Experimental Tests of General Relativity

Up to a few years ago, general relativity was verified by three tests: the
gravitational red shift, the deflection of light near massive bodies and the
planetary orbit effect on the planets. The first could also be explained, in
fact, without the use of the Einstein field equations. However, this picture
has been changed.

Gravitational Red Shift

Consider the clocks at rest at two points 1 and 2. The rate of change of
times at these points are then given by ds (1) = 900 (1)dt and ds (2) =
goo (2)dt. The relation between the rates of identical clocks in a gravi-
tational field is therefore given by goo (2) //900 (1). The frequency of an
atom, vo, located at point 1, when seen by an observer at point 2 is, hence,
128 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

given by
Jgoo (1)
U (6.77)
- Up Too (2)'
For a gravitational field like that of Schwarzschild, one therefore obtains
for the frequency shift per unit frequency:
Ov _ v - vpGm 1 1
(6.78)
Up
Up C2 (r1 T2

to first order in Gm/c2r. If we take rl to be the observed radius of the

Sun and r2 the radius of the Earth's orbit around the Sun (thus neglecting
completely the Earth's gravitational field), then OU/vp = -2.12 x 10-6.
This frequency shift is usually referred to as the gravitational red shift.
The gravitational red shift was tested for the Sun and for white dwarfs,
and it was suggested that it be tested by atomic clocks. The red shift was
also observed directly using the Mossbauer effect by Pound and Rabka,
and by Cranshaw, Schiffer and Whitehead. The latter employed Fe" and
a total height difference of 12.5 metres. A red shift 0.960.45 times the
predicted value was observed by them. Pound and Rabka's result is more
precise. They obtained a red shift 1.050.10 times the predicted value.

Effects on Planetary Motion

One assumes that test particles move along geodesics in the gravitational
field (see next subsection), and that planets have small masses as compared
with the mass of the Sun, thus behaving like test particles. Consequently, to
find the equation of motion of a planet moving in the gravitational field of
the Sun one has to write the geodesic equation in the Schwarzschild field. In
fact one does not need the exact solution (6.76) but its first approximation,
goo = 1 - 2Gm/r,
go, = 0, (6.79)
gr9 = -6,.,s - 2Gmx'x9/r3.
In the above equations the speed of light is taken as unity.
Using the approximate metric (6.79) in the geodesic equation (6.43)
gives (see Problem 6.24)
x - GmV 1
r
= Gm 2 (*2) V r - 2Gm 1 V I - 2 ic V r) * + T (x x)2 x1 (6.80)
6.1. ELEMENTS OF GENERAL RELATIVITY 129

where we have used three-dimensional notation and a dot denotes differen-

tiation with respect to t. Multiplying Eq. (6.80), vectorially, by the radius
vector x gives
//
xxx=-2Gm(x.V !)(xxic),
r
(6.81)
\\

thus leading to the first integral

x X is = Je-ZGm/r (6.82)

where J is a constant vector, the angular momentum per mass unit.

Hence the radius vector x moves in a plane perpendicular to the vector
J, as in Newtonian mechanics. Introducing in this plane polar coordinates
r, 0 to describe the motion of the planet, the equation of motion (6.80),
consequently, decomposes into

r - rc2 + Gm = Gm
3rz - 2r +2 G-m } , (6 . 83a)
r2 r2 l r JJJ

r2 = Je-2Gm/r, (6.83b)

where J is the magnitude of the vector J.

Introducing now the new variable u = 1/r one can rewrite Eqs. (6.83)
in terms of u (0):

U+u- J2 = Gm (_U '2 + 2u2 + 2 J2 u) . (6.84)

Here a prime denotes a derivation with respect to the angle q5.

Let us try a solution of the form
u = b (1 + (: cos aO). (6.85)

Here a is the eccentricity and a is some parameter to be determined and

whose value in the usual nonrelativistic mechanics is unity. The other
constant b is related to J in the nonrelativistic mechanics by Gm/J2 = b.
Using the above solution in Eq. (6.84) and equating coefficients of cos ao
gives
a2 = 1 - 2Gm (2b + Gm/ j2) . (6.86)
Substituting for Gm/J2 its nonrelativistic value b then gives a2 = 1-6Gmb,
or, to a first approximation in Gm,
a = 1 - 3Gmb. (6.87)
130 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

Successive perihelia occur when

(1 - 3Gmb) (2ir + 0O) = 2ir. (6.88)

Consequently, there will be an advance in the perihelion of the orbit per

revolution given by A0 = 67rGmb, or A0 = 6,7rGm/a (1 - e2) if we make use
of the nonrelativistic value of the constant b, and where a is the semimajor
axis of the orbit. Reinstating now c, the velocity of light, finally gives for
the perihelion advance
61rGm
A0 = (6.89)
c2a (1 - E2)
in radians per revolution.
We list below the calculated values of A0 per century for four planets:

Planet A0
Mercury 43.03"
Venus 8.60"
Earth 3.80"
Mars 1.35"

The astronomical observations for the planet Mercury give 43.11 0.45 sec
per century, in good agreement with the calculated value.

Deflection of Light
To discuss the deflection of light in the gravitational field we must again
solve the geodesic equation, but now with the null conditions ds = 0. Using
the appropriate solution (6.79) then gives for g,,,,dxdx" = 0

C1 + 2Gm) [,c 2Gm (x x)2J

is + = 1. (6.90)
r r3

Using polar coordinates r, q5, consequently, gives to the first approximation

in Gm
T2 + 4Gmr2
+ 1. (6.91)
r
Again changing variables into u (0) = 1/r, and using Eq. (6.83b), gives

u'2 + u2 + 2Gmu (2u'2 + u2) = J-2e4Gmu. (6.92)

6.1. ELEMENTS OF GENERAL RELATIVITY 131

Differentiation of this equation with respect to 0 gives

u" + u + Gm (2ui2 + 4uu" + 3u2) = 2GmJ-2, (6.93)

to the first approximation in Gm.

To solve Eq. (6.93) we note that in the lowest approximation one has

u2 J-2 - u2, (6.94)

U/1
-u. (6.95)
Using these values in Eq. (6.93) gives

u" + u = 3Gmu2, (6.96)

for the orbit of the light ray. In the lowest approximation u satisfies u"+u =
0, whose solution is a straight line

u= cos
R, (6.97)

where R is a constant. This shows that r = 1/u has a minimum value R at

0 = 0. Substituting into the right-hand side of Eq. (6.96) then gives

u1 +u= 3 Gm
R2
cos 2 (6 . 98)

The solution of this equation is

U =cosR + Gm
R2
(1 + s in
2 0)
. (6 . 99)

Introducing now Cartesian coordinates x = r cos 0 and y = r sin , the

above equation gives

x=R- Gm x2 + 2y2
R ,,,,,-x2+
y2 . (6.100)

For large values of I y I this equation becomes

R-2R l y (6 . 101)

Hence, asymptotically, the orbit of the light ray is a straight line in

space. This result is expected, since far away from the central mass the
132 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

space is flat. The angle 04 between the two asymptotes is, however, equal
to
AO=4 GR (6.102)

in units in which c is different from unity.

The angle AO represents the angle of deflection of a light ray in passing
through the Schwarzschild field. For a light ray just grazing the Sun Eq.
(6.102) gives AO = 1.75 sec. Observations indeed confirm this result; one
of the latest results gives 1.75 0.10 sec.

Gravitational Radiation Experiments

Weber has developed methods to detect gravitational waves that Einstein's
gravitational field equations predict. The experiment involves detectors at
opposite ends of a 1000 km baseline. Sudden increases in detector output
were observed by him roughly once in several days, coincident within the
resolution time of 0.25 seconds.
Weber's apparatus measures the Fourier transform of the Riemann ten-
sor. The method uses the fact that the distance i between two neighbour-
ing test particles, which follow geodesics, satisfies the geodesic deviation
equation
a2
+ R avp Aa7l vA13 = 0 , (6 . 103)
52
where \a is the tangent vector to one of the geodesics, and b/bs = A"V
is a directional covariant derivative. Weber measured the strains of a large
aluminium cylinder, having mass of the order 106 grams, by means of a
piezoelectric crystal attached to the cylinder which transforms the mechan-
ical movement into an electric current. The detector was developed for
operation in the vicinity of 1662 cycle/sec. A high frequency source was
developed for dynamic gravitational fields and the detector was tested by
doing a communication experiment with high frequency Coulomb fields.

Radar Experiment
Shapiro has designed a radar experiment to test general relativity by mea-
suring the effect of solar gravity on time delays of round-trip travel times of
radar pulses transmitted from the Earth toward an inner planet, i.e., Venus
or Mercury. The experiment is based on the phenomenon that electromag-
netic waves "slow down" in the gravitational field. Within the framework of
general relativity there should be an anomalous delay of 200 microseconds
6.1. ELEMENTS OF GENERAL RELATIVITY 133

in the arrival time of a radar echo from Mercury, positioned on the far side
of the Sun near the limb.
For example, if we calculate the proper time T at r = r2 for a radial
round-trip travel r2 -4 rl --> r2i with r2 > r1, of a radar pulse in the
Schwarzschild field, and subtract from T the corresponding value To when
the spherical mass m = 0, we find
4 r2
OT = C3 I In
\ 1
- r2 r 2
rl + O (m2) (6.104)

In gener a l on e finds

OT - 4 Gm (I n
e
r e + rpp - R/
(6 . 105)

where re is the Earth-Sun distance, rp the planet-Sun distance, and R the

Earth-planet distance.
Shapiro found that the retardation of radar signals are 1.020.05 times
the corresponding effect predicted by general relativity.

Low-Temperature Experiments
Schiff has proposed an experiment to check the equations of motion in
general relativity by means of a gyroscope, which is forced to go around the
Earth either in a stationary laboratory fixed to the Earth or a satellite. The
unique experiment is made possible by complete use of a low-temperature
environment, and the properties of superconductors, including the use of
zero magnetic fields and ultrasensitive magnetometry. Schiff has calculated,
using results obtained by Papapetrou for the motion of spinning bodies
in general relativity, that a perfect gyroscope subject to no torques will
experience an anomalous precession with respect to the fixed stars as it
travels around the Earth.

6.1.7 Equations of Motion

Geodesic Postulate
In the last subsection it was assumed that the planet's motion around the
Sun is described by the geodesic equation (6.43). The assumption that the
equations of motion of a test particle, moving in gravitational field, are
given by the geodesic equation is known as the geodesic postulate and was
suggested by Einstein in his first article on the general theory of relativity.
134 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

Eleven years later when Einstein and Grommer showed that the geodesic
postulate need not be assumed, but that it rather follows from the gravita-
tional field equations; this is a consequence of the nonlinearity of the field
equations along with the fact that they satisfy the four contracted Bianchi
identities (see Subsection 6.1.1). The discovery of Einstein and Grommer
is considered to be one of the most important achievements, and one of the
most attractive features of the general theory of relativity. Later on Infeld
and Schild showed that the equations of motion of a test particle are given
by the geodesic equation in the external gravitational field. This result,
however, does not differ from the geodesic postulate because, by definition,
a test particle has no self-field.

Equations of Motion as a Consequence of Field Equations

In order to establish the relation between the Einstein field equations and
the equations of motion one proceeds as follows. We have seen in Subsection
6.1.4 that because of the contracted Bianchi identity it follows that the
energy-momentum tensor TI" satisfies a generally covariant conservation
law of the form given by Eq. (6.51). Consequently, one obtains for the
energy-momentum tensor density T"'
0"T"'" = a T" + FPTo'p = 0, (6.106)

where Ti"
For a system of N particles of finite masses, represented as singularities
of the gravitational field, T" may be taken in the form
N
T" = > mgvAVA5A (X - ZA) . (6.107)
A=1

Here zA are the coordinates of the Ath particle. (Roman capital indices,
A, B, , run from 1 to N. For these indices the summation convention
will be suspended.) Also v = z" = dz"ldt (vy = zA = 1), and b is the
three-dimensional Dirac delta function satisfying the following conditions:

J (x) = 0; for x # 0, (6.108a)

f 6 (x - z) d3x = 1, (6.108b)

J f (x) 6 (x - z) d3x = f (Z), (6.108c)

6.1. ELEMENTS OF GENERAL RELATIVITY 135

for any continuous function f (x) in the neighbourhood of z. In Eq. (6.107),

MA is a function of time which may be called the inertial mass of the Ath
particle.
If we put the energy-momentum tensor density (6.107) into (6.106) and
integrate over the three-dimensional region surrounding the first singularity,
we obtain
dp
dt = fFo(x - z) d3x, (6.109)

where p = mv and F = -mr'pv`1vp, and where we have put, for

simplicity, m = ml, zw = z1 , v =V 1, and b (x - z) = b1 (x - z1).

Self-Action Terms
Equation (6.109) may be interpreted as an "exact equation of motion" of
the first particle. However, since the Christoffel symbols are singular at
the location of the particle, the equation contains infinite self-action terms.
However, it was shown by Carmeli that these terms can be removed as
follows.
Putting Eq. (6.107) into Eq. (6.106) we obtain

ao + an + 0,
A=1 A=1
(6.110)
where Latin indices run from 1 to 3. The first term on the left-hand side
of Eq. (6.106) can be written as
N N N
a0 E mAVAJA = a0 (mAVA) 5A + E MAVgaObA, (6.111)
A=1 A=1 A=1

with
aO5A = MA (x9 - zA) = -an&AVA (6.112)
Using the above results in Eq. (6.110), we obtain
N
d
E
A=1
( dtv A) + mAr"PVA'fJA JA = 0. (6.113)

Equation (6.113), which is identical with Eq. (6.106), is satisfied for any
spacetime point, since otherwise the Bianchi identity or the Einstein field
equations will not be satisfied.
136 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

We now examine the behavior of Eq. (6.113) in the infinitesimal neigh-

bourhood of the first singularity, which we assume not to contain any other
singularity. In this region bB (x - ZB) = 0 for B = 2, 3, , N. Hence Eq.
(6.113) gives for the conservation law near the first singularity
rd(mvw) 1
} mI apvavp 1 b (x - z) = 0. (6.114)
5l dt

Let us further assume that the Christoffel symbols near the first singularity
can be expanded into a power series in the infinitesimal distance r, defined
by r2 = (x' - z') (x' - z'), where z' = zj, in the vicinity of the first
particle. Then we have

rap --k rap +-k+1 rap + ... +0 rap + .. . (6.115)

where the indices written in subscripts on the left of a function indicate its
behavior with respect to r, and k is a positive integer.
For example oF1 is that part of the Christoffel symbol which varies as
ro, i.e., is finite at the location of the first particle. When one uses spherical
coordinates r, 0 and , one can write

-krap = r-kAap (6.116a)

k+1 rap = r-k+1 Bap (0, 0) , (6.116b)

orap =Da (0, 0), etc. (6.116c)

Terms like lI'a trap, etc., however, need not be taken into account when
one puts the above expansion into Eq. (6.114) since rib (x - z) = 0 for any
positive integer j. If we denote now mAapv'vp, by A, . we can write
Eq. (6.114) in the form

{r-k A +r -k+ 'Bw + + r-1Cw + Di } b (x - Z) = 0, (6.117)

where we have used the notation Di = d (mv) /dt + D.

In order to get rid of terms proportional to negative powers of r in Eq.
(6.117) we proceed as follows. Multiplying Eq. (6.117) by rk and using
rib (x - z) = 0 we obtain
A (0, ) 5 (r) = 0, (6.118)
6.1. ELEMENTS OF GENERAL RELATIVITY 137

the integration of which over the three-dimensional region yields, using

spherical coordinates,

ff A" (9,c,) sin OdOd-O J r26 (r) = 0. (6.119)

From the property of the delta-function

16(r) d3x =
sin OdOdo rzdr = 1, (6.120)

one obtains f 6 (r) rzdr =(4ir)-1. Hence we obtain

ff A (0, 0) sin Od0do = 0, (6.121)

independent of the value of the variable R. Thus the angular distribution

of A (0, ) is such that its average equals zero.
However, not only does the above equation hold, but also (s is any finite
positive integer)

a (r) = r_' ff A" (0, 0) sin 0d0d4 = 0, (6.122)

for small values of r as well as when r tends to zero, as can be verified

by using L'Hospital's theorem, for example. It follows then that a (r) is a
function of r whose value is zero for any small r, including r = 0. Using
the property of delta-function we obtain

fr26(r)dr= (4ir)-1 f (0) , (6.123)

for any continuous function of r. Since a (r) is certainly continuous, one

obtains
fr26 (r) a (r) dr = 0. (6.124)

Hence when one integrates Eq. (6.117) over the three-dimensional space,
there will be no contribution from the first term.
In order to show that the second term of Eq. (6.117) will also not
contribute to the three-dimensional integration of the same equation, we
multiply it by rk-1. We obtain now, after neglecting terms that do not
contribute,
{r-1A'`(0, 0)+B(0,0)}6(r)=0. (6.125)
138 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

Integration of this equation, again using spherical coordinates, shows that

the first term will not contribute anything because of Eq. (6.124), and we
are left with
ff B (0, 0) sin Od0do r2b (r) dr = 0. (6.126)
J
Hence we have
ff B" (0, ) sin Od0do = 0, (6.127)

independent of r. From this equation one obtains another one, analogous

to Eq. (6.124) but with B" instead of A":

f r2b (r) b (r) dr = 0, (6.128)

with
b (r) = r ff B" (0, 0) sin Od0do = 0. (6.129)

Proceeding in this way, one verifies that the angular distribution of all
functions A, BA, etc., is such that they all satisfy equations like Eqs.
(6.121) and (6.127). Hence it is clear that one obtains

Di (0, 0) b (r) d3x = 0, (6.130)

J
which gives
W
dt + my"vp or11pb (r) d3x = 0, (6.131)

or equivalently

'lik + vvp f (orkP - vkOr ) b (r) d3x = 0. (6.132)

Equation (6.132) is the "exact equation of motion".

Einstein-Infeld-Hoffmann Method
Having found the law of motion (6.131), one can now proceed to find the
equation of motion of two finite masses, each moving in the field produced
by both of them. In the following we find such an equation of motion in the
case for which the particles' velocities are much smaller than the speed of
light. Moreover, we will confine ourselves to an accuracy of post-Newtonian.
This means the equation of motion obtained will contain the Newtonian
6.1. ELEMENTS OF GENERAL RELATIVITY 139

equation as a limit, but is a first generalization of it. Such an equation was

first obtained by Einstein, Infeld, and Hoffmann. To obtain this equation
we solve the field equations and formulate the equations of motion explicitly
by means of an approximation method, the Einstein-Infeld-Hoffmann (EIH)
method, to be described below.
Let us assume a function developed in a power series in the parameter
. = 1/c, where c is the speed of light. One then has

0 =00+ 10+20+ - (6.133)

The indices written as left subscripts indicate the order of A absorbed by

the O's.
If a function 0 (x) varies rapidly in space but slowly with xa, then we
are justified in not treating all its derivatives in the same manner. The
derivatives with respect to x will be of a higher order than the space
derivatives. We thus write

ao (1o) = t+i',. (6.134)

That is, differentiation with respect to x raises the order by one. Thus if
the coordinates z9 of a particle are considered to be of order zero, i' will be
of order one, and P of order two. Using now the Newtonian approximation
mass x acceleration=mass x mass/(distance)2, we see the mass is of order
two. In all the power developments we take into account only even or only
odd powers of 1/c. (The expansion of the metric tensor, etc., in a power
series in c-2 (such as 0 = 00 + 20 + - - -, or 0 = 10 + 395 + .. ) corresponds
to the choice of the symmetric Green function, thus excluding radiation.)
Thus, because of the order with which we start m and z', we have
TOO =2 Too +4 Too + .. .

,rOn =3 TOn +6 TOn + .. .

Tmn =4 Tmn +6 Tmn + ... .

As to the metric tensor, we write

9v = 77v + h". (6.136)
9/&v = '7v + hv,
The gravitational field equations can be written as

I2 gapT , (6.137)
)
140 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

where T = T"g", and R,,,Q is the Ricci tensor. From the right-hand side
of the field equations it follows that Roo and R,,,.n (when m = n) start with
order two, Rn (when m # n) start with order four, while Rom starts with
order three. The lowest order expressions of the left-hand side are

Roo ti I hoo,ss,
Rom 2 (hom,ss - hos,ms - hms,os + hss,or), (6.138)

Rmn (hmn,ss - hms,ns - hns,ms - ho0,mn + hss,mn) i

2
where a comma denotes a partial derivative, 0,s = a .,O. Hence we have

hom = An + An +-) (6.139)

hmn = 2hmn + 4hmn + '

Newtonian Equation of Motion

We now find the equation of motion in the lowest (Newtonian) approx-
imation. We do it in such a way as to make the generalization to the
post-Newtonian approximation as simple as possible.
Because of Eqs. (6.137) and (6.138), the field equations of the lowest
order are in hoo,
2

22hoo,ss = K zT - 22700) = 22T00 = 2 E PAJA, (6.140)

J A=1

where, for simplicity, we have put AA = 2mnA. Hence the equation obtained
is
2

2hoo,ss = E AAbA (6.141)

A=1
The solution of this equation that represents two masses is
2
2h00 = -2G E /LArA1, (6.142)
A=1
6.1. ELEMENTS OF GENERAL RELATIVITY 141

where rA = W - zA) (x' - zs4). Using 2hoo in the equation of motion

(6.132), we obtain in the lowest (second) order for the equation of motion
of the first particle

z1 -GJ ak(2r21)8(x-z1)d3x=0. (6.143)

This gives
z1 Gazk z2 (6.144)
1

where z2 = (z7 - z2) (zj - z2). Equation (6.144) is, of course, the Newto-
nian equation of motion.

Einstein-Infeld-Hoffmann Equation
To find the equation of motion up to the fourth order, we must know besides
2hoo the functions 4hoo, 3hon and 2hmn The second and third functions are
easy to find. The left-hand side of the corresponding equations is written
out in Eq. (6.138), whereas the right-hand side is given by Eq. (6.137) and
it is -K > PAZ' 8A for the Om component, and z E / A6A for the mn
component. Therefore, for the 2hmn we have the equation

2hmn,ss - 2hms,ns - 2hns,ms + 2hss,mn - 2h00,mn = amn2ho0,ss, (6.145)

whose solution is
2hmn = 5mn 2 h00. (6.146)

The equation for 3hon is

An
AAzn

3hOn,ss - 3hOs,ns - 2hns,Os 2hss,nO = -2r.

Using the value of 2hmn in terms of the 2hoo found above, we obtain
2
3hOn,ss - 3hos,ns + 22hoo,no = -2K E ILAinjA. (6.148)
A=1

The solution of this equation is

2

3hon = 4G E AAZArA1. (6.149)

A=1
142 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

Calculation of 4hoo is somewhat more complicated. The relevant part

of 4hoo, for two masses, that contributes to the equation of motion of the
first particle, is

4hoo G {2G22r2 2 - 32i2i2r2-1 - 2r2,oo + 2G12 (zr2)-1 } . (6.150)

Using these values for 4hoo, 3hon, and 2h,,,.m in the equation of motion
(6.132) gives, for the two-body problem (Problem 6.25):

=2{l iiij 2i2i2 -4iii2 -42z z

3(11 z)
8z" 1

8 -3 Z
+ [4i1 (i2 - in) + 3ii i2 - 4i2i2] az/z) + 2 z2z2 (6.151)
i 8za az" }. 1 1 1

In Eq. (6.151) the Newtonian gravitational constant G was taken as equal

to 1. The equation of motion for the second particle is obtained by replacing
1,2, z1, z2 by 2,1,z2, z1, respectively.
Equation (6.151) is known as the Einstein-Infeld-Hoffmann equation of
motion, and is a generalization of the Newton equation. The essential rel-
ativistic correction may be obtained by fixing one of the particles. Writing
M for 2, neglecting 1 and ,i2, and using an obvious three-dimensional
vector notation, Eq. (6.151) simplifies to

z-MV (.) =M( (zz 4M ) V (.!) -4z (6.152)

-
where z denotes the three-vector 4.

6.1.8 Decomposition of the Riemann Tensor

The Riemann curvature tensor R1pya can be decomposed into its irreducible
components. These are the Weyl conformal tensor Cgp.ya, the tracefree Ricci
tensor Sap, and the Ricci scalar curvature R. The tensor Sap is defined by

Sap = Rap - I gapR, (6.153)

where Rap is the ordinary Ricci tensor.

6.2. THE CURVATURE SPINOR 143

The decomposition can be written symbolically as

R,,,,e,j = Cap,6 (DS,,,,, R. (6.154)

No new quantities can be obtained from any of the above three irreducible
components by contraction of their indices.
When written in full details, the decomposition (6.154) will then have
the form:

1
(9pv9a - 9pogav) R. (6.155)
12
It can also be written in the form:

1
+6 (9Pvga - 9P9av) R. (6.156)

6.2 The Curvature Spinor

We first derive the curvature spinor and investigate its properties. This is
done along the lines of deriving the Riemann curvature tensor. Instead of
applying the commutator of the covariant derivatives on a vector, however,
we apply it on a spinor.
In Subsection 5.4.4 we have seen how such a procedure works when the
commutator of the spin covariant derivatives were applied on a spinor. Here
we use the ordinary (spacetime) covariant derivatives.
If we differentiate covariantly the quantity vEQ, given by Eq. (5.52),
we obtain

V vvt4 = av (V 4) - rQvvCa, (6.157)

where
0S4 = aSQ - rQSP (6.158)
Substituting Eq. (6.158) in Eq. (6.157), the latter equation then yields

vvv 4 = avrQ,cP -
+r,\, rQX - (6.159)
144 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

Calculating now the same expression, but with the indices and v being
exchanged, and subtracting it from the expression (6.159), we then obtain
(V ,VY, - Vvy) SQ = -F Py(P (6.160)

where the mixed quantity F", is given by

FW, = rQ,v - rQy, + rQrBy - I QyrPB. (6.161)
In Eq. (6.161) a comma followed by a Greek letter indicates a partial
differentiation, f, = 8f/8x". The quantity will be referred to in
the sequel as the curvature spinor.
In the same way, if we apply the commutator (VvV - V, V,) to the
spinor Q we then obtain
(0v0 - VVv) SQ = F
CC C
PyCP. (6.162)

Equations (6.160) and (6.162) are analogous to the formulas for defining the
Riemann curvature tensor (see, for instance, Eisenhart). The occurrence of
the minus sign in the curvature spinor is just a matter of convention.
We may also apply the commutator of the covariant derivatives to prod-
ucts of spinors and spinors with more than one index, using a combination
of Eqs. (6.160) and (6.162). Thus, for instance, we obtain
(VvV - VV,) ((P?Q) = -F py(A77Q + FQ,,eP?IA, (6.163)

for arbitrary one-index spinors CP and 779. Likewise we obtain

(VvV - VV-) (P = FQA, P - F P1Ay(Q1 (6.164)

for an arbitrary spinor (P with two unprimed indices.

6.2.1 Spinorial Ricci Identity

The above formulas may be further generalized to spinors of higher orders
and to those with primed indices as well. Thus we obtain
00v) CPR... = FQ v(PR... + FAv(PR...

-F py(QR... - F Ry(PA... ' (6.165)

for an arbitrary spinor (PR:;: with unprimed indices. Likewise we obtain

(VvV - V, Vv) (PQ' = -F A y(AQ'

-A'
F Qiv(PA', (6.166)
p
6.2. THE CURVATURE SPINOR 145

(VLV - DOL) (PQ, = F A,C Qr - F' QrL(A,, (6.167)

(VV, - VV) CPQ' = FP Av(AQ' + F vCPA' (6.168)

for the arbitrary spinors (pQ,, (P, and (PQ' with mixed indices. Equation
(6.165) will be referred to as the Spinorial Ricci identity.

6.2.2 Symmetry of the Curvature Spinor

We now study the symmetry properties of the curvature spinor introduced
above. Later on we will relate it to the Riemann curvature tensor.
From Eq. (6.160) we obtain
(VV,. - V0,) Q = FPQILeP. (6.169)

We may, on the other hand, lower the free index Q in Eq. (6.162), thus
getting
(V LV - V,Ov) eQ = FQPLSP. (6.170)
Comparing now the last two equations we find that the curvature spinor
satisfies the property
FPQL = FQpL, (6.171)
namely, it is symmetric with respect to its two spinor indices P and Q.
By its definition, furthermore, it is antisymmetric in its spacetime tensorial
indices u and v, namely

FpQL = -FPQL (6.172)

To further study the symmetry properties of the curvature spinor we

define the spinor
'CD' = FPQLQAB'QCp',
FPQAB'CD' (6.173)

which is, of course, skew-symmetric under the exchange of the pairs of

indices AB' and CD'. Hence it may. be decomposed, similar to the spinor
equivalent of the electromagnetic field tensor given by Eq. (5.83).
Accordingly we may write

FPQAB'CD' = - (XPQACCB'D' + Q5PQB'D'EAC), (6.174)

where the minus sign is introduced for convenience, and where the two new
spinors XPQAC and OPQB'D' are defined by

XPQAC = - FpQAB'C B'

1 r

(6.175)
2
146 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

OPQB'D' _ -2FPQAB' D'. (6.176)

In the following we study the properties of the above two spinors. Before
doing so we relate the curvature spinor to the Riemann curvature tensor.
As we see, the curvature spinor has only six spinorial indices. Since the
Riemann tensor has four spacetime indices, its spinor equivalent will have
eight spinorial indices. We will see in the next section that the two spinors
X and 0 describe completely the Riemann spinor. It thus follows that the
curvature of spacetime is determined by a six-indices spinor (the curvature
spinor) and not by an eight-indices spinor (the Riemann spinor). In fact,
we will see that the Riemann spinor is obtained from the curvature spinor
and its complex conjugate.

6.3 Relation to the Riemann Tensor

Multiplying Eq. (6.166) by QPQ' and rearranging the indices we obtain

(V v0 - V ,V ) (PQ'ace
P Q1 D' C ' ) (CD'.
= - (F CpvOraPD' + F Q vcaQ
(6.177)
Hence we may write, since (pQ'Q.PQ' = (a is a vector,

(0v0 - VV,) (a = RCD'av(CD' = RPav(P, (6.178)

where use has been made of the notation

RCD av =- (F PJ.W a + aQ ) , (6.179)

and
RPav = RCD/avfCD'+ (6.180)
by Eq. (6.178). The tensor given by Eq. (6.180) is the Riemann curvature
tensor. The last two formulas give the relationship between the curvature
spinor and its complex conjugate on the one hand, and the Riemann cur-
vature tensor on the other hand.
From Eq. (6.179) we now obtain

RAB'EF'v = RCD avECAED'B'cr F'

= FAEvfB'F' + EAEFB'F'v. (6.181)

6.3. RELATION TO THE RIEMANN TENSOR 147

Equivalently, the latter equation may be written in the form

RAB'EF'MN'PQ' = FAEMN'PQ'EB'F' + EAEFB'F'N'MQ'P (6.182)
The left-hand side of the above formula is the spinor equivalent of the
Riemann curvature tensor.
We may also obtain equations for the curvature tensor. Multiplying Eq.
(6.181) by EB'F' we obtain

FAEv = 2RAB'EBv = 2QaAB'a E (6.183)

Likewise we obtain
1 RABAF'' = 2QcAB'aOA FiRapv, (6.184)

by multiplying Eq. (6.181) by eAE

6.3.1 Bianchi Identities

We may also write the Bianchi identities in terms of the curvature spinor.
From Eq. (6.183) we then obtain
V FAEpy + OpFAE-y. + OryFAEap

= 2Q1,AB'QvEB' (vaR",, +VPRvya+07Rvap) = 0, (6.185)

where the last equality is obtained from the ordinary (tensorial) Bianchi
identities,
vyR`vap + 0. (6.186)
Defining the dual to the curvature spinor by

*FPQap = 2 r-Eal3 wFPQlsv = I FpQvEv, (6.187)

the Bianchi identities may be written in the form

Op *FpQap = 0. (6.188)
The relations (6.185) and (6.188) may also be written in the forms
VAB'FPQCD'EF' + VCD'FPQEF'AB' + VEF'FPQAB'CD' = 0, (6.189)
and
VGH' *FPQEF'GH' (6.190)
= 0,
respectively, when written in spinor forms.
In the next section we further discuss the graviational field dynamical
variables.
148 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

6.4 The Gravitational Field Spinors

We are now in a position to find the spinors in terms of which the gravi-
tational field is described. We have already found in Chapter 5 the spinor
equivalent to the geometrical metric tensor g whose expression was shown
to be given by the flat spacetime metric
0 0 1

-1 0
9AB'CD' = EACEB'D' 0 -1
(6.191)
0
1 0

The rows and the columns of the 4 x 4 matrix (6.191) are labeled by the
pairs of indices AB' and CD', each taking the values (1, 2, 3, 4)=(00', 01',
10', 11).

6.4.1 Decomposition of the Riemann Tensor

We next discuss the Riemann curvature tensor and decompose its spinor
equivalent, which is given by

RAB'CD'EF'CH' = aAB'QCD'UEF'aCH'Rc,pya. (6.192)

Decomposing this spinor by the method of decomposing the spinor equiv-

alent of the electromagnetic field used in Chapter 5, we then obtain

RAB'CD'EF'GH' = (fACRPB'PD'EF'CH' + RAP'CPEF'CH'EB'D')

2

= 4EAC (RPB'PD'KF' RPB'PD'EL'G L EF'H')

+4 (RAP'CPKF HK I EEG +RAP'CPEL'G L'fF'H') EB'D' (6.193)

The proof of the above formula is left for the reader (see Problem 6.3).
To compare the above decomposition for the spinor equivalent of the
curvature tensor with that given in the last section for the same tensor, we
denote the last two terms on the right-hand side of Eq. (6.193) as follows:

L' (6.194)
XACEG = -4RAP'CPEL'G

P,
_ -4 RAP'CKF' K
H'- (6.195)
6.4. THE GRAVITATIONAL FIELD SPINORS 149

Using now the decomposition (6.182) for the spinor equivalent to the Rie-
mann curvature tensor in Eqs. (6.194) and (6.195), we get
1
XACEG = --FACEL'G L' (6.196)

OACF'H' 21 FACKF' H'- (6.197)

Comparing the last two formulas with Eqs. (6.175) and (6.176), we find
that they are identical. Hence Eqs. (6.194) and (6.195) are consistent with
our previous definitions for the same quantities XPQAB and OPQA'B' given
by Eqs. (6.175) and (6.176), respectively, when the decomposition (6.182)
is used.
The decomposition of the spinor equivalent of the Riemann curvature
tensor, given by Eq. (6.193), may be further simplified if we notice that the
first two terms on the right-hand side of that equation may be written in
terms of the complex conjugate of the spinors XABCD and OABC'D'. To see
this we use the fact that the Riemann curvature tensor is real, and therefore
it satisfies
R P K P
PB' D'KF' H' - R B'PD' F' KH'
_ KR P,
BP' D FK' H
K'

= -4XBDFH = -4XB'D'F'H', (6.198)

R P P L'
PB' D'EL'C
L' - RB'PD' L'E C . RBP'DP'LE' LG'
_ -4cBDE'G' = -40B'D'EG' (6.199)

by Eqs. (6.194) and (6.195). Accordingly we finally obtain

RAB'CD'EF'GH' _ -(XACEG EB'D' EF'H' + Q5ACF'H' EB'D' EEG

+EAC B'D'EC EF'C' +EAC EEG XB'D'F'H'), (6.200)

for the decomposition of the spinor equivalent to the Riemann curvature
tensor.
We next decompose the spinor equivalent to the dual of the curvature
tensor. If *R,,,p.ya is the dual to the Riemann curvature tensor, then its
spinor equivalent is given by

*RAB'CD'EF'GH' = i(XACEC EB'D' EF'H' - WACF'H' CB'D' -EEG

+EAC B'D'EG EF'C' - EAC EEC XB'D'F'H'), (6.201)

The proof of the above formula is given in Problem 6.4.
150 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

6.4.2 The Gravitational Spinor

The two spinors XABCD and 4ABC'D' uniquely determine the spinor equiv-
alent to the Riemann curvature tensor. The symmetry properties of the
spinor XABCD follow from the symmetry properties of the Riemann tensor
given by
Rc,p76 = -Rpa.y6 = -Rapa.y, (6.202a)

Rap-y6 = Ry6czp, (6.202b)

Rap-y& + Ra.y6p + Ra6p7 = 0. (6.202c)

Because of the relation R,,p.y6 = -Rpay6i for instance, we have

1 L 1 L
XCAEG = 4 RCP'APEL'G = 4-RA CP'EL'G

I P, L'
_ -4 RAP'C EL'G = XACEG (6.203)

In the same way, using the fact that R,,,p.y6 = we find that
XACEG is symmetric with respect to the two indices E and G, namely,
XACEG = XACGE Finally, using the fact that Rap76 = R.y6ap leads to the
symmetry of XACEG under the exchange of the first and second pairs of
indices, XACEC = XEGAC. Accordingly we have

XABCD = XBACD = XABCC, (6.204)

XABCD = XCDAB, (6.205)

which the spinor XABCD satisfies.

Similarly we find that the spinor ABC'D' is symmetric under the ex-
change of indices A and B and C' and D'. We have, moreover,
1 P,
cC'D'AB = OCDA'B' = -4RCP'D KA'KB'
1_ P K' _ K' P
_ -4RC'PD' K'A B = -4RK'A BC'PD' (6.206)

by Eq. (6.195) and using the symmetry of the Riemann tensor. Using now
the fact that the Riemann tensor is real and hence its spinor equivalent is
Hermitian, we then obtain

OC'D'AB = -4RAK'BKPC' D' = OABC'D' (6.207)

6.4. THE GRAVITATIONAL FIELD SPINORS 151

Summarizing the above results we find the following formulas:

OABC'D' = `VBAC'D' = cbABD'C', (6.208)

OABC'D' = v'C'D'AB- (6.209)

Equations (6.208) and (6.209) express the symmetry properties of the spinor
OABC'D'
Because of the symmetry properties (6.204) and (6.205), the spinor
XABCD behaves like a 3 x 3 symmetric complex matrix. This fact may
easily be seen since each pair of the indices AB and CD takes the three
values (1, 2, 3)=(00, 01=10, 11) and XABCD is unchanged under the ex-
change of AB with CD. Hence the spinor XABCD may have at most six
complex components. These components, however, are not entirely inde-
pendent. For if we calculate the trace of XABCD,
=EAC6BDXABCD, (6.210)
X = XAB AB

we find that A is a real quantity.

Reality of A
The reality of A may be seen using the symmetry property expressed by
*RP.PP = 0, which the dual to the Riemann curvature tensor satisfies. In
spinor calculus the above equation is given by
*REF' (6.211)
AB'EF'CD' _ 0.
Using the expression for the spinor equivalent to the dual of the Riemann
curvature tensor, given by Eq. (6.201), in Eq. (6.211), we then find
*REF'
AB'EF'CD'

(-X AEC EB'D' +OCAB'D' - D'B'AC+EAC XF'B'F'D') = 0.

(6.212)
The two terms with 0 and on the right-hand side of the above formula
cancel out because of Eqs. (6.208) and (6.209). Hence Eq. (6.212) reduces
to
X AEC EB'D' = CAC XF B'F'D', (6.213)

from which, by multiplying it by ED'B',

E 1 _ 1
X AEC = --EAC XF DF'D' = --EAC XF'DF
- D' (6.214)
152 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

Accordingly we obtain
XEAEC = 2EACX, (6.215)

where A is defined by Eq. (6.210). Multiplying now Eq. (6.215) by eAc

the latter equation then yields
A=X, (6.216)

namely, ) is real.
As a consequence of the reality of A, the spinor XABCD has only 11 in-
dependent real components rather then 12. In the sequel the spinor XABCD
is shown to describe the Weyl spinor plus the Ricci scalar curvature, and
it will be referred to as the gravitational spinor.
The spinor ABC'D', on the other hand, behaves like a 3 x 3 Hermitian
matrix. This fact may easily be seen if we write OABC'D' in the form of the
matrix
000 001 4'02 0ooo'o' 4ooo'i' 0ooi'i'
_ 010 011 012 = 0oio'o' 0oio'i' 0oii'i' (6.217)
020 021 22 0iio'o' 0iio'i' -0iii'i'
Hence the matrix elements satisfy 4mn = fin,,,,, with m, n = 0, 1, 2, by Eq.
(6.209), namely, the matrix P is Hermitian, V = -P.
Accordingly the spinor 4ABC'D' has three complex components 001,
002, '12 and three real components too, 01i, 022, namely, it has nine real
independent components. In the sequel the spinor cABC'D' is shown to
describe the tracefree Ricci tensor
1 (6.218)
S, = R - 4
6.4.3 The Ricci Spinor
We now calculate the Ricci spinor. It is given by

RAB'CD' = REF'AB'EF'CD' = eEGEF'H'

RGH AB EF CD' (6.219)

Using the expression (6.200) for the spinor equivalent to the Riemann cur-
vature tensor, we then obtain

RAB'CD' XEA C EB'D' - Q5CAB'D' - QD'B'AC + EAC XF'B'F D')

(6.220)
6.4. THE GRAVITATIONAL FIELD SPINORS 153

The second and third terms on the right-hand side of the above formula
are equal to each other by Eqs. (6.208) and (6.209). Moreover, from Eqs.
(6.215) and (6.216) we obtain
1
E (6.221a)
XEA C = 2 CAC,

FD' (6.221b)
XF' B' = Z1 EB'D' a,
Hence the Ricci spinor is given by

RAB'CD' = 2cACB'D' - A CAC EB'D' (6.222)

The Ricci scalar curvature is given by R = R Hence we have

R = RAB AB' = -4A. (6.223)

As a result, the spinor equivalent to the tracefree Ricci tensor is given by

SAB'CD' = RAB'CD' - 49AB'CD'R, (6.224)

or, using Eqs. (6.222) and (6.223), we obtain

SAB'CD' = 2cbACB'D' (6.225)

Hence the spinor ACB'D' is equal to one half the spinor equivalent of the
tracefree Ricci tensor.
The spinor equivalent of the Einstein tensor is given by

GAB'CD' = RAB'CD' - 19AB'CD'R

= 20ACB'D' +A EAC EB'D' (6.226)

6.4.4 The Weyl Spinor

The decomposition of the spinor equivalent to the Riemann tensor given
above is not complete since the spinor XABCD may be further decomposed
and related to the Weyl conformal spinor. To this end we write the spinor
XABCD in the form
1 1
XABCD = (XABCD + XACBD + XADBC) + (XABCD - XACBD)
3 3
154 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

1
+3 (XABCD - XADBC) (6.227)

Hence we may write

1 1
XABCD = ?GABCD+3 (XABCD - XACBD)+3 (XABCD - XADBC), (6.228)

where
'GABCD = (XABCD + XACBD + XADBC) (6.229)
3
We notice that the first expression in brackets on the right-hand side
of Eq. (6.228) is antisymmetric in the indices B and C. Hence using Eq.
(5.70), it may be written in the form
1 1 E (6.230)
3 (XABCD - XACBD) = 3XAE D EBC.
The last term of Eq. (6.228) may also be written as
1 1 (6.231)
(XABCD -XADBC) = (XABDC -XADBC) = 3XAE C EBD,
3 3
by Eqs. (6.204) and (5.70). Using now Eq. (6.221), furthermore, we finally
obtain the following for Eq. (6.228):

XABCD = PABCD + 6 (EACEBD + EADEBC) (6.232)

Since A is a real quantity, it follows that the spinor OABCD has 10 indepen-
dent real components.
The symmetry of the spinor ?PABCD may be found as follows. From Eq.
(6.232) we see that it satisfies the same symmetry as the spinor XABCD,
namely, 'OABCD ='GBACD = I)ABDC = ',CDAB In fact the spinor ',bABCD
is symmetric with respect to all of its four indices. For instance

(6.233)
V'ACBD =
3 (XACBD + XABCD + XADCB) ,
by definition. Hence
1
YACBD = 3 (XABCD + XACBD + XADBC) = VGABCD, (6.234)

where use has been made of XADCB = XADBC

The totally symmetric spinor ,bABCD has thus only five independent
complex components, as has been pointed out above. These components
6.4. THE GRAVITATIONAL FIELD SPINORS 155

are 1oooo, V)oooi, 1ooii, ' otii, and 01,11. These components are sometimes
denoted as follows:

V)o =' 0000, 01 =V1oool, 02 = y'ooil, 03 =V1oiii, 04 = '01111-

(6.235)
Using now the decomposition (6.228) for the gravitational spinor XABCD we
may then find the decomposition of the spinor equivalent to the Riemann
tensor given by Eq. (6.200). We obtain

RAB'CD'EF'GH' = -('YACEG EB'D' EF'H' +WACF'H' EB'D' EEG

+EAC B'D'EG EF'H' + EAC EEG 3;B'D'F'H')

- 6 [(EAEECG + EAGECE) EB'D' EF'H'

+EAC EEG (CB'F' ED'H' + EB'H' ED'F')]- (6.236)

Relation to the Weyl Tensor

We now show the relationship between the spinor 'ABCD and the spinor
equivalent to the Weyl conformal tensor Cap.ya. Let us denote the latter
spinor by CAB'CD'EF'CH' We show below that

CAB'CD'EF'CH' = - (1,ACEC EB'D' EF'H' + EAC EEG1bB'D'F'H') (6.237)

From Eq. (6.236) we see that the spinor (6.237) satisfies the same symme-
tries as those of the spinor equivalent to the Riemann curvature tensor. We
have to show, in addition, that the trace
cEF'CD'EF'GH' (6.238)
_ 0-
Indeed a direct calculation verifies that Eq. (6.238) is satisfied.
Hence the spinor IGABCD is equivalent to the Weyl conformal tensor and
is referred to in the sequel as the Weyl conformal spinor.
If C,,p.ya is the Weyl conformal tensor and *Cap.ya is its dual,

Coyd 1 --
= 2v
(
6:239)

then the spinor equivalent to the tensor

i*Cap,y,

C+ptia = Cap-Y& + (6.240)

156 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

is given by
CAB'CD'EF'GH' 2,bACEG EB'D' EF'H' (6.241)
The proof of the above formula is given in Problem 6.5.
As a consequence of the decomposition of the gravitational spinor 'OABCD
into the Weyl conformal spinor plus the Ricci scalar curvature, the curva-
ture spinor (6.174) may finally be written in the form
ll
FPQAB'CD' [PQAC + s (EPA EQC + EPC EQA)J EB'D' -OPQB' D' CAC.
(6.242)
Equation (6.242) describes the decomposition of the curvature spinor into
its irreducible components, namely, the Weyl spinor, the tracefree Ricci
spinor, and the Ricci scalar curvature. This is similar to the decomposi-
tion of the Riemann curvature tensor into its irreducible components (see
Subsection 6.1.8).
In analogy with the curvature spinor we may define the conformal spinor
by
1PPQ0 =''GPQABa C'apBC'. (6.243)
Under the conformal transformation gv = e2pgv, the matrices a trans-
form into & given by
&J, (x) = epa (x) , (6.244a)
& (x) = e-pa W, (6.244b)
We now find the transformed components of the conformal spinor " PQp
under the conformal transformation. From Eq. (6.241) we find that

,1PQAB = 8CPC'QC'AF'BF' (6.245)

Hence the conformal spinor, by Eq. (6.243), is given by

D'apBD'
-8CPC'QC'AF'BF'C
aaBF a D 0')3 ED' (6.246)
_ _8C+vKaaPC'av4 aKAF
Since C+vKa = C+vKa, and because the expression (6.246) includes an
equal number of terms of the matrices a having covariant and contravari-
ant spacetime tensor indices, we find that
GPQp = , PQp (6.247)
Therefore the Weyl spinor is invariant under the conformal transformation.
6.5. PROBLEMS 157

6.4.5 The Bianchi Identities

The Bianchi identities may be written in terms of the dual to the Riemann
tensor in the form given by
VA* Ra3-p=0.
(6.248)

The spinor equivalent to this equation is given by

VGH'*RAB'CD'EF'GH'
=0 (6.249)

Here the covariant differentiation operator VGH' is defined by Eq. (5.61).

Using the expression (6.201), for the spinor equivalent of the dual to the
Riemann curvature tensor, in Eq. (6.249) we obtain
V F' XACEG EB'D' - VE ' cACF'H' EB'D' + VG
F' c'B'D'EG
H' (6.250)
-EACVE XB'D'F'H' = 0.
Multiplying the above equation by EB'D' then gives
V F' XACEG - VEH' 'ACF'H' = 0. (6.251)
Equations (6.251) are the Bianchi identities in spinor calculus.
In the next chapter the spinors of gauge fields are discussed.

6.5 Problems
6.1 Find the expressions for the differential operators

V(AC) = (VAE'VC ' +VCE'VA ' ) , (1)

1

V(B'D') = (VEB,V D, +VED'V B,), (2)

2
when applied on an arbitrary unprimed one-index spinor Q.
Solution: By Eqs. (6.170) and (6.171) we have
(VCD'VAB' -VAB'VCD')eQ = FPQAB'CD'eP (3)
Using the decomposition of the commutator operator given in Problem 5.1,
and using the decomposition of the curvature spinor given by Eq. (6.174),
we then obtain
[ED'B'V (AC) + ECAV (B'D')] Q = - (XPQAC EB'D' + cPQB'D'EAC) C-
(4)
158 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

Multiplying now the latter equation by eD'B' and eCA we obtain

O(AC)CQ = XPQACCP, O(B'D')CQ = OPQB'D'CP, (5)

In the same way we obtain the corresponding results when the operators
(1) and (2) apply on a primed-index spinor ?7Q'. We obtain

O(AC)77Q' - 4'P'Q'ACry7P/ O(B'D')rnQ' = XP'Q'B'D'fl (6)

6.2 Show that
O(AB)(QDE'

= XPQAB(PDE'

+X PD
AB(QPE' +OE
P'ABCQDP', (1)
and
PDE' + PE' +
O(A'B')(Q DE' = rPQA'B'( PA'B'(Q XP E' A'B'(Q DP' (2)

Solution: Equations (1) and (2) are direct generalizations of the results of
Problem 6.1 and are left to the reader for verification.
6.3 Prove Eq. (6.193) for the decomposition of the Riemann curvature
tensor.
Solution: Equation (6.193) is a straightforward result of the application
of Eq. (5.70) and is left to the reader for verification.
6.4 Find the spinor equivalent to the dual of the Riemann curvature tensor.
Solution: The spinor equivalent to the dual of the Riemann curvature
tensor is defined by

*RAB'CD'EF'CH' = 0rAB'UCD'47EF'aCH' *Rapryb, (1)

where *R,py6 is the dual to the Riemann curvature tensor given by

*Rap.ya = 2/
1
(2)

The latter formula may also be written in the form

*Rap7a 1
= a'
(3)

and therefore in spinor notation in the form

*RAB'CD'EF'CH' = 2RAB'CD'KL'MN'EEF''GH'' (4)

6.5. PROBLEMS 159

Using now the expression for the spinor equivalent to the Riemann
curvature tensor given by Eq. (6.200) and the expression for the spinor
EEF'CH given by Eq. (5) of Problem 5.3 in the above formula, we then
obtain

*RAB'CD'EF'GH' = i(XACEG EB'D' EF'H' - OACF'H' EB'D' EEG

EF'H' - EAC EECXB'D'F'H') (5)

6.5 Find the spinor equivalent to the tensor

C+)37a = Ca.Arya + i*Caprya, (1)

where *C,,,p ya is the dual to the Weyl conformal tensor C','p.ya.

Solution: The spinor equivalent to the Weyl conformal tensor is given
by Eq. (6.237). The spinor equivalent to the dual of the Weyl tensor
may be obtained from that of the Riemann curvature tensor, given by Eq.
(5) of Problem 6.4, by replacing the spinor XABGD by 1,ABCD and taking
OABC'D' = 0,

*CAB'CD'EF'GH' = 2 ('YACEG EB'D' EF'H' - EAC EEGY'B'D'F'H') (2)

We consequently obtain

CAB'CD'EF'GH' _ -20ACEC EB'D' EF'H' (3)

6.6 Find the spinor equivalent to the tensor

R+P7a = Rap,6 + i *Rap yb. (1)

Solution: Using Eqs. (6.200) and (6.201) we obtain

RAB'CD'EF'GH' _ -2 (XACEG EB'D' EF'H' +EAC'B'D'EG EF'H') (2)

6.7 Find the expression for the spinor *FPQAB'CD', the dual to the spinor
FPQAB'CD', in terms of the Weyl conformal spinor, the tracefree Ricci
spinor, and the Ricci scalar curvature. Show that Eq. (6.190) is identical
to the Bianchi identities (6.251).
Solution: The spinor *FPQAB'CD' is defined by
1 KLMN'
*FPQAB'CD' = 2EAB'CD'FPQKL'MN' (1)
160 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

Using Eq. (2) of Problem 5.3 we then obtain

*FPQAB'CD' = iFPQAD'CB' = i (XPQAC EB'D' - OPQB D' CAC) (2)

Using the above result in Eq. (6.190), we then obtain

VCD'
*FPQAB'CD' _ _i (OCB'XPQAC - VAD'OPQB'D') = 0. (3)

Equation (3) is identical to the Bianchi identities (6.251).

6.8 Discuss the physical and geometrical meaning of the field equations
V"FpQ" = 41rJpQ`,
(1)

V"
*FpQ'" = 0,
(2)
where *FpQ'" is the dual to FpQ``" and JpQh represents the energy-
momentum tensor, as possible field equations for the theory of gravitation.
Solution: The solution is left for the reader.
6.9 Write the Einstein gravitational field equations in the presence of an
electromagnetic field using the spinor calculus.
Solution: The Einstein field equations in the presence of an electromag-
netic field have the form R" = (8.7rG/c4) T" since R = 0. The equivalent
equations, using spinor calculus, are given by
/, 4irC
WACB'D' = TAB'CD', (1)
C4

where OACB'D' is the tracefree Ricci spinor. Using now Eq. (6.217) and
Eq. (6) of Problem 5.10, we then obtain
2G (2)
mn = C4 0"01 "

for the Einstein field equations. Here m, n = 0, 1, 2.

6.10 Prove the transformation laws (6.15) and (6.16) of the Christoffel
symbols of the first and second kinds.
Solution: Using the transformation laws for the metric tensor leads to
Eqs. (6.15) and (6.16).
6.11 Prove Eq. (6.24).
6.5. PROBLEMS 161

Solution: The solution is left for the reader.

6.12 Show that the covariant derivatives of the tensors Tap, Tap and T
are given by
V-YTap= p - I,a-Y Tap,

& Tap
V.,T`p = + ra,rTap + I' T`6,

O-fT nciP + rayTj p - r0-Y a.

From this find the general rule for covariant differentiation.
Solution: The solution is left for the reader.
6.13 Show that the covariant differentiation of the sum, difference, outer
and inner products of tensors obeys the usual rules of ordinary differentia-
tion.
Solution: The solution is left for the reader.
6.14 Generalize Eq. (6.30) for a tensor T,,..
Solution: The solution is left for the reader.
6.15 If Tap is the curl of a covariant vector, show that

V ryTap + V aTpry + V pT7a = 0,

and that this is equivalent to

ayTap + aaTpry + apT7a = 0.

Solution: The solution is left for the reader.

6.16 Show that the divergence 0V of the vector V is given by
__ 1 a
DIA
a2 \vV J)

Also show that for a skew-symmetric tensor Fap the covariant divergence
is
ppFap = 1 a
V ga2p (Fap,f g) .
Solution: The solution is left for the reader.
162 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

6.17 Find the expression for the Riemann tensor From it prove
Eqs. (6.32).
Solution: The solution is left for the reader.
6.18 Show that a curve with a covariantly constant tangent vector is nec-
essarily geodesic.
Solution: Let the curve be denoted by xa = xa (s) and the tangent vector
by dxa/ds. If the tangent vector is covariantly constant, then
dxc' ()
ds 0' 1

or explicitly
a v

8x ds
+ r" ds = 0. (2)

Multiplying Eq. (2) by dx/ds and using the identity

dx 8 d
ds 8x = Ws-

gives
d2xa dx dx" _ 0.
+ r " ds (4)
ds2 ds
6.19 Discuss the constancy of the weak and gravitational coupling con-
stants.
Solution: The solution is left for the reader.
6.20 Use the geodesic equations, Eq. (6.43), to determine the force per unit
mass on a body at rest, and show that it is given by F` = -c2I oo where
i = 1, 2, 3. In the weak field approximation gia are very close to the Lorentz
metric, and for a time- independent metric F` = c2I'0io (c2/2) 8i9oo Show
that in the weak field case Eq. (6.50) reduces to the Poisson equation (6.49)
where goo 1 + 20/c2. From this show that the constant t in Eq. (6.50)
is given by is = 8.7rG/c4.
Solution: The solution is left for the reader.
6.21 Prove Eqs. (6.58) and (6.59).
Solution: The solution is left for the reader.
6.22 Derive the gravitational field equations (6.50) using the calculus of
variation by treating both g" and I'.p as independent variants, and ob-
tain thereby equations that determine both objects. Such a procedure is
6.6. REFERENCES FOR FURTHER READING 163

known as the Palatini formalism. The procedure is analogous to the one

employed in deriving the electromagnetic field equations from a variational
principle where both the field f" and the potential A are variants of an
action principle.
Solution: The solution is left for the reader.
6.23 Find the energy-momentum tensor T" for: (1) a system of neutral
particles of inertial mass M (function of time); (2) the electromagnetic field;
and (3) a scalar field 0. Show that they are given by:

(1) Ti"' = E MiAi'S (x - z),

(2) T" =
47r { 4g"fo rP - .fief" 1r

(3) T" = a0a"Y' - 2g" (DaOV - m202)

Solution: The solution is left for the reader.
6.24 Use the approximate metric (6.79) into the geodesic equation (6.43)
to show that the equation obtained is (6.80).
Solution: The solution is left for the reader.
6.25 Prove Eqs. (6.150) and (6.151).
Solution: The solution is left for the reader.

6.6 References for Further Reading

W.L. Bade and J. Jehle, An introduction to spinors, Revs. Mod. Phys. 25,
714-728 (1953). (Sections 6.2-6.4)
B. Bertotti, D. Brill and R. Krotkov, Experiments on gravitation, in: Grav-
itation: An Introduction to Current Research (L. Witten, Editor), (John
Wiley, New York, 1962). (Section 6.1)
G. Birkhoff, Relativity and Modern Physics (Harvard University Press,
Cambridge, Mass., 1923). (Section 6.1)
M. Carmeli, Group Theory and General Relativity (McGraw-Hill, New York,
1977).
M. Carmeli, Classical Fields: General Relativity and Gauge Theory (John
164 CHAPTER 6. THE GRAVITATIONAL FIELD SPINORS

Wiley, New York, 1982).

M. Carmeli, Equations of motion without infinite self-action terms in gen-
eral relativity, Phys. Rev. 140, B1441 (1965). (Section 6.1)
M. Carmeli, E. Leibowitz and N. Nissani, Gravitation: SL(2,C) Gauge The-
ory and Conservation Laws (World Scientific, Singapore, 1990). (Sections
6.2-6.4)
T.F. Cranshaw, S.P. Schiffer and A.B. Whitehead, Phys. Rev. Letters 4,
163 (1960). (Section 6.1)
R.H. Dicke, in: Relativity, Groups and Topology (C. DeWitt et al., Eds.)
(Gordon and Breach, New York, 1964). (Section 6.1)
A. Einstein, Ann. Phys. 49, 761 (1916); English translation in: The Prin-
ciple of Relativity (Dover, New York, 1923). (Section 6.1)
L.P. Eisenhart, Riemannian Geometry (Princeton University Press, New
Jersey, 1949). (Section 6.1)
L. Infeld and A. Schild, Rev. Mod. Phys 21, 408 (1949). (Section 6.1)
L. Infeld and B.L. van der Waerden, The wave equation of the electron in
the general relativity theory, Sb. preuss. Akad. Wiss., Phys.-mat. K1. 380
(1933). (Sections 6.2-6.4)
A. Papapetrou, Proc. R. Soc. London (A) 209, 248 (1951). (Section 6.1)
R. Penrose, A spinor approach to general relativity, Ann. Phys. (N. Y.) 10,
171-201 (1960). (Sections 6.2-6.4)
F.A.E. Pirani, Introduction to gravitational radiation theory, in: Lectures
on General Relativity (1964 Brandeis Summer School) (Prentice-Hall, En-
glewood Cliffs, New Jersey, 1965). (Section 6.1)
R.V. Pound and G.A. Rabka, Jr., Phys. Rev. Letters 4, 337 (1960). (Sec-
tion 6.1)
P.G. Roll, R. Krotkov and R.H. Dicke, Ann. Phys. (N. Y.) 26, 442 (1964).
(Section 6.1)
L.I. Schiff, Proc. Natl. Acad. Sci. 46, 871 (1960). (Section 6.1)
I.I. Shapiro, General Relativity and Gravitation 3, 135 (1972). (Section 6.1)
A. Trautman, Foundations and current problems of general relativity, in:
Lectures on General Relativity (1964 Brandeis Summer School) (Prentice-
6.6. REFERENCES FOR FURTHER READING 165

Hall, Englewood Cliffs, New Jersey, 1965). (Section 6.1)

0. Veblen and J. von Neumann, Geometry of Complex Dynamics (The
Institute for Advanced Study, Princeton, New Jersey, 1958). (Sections 6.2-
6.4)
J. Weber, Gravitational radiation experiments, in: Relativity (M. Carmeli,
S.I. Fickler and L. Witten, Eds.), (Plenum Press, New York, 1970). (Sec-
tion 6.1)
S. Weinberg, Gravitation and Cosmology (John Wiley, New York, 1973).
(Section 6.1)
Chapter 7
The Gauge Field Spinors
After discussing the electromagnetic and the gravitational fields in the last
two chapters, we now formulate the gauge field dynamical variables in a
spinorial form. Following a brief review of the Yang-Mills theory, the spinors
equivalent to the gauge potential and the gauge field strength are writ-
ten down. Then, in analogy to the electromagnetic field, the gauge field
strength spinor is decomposed. Likewise, the expression for the energy-
momentum tensor of the gauge field is given in a spinorial form. This is
subsequently followed by formulating the gauge field variables as spinors in
the interior spaces of both the groups SL(2,C) and SU(2). The chapter is
then concluded with formulating and analyzing the geometry of the gauge
field dynamical variables.

7.1 The Yang-Mills Theory

In this section a brief review of the Yang-Mills theory is given.

7.1.1 Gauge Invariance

In ordinary gauge invariance of a charged field which is described by a com-
plex wave function Vi, a change of gauge means a change of phase factor
Vi - 'i', Eli' = (exp ia) ii, a change that is devoid of any physical conse-
quences. Since 0 depends on spacetime points, the relative phase factor
of bi at two different points is completely arbitrary and a is, accordingly,
a function of spacetime. In other words, the arbitrariness in choosing the
phase factor is local in character.

167
168 CHAPTER 7. THE GAUGE FIELD SPINORS

To preserve invariance it is then necessary to counteract the variation of

the phase a with spacetime coordinates by introducing the electromagnetic
potentials A, (x) which change under a gauge transformation as
1 (9a
A-A+eax'
and to replace the derivative of 0 by a "covariant derivative" with the
combination (a, - ieA,) 0.

7.1.2 Isotopic Spin

An isotopic spin parameter was first introduced by Heisenberg in 1932 to
describe the two charge states, namely neutron and proton, of a nucleon.
The idea that the neutron and proton correspond to two states of the same
particle was suggested at the same time by the fact that their masses are
nearly equal, and the light stable even nuclei contain equal numbers of
them.
Later on it was pointed out that the p - p and n - p interactions are
approximately equal in the 1S state, and consequently it was assumed that
the equality holds also in the other states available to both the n - p and
p - p systems.
Under such an assumption one arrives at the concept of a total isotopic
spin which is conserved in nucleon-nucleon interactions. Experiments on
the energy levels of light nuclei strongly suggest that this assumption is
indeed correct. This implies that all strong interactions, such as the pion-
nucleon interaction, should also satisfy the same conservation law. This,
and the fact that there are three charge states for the pion can be coupled
to the nucleon field singly, lead to the conclusion that pions have isotopic
spin unity. A verification of this conclusion was found in experiments which
compare the differential cross-section of the process n + p -+ -7ro + d with
that of the perviously measured process p + p -4 ir+ + d.

7.1.3 Conservation of Isotopic Spin and Invariance

The conservation of isotopic spin is identical with the requirement that all
interactions be invariant under isotopic spin rotation, when electromagnetic
interactions are neglected. This means that the orientation of the isotopic
spin has no physical significance.
Differentiation between a neutron and a proton is then an arbitrary
process. This arbitrariness is subject to the limitation that once one chooses
7.1. THE YANG-MILLS THEORY 169

what to call a proton and what to call a neutron at one spacetime point,
one is then not free to make any other choices at other spacetime points.
It also seems not to be consistent with the localized field concept which
underlies the usual physical theories.

7.1.4 Isotopic Spin and Gauge Fields

The possibility of requiring that all interactions to be invariant under in-
dependent rotations of the isotopic spin at all spacetime points, so that the
relative orientation of the isotopic spin at two spacetime points becomes
physically meaningless, was accordingly explored by Yang and Mills. They
introduced isotopic gauge as an arbitrary way of choosing the orientation of
the isotopic spin axes at all spacetime points, in analogy with the electro-
magnetic gauge which represents an arbitrary way of choosing the complex
phase factor of a charged field at all spacetime points.
This suggests that all physical processes, which do not involve the elec-
tromagnetic field, be invariant under the isotopic gauge transformation
Vi - iii', i,b' = S_1W^1', where S represents a spacetime dependent isotopic
spin rotation which is a 2 x 2 unitary matrix with determinant unity, i.e.,
an element of the group SU(2) discussed in Chapter 1.
In an entirely similar manner to what is done in electromagnetics, Yang
and Mills introduced a potential B in the case of the isotopic gauge trans-
formation to counteract the dependence of the matrix S on the spacetime
coordinates.
Accordingly, and in analogy with the electromagnetic case, all deriva-
tives of the wave function 0 describing a field with isotopic spin z should
appear as "covariant derivatives" of the form (8. - iB) ii, where B. are
four 2 x 2 Hermitian matrices. The field equations satisfied by the twelve
independent components of the B potential, which is called the b potential,
and their interaction with any field having an isotopic spin, are fixed just
as in the electromagnetic case.

7.1.5 Isotopic Gauge Transformation

Under an isotopic gauge transformation, a two-component wave function '0
describing a field with isotopic spin z, transforms according to
10 = SVi' (7.2)
Invariance then requires that the covariant derivative expression transforms
as
S (8 - 2B) 1'1' = (e - ZB,,) Y'. (7.3)
170 CHAPTER 7. THE GAUGE FIELD SPINORS

When combined with Eq. (7.2), we obtain the isotopic gauge transforma-
tion of the 2 x 2 potential matrix B,.:

B' = S-1B AS+i5-101S. (7.4)

In analogy to the procedure of obtaining gauge invariant fields in the

electromagnetic case, Yang and Mills defined their field as

F., =a (7.5)

where the commutator is given by

[B, B.] = Bit B,, - B,,B (7.6)

Under the transformation (7.2) the 2 x 2 field matrix (7.5) transforms as

F,;,,, = S-1Fu,,S. (7.7)

Now Eq. (7.4) is valid for any S and its corresponding B,. Furthermore,
the matrix S-18S/8x appearing in Eq. (7.4) is a linear combination of the
isotopic spin "angular momentum" matrices T', i = 1, 2, 3, corresponding
to the isotopic spin on the field ' under consideration. Here T' = "ai,
where o' are the three Pauli spin matrices.
Accordingly, the matrix B,. itself must also contain a linear combination
of the matrices Ti; any part of B in addition to this, denote it by BA,
is a scalar or tensor combination of the T's, and must transform by the
homogeneous part of (7.4),

B = S-1B,S. (7.8)

Such a field is extraneous and was allowed by the very general form we
took for the B potential, but is irrelevant to the question of isotopic gauge.
Therefore, the relevant part of the B potential can be written as a linear
combination of the matrices T:

B = 2b N, T, (7.9)

where bold-face letters denote 3-component vectors in the isospin space.

The isospin-gauge covariant field matrices F, can also be expressed as
a linear combination of the T's. One obtains

-F = 2f T, (7.10)
7. 1. THE YANG-MILLS THEORY 171

where
f" =
8b - 8b" + 2b x b,,. (7.11)
8x" 8x
One notices that f transforms like a vector under an isotopic gauge trans-
formation. The corresponding transformation of b, is cumbersome. Under
infinitesimal isotopic gauge transformations,

S = 1 - 2iT 6w, (7.12)

however, one obtains

b=b-2bx6w+axw. (7.13)

7.1.6 Field Equations

In analogy to the electromagnetic case one can write down an isotopic gauge
invariant Lagrangian density:

L = -1TrF0"F" _ -4f" . f"`". (7.14)

One can also include a field with isotopic spin 1 to obtain the following
total Lagrangian density:

L = -8 TrF"F" - P1y (8 - iBa) i - mOo. (7.15)

The equations of motion obtained from the Lagrangian (7.15) are

8f"`"
-2(b" x f")+J= 0 , (7.16)
8x"

1y (a - 2iT b) V) + mV, = 0, (7.17)

where
J = 2iV'y T'. (7.18)

Continuity equation
Since the divergence of J does not vanish, one may define

= J - 2b" x f", (7.19)

172 CHAPTER 7. THE GAUGE FIELD SPINORS

which leads to the equation of continuity,

aJ
=0 (7 . 20)
ax
Equation (7.20) guarantees that thef0d3x,
total isotopic spin

T= (7.21)

Nonlinearity of the Field Equations

Equation (7.19) shows that the isotopic spin arises from both the spin-1
field J" and from the b, potential itself. This fact makes the field equations
for the B potential nonlinear, even in the absence of the spin-2 field. The
situation here is different from that of the electromagnetic case whose field
is chargeless, and hence satisfies linear equations.

7.2 Gauge Potential and Field Strength

The spinors equivalent to SU(2) gauge potential ba and gauge field strength
fav are complex functions, which are obtained from the potential and the
field strength in the same way that the comparable spinors are obtained in
the theory of electrodynamics (see Section 5.4). The gauge potential and
field strength are related by the equation
fav = avbau - abav + 9eabcbbbcve (7.22)
where g is a coupling constant and Cabc is the skew-symmetric tensor defined
by E123 = 1. In the above quantities the indices a, b, c = 1, 2, 3 are SU(2) la-
bels describing the inner space degrees of freedom, whereas , v = 0, 1, 2, 3,
are spacetime indices.

7.2.1 The Yang-Mills Spinor

The spinor equivalent to the gauge potential is given by
baAB' = aAFiB'baF&) (7.23)

whereas that equivalent to the gauge field strength is given by

W
faAB'CD' = o`AB'o'CD'fav (7.24)
7.2. GAUGE POTENTIAL AND FIELD STRENGTH 173

Since the potential ba,. is real, its spinor equivalent baAB' is Hermitian,
namely,
baAB' = baB'A (7.25)

Accordingly ba00' and ball' are real quantities, whereas baol' and balo' are
complex quantities, conjugate to each other,

ba10' = ba0'1 = ba01'. (7.26)

Decomposition of the Spinor Equivalent

In analogy to the decomposition given by Eq. (5.83) of the spinor equivalent
to the electromagnetic field tensor, the spinor equivalent to the gauge field
strength may be decomposed. We then obtain

.faAB'CD' = XaAC EB'D' + EACXaB'D', (7.27)

where
1 CB'D'faAB'CD' (7.28)
XaAC = XaCA =

Since the spinor XaAB is symmetric in its spinor indices A and B, it has
3 x 3 complex components: Xaoo, XaOl = XalO, and Xall, with a = 1,2,3.
These nine complex components are equivalent to the 18 real components
of the field strength fav

The Yang-Mills Spinor

The gauge field spinor XaAB will be referred to in the sequel as the Yang-
Mills spinor. Its role is analogous to the Maxwell spinor OAB in electro-
magnetics (see Section 5.4). We will also, sometimes, use the notation

XaO = XaOO

Xal = Xa01 = Xa10 (7.29)

Xa2 = Xall

in analogy to the Maxwell spinor.

We may also find the spinor equivalent to the tensor * faj,,,, where
1
*.fav = 2 V -JElhvP0' aPa, (7.30)
174 CHAPTER 7. THE GAUGE FIELD SPINORS

which is the dual to the gauge field strength fa,,. We then find
1 KL'MN'
*faAB'CD' = 2EAB'CD faKL'MN', (7.31)

or, using Eq. (2) of Problem 5.3,

*faAB'CD' = ifaAD'CB' = i (EACLB'D' - XaAC EB'D') (7.32)

The spinor equivalent to the tensor

f+po
a = fapa + i *fapa (7.33)

is consequently given by

faAB'CD' = faAB'CD' - faAD'cB' = 2XaACEB'D'r (7.34)

where use has been made of Eq. (7.27).

Likewise, we may find the spinor equivalent to the tensor
fapo = fapa - 2 *fapo (7.35)

We then obtain

faAB'CD' = faAB'CD' + faAD'CB' = 2EACXoB'D' (7.36)

Accordingly we have

faAB'CD' = 2 (faAB'CD' + faAB'CD') (7.37)

and
*fo,AB'CD' = ZfaAB'CD', (7.38)
for the duals of fAB'CD'

7.2.2 Energy-Momentum Spinor

The energy-momentum tensor of a gauge field is given by

Tv =
T7r
(49vfaapfaap - faafava) , (7.39)

and is in complete analogy to that of the electromagnetic field. It may also

be written in the form

Tv =
87r
(faafav a +* faafav a, (7.40)
7.2. GAUGE POTENTIAL AND FIELD STRENGTH 175

and is, as can easily be seen, traceless,

T'-'=0, (7.41)

just as the case in electrodynamics.

The spinor equivalent to the energy-momentum tensor of the gauge field
is then given by

TAB'CD' =
8 (faAB'EF'faCD'IF' + *faAB'EF' *faCD'IF) (7.42)

Using now the expressions for f and *f given by Eqs. (7.27) and (7.32) in
Eq. (7.42), we then obtain
1 _
TAB'CD' = rXaACXaB'D'- (7.43)

If we denote the above spinor by

Tmn =TA+C,B'+D', (7.44)

with m, n = 0, 1, 2, Eq. (7.43) will then have the form

Tmn = TXamXan-
1 (7.45)

This form for the energy-momentum tensor is in complete analogy to that

of the electromagnetic field (see Eq. (6) of Problem 5.10).

The Einstein Field Equations

The Einstein field equations with the above energy-momentum tensor have
the form
(7.46)
R ,AV= CT11-1
T
since the Ricci scalar curvature R = - (8irG/c4) T = 0 by Eq. (7.41).
The equivalent gravitational field equations, using the spinor notation, are
given by
4irG
OACB'D' = TAB'CD', (7.47)
C4

where use has been made of Eqs. (6.222) and (6.223). Using now Eqs.
(6.217) and (7.40), we then obtain for the Einstein field equations in the

O(7.48)
presence of a gauge field the following:
2G
mn = C`1 XnmXan-
176 CHAPTER 7. THE GAUGE FIELD SPINORS

Here m, n = 0, 1, 2, and a = 1, 2, 3. We notice that Eq. (7.48) is in complete

analogy to Eq. (2) of Problem 6.9 for the case of the Einstein equations in
the presence of an electromagnetic field.

7.2.3 SU(2) Spinors

So far we have described the gauge field and potential in terms of SL(2,C)
spinors, leaving the SU(2) inner space degree of freedom indices unchanged.
We now develop an SU(2) spinor calculus to take care of that degree of free-
dom. The Yang-Mills spinor XaAB, for instance, will thus be described as a
quantity having two SL(2,C) spinor indices and two SU(2) spinor indices,
XMNAB, where M, N = 0, 1 also.
The relationship between isospinors and isovectors is as follows. An
isospin-2 object is described by an SU(2) spinor with one index. An exam-
ple of this is the proton and the neutron which are described collectively by
the spinor Om (it is, in addition, a four-component Dirac spinor in configu-
ration space). An isospin-1 object is described by a two-index SU(2) spinor
which is symmetric in the two indices. An isospin-T object is described
by a totally symmetric spinor having 2T indices. It therefore has 2T + 1
independent components.
The correspondence between isovectors and isospinors is achieved by
means of the Pauli spin matrices. The spinor equivalent to the vector Sa is
given by
C MN = aM ca, (7.49)
whereas the isovector equivalent to the spinor CM is given by
C.
= MNCNM. (7.50)
Here aaM are the usual three Pauli matrices divided by f :
N_ 1 0 1 N_ 1 O i
1M 72= ( 1 0 '2M 72 -i 0

(7.51)
a3M ( 0 1 )
7.2.4 Spinor Indices
The SU(2) spinor indices for the Pauli matrices are chosen in such a way
that the spinor equivalent to the isovector is symmetric when both indices
are upper or lower:
CMN = CM EPN = QaM EPNCa = QaMNCa = CNM, (7.52)
7.3. THE GEOMETRY OF GAUGE FIELDS 177

MN = EMPcPN = EMPQaPN',a = OraMNra

= NM.
ofSthe (7.53)
Equations (7.52) and (7.53) are the consequence symmetry of the
matrices cYaMN and a N. In fact we have

91MN =
1-1
(
0 1
)
, Or2MN =
1 r -i
O
Ol
i

1 0 11 (7.54)
0'3MN = 1 0J,
and
QMN = MN 0
-i
1
; 2 ( 0 0)
1
QZ
2 0

UM N
3
= i 0 -1 ), (7.55)
72= f -1 0

as compared to our previous presentation for the Pauli matrices in the

SL(2,C) spinor calculus given by Eqs. (5.46).
The Yang-Mills spinor XaAB, for instance, will be presented by the
mixed SU(2) and SL(2,C) spinor XMNAB It is symmetric with its two
kinds of indices, namely,

XMNAB = XNMAB, XMNAB = XMNBA (7.56)

We also notice that in the SU(2) spinor calculus there are no primed indices,
and that we raise and lower the indices with eMN and eMN just as for the
SL(2,C) spinor case.
In the next section we give the geometry of the Yang-Mills fields.

7.3 The Geometry of Gauge Fields

We now give to the gauge field dynamical variables a geometrical descrip-
tion.

7.3.1 Four-Index Tensor

Let fav be the gauge field strengths. Here u, v = 0, 1, 2,3 are the spacetime
indices, and a is the isospin index taking the values 1, 2, 3. From the field
strengths we define the four-index tensor

Rpvpa = -fapvfapo (7.57)

178 CHAPTER 7. THE GAUGE FIELD SPINORS

The tensor Rvpa which is an SU(2) invariant, satisfies the symmetry prop-
erties
Rvpo = -Rv,cpa = -Rpvap = +Rpav (7.58)
Hence the tensor R,,,upa is skew-symmetric in each pair of the indices v
and pa, and is symmetric under the exchange of these two pairs of indices
with each other. These symmetry properties are the same as those of the
Riemann curvature tensor, or the Weyl conformal tensor, known from the
geometry of curved spacetime (see the book of Eisenhart).
It will also be useful to define another tensor Rvpo,, which is also an
SU(2) gauge invariant, by
Rvpo = -fav *fapo, (7.59)

Here * fapo, is the dual to the tensor fapo,

*fapo, = 2V gepv{+ufav. (7.60)

The tensor R*,,,po has the same symmetry properties as those of Rvpo,
namely
* * *
R vpo = R-vap =+Rp * aJv
(7.61)
From the two tensors Rp,,P, and Rvpo, we may then define the complex
tensor
Rvaa = Rvpo + 2Rvpa = favfapv
(7.62)
where
fapo -fapo + 2 fapo (7.63)

The new tensor Rvpo also satisfies the same symmetry properties of Rvpo
and Rvpo,
Rvpo = -Rvpa = -Rvap = +Rpav (7.64)

From the tensor Rvpo, we may define the Ricci tensor Rap = Raaap
and the Ricci scalar curvature R = R.
Since the tensor Rap.ya has the same symmetry properties (except for
the cyclic identity) as those of the Riemann curvature tensor, we may de-
compose it as follows:

Rpav = CPav + 2 ( 9pRav - 9PLRak - 9aRpv + 90vRPA)

1
+1 (9pv9a - 9p90-v) R, (7.65)
7.3. THE GEOMETRY OF GAUGE FIELDS 179

or in the alternative, but equivalent, form

RPov = CPov + (gpwsav - 9pvso - 90-4v + 9avsp l

2

1
(7.66)
12 (9Pt'9a - 9p9ov) R.
Here Sv is the tracefree Ricci tensor,

Sv = Rv - 49v
1 R, (7.67)

which satisfies S = 0.
Contracting now either Eq. (7.65) or Eq. (7.66) with respect to the
indices p and , we find that the trace of the tensor CPQ vanishes, CP,,PP =
0. Hence Eqs. (7.65) and (7.66) express the fact that the tensor R,,,p.ya
decomposes into its irreducible components.

7.3.2 Spinor Formulation

The above results may easily be put into the spinor language. The spinor
equivalent to the Yang-Mills field strength fav is given by (see Section 7.2)

faAB'CD' = AB'0vCD'fave (7.68)

faAB'CD' = EACXaB'D' + XaAC EB'D', (7.69)

where XaAC = ZB'D'faAB'CD' The spinor equivalent to the tensor *f,,,,
the dual to fa,,, is given by

*faAB'CD' = i (EACXaB'D' - XaAC EB'D') (7.70)

Subsequently the spinors equivalent to the tensors and R,pQ

may be found. So may the spinor equivalent to f :p,,, which, by Eq. (7.34),
is given by
f AB'CD' = 2XoAC EB'D' (7.71)
As a result, the spinor equivalent to the tensor is given by

RAB'CD'EF'GH' = -faAB'CD'faEF'GH' (7.72)

Using now Eq. (7.69), we then obtain

RAB'CD'EF'GH' = -(CACEG EB'D' EF'H' +EEG(ACF'H' EB'D'

180 CHAPTER 7. THE GAUGE FIELD SPINORS

+EACCB'D'EG EF'H' + EAC EEG(B'D'F'H') (7.73)

In Eq. (7.73) the two spinors (ABCD and (ABC'D' are defined by

eABCD = XaABXaCD, (7.74)

and
(ABC'D' = XaABXaC'D', (7.75)
respectively.
From the definition of the spinor (ABCD we see that it satisfies the
following symmetry properties:

(ABCD = (BACD = (ABDC = eCDAB (7.76)

Hence it can be decomposed into the sum of a totally symmetric spinor

77ABCD and a scalar P,
P
(ABCD = 7IABCD + (EAC EBD + CAD EBC) (7.77)
s
Here the scalar P is the trace of the spinor (ABCD,
AB 1
P = (AB = 4fav (faL + Z* fav) . (7.78)

A simple calculation, moreover, shows that

(ACCB
P (7.79)
= 2
EAB

7.3.3 Comparison with the Gravitational Field

The spinor (ABCD resembles in its properties the gravitational field spinor
XABCD, which combines the Weyl conformal spinor and the Ricci scalar
curvature (see Chapter 6). The difference between the two spinors is only in
their trace structure, the trace of the gravitational field spinor is XABAB =
-R/4, where R is the Ricci scalar curvature, which is a real quantity.
Here, however, the invariant P is a complex function. The role of P in
gauge fields, nevertheless, seems to be similar to that of the cosmological
constant of general relativity theory.
The spinor 77ABCD in Eq. (7.77), on the other hand, is a totally sym-
metrical spinor in all of its four indices and is given by
1
77ABCD = ((ABCD + (ACBD + (ADBC) (7.80)
3
7.3. THE GEOMETRY OF GAUGE FIELDS 181

It is therefore completely analogous to the Weyl conformal spinor, and

has only five independent complex components: 170 = 'goooo, 77i = rlooot,
772 = 770011, 773 = 770111, 774 = 771111
The other spinor (ABC'D' appearing in Eq. (7.73), defined by Eq.
(7.75), satisfies the same symmetries that the tracefree Ricci spinor OABC'D'
satisfies, namely,

(ABC'D' = (BAC'D' _ (ABD'C' = (C'D'AB (7.81)

It therefore has nine real independent components. The spinor (ABC'D'

is, moreover, irreducible. Its physical meaning lies in the fact that it is
proportional to the energy-momentum tensor of the Yang-Mills field (see
details in Section 7.2).
From the spinor RAB'CD'EF'GH' given by Eq. (7.73) we may define the
Ricci spinor RCD'GH' = REF'CD'EF'GH' We then find that

RCD'GH' = 2(CGD'H' - 2 (P + P) ECG ED'H' (7.82)

We also find for the Ricci scalar curvature

R=R GHGH' = -2 (P + P) . (7.83)

We then find the following expressions:

1
CAB'CD' = RAB'CD' - 2EAC EB'D'R = 2(ACB'D' + (P+ P) EAC EB'D',
2
(7.84)

SAB'CD' = RAB'CD' - 4REAC

1 EB'D' = 2(ACB'D', (7.85)

for the Einstein spinor and the tracefree Ricci spinor, respectively.
We now find the spinor equivalent to the tensor Ra,-,, defined by Eq.
(7.59). It is given by

RaAB'CD'EF'GH' = -.foAB'CD' *.faEF'GH' (7.86)

Using Eqs. (7.69) and (7.70) we obtain

RaAB'CD'EF'GH' = Z((ACEGEB'D' EF'H' - EEG(ACF'H' EB'D'

+EAC(B'D'EG EF'H' -EAC EEG(B'D'F'H'). (7.87)

182 CHAPTER 7. THE GAUGE FIELD SPINORS

7.3.4 Ricci and Einstein Spinors

The Ricci spinor, Ricci scalar curvature, Einstein spinor, and tracefree Ricci
spinor are subsequently given by

RCD'GH' = 2
(P -' l ECG ED'H', (7.88)

R*=2i(P-P), (7.89)

GAB'CD' _ -2 (P - P) CAC EB'D', (7.90)

S'A* B'CD' 0, (7.91)

respectively.
Finally, the spinor equivalent to the complex tensor defined by
Eq. (7.62), is given by

RAB'CD'EF'CH' = RAB'CD'EF'GH' + iRAB'CD'EF'GH' (7.92)

We then find that

RAB'CD'EF'GH' = -2 ((ACED EB'D' +EAC(B'D'EG) EF'H' (7.93)

The Ricci spinor, Ricci scalar curvature, Einstein spinor, and tracefree Ricci
spinor are then given by

RCD'GH' = 2(CGD'H' - PECC ED'H', (7.94)

1Z = -4P, (7.95)

GCD'GH' = 2(CGD'H'' + PECC ED'H', (7.96)

SCD'GH' = 2(CGD'H', (7.97)

respectively.
A fourth spinor which can be constructed out of the Yang-Mills spinor
is given by
XABCDEF = EabcXaABXbCDXcEF (7.98)

It satisfies the following symmetry:

XABCDEF = XBACDEF = XABCCEF = XABCDFE (7.99)

In addition, the spinor XABCDEF keeps or changes its sign, depending
upon whether the pairs of indices AB, CD, EF are an even or an odd
7.4. REFERENCES FOR FURTHER READING 183

permutation of the pairs of numbers 00, 01(=10), 11, and zero otherwise.
Hence it can be decomposed as follows:

XABCDEF =
Q (CAC CBE EDF+CAF EBC CDE+EAC EBF EDE+EAE EBC EDF
24

+EAD EBF CCE + CAD CBE ECF + CAF EBD ECE + CAE 6BD CCF), (7.100)
where Q is a complex quantity, the trace of the spinor XABCDEF:
Q = XACCEEA = CCBCED6 (7.101)
XABCDEF

Finally, two more mixed-indices spinors, with unprimed and primed

indices, can be defined as follows:

'bABCDE'F, = EabcXaABXbCDXcE'F', (7.102)

OABC'D'E'F' = CabcXaABXbC'D'XcE'F' (7.103)

The relationship between them can easily be shown to be given by

'ABCDE'F' _ E'F'ABCD' (7.104)

0ABC'D'E'F' _ C'D'E'F'AB' (7.105)

7.4 References for Further Reading

M. Carmeli, Classification of classical Yang-Mills fields, in: Differential Ge-
ometrical Methods in Mathematical Physics, K. Bleuler, H.R. Petry and A.
Reetz, Editors, 105-149 (Springer-Verlag, Heidelberg, 1978). (Sections 7.2,
7.3)
M. Carmeli, Classical Fields: General Relativity and Gauge Theory (John
Wiley, 1982).
M. Carmeli and M. Fischler, Classification of SU(2) gauge fields: Lorentz-
invariant versus gauge-invariant schemes, Phys. Rev. D 19, 3653-3659
(1972). (Sections 7.2, 7.3)
L.P. Eisenhart, Riemannian Geometry (Princeton University Press, New
Jersey, 1949). (Section 7.3)
R. Jackiw, C. Nohl and C. Rebbi, Conformal properties of pseudoparticle
configurations, Phys. Rev. D 15, 1642-1646 (1977). (Section 7.2)
184 CHAPTER 7. THE GAUGE FIELD SPINORS

R. Jackiw and C. Rebbi, Conformal properties of a Yang-Mills pseudopar-

ticle, Phys. Rev. D 14, 517-523 (1976). (Section 7.2)
R. Jackiw and C. Rebbi, Spinor analysis of Yang-Mills theory, Phys. Rev.
D 16, 1052-1060 (1977). (Section 7.2)
C.N. Yang and R.L. Mills, Conservation of isotopic spin and isotopic gauge
invariance, Phys. Rev. 96, 191-195 (1954). (Section 7.1)
Chapter 8

The Euclidean Gauge

Field Spinors
In the last chapter we discussed the gauge field spinors in the flat Minkowskian
spacetime. In this chapter we extend the discussion to the Euclidean gauge
field spinors. First we discuss the Euclidean spacetime in general terms.
This is followed by writing down the Dirac equation in this spacetime, and
discussing the matrices involved in this formalism. The spinor formula-
tion of the Euclidean gauge fields is subsequently given. This includes the
0(4) two-component spinors. The chapter ends with the discussion of the
self-dual and anti-self-dual fields appearing in the theory.

8.1 Euclidean Spacetime

We are now in a position to formulate the Euclidean gauge field theory for
isospin in terms of quantities which are multispinors of the group product
0(4) xSU(2), where 0(4) is the four-dimensional rotation group. The group
0(4) replaces here the group SL(2,C) employed in the previous chapters
for gauge fields and gravitation. Hence instead of dealing with quantities
defined in the Minkowskian or the Riemannian spacetimes, as has been
done so far, we will be dealing with quantities defined in the Euclidean
four-dimensional spacetime.

185
186 CHAPTER 8. THE EUCLIDEAN GAUGE FIELD SPINORS

The Groups 0(4) and SU(2)xSU(2)

We will also use the fact that the group 0(4) may be written as the product
of two SU(2) groups, 0(4)=SU(2)xSU(2). Hence in the spinorial formula-
tion described below the 0(4) quantities, with which we are concerned in the
four-dimensional Euclidean spacetime, may be designated by SU(2) xSU(2)
representation labels. Moreover, the additional internal isospin space SU(2)
gauge group also gives rise to such SU(2) labels. Accordingly all quantities
of interest in the four-dimensional Euclidean spacetime with SU(2) inter-
nal symmetry are actually three SU(2) multispinor quantities. A certain
simplification is achieved sometimes when the various SU(2) groups are
coupled to each other.
The spinorial method makes use of 2 x 2 matrices to be described below
which are Euclidean analogues of the 2 x 2 SL(2,C) matrices encountered
in the study of the Lorentz group. We subsequently present the spinorial
formulation of the SU(2) gauge theory. But we first present the Dirac
equation in the Euclidean spacetime.

8.1.1 The Euclidean Dirac Equation

Our starting point is the four-dimensional Euclidean spacetime gauge co-
variant Dirac equation
/' 0. (8.1)
Y'1 ='y (8 - i9B{a)

Here ?'l is a gauge covariant derivative of the spinor Vi. Under the in-
finitesimal SU(2) transformation with generators 9a, where a = 1, 2, 3, the
spinor 0 transforms according to some representation of the group SU(2),
namely,
5?i) = iTallbOa. (8.2)

The matrices Ta describe the infinitesimal generators of the group SU(2)

and satisfy
[Ta, Tb] = ZEabcTc, (8.3)

with a, b, c = 1, 2, 3. In Eq. (8.1) B. is the Yang-Mills gauge potential in a

Hermitian matrix representation given by

B = baTa. (8.4)
8.1. EUCLIDEAN SPACETIME 187

Gauge Potential and Gauge Field

We assume that from the gauge potential B. we can define a well-behaved
gauge field strength matrix

Fv=avB,4-a3B"+i[B,,By], (8.5)

so that the action integral is finite. This requirement then implies that the
Pontrjagin index, defined by

q 32;x2 f *fav ff'd4x,

is an integer. Here *fa{cv is the dual to fav, and fav is the gauge field
strength. Also, in Eq. (8.5) and the rest of this chapter the coupling
constant g is taken as unity.
Now 0 is also a four-component spinor in the Euclidean space. The
4 x 4 Dirac matrices y satisfy the Euclidean anticommutation relations
{'yr`, y"} = 2b". (8.7)

The metric here is 5N," and the signature is (+, +, +, +) so that there is no
distinction between upper and lower space indices.

The y Matrices
A convenient realization of the y matrices is given by

10 -ick
0 Ya= 10
0 I
(8.8)
'Yk= k 2 k
,

where ak (k = 1, 2, 3) are the three Pauli matrices given by Eq. (7.51) and
I is the unit 2 x 2 matrix. With the above realization for the matrices y,
we define the matrix y5 by

'Y5 = 'Y1727374 =
)
The matrix y5 consequently anticommutates with the matrices

{7,'Y5} = 0, (8.10)

with = 1, 2, 3, 4. As a result, y5 also anticommutates with the Hermitian

Dirac differential operator iy (a,, - iB,).
188 CHAPTER 8. THE EUCLIDEAN GAUGE FIELD SPINORS

Hence if we consider the full eigenvalue spectrum through the equation

of motion
i-y' (a - iB)OE = E1,bE, (8.11)
then the matrix ry5 transforms the spinor OE into the spinor ?,b_E, namely,
i.y ((9 - iB)'-E = -EO-E, (8.12)
with ,b_E = 750E. On the other hand, the zero-eigenvalue modes may be
chosen as the eigenstates of ry5, and they have either positive or negative
chirality.

8.1.2 Algebra of the Matrices s

Equations (8.8) show that the -y matrices may also be presented in the form
S
y= S
O (8.13)

where the matrices s and st are defined by

S = (-iak, I) , (8.14)

St = = (iQk, I), (8.15)

These 2 x 2 matrices will be used in the sequel, and some of their properties
are given in the following.
We first have
s s, + st s = 8,,, (8.16a)

S'US, + b,,. (8.16b)

Hermitian Spin Matrices

We then define the Hermitian spin matrices

sv =
2i
(St S - st s) , (8.17a)

s 1 (S's - (8.17b)

which satisfy
2st S. = bv + 2is,,, (8.18a)
8.1. EUCLIDEAN SPACETIME 189

2s, sy = b + 2is,,. (8.18b)

One then finds that in terms of the Pauli matrices we have

Sij = Sij = -2fijkak, (8.19a)

Si4 = -Si4 = `Z ai,

(8.19b)

with i, j, k = 1, 2, 3 and 6123 = +1. Moreover, under the duality transfor-

mation we have
*St
*Sv = -S AV, ,, = Sv, (8.20)

where
Sv = 12EvpQS po'

*t =12fvpQS tpQ
SAL

Commutation Relations
The spin matrices satisfy the 0(4) commutation relations given by

2[Sa,SuA = bvaSp - bvSap + aapSv - bpSvaI (8.21a)

i [s ,s,p] = bvasp - bapsv - bNpS (8.21b)

Moreover, the products of three s matrices are equal to linear combination

of s matrices,

S Sa = (StSv)t = 1 (bvaS - baSv + fva,6S'6)I (8.22a)

2i

Svsa = (sasv)t = (5Y-s - baSv - fvapStp). (8.22b)

2i

Finally, the following identities

aksvak = -s,, (no summation on k), (8.23a)

akSvak = -S (no summation on k), (8.23b)

SakS = 0, (8.23c)

between the Pauli matrices and the s matrices, may be verified.

190 CHAPTER 8. THE EUCLIDEAN GAUGE FIELD SPINORS

The eigenvalue equation (8.11) may be written in terms of the s matrices

also. It then has the form

. (8.24)
Lt 0 V) B_ E \ VE /
This is a two-component spinor form which exhibits the chiral structure.
Here the operators L and Lt are defined by

L = s (2a + B) , (8.25a)
Lt = st (ia. + BN,) , (8.25b)

The zero-eigenvalue modes then satisfy the equations

LVi- = 0, Ltii+ = 0. (8.26)

Here 7(i+ and 0- are two-component spinors which also carry an isospin
label according to the representation (8.3).

8.2 The Euclidean Gauge Field Spinors

We are now in a position to present the spinorial formulation of the Eu-
clidean Yang-Mills theory with the internal gauge group SU(2). Accord-
ingly all quantities will have spinor indices A, B, C, taking the values 0,1.
As has been mentioned above, these two-component spinors represent the
two SU(2) groups in terms of which the Euclidean group 0(4) is presented
according to 0(4)=SU(2)xSU(2).

8.2.1 0(4) Two-Component Spinors

An 0(4) two-component spinor is accordingly denoted by (A when having
one superscript index or by (A when having one subscript index. The spinor
equivalent to an 0(4) tensor is obtained like in the case of the group SL(2,C)
(see Section 5.2). To each tensorial 0(4) index , v, , there correspond
two spinorial indices which are now denoted by AA', BB', .. The spinor
equivalent to the vector VN, is denoted by VAN and that equivalent to the
tensor T,,, is denoted by TAA'BB', for instance. The tensors and spinors
are related by means of the s matrices. The vector V. and the tensor T,,
on the other hand, may be recovered from the spinors VAA, and TAA'BB'
by using the properties of the s matrices.
8.2. THE EUCLIDEAN GAUGE FIELD SPINORS 191

Matrix Elements
We denote the matrix elements of the matrix s by

s)... =sAA'. (8.27)

We also denote the matrix elements of the matrix st by

(st)
...
- stA'A (8.28)

Accordingly we have

1 1 0 -i
( -i
sAA' = o (8.29a)

2
sAA'
__ 1 ( D 1
(8.29b)
1\ -1 0

g
sAA,
1 -i0 0
i (8.29c)
V`
l rl
sAA'
9
7I\0 1
1
J,
(8.29d)

and
$tA'A = 1 0
i (8.30a)
( z

0
S2 (8.30b)
1

S3
= 1 i
(8.30c)
3
72(0
$tA'A = 1 1
4 (8.30d)
VL (
Superscript and Subscript Indices
= the subscript 0(4)
There is no distinction between the superscript and
tensorial indices, namely, (s) = (s) and (silt) (s) . The spinor
equivalent to the vector V. is thus given by

VAA' AA' V
192 CHAPTER 8. THE EUCLIDEAN GAUGE FIELD SPINORS

1 ( -iV3+V4 -iV1+V2 (8.31)

I\ -iV1 - V2 iV3 + V4

The spinorial indices A, A', may be raised or lowered according to the

ordinary rules for the SL(2,C) spinors, given in Subsection 5.2.2, namely,
CA = EABCB, (8.32)
CA = CBEBA-

Hence if we define the spinors

SAA' = EE
AB A'B' SBB i , (8.33)

tB'B 6B'A'EBA, (8.34)

SAA = SI

we then find that

- tA'A _ ( l
AA'
S JA - (S /...
(8 . 35)

S AA = S A A ' = (S ) ... (8.36)

From Eqs. (8.16) we also obtain

AA' (8.37)
S SvAB' + SA 4' SAB' = bB' bv,

SAA'SBA + SvAA'SBA = ABbv, (8.38)

From the above formulas one then obtains the followirig:

SAA' SAA = av, (8.39)

SAL SBB' = aBaB', (8.40)

SAA'SBB, = EABEA,B'. (8.41)

Using now Eq. (8.40), for instance, we obtain

VAA'SA' = VaSAA'SA' = Va<SA = V, (8.42)

for the relationship between an 0(4) vector and its spinor equivalent.
8.2. THE EUCLIDEAN GAUGE FIELD SPINORS 193

8.2.2 Self-Dual and Anti-Self-Dual Fields

We may now write the spinors for the gauge potential b and the gauge
field strength f a,, in the Euclidean space. They are given, respectively, by

bMNAA' = baQaMNSg,A', (8.43)

and
fMNAA'BB' = fa aaMNSAA'SBB' (8.44)
where vaMN are given by Eqs. (7.54). The pair of indices MN are internal
SU(2) spinor indices, whereas AA' and BY are 0(4) spinor indices. Both
the gauge potential and the gauge field strength are symmetric in their
SU(2) spinor indices M and N.
Obviously the field strength spinor (8.44) is skew-symmetric in the pair
of indices AA' and BY. The above spinors are related by

fMNAA'BB' = aBB'bMNAA' - aAA'bMNBB' +2b(MPAA'b'N)BB', (8.45)

where brackets indicate symmetrization,
1
((AB) = (8.46)
- (CAB + (BA),
and the differential operator aAA' = SAA,a.
Now because of its antisymmetrical property, the gauge field strength
spinor can be split into two parts as follows:

fMNAA'BB' = CABfMNA'B' +fMNABEA'B', (8.47)

where
1
fM+NA'B' _ -JMNAA'AB'" (8.48a)

fMNAB = Z1 fMNAA'BA (8.48b)

Here fM+NA'B' is symmetric under the exchange of the indices A' and B'
since
1 1 1
A
fMNB'A' _ -fMNAB' A' _ -- JMNA'AB'_
A -JMNAA' B' = fMNA'B'-
(8.49)
Likewise, fMNAB is symmetric under the exchange of the indices A and B.
Finally, from Eq. (8.45) it follows that
A
fMNAB' = aA(A'bMN B,) + b(MPAA'b N)AB', (8.50a)
194 CHAPTER 8. THE EUCLIDEAN GAUGE FIELD SPINORS

A'+ b(MPAA' b N)BA' (8.50b)

fMNAB = a(AA, bMNB)
Furthermore, under the duality transformation we find

*fMNAA'BB' = JMNAB'BA', (8.51)

which in terms of f + and f- can be written as

*
fMNAB' = fMNAB" (8 . 52a)

*fMNAB = -fMNAB. (8.52b)

Hence for self-dual fields the expression fMNAB must vanish, whereas
for anti-self-dual fields the expression fMNAB' must vanish.

8.3 Problems
8.1 Verify Eqs. (8.16), (8.17) and (8.18).
Solution: These equations are direct consequences of the definition of the
s matrices and are left to the reader for verification.
8.2 Verify Eqs. (8.21) and (8.22).
Solution: Equations (8.21) and (8.22) are left to the reader for verification.

8.4 References for Further Reading

M.F. Atiyah, N.J. Hitchin and I.M. Singer, Proc. Nat. Acad. Sci. (USA)
74 (1977). (Section 8.1)
M.F. Atiyah, V. Patodi and L. Singer, Math. Proc. Camb. Philos. Soc.
77, 43 (1975); 78, 405 (1975); 79, 71 (1976). (Section 8.1)
L.S. Brown, R.D. Carlitz and C. Lee, Massless excitations in pseudoparticle
fields, Phys. Rev. D 16, 417-422 (1977). (Section 8.1)
M. Carmeli, Classical Fields: General Relativity and Gauge Theory (John
Wiley, 1982).
R. Jackiw, C. Nohl and C. Rebbi, Conformal properties of pseudoparticle
configurations, Phys. Rev. D 15, 1642-1646 (1977). (Section 8.1)
R. Jackiw and C. Rebbi, Conformal properties of a Yang-Mills pseudopar-
ticle, Phys. Rev. D 14, 517-523 (1976). (Section 8.1)
8.4. REFERENCES FOR FURTHER READING 195

R. Jackiw and C. Rebbi, Spinor analysis of Yang-Mills theory, Phys. Rev.

D 16, 1052-1060 (1977). (Sections 8.1, 8.2)
F.R. Ore, Jr., Quantum field theory about a Yang-Mills pseudoparticle,
Phys. Rev. D 15, 470-479 (1977). (Section 8.1)
Index
(Page numbers in italics refer to Bhattacharjee, R., 107
publications cited in the references.) Bianchi identities, 117-118, 135,
147, 157
contracted, 118, 121, 134, 147
tensorial, 147
Absolute rest, 36 Birkhoff, G., 127, 163
Acceleration, 118, 119, 139 Bjorken, J.D., 107
gravitational, 118 Bleuler, K., 183
radial, 126 Boener, H., 81
Action integral, 121, 187 Bohm, D., 62
Angles, Euler, 4-6 Boost, 41, 47
Born, M., 62
Angular momentum,
Brauer, R., 82
isotopic spin, 170
Brill, D., 163
per mass unit, 129
Brown, L.S., 194
Approach,
infinitesimal, 47-54
Carlitz, R.D., 194
Atiyah, M.F., 194
Carmeli, M., 10, 32, 62, 63, 82,
Atoms, 119
107, 135, 163, 164, 183,
frequency of, 127
194
frequency shift of, 127, 129
Cartan, E., 65, 82
gold and aluminium, 119 Cauchy-Buniakovsky inequality, 28
Automorphism, 3 Cauchy condition, 28
Charge, 99
Bade, W.L., 107, 163 density of, 85
Banach space, 27-29 states of,
operators in, 30 nucleon, 168
Barut, A.O., 62 Charged particles,
Basis, 12, 13, 17, 52 Lagrangian density of, 84
canonical, 16-17, 20, 23, 52 Chaudhury, T., 107
Bertotti, B., 163 Chevalley, C., 10, 82

197
198 INDEX

Chirality, 188, 190 Coset, 3

Christoffel symbols, 113-114, 117, left, 3
125, 135, 136 right, 3
Circumference, 123 Cosmic time, 41, 42, 43
Clocks, 38, 127 Cosmological constant, 121, 180
atomic, 128 Cosmology, 43
Cohen, J.M., 107 Coulomb field,
Collinson, C., 105 high frequency, 132
Condition, Cauchy, 28 Covariant derivatives, 87, 114-115,
Conditions, 132, 143, 144, 168, 169
unitary, 53-54 gauge, 186
Cone, operator of, 93
galaxy, 43-44 ordinary coordinate, 94, 143
symmetry axis of, 44 spin, 93-94
light, 44, 91 spinor, 92-95
negative, 60 Covariant differentiation, 86, 114-
positive, 60, 61 115
Constant, Cranshaw, T.F., 128, 164
coupling, 187 Cross-section,
Continuity equation, 172 differential, 168
Coordinate, 42 Current density,
Cartesian, 127, 131 vector, 85
spacetime, 94, 168, 169
spatial, 36, 38, 41, 44, 45 Davis, T.M., 105
spherical, 136, 138 Deflection of light, 130-132
time, 36, 37, 38, 41, 44 angle of, 132
translation of, 37 Dicke, R.H., 119, 164
Coordinate system, 111, 112, 114, Dirac, P.A.M., 82, 108
118, 120, 121, 124 Dirac delta function, 134, 137
Cartesian, 5, 93 Dirac equation, 98-99, 186
geodesic, 117, 121 Euclidean, 186-188
inertial, 36-37, 38, 39, 44 gauge covariant, 186
orientation of, 38 Direct sum, 52
origin of, 38, 44 Direction, 111
Correspondence, Distance, 43
between isovectors and isospinors, Earth-planet, 132
176 Earth-Sun, 133
between spinors and tensors, from the center, 123
88 infinitesimal, 136
Corson, E.M., 108 planet-Sun, 133
INDEX 199

Drell, D., 107 Electromagnetic potential, 95, 99

Electromagnetics, 169, 173
Earth, Electromagnetic spinor,
gravitation of, 118 decomposition of, 96-97
motion of, 36 Energy and momentum,
orbit radius of, 128 conservation of, 121
Schwarzschild radius of, 126 Energy-momentum tensor,
Eccentricity, 129 zero, 105
Eigenvalue equation, 190 Eotvos, 118, 119
Eigenvalue spectrum, 188 Equations, Einstein-Dirac, 105, 106,
Einstein, 35, 36, 63, 120, 133, 134, 107
139, 164 Equations of motion, 133-142, 171,
Einstein-Dirac equations, 105, 106, 188
107 Newtonian, 140-141
Einstein- Infeld- Hoffmann equation, Euclidean space, 13, 16, 27, 28,
141-142 29, 187, 193
Einstein-Infeld-Hoffmann method, n-dimensional, 15
138-140 Euler angles, 4-6, 8, 9, 14, 17, 21,
Einstein's field equations, 109, 120- 22, 27
121, 123, 124, 126, 127, Events, 44
132, 134, 135, 175 Examples of spinor representation,
deduction from variational prin- 71-73
ciple, 121-122 Experiment,
exact solution of, 123-127 Eotvos, 118-119
with electromagnetic field, 175- gravitational radiation, 132
176 low-temperature, 133
Einstein-Maxwell field equations, Michelson and Morley, 36, 38
coupled, 123 Null, 118-119
Einstein summation convention, 110 radar, 132-133
Einstein tensor,
the spinor equivalent to, 153 Factor group, 1, 3
Eisenhart, L.P., 10, 144, 164, 178, Fickler, S.I., 165
183 Field,
Electric field, 84 anti-self-dual, 193-194
Electrodynamics, 36, 172, 175 classical, 36
laws of, 36 Coulomb, 132
Electromagnetic field, 35, 36, 95, covariant isospin-gauge, 170
96, 122, 169, 174, 175 electromagnetic, 35, 36, 169
antisymmetry property of, 96 gauge invariant, 169
Lagrangian density for, 83, 84 Maxwell, 87
200 INDEX

self-dual, 193-194 distribution of, 43

spin-2,172 locations of, 43
Field equations, 86, 122, 134, 140, Galaxy cone, 43-44
169, 171-172 Galileo, 118
nonlinearity of, 172 Gauge,
Field strength, 172-177 electromagnetic, 169
gauge, 167, 172, 187, 193 isotopic, 169, 170
dual to, 173-174, 178 Gauge fields, 169, 174, 175, 176,
spinor equivalent to, 167, 180, 185, 187
172, 192 geometry of, 177-183
Yang-Mills, Gauge invariance, 167
spinor equivalent to, 179 Gauge theory,
Fierz, M., 108 SU(2),
Fischler, M., 183 spinorial formulation of, 186
Four-current density, 84 Gauss' theorem, 122
Four-vectors, 55 Gelfand, I.M., 32, 33, 53, 63, 82
Minkowskian space of, 55 General relativity,
Fourier transform, experimental tests of, 127-133
generalized, 79 original formulation of, 120
Frame, General relativity theory, 88, 119,
accelerated, 119 124, 132, 133, 134, 180
French, A.P., 63 classical, 120
Function, elements of, 109-143
continuous, 31, 50, 57 necessary condition of, 118
gradient of, 111 sufficient condition, 119
numerical, 29, 31 Generators,
continuous, 31 infinitesimal, 14-15
operator, 31, 50 commutation relations, 16
continuous, 31 Geodesic devitation equation, 132
scalar, 111 Geodesic equations, 116, 126, 128,
vector, 31, 50 130, 133, 134
Functional, Geodesic postulate, 133-134
bounded linear, 30, 31 Geodesics, 116-117, 128, 132
linear, 29 Geometry,
Functions, Riemannian, 110-118, 120
basis, 22 Ghost neutrinos, 105
Riemannian plane of, 60 Goldstein, H., 11
space of, 23 Graev, M.I., 32, 33, 63, 82
Gravitation, 120, 185
Galaxies, Einstein's constant of, 121
INDEX 201

field of, 119 infinitesimal matricies of, 48

Newtonian theory, 120, 121 irreducible representations,
Newton's constant of, 121, 126, 51, 52, 54
142 linear representation, 49
potential of, 120 Lorentz, 35-63, 70, 88, 186
presence of, 87 covering group of, 70
waves of, 132 homogeneous, 46, 56, 57
Gravitational field, 95, 119, 123, inhomogeneous, 46
127, 128, 130, 133, 134, orthochronous, 46, 47
147, 180-181 proper, 46, 47, 56, 57
clocks in, 127 proper orthochronous, 46,
curvature tensor, 95 51, 54, 57, 58, 59
deflection of light in, 130 subgroups of; 46-47
dynamic, 132 0(4), 185, 186, 190
Earth's, 128 0(4)xSU(2), 185
Einstein's equations of, 132 order of, 2
equations of, 109, 120-122, 125, Poincare, 46
134, 139, 175 pure rotation, 1, 4-6, 13
spinor equivalent to, 175 three-dimensional, 13, 51
external, 134 rotation, 1, 7, 48
in vacuum, 126 four-dimensional, 185
Lagrangian for, 121 simple, 3
slowing down of electromag- SL(2,C), 35-63, 65, 66, 77, 78,
netic waves, 132 83, 88, 93, 185, 186, 190
Sun's, 128 center of, 58, 59
weak, 121 inhomogeneous, 59
Gravitational red shift, 127-128 representation of, 75, 77, 78,
Green function, 139 93
Grommer, 134 space of representation, 66
Group, 1-2 spinor representation, 65-
abelian, 2 73, 77, 93
basic one-parameter, 15, 50 subgroups of, 59-60
center of, 58, 59 unity element of, 67
compact, 22 SL(2,C)/Z2i 59
factor, 3 SL(2,R), 59, 60
finite, 2 spinor representation, 19-20
Galilean, 37 SO(3), 4-6, 7, 8, 9, 51
infinite, 2 invariant integral over, 8
L, 46, 47, 51, 52, 53, 54, 56, representation of, 13-18, 23
57, 58, 59 SU(1,1), 59, 60
202 INDEX

SU(2), 6-7, 65, 78, 169, 190 129, 79-80

complexification of, 76 L2s(SU(2)), 78, 79, 80
generators of, 186 orthogonal sum of, 32
internal gauge, 190 Hitchin, N.J., 194
invariant integral, 8-9 Hoffmann, 139
irreducible representation of, Homomorphism, 1, 3-4
76, 79 between SO(3) and SU(2), 6-
operators of, 18 7, 13
parametrization of, 59 between the Lorentz group and
representation of, 13-18, 19- SL(2,C), 54, 88
20, 21, 76, 186 continuous, 56
SU(2) gauge, 186, 190 kernel of, 4, 58-59
SU(2)xSU(2), 186, 190 natural, 4
subgroup of, 3 of SL(2,C) on L, 56-58
translational T4, 59 Hubble's constant, 41
triangular matrices, 60 Hubble's time, 41
U(1)xT2, 60 Hyperboloid,
unimodular unitary, 13 single-sheeted, 60, 61
Group element, 1, 2 two-sheeted, 60, 61
Hermitian conjugate of, 55,
56 laglom, A.M., 53, 63
inverse of, 2 Identity, 2
left inverse of, 2 left, 2
negative of, 2 right, 2
right inverse of, 2 Inertia,
Group elements, 1, 2, 3, 23 law of, 36
addition of, 2 Infeld, L., 88, 108, 134, 139, 164
central, 58, 59 Infinite-dimensional spinors, 77-80
equivalent, 3 Interactions,
multiplication of, 2 electromagnetic, 168
product of, 2, 25 n - p, 168
Group theory, introduction to, 1- nucleon-nucleon, 168
10 pion-nucleon, 168
Gyroscope, p - p, 168
anomalous precession of, 133 strong, 168
Interval, 123
Intrinsic spin structure, 97-98
Harish-Chandra, 53, 63 Invariance,
Heisenberg, 168 Galilean, 38
Hilbert space, 29-30 Lorentz, 38
INDEX 203

Invariant, 111, 112 Length,

Invariant integral, 1, 8-9 element of, 112
over SO(3), 8 extremal,
over SU(2), 9, 14 curves of, 116
Inverse, 2 Levi-Civita metric spinors, 88, 89,
left, 2 92
right, 2 Levi-Civita symbol, 49
Isomorphism, 1, 3-4 L'Hospital's theorem, 137
between L and SL(2,C)/Z2, Line of nodes, 5
58-59 Linear operators, 11-12
natural, 4 addition of, 11
standard, 60 multiplication of, 11-12
Isospin, 185, 190 Light,
Isospinor, 176 deflection of, 127, 130-132
Isovector, 176 null propagation of, 42
spinor equivalent to, 176 propagation of, 42
propagation velocity, 36
velocity in empty space, 36
Jackiw, R., 183, 184, 194, 195
Light pulse, 44
Jackson, J.D., 108
Light signals, 37
Jacobian, 110, 112
Jehle, J., 107, 163
Magnetic field, 84
Malin, S., 10, 62, 82
Kahan, T., 108 Mapping, 3, 4
Kleinert, H., 62 between SL(2,C) and L, 58
Kronecker delta function, 112 between SU(1,1) and SL(2,R),
Krori, K.D., 107 60
Krotkov, R., 163, 164 Mass, 99, 131, 138, 139, 140, 168
distance from, 126
Lagrange equation, 117 gravitational, 119
Lagrangian, 117, 121, 122, 171 inertial, 119, 135
Lagrangian density, 83, 84, 86, 122, of the Sun, 128
171 spherical, 133
charged particles, 83, 84 with spherical symmetry, 124
electromagnetic field, 83 Matrices,
isotopic gauge invariant, 171 Di,21,22
total, 171 Dirac, 187
Laws of nature, 119 y, 187-188
Lee, C., 194 group of, 31
Leibowitz, E., 107, 164 Hermitian, 60, 61, 88, 89, 169
204 INDEX

space of, 55, 60 orthogonal, 5, 14

unimodular, 61 unit, 45, 55, 88, 90
infinitesimal, 15, 18, 48-49, 51, unitary, 7, 17, 21, 22, 169
73 Maxwell, 36
commutation relations of, Maxwell equations, 83, 84, 85, 87,
15, 16, 49, 51 98-99, 123
isospin-gauge covariant field, alternative form of, 85
170 generalization into curved space-
isotopic spin "angular momen- time, 86-87
tum", 170 in the presence of gravitation,
Lorentz, 47-48 87
infinitesimal, 47-49 spinor version of, 99
orthogonal, 4, 5, 6, 45 Maxwell field, 87
Pauli spin, 6, 18, 55, 56, 73, Maxwell spinor, 97
88, 90, 170, 176, 177, 187, Maxwell tensor, 86
189 Maxwell's theory, 36, 83-87
symmetry of, 176, 177 Measure, 8, 9
properties of D2, 21-22 Measuring rods, 38
real orthogonal, 4, 5 Metric, 115, 116, 123, 124, 125,
a, 89-90, 92 128, 187
covariant derivatives of, 92 components of, 126
SIA 1 conformally flat, 116
algebra of, 188 flat-space, 44, 123, 126, 127
elements of, 191 flat spacetime, 148
products of, 189 g, 90-91
SL(2,C), 186 geometrical, 120
spin, 188 isometric,
commutation relations of, operator, 31
189 Minkowskian, 90-91
unitary, 6, 7, 17, 21 Schwarzschild, 127
Matrix, spherical symmetry of, 123,
eigenvalues of, 58 126
'Y5, 187 Metric tensor, 88, 89, 112-116, 120,
eigenstates of, 188 139
Hermitian, 6, 55, 152 cofactor of, 112
Hermitian conjugate of, 7, 55, contravariant, 124
56 covariant, 124
A, 39-41, 55, 56, 57, 58 curved spacetime, 86
inverse, 40 expansion of, 139
normalized eigenvector, 58 flat spacetime, 90, 148
INDEX 205

geometrical, 88, 89, 120 Non-ghost neutrinos, 107

spinor equivalent to, 90, 148 Nonrelativistic mechanics, 129
Minkowskian, 90, 91 Norm, 27-28, 30, 50
transformation law for, 113 Normal subgroup, 1, 3
Miller, A.I., 63 Nucleon,
Mills, R.L., 169, 170, 183, 184 field of, 168
Minlos, R.A., 33 light, 168
Morris, P., 105 Null conditions, 130
Moses, 33 Null experiment, 118-119
Mossbauer effect, 128 Michelson-Morley, 38
Motions, 60, 61
Euclidean, 60 Operator, 13
transitive group of, 61 adjoint, 16, 32, 53-54
Lobachevskian, 60-61 angular momentum, 26-27
planet's, 133 basic infinitesimal, 16
rectilinear, 37 bounded, 30, 32
spinning bodies, 133 norm of, 30
translational, 37 continuous, 30
uniform, 37 covariant derivative, 93, 157
Multiplication, differential, 24-26
associative, 2 eigenvalue of, 52
group, 58, 60 normalized eigenvectors of, 52
group elements, 2, 3 Hermitian, 16, 32
Multispinor, 185, 186 Dirac, 187
infinitesimal, 25, 49-52, 53, 74-
Naimark, M.A., 10, 33, 53, 63, 82 75
von Neumann, J., 108, 165 inverse, 32
Neutrino equation, 98 isometric, 30
Neutrinos, linear, 13-14, 30, 31, 50
ghost, 105 projection, 32
non-ghost, 107 skew-symmetric, 16
Neutron, 168, 169, 176 spinor,
Newton, 118 matrix elements of, 75-76
Newtonian approximation, 139, 140 unitary, 13, 16, 32, 53
Newtonian equation, 138-139, 142 Operator function, 12
Newton's law of motion, 126 continuous, 31
Newton's laws of mechanics, 36, Operators,
38, 129 angular momentum, 26-27
Nissani, N., 107, 164 basic infinitesimal, 16, 50
Nohl, C., 183, 194 continuous, 12
206 INDEX

orthogonal sum of, 32 homogeneous, 19, 75, 76

Ore, F.R., Jr., 195 Jacobi, 21
Orthogonal sum, 32 operators on, 69, 71
"Orthogonality" condition, 39 product by a number, 66
space of, 19, 65-66, 72, 75, 77
Palatini formalism, 163 dimension of, 66
Papapetrou, A., 133, 164 Pontrjagin, L., 10, 22, 33
Parametrization, 9, 16 Pontrjagin index, 187
Particles, 71 Post-Newtonian approximation, 138-
motion of, 140, 141 140
system of, 134 Potential, 176
test, 128, 132, 133 B, 169-171
velocities of, 138 bj,, 169-171
Patodi, V., 194 electromagnetic, 95, 99, 168
Pauli, W., 108 spinor equivalent to, 95
Pauli matrices, 18 gauge, 172-177, 187, 193
Pauli spinor, 98 spinor equivalent to, 167,
Penrose, R., 108, 164 172, 192
Perihelion, 130 Yang-Mills, 186
orbit, 130 scalar, 85
advance of, 130 vector, 85
Periodicity conditions, 9 Pound, R.V., 128, 164
Petry, H.R., 183 Prerelativity, 38
Phase factor, 167-169 Principle of equivalence, 109, 118-
Pion, 168 119
three charge states of, 168 Principle of general covariance, 109,
Pirani, F.A.E., 164 119-120
Plane symmetry spacetime, 106 Principle of relativity, 36, 38
Planetary motion, general, 120
effects of general relativity, 127, special, 119, 120
128-130 Product, 2, 18, 19
Planets, Proton, 168, 169, 176
motion of, 128, 129, 130 Pure rotation group, 4-6
Poincare group, 46 Pyatetskii-Shapiro, 1.1., 32
Poisson equation, 121
Polynomials, 19, 69, 71, 72, 73 Quadratic differential form, 112
addition of, 66
coefficients of, 71, 72, 77, 79 Rabka, G.A., 128, 164
expansion coefficients of, 76 Radar pulses,
expansion into powers, 75 time delay of, 132-133
INDEX 207

Radius, of SU(2), 13, 14, 17, 18, 19,

Earth's orbit, 128 21
Schwarzschild, 126 irreducible, 79
Sun, 128 spinor, 19-20
Ray, unitary irreducible, 76
light, single-valued, 13, 21, 58
orbit of, 131, 132 space of, 12, 17, 52, 66, 71,
one-index spinor, 91 72, 79, 80
Ray, J.R., 105 spinor, 19-20, 20, 65, 68, 69,
Rebbi, C., 184, 194, 195 70, 71, 77
Rechenick, K.R., 107 dimension of, 68, 71
Redshift, 43 examples of, 71-73
Reetz, A., 183 operators of, 73-77
Reflection, 4 realization of, 66-68
Representation, 12, 13, 30-31 unitary, 13, 16, 31-32, 53, 78
weight of, 17, 19, 20, 23, 51
canonical basis for, 20
Representations,
continuous, 12, 13, 31
, double-valued, 13, 14
dimension of, 12
finite-dimensional, 12, 16, 77
double-valued, 13, 14, 21, 58
equivalent, 12
finite-dimensional, 11, 20, 50, infinite-dimensional, 27-32
52
equivalent, 58
of SL(2,C), 68, 77 matrix elements of, 18-22
of SU(2), 20 orthogonal sum of, 32
general definition, 30-31 principle series of, 54, 77-78,
infinite-dimensional, 11, 49, 50, 80
51, 52, 77 uniquely determined, 16
infinitesimal operator of, 50, unitary, 13
51 Representation theory, 11-33, 65
irreducible, 12, 13, 14, 17, 20, Ricci identity,
23, 30, 52, 54, 78 spinorial, 144-145
of L,
Ricci scalar, 115, 121
irreducible, 51, 52 Ricci scalar curvature, 123, 142,
of SL(2,C), 67, 75 152, 153, 156, 175, 178,
finite-dimensional, 68, 77 181, 182
spinor, 73 Ricci tensor, 115-116, 140, 142,
of SO(3), 16-21, 23 178
finite-dimensional, 16 tracefree, 142, 152, 153, 179
irreducible, 17, 23, 51, 52 spinor equivalent to, 153
uniquely determined, 16, 17 Roll, P.G., 164
208 INDEX

Rotation, 4, 5, 6, 7, 13, 14, 15, 20, complementary, 54

24, 25, 26, 45, 57, 60 convergent, 29
angle of, 13, 24, 25, 27 sum of, 32
differential operator, 22-27 principal, 54
direction of, 13, 27 Set,
infinitesimal, 22 base vectors, 52
infinitesimal matrices, 18 closed, 29
matrix of, 24, 47 bounded, 31
orthogonal, 37 closure of, 29
pure, 4 dense, 29
spatial coordinates, 41, 45 Shapiro, 1.1., 132, 133, 164
three-dimensional, 4, 41 Shapiro, Z.Ya., 33, 33
unitary matrix, 17-18 Signature, 187
Rotation group, 1, 4-6 Simultaneity, 37
Riihl, 63, 75, 82 Singer, I.M., 194
Singer, L., 194
Scalar function, Space, 11, 12
gradient of, 111 Banach, 27-29, 30, 31
Scalar potential, 85 norm in, 50
Scalar product, 13, 27, 29, 30, 53, operators in, 30
55, 78, 79 subspace of, 50
Schiff, L.I., 133, 164 Cartesian, 41
Schiffer, S.P., 128, 164 complete, 30
Schild, A., 134, 164 conformal, 116
Schwarzschild, 123 conjugate, 30, 31
Schwarzschild field, 123-127, 132, dimension of, 69
133 dual of distance and velocity,
Schwarzschild metric, 127 43
Schwarzschild radius, 126 Euclidean, 13, 16, 27, 28, 29,
Earth, 126 187, 193
electron, 126 n-dimensional, 15
Sun, 126 finite-dimensional, 12, 13, 31,
Schwarzschild solution, 109, 123- 68
127 flat, 123, 126, 127, 132
Self-action terms, 135-138 line element of, 123
Sequence, four-dimensional, 42, 110, 115
convergent in the norm, 28 4-dimensional flat, 41
fundamental, 28 Hilbert, 29-30, 32, 53, 78, 79
Series, lz', 79-80
absolutely convergent, 29 L22'(SU(2)), 78, 79, 80
INDEX 209

orthogonal sum of, 32 Riemannian, 185

infinite-dimensional, 31, 49 V4, 110, 111
inner, Special relativity, 35-44, 119, 120
degree of freedom, 172, 176 postulates of, 35-37
isospin, 170 principles of, 35
internal, 186 Special unitary group, 6-7
12, 28, 29 Speed of light, 36, 93, 124, 128,
linear, 11, 19, 27, 29, 30, 66, 130, 138
67, 68, 75 constancy of, 38
Lobachevskian, 61 in vacuum, 41, 99
m-dimensional, 31 invariance of, 44
Minkowskian, 55, 60 Spin,
non-Euclidean, 135 1/2, 98
normed, 27-29, 30, 31 isotopic, 168, 169
complete, 28, 29 half-integer, 14, 71, 88
polynomials, 65-66, 71 integer, 71, 88
dimension of, 69 intrinsic structure, 97-98
unit operator in, 66 isotopic, 168-169, 172
R, 27, 28, 29, 30, 31, 32 conservation and invariance
reflexive, 31 of, 168-169
Riemannian, 123 conservation law of, 168-169
three-dimensional, 137 orientation of, 169
Space inversion, 47 rotation of, 169
Spaces, total, 168, 172
conformal, 116 Spin affine connections, 94
Spacetime, 169 Spin covariant derivative, 93-94
curvature of, 146 spin form, 93
curved, 86, 119, 120 vectorial form, 93
geometry of, 178 Spinor affine connection, 92-93
Maxwell's equations in, 86- Spinor ray, 91
87 Spinors, 83, 88, 91, 95
spinors in, 88-91 complex conjugate of, 91, 97
Euclidean, 185-190 components of, 71
four-dimensional, 185, 186 conformal, 155, 156
flat, 93 covariant derivatives, 92-95
metric tensors of, 90 curvature, 143-146, 147, 156
functions of, 77, 93 dual to, 147
Minkowskian, 42, 185 symmetry of, 145-146
plane symmetry, 106 Dirac, 176
points of, 167, 169 Einstein, 181-182
210 INDEX

electromagnetic field, 95-99 0(4), 190-192

decomposition of, 96-97 180
electromagnetic potential, 95 Weyl, 152, 153-156
180, 181 conformal, 153, 155, 156,
energy-momentum, 174-176 180
field strength, 193 Yang-Mills, 172-174, 176, 177,
four-component, 98, 99, 176, 182
187 181
gauge field, 167-184 Standard isomorphism, 60
Euclidean, 185-195 Subgroup, 1-3
gravitational, 150-152, 155, 156 invariant, 3
gravitational field, 109-165, 180 normal, 1, 3
trace of, 180 one-parameter, 73
Hermitian, 91, 95 of L, 73
indices of, 88, 89, 176-177 of SL(2,C), 73
contraction of, 89 Subspace,
primed, 89, 91 closed, 30, 31, 32
raising and lowering, 88, 89 finite-dimensional, 30
unprimed, 91 invariant, 12, 30
infinite-dimensional, 77-80 null, 12
isovector equivalent, 176 Summation convention, 44
Levi-Civita metric, 88, 89, 92 Sun,
Maxwell, 97, 173 gravitational acceleration to-
mixed-indices, 183 wards, 119
one-index, 144 Superconductors,
order of, 70 properties of, 133
Pauli, 98
product of, 91, 144 Tensor density, 86, 112
OE, 188 weight of, 86, 112
Ricci, 152-153, 181, 182-183 Tensors, 88, 111-112
tracefree, 156, 181, 182 conformal, 95
Riemann, 146 contraction of, 112
set of, 91 contravariant, 112
SL(2,C), 176, 177, 192 covariant, 112
SU(2), 176, 177 curvature, 98, 143, 144, 145,
tensor equivalent to, 90, 91 146-147, 148-149, 150, 178
tracefree Ricci, 156 density of, 86
transformation law, 77 derivative of, 114
two-component, 65-82, 83, 88, Einstein, 95, 116, 125
93, 98, 190 spinor equivalent to, 153
INDEX 211

electromagnetic field, 84, 95, symmetry of, 150, 151

97, 145 spinor equivalent of, 89, 91
dual to, 84, 85, 97 skew-symmetric, 112
spinors equivalent of, 96, 145, symmetric, 112
148, 172 Tve
energy-momentum, 121, 122- spinor equivalent to, 190
123, 127, 134, 174, 175, tracefree Ricci, 142, 152, 179
181 spinor equivalent to, 153
density of, 134 Weyl, 116, 155
electromagnetic, 122-123, 174, Weyl conformal, 116, 142-143,
175 155-156, 178
gauge field, 174-175 spinor equivalent to, 155-
of Yang-Mills field, 181 156
trace of, 123 Time,
zero, 105 rate of change, 127
four-index, 177-179 Time reversal, 47
indices of, 88, 89 Transformation, 5, 18, 19
Levi-Civita, 97 conformal, 156
density, 97 coordinates, 110-111, 117, 119,
weight of, 97 124, 126, 127
Maxwell, 86 cosmological, 41-44
metric, 89, 112, 116, 120, 124, interpretation of, 43
139 duality, 189, 194
mixed, 112 g,,, 113
0(4), Galilean, 37-38,
indices of, 191-192 gauge, 168
spinor equivalent of, 190 isotopic, 169-171
Ricci, 104, 115-116, 140, 142, identity, 57
178-179 A, 39, 40, 55, 56, 57
Riemann, 115-116, 118, 142- linear, 4, 18, 19, 44, 80
143, 146, 149, 150, 151 Lorentz, 38-41, 42, 43, 45, 46,
Riemann curvature, 98, 143, 61, 172
144, 145, 146-147, 148, derivation of, 38-41
149, 150, 178 determinant of, 45
decomposition of, 142-143, four-dimensional, 45
148-149, 155 improper, 45, 46, 47
dual of, 149, 151, 157 inhomogeneous, 46
spinor equivalent of, 146, inverse, 40, 41
148, 149, 150, 151, 152, matrix of, 39, 40, 47
153, 155 orthochronous, 46
212 INDEX

proper, 45, 46, 47 Wave function,

Mobius, 66, 67 derivatives of, 169
image of, 67 two-components, 169
Translations, 46 Weber, J., 132, 165
origins of systems, 37 Weil, A., 10
SU(2), Weinberg, S., 165
time coordinate, 37 Weyl, H., 14, 82
Trautman, A., 164 Weyl's method, 13, 16
Two-body problem, 142 Whitehead, A.B., 128, 164
Wigner, E.P., 33
Witten, L., 163
Universe,
age of, 41
expansion of, 41-43 Yang, C.N., 169, 170, 184
homogeneous and isotropic, 41 Yang-Mills field, 181
geometry of, 177-183
Yang-Mills theory, 167-172
Euclidean, 190-194
Variational problem, 116
Veblen, 0., 108, 165
Zero, 2
Vector current density, 85
Zero-eigenvalue modes, 190
Vector function,
Zero energy-momentum tensor, 105
continuous, 31
Vector potential, 85
Vectors, 190
contravatiant, 110-111, 114
covariant, 111, 114, 115
curl of, 114
0(4), 192
spinor equivalent to, 191,
192
Velocity, 37, 42, 43, 44
components of, 38
coordinate system, 38, 41
light propagation, 38
receding of galaxies, 41
relative, 39
Vilenkin, N.Ya., 33, 63, 82

Van der Waerden, B.L., 10, 82,

88, 108, 164
 AN INTRODUCTION

SPINORS ARE USED EXTENSIVELY IN

0 PHYSICS. IT 15 WIDELY ACCEPTED THAT
THEY ARE MORE FUNDAMENTAL THAN
eOI"r#b Y
TENSORS, AND THE EASY WAY TO SEE
THIS IS THROUGH THE RESULTS OBTAINED IN GENERAL RELATIVITY THEORY
BY USING SPINORS - RESULTS THAT COULD NOT HAVE BEEN OBTAINED BY
USING TENSOR METHODS ONLY.

THE FOUNDATION OF THE CONCEPT OF SPINORS IS GROUPS; SPINORS

APPEAR AS REPRESENTATIONS OF GROUPS. THIS TEXTBOOK EXPOUNDS THE
RELATIONSHIP BETWEEN SPINORS AND REPRESENTATIONS OF GROUPS. AS
IS WELL KNOWN. SPINORS AND REPRESENTATIONS ARE BOTH WIDELY USED
IN THE THEORY OF ELEMENTARY PARTICLES.

THE AUTHORS PRESENT THE ORIGIN OF SPINORS FROM REPRESENTATION

THEORY. BUT NEVERTHELESS APPLY THE THEORY OF SPINORS TO GENERAL
RELATIVITY THEORY, AND PART OF THE BOOK IS DEVOTED TO CURVED
SPACE-TIME APPLICATIONS.

BASED ON LECTURES GIVEN AT BEN GURION UNIVERSITY, THIS TEXTBOOK

IS INTENDED FOR ADVANCED UNDERGRADUATE AND GRADUATE STUDENTS IN
PHYSICS AND MATHEMATICS. AS WELL AS BEING A REFERENCE FOR
RESEARCHERS.

1380 he

www.worldscientific.com L