178 57 36MB
English Pages 1378 [1384] Year 1996
Wolfram Neutsch Coordinates
Wolfram Neutsch
Coordinates
w DE
Walter de Gruyter · Berlin · New York 1996
Author Priv.-Doz. Dr. Wolfram Neutsch Institut für Astrophysik der Universität Bonn Auf dem Hügel 71 53121 Bonn
Figure in cover illustration by kind permission of Oxford University Press, from Ray d'Inverno: Introducing Einstein's Relativity, Clarendon Press, Oxford, 1992
Library of Congress Cataloging-in-Publication Data Neutsch, W. Coordinates / Wolfram Neutsch. p. cm. Includes bibliographical references (p. ISBN 3-11-014852-8 (alk. paper) 1. Coordinates. I. Title. QA556.N48 1996 516'.16-dc20
—
) and index.
96-24425 CIP
Die Deutsche Bibliothek — Cataloging-in-Publication Data Neutsch, Wolfram: Coordinates / Wolfram Neutsch. — Berlin ; New York : de Gruyter, 1996 ISBN 3-11-014852-8
© Copyright 1996 by Walter de Gruyter & Co., D-10785 Berlin All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording or any information storage and retrieval system, without permission in writing from the publisher. Printed in Germany. Printing: WB-Druck, Rieden. — Binding: Heinz Stein, Berlin. Cover design: Hansbernd Lindemann, Berlin.
Preface Some time ago, and other
I began to collect systematically transformation
important
formulas
information about certain coordinate frames which I
found useful when dealing with mathematical and physical problems. Unfortunately, nomenclature,
in the available literature there is no universally
accepted
and no comprehensive account is readily at hand. This caused
me to prepare a number of tables which initially had been intended f o r my personal usage in a more systematical way (and with uniform terminology) in order to publish them. At the same time, I extended the somewhat scattered data, thus attempting to get a more complete presentation. It soon turned out that a simple compilation of the basic equations would not
be s u f f i c i e n t
for
the
practitioner;
a theoretical
introduction
is
in-
dispensable. For the same reason I also included a number of novel developments which are only loosely connected with the main topic of these books, among them a great many quadrature methods, some of which are new. Over the years, the short list of reference frames grew into the two heavy volumes of the present monograph.
In spite of
its size,
it is believed to
be welcomed as a handy companion f o r all who have to apply coordinates. It is my pleasant duty to thank all those who contributed to this
project
by countless critical remarks, a lot of new thoughts and insights, and r e f erences to books and research papers which I had overlooked. The number of single pieces of advice is much too large to mention them separately.
Without wanting to b e l i t t l e the others,
I would like to name only
Horst Fichtner and Klaus Scherer to whom I am most g r a t e f u l f o r their untiring willingness to discuss various aspects of the present book. I f e e l under an obligation to Reinhold Schaaf who drew my attention to an e r r o r in an e a r l i e r publication. My special thanks go to Günter Lay f o r his invaluable help (not only) concerning questions of
computing.
By f a r the greatest support, however, I owe to my mother, Ingeborg Neutsch, who during the preparation of the manuscript patiently bore my whims w i t h out complaining. I furthermore want to apologize to her f o r the chaos which I temporarily
produced by paper
lying around in her
f o r g i v e me, and thanks f o r everything!
living
room.
Please
Volume 1: Theory 0. Introduction
1
Α. Foundations 1. Historical development of the coordinate concept
1.1. Geography
9
9
1.2. Affine and projective geometry
15
1.3. Differential geometry
20
1.4. Geodesy and cartography
24
1.5. Special coordinates
34
2. Notation and conventions
2.1. Sets and topological spaces
38
38
2.2. Groups
42
2.3. Matrices
48
2.4. Multilinear forms sind tensors
56
2.5. Polynomials
65
B. Geometry 3. Manifolds
3.1. Continuous coordinate systems
70
70
3.2. Smooth manifolds
73
3.3. Curves and tangent spaces
76
3.4. Vector bundles
80
3.5. Differential and exterior derivation
85
3.6. Partition of unity
94
3.7. Oriented manifolds
102
3.8. Integration of differential forms
106
3.9. Stokes' theorem
110
4. Riemannian spaces
113
4.1. The metric tensor
113
4.2. Christoffel symbols and covariant derivation
117
4.3. Normal coordinates
123
4.4. Curvature
128
4.5. Volume
136
VIII
4.6. Duality 4.7. Classical vector analysis 5. Applications to physics
145 149 160
5.1. Mechanics
160
5.2. Hydrodynamics
166
5.3. Relativity
172
5.4. Electromagnet ism
175
5.5. Optics
182
5.6. Quantum mechanics
187
6. Complex analysis
190
6.1. Elementary properties of complex numbers
190
6.2. Convergence of function series
193
6.3. Power series
198
6.4. Analytical continuation
205
6.5. Holomorphic functions
211
6.6. Angle-preserving transformations
225
6.7. Lie groups
233
6.8. Lie algebras
237
7. Projective Geometry
246
7.1. Affine and projective coordinates
246
7.2. Projectivities
255
7.3. The cross ratio
260
7.4. B6zout' s theorem
268
7. 5. Planar algebraic curves
279
7.6. Stereographic projection
287
7.7. Higher dimensional spaces
293
C. Rotations 8. Orthogonal groups
295
8.1. Isometries and Euclidean transformations
295
8.2. The exponential mapping
301
8.3. Rational parametrization
311
9. Linear transformations of complex spaces
318
9.1. Pauli matrices
318
9.2. Cayley-Klein parameters
322
9.3. The angular momentum algebra
327
9.4. Gaussian mutations of space
336
IX
9.5. Euler angles 9.6. Elementary geometry of the Riemann number sphere 10. Quaternions 10.1. The skew field of quaternions
339 342 353 353
10.2. Left and right multiplication
359
10.3. Rotations of the quaternion algebra
362
10.4. Representation by complex matrices
367
10.5. Finite groups of quaternions
371
11. Octaves
379
11.1. Doubling method of Cayley and Dickson
379
11.2. Alternative division algebras
389
11.3. The theorem of Hurwitz
404
11.4. Quadratic algebras
411
11.5. Parallelizability and regular vielbeine
425
12. Hopf mappings
429
12.1. Homotopy groups of spheres
429
12.2. Homotopy invariants
433
12.3. Duplication of angles and classical Hopf fibration
438
12.4. Generalizations
444
12.5. Geometric peculiarities 13. Spinors
448 452
13.1. Schur extensions of the symmetric groups
452
13.2. Spin groups
466
13.3. Graßmann algebras
470
13.4. Clifford algebras
476
13.5. Dirac matrices
486
14. Lorentz transformations
492
14.1. The Poincar6 group
492
14.2. Boosts
499
14.3. Complex Lorentz matrices
508
14.4. Description by quaternions
S16
14.5. Representation theory of the Lorentz group
522
D. Reflections 15. Coxeter groups
528
15.1. Discrete symmetries
528
15.2. Invariant theory; Molien's theorem
532
χ
15.3. Polynomial Invariant rings
538
15.4. Reflection groups
550
15.5. Fundamental systems
558
15.6. The Euclidean Coxeter groups
566
16. Invariant rings of finite Weyl groups
574
16.1. Root systems
574
16.2. Weyl groups
589
16.3. Elementary symmetric functions
599
17. Basic invariants
605
17.1. Generic Weyl groups (A,IB,ID)
605
17.2. Weyl groups of type IE
607
17.3. Weyl groups of types F,(B,IH, D 17.4. Basic degrees
623 635
Volume 2: Applications E. Lattices 18. Elliptic functions and modular forms
637
18.1. Doubly periodic functions
637
18.2. Weierstraß functions
643
18.3. The modular group
654
18.4. Modular forms
661
18.5. Cusp forms
672
18.6. Modular functions; the J-invariant
678
19. Euclidean lattices
688
19.1. Foundations of lattice theory
688
19.2. Theta functions
698
19.3. The Gösset lattice
712
19.4. Niemeier lattices
717
20. Linear codes
726
20.1. Encoding of information
726
20. 2. The Hexacode
730
20.3. The binary Golay code
738
20.4. Miracle octad generator
751
20.5. Alternative constructions of the Golay code
757
20.6. Steiner systems and Mathieu groups
767
XI
21. The Leech lattice
781
21.1. Uniqueness of the Leech lattice
781
21.2. Explicit representations
789
21.3. Aut omorphi sms
792
21.4. Holes in the Leech lattice
799
21.5. Invariants of the Conway group
806
21.6. Lorentz lattices and sporadic groups
819
F. Spheres 22. Harmonic functions
837
22.1. The theorems of Green
837
22.2. Dirichlet's principle
844
22.3. The Poisson integral
854
22.4. Potential functions
862
23. Spherical surface functions
866
23.1. Spherical harmonics
866
23.2. Legendre polynomials
870
23.3. Orthogonal functions on the 2-sphere
881
23.4. C1 ebsch-Gordan coefficients
897
24. Lattice integration
913
24.1. Integration points and weights
913
24.2. Numerical integration after Gauß
920
24.3. Simple examples
928
24.4. Integration by means of the Gösset lattice
935
24.5. Transfer via Hopf mappings
941
25. Spherical designs
945
25.1. Integration methods with equal weights
945
25.2. Tight designs
951
25.3. Optimal integration on the 2-sphere
956
25. 4. The hypersphere
966
25.5. Very high precision numerical integration
976
25.6. Four-dimensional root systems and quaternions
981
G. Coordinate systems 26. Linear and reducible coordinates 26.1. Cartesian and oblique linear frames
985 985
XII
26.2. Polar coordinates
991
26.3. Classical cylindrical coordinates
995
26.4. Separable isothermal systems 27. Three-dimensional Stäckel coordinates
997 1006
27.1. Rotational symmetry
1006
27.2. Stäckel spaces
1015
27.3. Classification of Stäckel-type metrics
1020
27.4. Bipolar coordinates
1029
28. Confocal Coordinates
1037
28. 1. Orthogonal families of quadrics
1037
28.2. Metric and differential operators
1045
28.3. Separation of the potential equation
1053
28.4. Lam£ functions
1058
28.5. Geodesies on the ellipsoid
1064
29. Gauß-Krüger Coordinates
1068
29.1. Soldner's parameterization of the spheroid
1068
29.2. Series expansions
1077
29.3. Transversal Mercator projection
1083
29.4. Besse 1's figure of the Earth
1095
29.5. Recent models
1100
30. Coordinates for special applications
1104
30.1. Clairaut coordinates
1104
30.2. Roche coordinates
1109
30.3. Planar Weyl coordinates
1113
30.4. Weyl coordinates in three-dimensional space
1118
30.5. Higher dimensional Weyl coordinates
1126
H. Tables Calculation and organization of the tables 2
Coordinates in IR
1132 1136
Cartesian coordinates
1136
Polar coordinates
1139
Parabolic coordinates
1142
Elliptic coordinates
1145
Confocal coordinates
1148
Bipolar coordinates
1152
Digonal coordinates
1156
XIII
Trigonal coordinates
1159
Tetragonal coordinates
1162
Pentagonal coordinates
1165
Hexagonal coordinates 3
1168
Coordinates in IR
1171
Cartesian coordinates
1171
Cylindrical coordinates
1174
Polar coordinates
1178
Geographic coordinates
1182
Coordinates of the parabolic cylinder
1186
Coordinates of the elliptic cylinder
1190
Confocal cylindrical coordinates...
1194
Coordinates of the circular paraboloid
1198
Coordinates of the elliptic paraboloid
1202
Ellipsoidal coordinates (prolate)
1207
Ellipsoidal coordinates (oblate)
1211
Spheroidal coordinates (prolate)
1215
Spheroidal coordinates (oblate)
1219
Conical coordinates
1223
Confocal coordinates
1228
Bicylindrical coordinates
1234
Bispherical coordinates
1238
Torus coordinates
1243
Tetrahedral coordinates
1248
Octahedral coordinates
1253
Icosahedral coordinates
1259
Coordinates in IR4
1265
Cartesian coordinates
1265
Cylindrical coordinates of type (1,3)
1268
Cylindrical coordinates of type (2,2)
1272
Polar coordinates
1276
Geographic coordinates
1280
Double polar coordinates
1284
Confocal coordinates
1288
Appendix References
1297
Index
1322
Der N i e d e r r h e i n e r w e i ß nichts,
kann aber a l l e s
erklären. Hanns D i e t e r
HUsch
0. Introduction
C o o r d i n a t e s a r e ubiquitous in science and t e c h n o l o g y .
T h e r e is v i r t u a l l y
question
use.
of
solvability tion.
importance of
which
a particular
does not
problem
require
depends
strongly
A c l e v e r t r i c k may lead to new insights,
thematical
description
complicates
the
their
on the
the
parametriza-
w h i l e an i n a p p r o p r i a t e
investigation
or
no
As a r u l e ,
even
makes
ma-
it
im-
possible. Therefore,
i t is e s s e n t i a l
to be w e l l
i n f o r m e d about the a v a i l a b l e
Quite o f t e n ideas which at a f i r s t glance look a bit " e x o t i c " a r e helpful,
at
least
if
they
are
flexible
enough
to
be
adapted
frames. extremely
to
changing
conditions. The v a r i e t y
of
all
possible
the p a s t ) is immense;
reference
frames
(and even those
in
e v e r y d i f f e o m o r p h i s m of an a r b i t r a r y space o n t o some
r e g i o n in IRn or Cn d e f i n e s an n-dimensional mapping, nate
published
i. e.
a smooth
coordi-
system.
F o r t h i s r e a s o n a comprehensive r e p r e s e n t a t i o n of all imaginable zations
is t o t a l l y
out of
question,
parametri-
unless we a r e content w i t h g e n e r a l explicit.
finitions
and equations which a r e not v e r y
however,
these a r e of no use; he or she demands f o r m u l a s which a r e r e a d i l y
at hand and may be employed w i t h o u t long p r i o r
For
the
de-
calculations.
practitioner,
Thus,
a
re-
s t r i c t i o n t o the most i m p o r t a n t special systems is unavoidable. Universal
rules
appropriate
for
allowing
us to decide,
a given
purpose,
which
coordinate
do not e x i s t .
Rather,
the s k i l l f u l n e s s w i t h which one a t t a c k s a p r o b l e m . charm of
our f i e l d l i e s just in this " a r t i s t i c
The p r o p e r use of alternatives tively
which w e want to choose.
d i f f i c u l t t o d e r i v e f r o m the l i t e r a t u r e
are
On the o t h e r hand,
of
sufficiently
Unfortunately,
reasonably
it
complete
is
but may be i m p o r t a n t on special o p p o r t u n i t i e s .
can be g i v e n .
of the
many rela-
informa-
t i o n c o n c e r n i n g t y p e s of c o o r d i n a t e s which a r e employed c o m p a r a t i v e l y ly,
most
is a m a t t e r
freedom".
c o o r d i n a t e s r e q u i r e s the knowledge
between
systems it
S e v e r a l reasons f o r
rarethis
0.
2
In the f i r s t place,
Introduction
only a few systems are documented well enough;
formularies and t e x t books r e s t r i c t to the most common ones,
most
while more
" e x o t i c " ideas are discussed only occasionally and rudimentarily,
e. g.
by
doing no more than j u s t offering the definitions. Before we can apply those coordinates,
we have to f a c e more or less involved auxiliary calculations.
This is annoying and acts as a deterrent. In addition, each author has his or her own scheme of notation. This makes comparisons between d i f f e r e n t sources laborious. The above-mentioned reasons led me already a couple of years ago to compile the basic information on those coordinates which I found advantageous f o r my mathematical,
physical, and astronomical studies.
In the beginning,
I collected data a r b i t r a r i l y ,
but l a t e r a more and more
systematic treatment emerged. I always kept an eye on c l a r i t y and uniformity of notation.
After a while I developed a scheme which in my opinion
comes quite close to these ideals. Several colleagues prompted me to extend the mass of data and to publish them in concise form. The question
arose which degree of explicitness
is adequate.
Initially
I
tended to edit no more than a simple formulary, amended by a few comments and a t h e o r e t i c a l introduction. The bulk of the work was planned to consist of t a b l e s of the most important f a c t s like definitions, tial operators,
metrics,
differen-
etc. f o r as many systems as possible.
My own experience and, even more, numerous talks with colleagues from d i f f e r e n t branches, however, soon taught me that this kind of presentation is inadequate f o r several reasons: (1) A simple listing is of no value without some advice which coordinates should be applied under given circumstances. The most important c r i t e rion is the agreement of the symmetry properties of the problem to be solved and its required parametrization.
Accordingly,
I put the empha-
sis j u s t on this aspect. (2) Many mappings elude the s t r i c t schematization of a pure collection tables.
Therefore
a detailed evaluation
of more theoretical
ideas
of is
required. (3) Although most of the tabulated coordinates r e f e r to Euclidean spaces 2 3 4 (usually IR , R and IR ), charts of other domains are also of great int e r e s t . For instance, a discussion of our present topic would be incomplete without a thorough investigation of rotations or, to speak somewhat more generally,
isometries of the spaces in question.
Representa-
0.
Introduction
3
t i o n t h e o r y and i t s p r a c t i c a l c o n s e q u e n c e s t h e r e f o r e p l a y a m a j o r in t h i s m o n o g r a p h ;
This l e a d s us t o an overview of t h e c o n t e n t s . into eight p a r t s ,
role
they may even be c o n s i d e r e d a s a c e n t r a l t o p i c . The w o r k a s a w h o l e
splits
t h e f i r s t seven of which (A t h r o u g h G) c o n s t i t u t e a t e x t
book w h i l e t h e t a b l e s f o r m t h e l a s t one (H). P a r t A is d e d i c a t e d t o r i c a l and t e c h n i c a l p r e p a r a t i o n s , its various shades,
histo-
p a r t s Β t o Ε develop t h e t h e o r y p r o p e r in
p a r t F c o n t a i n s some i m p o r t a n t a p p l i c a t i o n s
concerning
t h e t r e a t m e n t of s p h e r i c a l h a r m o n i c s which a r e of wide use in s c i e n c e .
Fi-
n a l l y , p a r t s G and Η f o r m an i n t e g r a t e d whole. In t h e f o r m e r , t h e n e c e s s a r y e x p l i c a t i o n s t o t h e t a b l e s a r e t o be f o u n d . Both p a r t s , nates.
G and H, a r e r e s t r i c t e d t o t h e d i s c u s s i o n of c o n c r e t e
All f r a m e s l i s t e d
in p a r t
Η are described
coordi-
in m o r e o r l e s s
The i n t e r r e l a t i o n b e t w e e n p a r t s G and Η is n o t 1: 1, h o w e v e r .
detail.
The t h e o r e t i -
cal e v a l u a t i o n c a r r i e s much m o r e i n f o r m a t i o n t h a n t h e t a b l e s .
As i n d i c a t e d
above, a n u m b e r of s p e c i a l c o o r d i n a t e s could not be a d a p t e d t o t h e c o n c e p t of p a r t H; t h u s t h e y a r e m e n t i o n e d only in p a r t G. On t h e o t h e r h a n d , I i n cluded e x a m p l e s w h e r e v e r
possible.
F o r i n s t a n c e , t h e t h o r o u g h t r e a t m e n t in c h a p t e r 29 of t h e G a u ß - K r i i g e r map ( a l s o known a s t r a n s v e r s a l M e r c a t o r p r o j e c t i o n ) which is of w i d e s p r e a d use in geodesy and s u r v e y i n g h a s been e n r i c h e d by many n u m e r i c a l s e r i e s d e v e l o p m e n t s . To a s s i s t t h e r e a d e r , t h e c o n c r e t e n u m b e r s d e r i v e d f r o m t h e m o s t common s p h e r o i d a l m o d e l s of t h e E a r t h (among t h e m t h o s e of Bessel, H a y f o r d , and K r a s s o w s k i ) a r e s t a t e d a l s o . F u r t h e r m o r e a f e w o t h e r e x c e p t i o n a l p a r a m e t r i z a t i o n s like Roche and C l a i raut
c o o r d i n a t e s a r e described as well,
though their relatively
domain of a p p l i c a t i o n does not j u s t i f y a m o r e e x t e n s i v e In c l e a r c o n t r a s t t o t h i s ,
restricted
presentation.
t h e Weyl or C o x e t e r c o o r d i n a t e s which have been
d i s c o v e r e d only r e c e n t l y ( d u r i n g t h e l a s t d e c a d e ) will r e c e i v e g r o w i n g t e n t i o n in t h e n e a r f u t u r e .
They a r e a l w a y s s u i t a b l e if we a r e
w i t h a g e o m e t r y which p o s s e s s e s a n o n t r i v i a l f i n i t e g r o u p of
at-
concerned
symmetries.
This o f t e n o c c u r s in s o l i d s t a t e p h y s i c s ( c r y s t a l l o g r a p h y ) and e l s e w h e r e . Up t o now no c o m p r e h e n s i v e a c c o u n t of Weyl c o o r d i n a t e s h a s been p u b l i s h e d . All a v a i l a b l e
i n f o r m a t i o n is contained
correspondingly d i f f i c u l t to obtain.
in a f e w m a t h e m a t i c a l
papers
and
In t h e p r e s e n t book we a t t e m p t f o r t h e
f i r s t t i m e t o e s t a b l i s h a sound t h e o r y of t h i s c a t e g o r y of c h a r t s .
Except
f o r c e r t a i n known f a c t s , some new r e s u l t s have been f o u n d which open a l t e r n a t i v e p a t h s and a r e a d v a n t a g e o u s in c o m p a r i s o n t o t h e m o r e methods.
conventional
4
0.
Introduction
Anyway, t h e a s s e r t i o n of many a u t h o r s t h a t a couple of c o o r d i n a t e
systems
a r e by t h e m s e l v e s much m o r e i m p o r t a n t t h a n t h e o t h e r s b e c a u s e t h e y may be applied
more
universally,
very f e w i n s t a n c e s ,
e.g.
has to
be q u a l i f i e d .
Cartesian
or
It
is c e r t a i n l y
true
(with some r e s t r i c t i o n s )
in
a
spherical
coordinates. Beyond t h i s ,
in my opinion t h i s s t a t e m e n t
seems t o be an i n t e r c h a n g e
c a u s e and e f f e c t . If one is c o n f r o n t e d w i t h a c o n c r e t e m a t h e m a t i c a l lem,
it is q u i t e n a t u r a l t o a p p l y t h e b e s t known p r o c e d u r e s i n s t e a d of
ing t o new t e r r i t o r y . ches a r e p r e f e r r e d ,
Thus t h e m o r e e x p l i c i t l y d e s c r i b e d "simple" even if o t h e r p o s s i b i l i t i e s might be b e t t e r .
of
probgo-
approaIt is one
of t h e g o a l s of t h e s e t w o volumes t o achieve a s o m e w h a t m o r e b a l a n c e d p r e sentation. Coming back t o t h e t h e o r e t i c a l p a r t s A t o F, some r e m a r k s a r e n e c e s s a r y : At t h e h e a r t of t h e i n t r o d u c t o r y p a r t A l i e s c h a p t e r 2,
where I collected
t h o s e f u n d a m e n t a l d e f i n i t i o n s and t h e o r e m s which a r e p r o b a b l y known t o many readers,
b u t may be a l i t t l e b i t u n f a m i l i a r t o o t h e r s .
The aim of t h i s i n s e r t i o n is t w o f o l d . F i r s t , tion,
we p r e s e n t t h e b a s i c i n f o r m a -
which m a i n l y c o n s i s t s of w h a t is c a l l e d m a t h e m a t i c a l
"folklore",
c o m p r e s s e d f o r m t o avoid i n t e r r u p t i o n s of t h e t e x t p a s s a g e s t o f o l l o w .
in La-
t e r - o n t h e s e s e c t i o n s of t h e t e x t a r e c i t e d w i t h o u t f u r t h e r e x p l a n a t i o n . Beside t h i s , t h e n o t a t i o n is d e f i n e d and e x p l a i n e d . The n o m e n c l a t u r e e s s e n tially f i t s into e a r l i e r systems derived f r o m the relevant l i t e r a t u r e .
Only
o c c a s i o n a l l y d e v i a t i o n s or a m e n d m e n t s t u r n e d out t o be n e c e s s a r y . They a r e all s t a t e d
explicitly.
Whenever a l t e r n a t i v e a s s i g n m e n t s of s y m b o l s a r e in u s e , I t o o k one of t h e m according to two d i f f e r e n t c r i t e r i a . was p r e f e r r e d . suggested
Sometimes,
however,
Normally the more "abundant" notation f o r r e a s o n s of e f f i c a c y o t h e r
choices
themselves.
In any c a s e I c a r e f u l l y o b s e r v e d t h e d e m a n d s of f a i t h f u l n e s s and
retainabi-
lity. R e a d e r s who a r e f a m i l i a r w i t h t h e r e l e v a n t i d e a s may skip c h a p t e r 2 w i t h o u t serious loss,
but should go back if they a r e in doubt a b o u t t h e m e a n i n g of
a p a r t i c u l a r concept.
Those who a r e n o t so w e l l - v e r s e d in a c e r t a i n m a t h e -
m a t i c a l f i e l d a r e a d v i s e d t o s t u d y t h e s e c t i o n s r e g a r d i n g it c a r e f u l l y . t e r w a r d s t h e y w i l l be p r e p a r e d t o f o l l o w t h e d i s c u s s i o n w i t h o u t (and w i t h o u t r e f e r r i n g t o o t h e r
Af-
difficulty
sources).
The p r i m e t o p i c s of p a r t Β a r e t o d e f i n e t h e t e r m " c o o r d i n a t e s " and t o p r o vide a couple of
elementary geometric tools.
Here I t o o k a v e r y
general
0.
Introduction
5
p o i n t of view in o r d e r t o leave s p a c e f o r a l l i t s v a r i o u s f a c e t s ,
even if
f o r m o s t a p p l i c a t i o n s a n a r r o w e r f o r m u l a t i o n would have s u f f i c e d . A r e s t r i c t i o n t o , e. g. , r e a l or c o m p l e x c o o r d i n a t e s would mean t h e l o s s of several extremely interesting parametrizations f o r c e r t a i n s c i e n t i f i c or technical
which a r e a l s o v e r y u s e f u l
problems.
C o n c e r n i n g t h e c o n t e n t s of p a r t B, we may be b r i e f . (differential)
geometric,
i.e.
analytical,
It l a r g e l y g i v e s
foundations
without
the
neglecting
a f f i n e and p r o j e c t i v e g e o m e t r y . Here and in o t h e r p a s s a g e s of t h e book we go as f a r a s j u s t i f i a b l e t h e l i m i t a t i o n s of s p a c e . Where it w a s p o s s i b l e ,
under
I f u r t h e r m o r e included
f o r m a t i o n which is s o m e w h a t o u t s i d e t h e r e a l m of p r a c t i c a l n e c e s s i t y . is i n t e n d e d t o make f u t u r e d e v e l o p m e n t s e a s i e r and t o widen t h e
This
horizons.
It w a s ( r a r e l y ! ) n e c e s s a r y t o r o u n d o f f t h e p r e s e n t a t i o n by s k e t c h i n g c i a l r e s u l t s of r e c e n t r e s e a r c h ,
cru-
though the p r o o f s a r e too d i f f i c u l t or
volved t o p r e s e n t t h e m in t h i s book.
I believe t h a t such o c c a s i o n a l
e x c u r s i o n s a r e i n d i s p e n s a b l e f o r a deeper
in-
short
understanding.
In p a r t C we c o n s i d e r p a r a m e t r i z a t i o n s of r o t a t i o n s , i n t r o d u c t i o n of c o o r d i n a t e s on o r t h o g o n a l g r o u p s . discussed,
in-
and in p a r t i c u l a r
the
A number of m e t h o d s
is
t h e bulk of which a r e of a l g e b r a i c o r i g i n .
By t h e way, d u r i n g t h e w r i t i n g of t h e m a n u s c r i p t t h e w e i g h t i n g s h i f t e d f r o m a mainly analytical to a more algebraic i n t e r p r e t a t i o n . p e c t is c a u s e d by t h e i n s i g h t t h a t
the most
T h i s c h a n g e of
"elegant" mathematical
asideas
a r e a l w a y s t h o s e which p o s s e s s s p e c i a l s y m m e t r i e s . This does n o t mean a n y t h i n g e l s e t h a n t h e e x i s t e n c e of a " l a r g e " g r o u p of i n v a r i a n c e
operations.
Beside a f e w d i f f e r e n t d e s c r i p t i o n s of t h e r o t a t i o n g r o u p s in IRn f o r
arbi-
t r a r y ( f i n i t e ) n, we a l s o d i s c u s s some m e t h o d s which a r e i m p o r t a n t t o t h e practitioner,
b u t a p p l i c a b l e only t o t h e most o f t e n needed low
dimensions
2, 3, and 4. They a r e n o t e x t e n s i b l e t o h i g h e r d i m e n s i o n a l s p a c e s since t h e y depend on g r o u p t h e o r e t i c a l p e c u l i a r i t i e s of t h e complex n u m b e r s , q u a t e r n i o n s o r C a y ley n u m b e r s ( o c t a v e s ) . Here I added a couple of r e m a r k s on f u r t h e r i n v e s t i g a t i o n s and d e e p e r r e s u l t s in t o p o l o g y and t h e n e i g h b o u r i n g b r a n c h e s . The l a s t t h r e e c h a p t e r s of p a r t C c o n t a i n s e v e r a l c o n c e p t s which a r e very u s e f u l b u t n o t widely known o u t s i d e
mathematics.
In c h a p t e r 12 we c o n s i d e r t h e s o - c a l l e d Hopf m a p p i n g s . Up t o now, t h e i r r e al m e a n i n g s e e m s t o have been acknowledged only by t h e ( a l g e b r a i c ) gists.
(Later,
topolo-
in c h a p t e r 24, we s h a l l e x p l o i t t h e m t o c o n s t r u c t high
cision numerical i n t e g r a t i o n
methods).
pre-
0.
6
In the sequel,
a thorough investigation of the nontrivial coverings of
thogonal and related groups f o l l o w s . spin groups.
Introduction
or-
These are known to the physicist
as
We c l a s s i f y the real and complex C l i f f o r d algebras and apply
them to spinors.
As a by-result
we get the representation theory of
the
three-dimensional rotation group. This is not merely fundamental f o r quantum mechanics; there are applications in many other f i e l d s , a f e w of which are mentioned in the text. The remainder of part C is a chapter about Lorentz transformations.
Here,
except f o r the very well known real forms we also give somewhat less common descriptions by means of complex matrices and quaternions. Among the isometries of Euclidean spaces, tions,
not only the rotations,
i.e.
the orthogonal
transforma-
but also the r e f l e c t i o n s are of utmost
inte-
rest. T h e r e f o r e the leitmotif of part D is a combination of representation theory of Coxeter groups (which are generated by r e f l e c t i o n s ) and invariant theory (Molien's
theorem).
with the outcome;
Mathematicians I nevertheless
or
at
least
algebraists
are
familiar
attempted to give a self-contained
and
comprehensive account although this goes a bit beyond the limitations kept elsewhere in this monograph. To be precise, I not only included a c l a s s i f i cation of the Weyl representations of all f i n i t e Coxeter groups; compiled a complete list of the polynomial invariants,
I moreover
sometimes in several
equivalent ways. These occasionally longish formulas are the basis of several later
investi-
gations.
integra-
We need them to construct extremely e f f i c i e n t numerical
tion methods on spheres (chapters 24 and 25, see below). The explanations
in this part
are a l i t t l e
more involved than
but I tried to present them in an easily understandable f o r m .
elsewhere, This should
allow readers who are not so well-versed mathematically or more inclined to practical applications to find their way to this fascinating subject. Lattice theory as developed largely by elementary
means in part Ε is
of
fundamental importance f o r the chapters to f o l l o w . Unfortunately here again a f u l l discussion has to be ruled out because of only the most significant ideas can be dealt with.
its enormous extent,
so
Among them is a short
account of the complex-analytical properties of theta functions and a concise treatment of some important lattices,
particularly those discovered by
Gösset and Leech. A f e w remarks on recent progress in sporadic group r e search f o l l o w (without p r o o f s ) . Virtually all terms discussed previously are applied in part F to the prob-
0.
Introduction
7
lem of a p p r o x i m a t i n g s p h e r i c a l f u n c t i o n s by s t a n d a r d t y p e s ( s p h e r i c a l monics)
and
to
mechanical
integration.
The
cross
relations
to
har-
potential
t h e o r y (key w o r d : h a r m o n i c f u n c t i o n s ) a r e shown and d e s c r i b e d in much d e tail. We a l s o d i s c u s s t h e s t r u c t u r e c o n s t a n t s ,
known f r o m q u a n t u m
mechanics,
which a r i s e d u r i n g t h e r e d u c t i o n of t e n s o r r e p r e s e n t a t i o n s of t h e
angular
momentum a l g e b r a ( C l e b s c h - G o r d a n c o e f f i c i e n t s and t h e i r r e l a t i v e s , t h e Wigner symbols, cf. section 23.4). The c u l m i n a t i o n p o i n t of t h e s e c o n s i d e r a t i o n s a r e t h e two f i n a l c h a p t e r s of p a r t F, which a r e c o n c e r n e d w i t h t w o a l t e r n a t i v e n u m e r i c a l i n t e g r a t i o n m e t h o d s on s p h e r e s (of v a r i o u s d i m e n s i o n s ) . The f i r s t ( c h a p t e r 23) is f o u n d e d on g e o m e t r i c and a l g e b r a i c p r o p e r t i e s of special lattices; last few years.
m e t h o d s of t h i s c l a s s have been developed only w i t h i n t h e They t u r n out t o be very u s e f u l .
Another a p p r o a c h ,
s t e m m i n g f r o m c o m b i n a t o r i c s (= f i n i t e g e o m e t r y ) ,
the recently discovered so-called spherical designs.
By t h i s one m e a n s
t e g r a t i o n p r o c e d u r e s w i t h equal w e i g h t s f o r a l l p o i n t s . configurations, exhibiting complete symmetry,
is via
A s u b s e t of
is p a r t i c u l a r l y
in-
these
valuable.
Many of t h e c o n s t r u c t i o n s in c h a p t e r 25 have been a v a i l a b l e u n t i l t o d a y o n ly in m a t h e m a t i c a l
journals,
but not in t e x t books.
O t h e r s have been
in-
v e n t e d newly f o r t h e p r e s e n t m o n o g r a p h . Some of t h e m p o s s e s s d e g r e e s of a p p r o x i m a t i o n which s u r p a s s t h o s e of t h e o l d e r ones, t h u s d e m o n s t r a t i n g t h e i r high q u a l i t y . As i n d i c a t e d above,
it is i n e v i t a b l e t h a t t h e v a r i o u s c o n t r i b u t i o n s have t o
be h a n d l e d on d i f f e r e n t levels.
In c o n n e c t i o n
with
s i m p l e and d i f f i c u l t q u e s t i o n s t o be c o n s i d e r e d .
coordinates
there
This need not d e t e r
dy; even t h e m o r e a d v a n c e d p a s s a g e s should be a c c e s s i b l e t o wide
are
anybosections
of t h e r e a d e r s h i p . One may skip o c c a s i o n a l deeper a r g u m e n t s or g l a n c e t h r o u g h t h e m
without
loss. Both v o l u m e s of t h i s m o n o g r a p h a r e d i r e c t e d a t s t u d e n t s of science
or
especially
technical at
disciplines
practitioners
as well
(engineers)
They a r e w r i t t e n f o r p r i v a t e s t u d y ,
as l e c t u r e r s
of
mathematics,
these
in d i f f e r e n t b r a n c h e s
of
fields
and
industry.
but a r e a t l e a s t p a r t i a l l y a p t a l s o t o
support appropriate university courses. Above a l l t h e p a r t s c o n c e r n e d w i t h s p e c i a l s y s t e m s and t h e f o r m u l a r y may a l s o s e r v e a s a g u i d a n c e t o decide which c o o r d i n a t e s should be used f o r a particular
purpose.
I tried
hard
to present
all
information
as
explicitly
δ
0.
Introduction
a s p o s s i b l e in a c c o r d a n c e w i t h t h e l i m i t e d s p a c e . T h e r e f o r e in any c a s e a f t e r t h e d e f i n i n g t r a n s f o r m a t i o n s ( f r o m which e v e r y t h i n g e l s e w a s d e r i v e d ) t h e c o n t r a - and c o v a r i a n t m e t r i c t e n s o r s , f e r e n t i a l s of t h e c o o r d i n a t e s ,
the d i f -
and t h e volume e l e m e n t a r e given.
They a r e
f o l l o w e d by t h e C h r i s t o f f e l s y m b o l s ( a f f i n e c o n n e c t i o n ) and t h e m o s t i m p o r tant differential operators, (div) and L a p l a c i a n omitted,
curl (rot),
divergence
s i n c e it would r e q u i r e very much r o o m w h i l e being of
low i n t e r e s t . the
namely g r a d i e n t ( g r a d ) ,
(Δ). Only f o r t h e f o u r - d i m e n s i o n a l f r a m e s t h e c u r l
If n e c e s s a r y ,
Christoffel
is
relatively
it can be deduced w i t h a l m o s t no e f f o r t f r o m
coefficients;
the
rules
required
for
obtaining
it
are
in-
cluded. T h e r e is no need t o e m p h a s i z e t h a t I took g r e a t p a i n s t o e l i m i n a t e t o t h e b e s t of my a b i l i t y .
errors
All c a l c u l a t i o n s have been c a r r i e d out w i t h
the
h e l p of c o m p u t e r a l g e b r a p r o g r a m s (Reduce) w h e n e v e r t h i s w a s f e a s i b l e . The r e s u l t s were t r a n s f e r r e d directly into the t e x t f i l e s . Only t h e l a y o u t h a s been improved by hand,
and I s t r i c t l y avoided t o
s c r i b e f o r m u l a s o r l a r g e n u m b e r s and copied them Clearly,
automatically
tran-
instead.
many b l u n d e r s will have s u r v i v e d , t h i s is i n e v i t a b l e .
Of c o u r s e ,
a l l c a l c u l a t i o n s w e r e checked i n d e p e n d e n t l y if p o s s i b l e ,
for
ex-
a m p l e by c o m p a r i s o n w i t h t h e l i t e r a t u r e (on t h i s o c c a s i o n I f o u n d n u m e r o u s i n c o r r e c t s t a t e m e n t s in s e v e r a l p u b l i c a t i o n s ) or by r e d o i n g w i t h
independ-
ent
numerical
codes
or
alternative
methods.
Sometimes
specialization
or
t r e a t m e n t of s i n g u l a r c a s e s t u r n e d out t o be u s e f u l . I hope and am q u i t e c o n f i d e n t t h a t t h e number of r e m a i n i n g m i s t a k e s is a c c e p t a b l e and t h a t t h i s book will be welcomed a s a handy and c o n v e n i e n t c o m panion.
A. Foundations 1. Historical Development of the Coordinate Concept 1.1. Geography During the Neolithic age the transition f r o m nomadic to settled f o r m of e x istence required to divide the available land into parcels which were ed by individual persons or families. estate boundaries precisely.
seiz-
This made it necessary to define the
Simple land registers or f i e l d maps which pic-
t o r i a l l y represent the distribution of property do not s u f f i c e ;
one rather
needs an exact and unique quantitative specification. It is no longer possible to say when and in which culture area this was done f o r the f i r s t time,
but it is j u s t i f i e d to assume that a gradual deve-
lopment has taken place,
leading f r o m a more descriptive f o r m to a tabular
recording of important sites,
as e. g.
the corners of the plots of land or
conspicuous natural formations etc. People
soon
started
having
the
self-suggesting
thought
to
distinguish
pair of mutually orthogonal axes to which the positions are r e f e r r e d .
a
The
Cartesian coordinates defined in this way are not only the simplest method which
serves
the
purpose;
they
are
also
extremely
handy
and
flexible.
T h e r e f o r e over the millennia alternative ideas merely played a minor r o l e . Cartesian frames,
however,
t r i c t e d to a limited region,
are applicable only if the mapped area is r e s e. g.
single villages,
towns or districts.
With
growing skill and experience in shipbuilding and navigation more and more extended voyages of discovery and trade asked f o r a parametrization of the whole surface of the Earth with a single coordinate system. Probably relying on Chaldean sources and confirmed by the circular shapes of the horizon and the t e r r e s t r i a l shadow during lunar eclipses,
Pythagoras
(ca. 580 - ca. 500 BC) taught that the Earth is spherical. This f o r b i d s the use of Cartesian coordinates as soon as the curvature becomes noticeable. Instead, one applies a description with the help of two angles. Their d e f i nition depends on the rotation of the Earth. known in classical Greece,
This phenomenon was already
but lacked a rigorous proof until,
in 1851, Jean
Bernard L6on Foucault (1819-1868) carried out his famous pendulum experi-
10
ment.
1. H i s t o r i c a l Development o f the Coordinate Concept
The r o t a t i o n axis is f i x e d in space and physically
antipodal
points
on the t e r r e s t r i a l
surface
(North pole
distinguishes and South
two
pole).
The equatorial plane l i e s symmetrically between them. It passes through the centre of the Earth. As the f i r s t coordinate of some point Ρ on the s u r f a c e of the Earth we c o n v e n t i o n a l l y take i t s geographic latitude φ. It is the angle enclosed by the direction
f r o m the E a r t h ' s centre to Ρ and the equatorial
plane,
counted
p o s i t i v e to the North. The lines of constant φ are the circles of latitude, among them the equator (φ = 0 ) .
The orthogonal t r a j e c t o r i e s are also c i r c l e s ,
namely a l l which go
through both poles. They are called circles of longitude or meridians. One of
them is selected
arbitrarily.
It
the geographic
gets
longitude
0.
Then the longitude λ of any point Ρ is given by the angle between the prime meridian (λ = 0 ) but not a l w a y s ,
and the
circle
of
longitude
passing
through
P.
Usually,
Eastern longitudes are considered as p o s i t i v e .
The positioning of the z e r o meridian is not r e s t r i c t e d by any natural ference;
one may d e f i n e it at w i l l .
pre-
For that reason in the past a g r e a t
va-
r i e t y of d i f f e r e n t conventions has been applied. Most countries (but by no means a l l ) r e f e r r e d the geographic longitude to a w e l l d e f i n e d point which a l l o w e d f o r precise measurements. N o r m a l l y the meridian of
an important o b s e r v a t o r y ,
was e l e c t e d .
Nevertheless,
for
instance that of
some s t a t e s p r e f e r r e d other,
the capital
city,
less convenient
ar-
rangements. Only on the International Meridian Conference the z e r o
(Washington,
October
It passes through a passage
line was f i x e d bindingly.
(meridian circle) at the Royal Observatory of
Greenwich.
We c a l l the coordinate f r a m e consisting of λ and φ geographic. r a t u r e the term "polar coordinates" is used also, exclusively
to
a slight
modification
of
1884)
instrument
In the
lite-
but we shall r e s t r i c t
the geographic
coordinates,
is found if we r e p l a c e the latitude φ by the polar angle 90°-
The
latter
is also known as the "colatitude". Latitude measurements can be c a r r i e d out by simple means; e. g. , it s u f f i c e s to d e r i v e the elevation of the North pole which equals φ f r o m the daily tremes (upper and lower culminations) of circumpolar Another
method
which
is
mainly
applied
to
("shoot") the Sun or a star with a sextant.
nautical
ex-
stars. tasks
is
to
observe
The aim is to f i n d the
angle
f o r m e d by the d i r e c t i o n to the c e l e s t i a l body and to the horizon. When w e neglect several small c o r r e c t i o n s like refraction, which is caused
1.1.
Geography
11
by t h e d i f f r a c t i o n of
light
in t h e
atmosphere,
or the d i f f e r e n c e
t h e visual a n d t h e mathematical horizon ( s e e b e l o w ) ,
between
w h i c h d e p e n d s on
the
h e i g h t of t h e o b s e r v e r a b o v e s e a l e v e l , w e i m m e d i a t e l y g e t t h e e l e v a t i o n of the sighted If
its
object.
position
in t h e
sky
is
tabulated
in
the
l a t i t u d e is derived easily f r o m the culmination On t h e o t h e r h a n d ,
(almanacs),
yearbooks
the
height.
b e c a u s e of t h e a r b i t r a r y f i x i n g of t h e z e r o m e r i d i a n t h e
l o n g i t u d e c a n b e d e t e r m i n e d o n l y by c o m p a r i s o n of t h e r e s u l t s of t w o tions.
This
centuries,
requires
a precise
knowledge
of
the
i t c o u l d n o t be a c h i e v e d t o e v e r y o n e ' s
In p a r t i c u l a r
the transport
of
actual
time.
sta-
earlier
satisfaction.
c l o c k s by s h i p w a s e x t r e m e l y
T h e r e f o r e in 1714, on N e w t o n ' s s u g g e s t i o n ,
In
problematic.
the English p a r l i a m e n t o f f e r e d a
p r i z e of 2 0 0 0 0 p o u n d s s t e r l i n g f o r a m e t h o d w h i c h a l l o w e d t o m e a s u r e i n t e r continental
longitude
differences
(Europe-America)
with
an
error
of
not
more than half a degree. A f t e r long discussions, son ( 1 6 9 3 - 1 7 7 6 ) ,
t h e a w a r d w a s g i v e n t o t h e c l o c k - m a k e r J o h n Harri-
who had c o n s t r u c t e d
a chronometer w h i c h w a s f i x e d in
a
C a r d a n i c f r a m e a n d w o r k e d q u i t e p r e c i s e l y even on h e a v y s e a s . Up t o t h e n o n e h a d t a k e n r e s o r t t o d i f f e r e n t p r o c e d u r e s of w h i c h w e s h a l l m e n t i o n o n l y a f e w of t h e m o s t i m p o r t a n t On s o l i d g r o u n d
there
is the possibility
ones. to carry
of m a n y p o i n t s by trlangulation. To t h i s e n d ,
out relative
positioning
one d e f i n e s a net of trian-
gles in w h i c h one s i d e a n d a l l a n g l e s c a n be e v a l u a t e d .
If
the radius
t h e E a r t h i s k n o w n ( i t c a n be c a l c u l a t e d f r o m t h e d a t a t h e m s e l v e s ) ,
of
longi-
t u d e d i f f e r e n c e s may be deduced as well. At s e a ,
however,
t h i s c a n n o t be d o n e s i n c e f i x e d p o i n t s a r e l a c k i n g .
o n e m a d e u s e of c e r t a i n
astronomical
places (nearly) at the same time.
phenomena which occur
Valuable are,
in
among o t h e r s ,
Here
different
stellar oc-
cultations by t h e Moon a n d eclipses of t h e Jovian satellites w h i c h h a v e b e c o m e o b s e r v a b l e a f t e r t h e i n v e n t i o n of t h e
telescope.
With t h e c o n s t r u c t i o n of electric telegraphs a b o u t 1830 by C a r l Gauß (1777-1855) a n d Wilhelm E d u a r d Weber (1804-1891) a n d n e a r l y
Friedrich simultane-
o u s l y by P a w e l L w o w i t s c h Schilling (1786-1837) t i m e s i g n a l s c o u l d b e t r a n s f e r r e d over large distances,
and the d i f f i c u l t y was overcome
Today longitude measurements, the poles,
e x c e p t in t h e i m m e d i a t e v i c i n i t y of o n e of
a r e a s p r e c i s e a s t h o s e of t h e l a t i t u d e .
t h i s p o i n t in s e c t .
1.4,
completely.
We s h a l l c o m e b a c k
w h e r e we discuss some r e c e n t l y developed
which a r e applied mainly to geodesy.
to
methods
12
1. H i s t o r i c a l Development o f the Coordinate Concept
In astronomy a number of reference frames are needed to describe positions.
From a f o r m a l mathematical point of view,
stellar
they are all equiva-
lent to the geographic coordinates. T h e r e f o r e we shall just summarize them without f u r t h e r comment. More detailed information is found in the specialist literature,
f o r instance Bucerius [1967] or Neutsch & Scherer [1992].
The astronomical coordinates are roughly subdivided into topo-, heliocentric ones,
geo-,
and
according as the reference point f r o m where the angles
are counted lies at the position of the observer,
the Earth's or the Sun's
centre. The most easily realized topocentric coordinates are given by the horizontal system.
It is based on the direction of
local g r a v i t y ,
cally downward to the nadir. Diametrically opposed to this, the observer,
pointing
verti-
straight above
is the zenith. The plane through the origin which is perpen-
dicular to the line connecting zenith and nadir defines the
(mathematical)
horizon. It d i f f e r s f r o m the apparent horizon at sea by the perspective e f f e c t due to the curvature of the Earth's surface. On the continents additional
devi-
ations are produced by the local bumpiness of the terrain. For astronomical purposes the mathematical horizon alone is decisive. The angle between the horizon and some heavenly body is its elevation h, which sometimes is replaced by the zenith distance ζ = 90°-h. coordinate is the orientation angle along the horizon.
The second
It is known as the
azimuth A. Unfortunately
in the various branches of
profession which are
with navigation several d i f f e r e n t conventions are in use. Thus,
concerned one has to
be careful when comparing results f r o m external sources. In astronomy i t s e l f the azimuth is reckoned f r o m the Northern point at the horizon via East, South, and West, while f o r example navigators occasionally p r e f e r to count f r o m either North or South to East or West. The position Ν 35° W
hence
corresponds
to
the
azimuth
A = -35°
or,
equivalently,
A = 315°. In aeronautics and aviation even other rules are employed,
etc.
A d e f i n i t e disadvantage of the horizontal system is that the coordinates of a star change permanently because of the rotation of the Earth, even if we neglect the minor positional shifts produced by proper motion, and parallax.
(They are in the order of arc seconds).
ready in antiquity other reference systems have been introduced, oriented according to the direction of tension
intersects
the
fictitious
the Earth's rotation
celestial
sphere
aberration,
For this reason
in
axis.
North
al-
which are and
Its
ex-
South
1.1.
Geography
13
Poles;
t h e g r e a t c i r c l e which is s i t u a t e d s y m m e t r i c a l l y t o t h e s e t w o p o i n t s
is t h e
(celestial) equator.
The d e v i a t i o n of a s t a r f r o m t h i s line,
reckoned positive to the North,
is
i t s declination δ. The a n g l e c e n t r e d a t t h e N o r t h Pole which e m b r a c e s t h e d i r e c t i o n t o t h e o b j e c t and t h e m e r i d i a n is t h e hour angle β. It
increases
in p r o p o r t i o n t o t h e t i m e while t h e d e c l i n a t i o n is i n d e p e n d e n t of t i m e and t h e l o c a t i o n of t h e o b s e r v e r .
The rotating equatorial system t h u s
defined
t h e r e f o r e is much s i m p l e r t h a n t h e h o r i z o n t a l f r a m e . It h a s been known a l r e a d y t o Hipparchus (ca. 180 - ca. 125 BC). He a l s o did t h e n e x t s t e p and r e p l a c e d t h e h o u r a n g l e by a q u a n t i t y
which
is i n v a r i a n t and d e p e n d s on t h e s t e l l a r p o s i t i o n a l o n e . To t h i s end he s u b t r a c t s f r o m t h e h o u r a n g l e f i r s t t h e s o - c a l l e d sidereal time. T h i s i s a u n i f o r m m e a s u r e of t i m e which i n c r e a s e s by 360° d u r i n g one r e v o l u t i o n p e r i o d of t h e E a r t h ( c l o s e t o 23 h 56 m 04 s ). s t a n t in t i m e ,
but varies f r o m place to place.
t h e g e o g r a p h i c a l l o n g i t u d e of t h e o b s e r v a t o r y , of t h e s t a r ,
we g e t a p r o p e r
subtract
coordinate
n a m e l y i t s right ascension a.
An a r b i t r a r y a d d i t i v e c o n s t a n t , r i d i a n on t h e E a r t h , The l a t t e r
The d i f f e r e n c e is c o n -
If we f u r t h e r m o r e
is t h e
s i m i l a r t o t h e e s t a b l i s h i n g of t h e z e r o m e -
is f i x e d such t h a t t h e vernal equinox l i e s a t α = 0.
point
on t h e
ecliptic,
i. e.
the
apparent
orbit
Sun, w h e r e i t moves f r o m t h e s o u t h e r n t o t h e n o r t h e r n h e m i s p h e r e .
of
the
In o u r
t i m e s t h i s h a p p e n s every y e a r a b o u t March 20. Right a s c e n s i o n and d e c l i n a t i o n t o g e t h e r f o r m t h e (inertial) equatorial system. E x c e p t f o r t h e a b o v e - m e n t i o n e d r e f e r e n c e f r a m e s s e v e r a l o t h e r s a r e in c o m mon u s e , f o r i n s t a n c e t h e ecliptic coordinates. They d i f f e r f r o m t h e e q u a t o r i a l o n e s only in t h e choice of t h e f u n d a m e n t a l p l a n e which in t h e s e n t c a s e is t h e e c l i p t i c . latitude.
pre-
The p a r a m e t e r s a r e c a l l e d e c l i p t i c l o n g i t u d e
The f o r m e r is a g a i n c o u n t e d w i t h r e s p e c t t o t h e v e r n a l
and
equinox,
in t h e d i r e c t i o n of t h e ( a p p a r e n t ) s o l a r m o t i o n . Both s y s t e m s a r e i n c l i n e d t o each o t h e r by an a n g l e of a b o u t 2 3 ° 2 6 ' 2 1 " (in the year 2000),
t h e obliquity of the ecliptic. We s h a l l n o t d i s c u s s
here
t h e f o r m u l a s r e q u i r e d f o r t h e t r a n s i t i o n f r o m one s e t of c o o r d i n a t e s t o t h e o t h e r ; t h e y can be looked up in N e u t s c h & S c h e r e r [1992] o r , w i t h n u m e r i c a l e x a m p l e s and d e t a i l e d i n s t r u c t i o n s , in C u r n e t t e & Woolley [19741. Among m a ny a l t e r n a t i v e s ,
t h e y may a l s o be o b t a i n e d f r o m t h e i n v e s t i g a t i o n s of
part
Β of t h i s book. By t h e way, in t h e l i t e r a t u r e j u s t c i t e d one may f i n d m o r e p r e c i s e i n f o r m a t i o n on t h e t e m p o r a l
v a r i a t i o n of
the position
of t h e e q u i n o x e s .
This
is
14
1. H i s t o r i c a l Development of t h e C o o r d i n a t e Concept
c a u s e d by t h e precessional motion of t h e E a r t h ' s a x i s , Hipparchus.
also discovered
T h e s a m e h o l d s t r u e f o r t h e v a r i a b i l i t y of t h e e c l i p t i c a l
quity.
It g o e s b a c k t o g r a v i t a t i o n a l perturbations by Moon a n d Sun.
Since
all
astronomical
coordinate
systems,
including
( t h e y r e f e r t o t h e s y m m e t r y p l a n e of o u r G a l a x y ) ,
the
galactic
a r e of t h e s a m e
t h e s e s h o r t a n d r u d i m e n t a r y r e m a r k s s h o u l d be s u f f i c i e n t .
by
obli-
ones type,
1.2.
A f f i n e and P r o j e c t i v e
Geometry
15
1.2. Affine and Projective Geometry
In h i s p r i n c i p a l
work
"Elements", Euclid (Eukleides, c a .
3 0 0 BC)
develops
geometry a s a s e l f - c o n t a i n e d l o g i c a l s y s t e m of t h e o r i e s w h i c h is b u i l t on a f e w b a s i c a s s u m p t i o n s (axioms). A m o n g t h e 13 b o o k s of t h e " E l e m e n t s " m a i n l y t h e f i r s t f o u r a r e of to us.
T h e y p o s s i b l y go b a c k t o e a r l i e r
of Chios ( 2 n d h a l f
authors,
of t h e 5 t h c e n t u r y BC),
in p a r t i c u l a r
interest
Hippocrates
and a r e concerned with t h e
geo-
m e t r y in t h e (Euclidean) plane. R u l e r and c o m p a s s e s a r e t h e only a l l o w e d m e a n s to c o n s t r u c t new g e o m e t r i c objects,
a n d t h e y h a v e t o b e a p p l i e d in a d e f i n i t e w a y ( n a m e l y in o r d e r
d r a w s t r a i g h t l i n e s o r c i r c l e s of w h i c h t h e c e n t r e a n d o n e p e r i p h e r a l a r e given).
Coordinates are not introduced.
T h e c o r r e c t n e s s of a
t i o n i s u s u a l l y p r o v e n w i t h t h e h e l p of c o n g r u e n c e t h e o r e m s f o r Only R e n i
Descartes ( 1 5 9 6 - 1 6 5 0 )
quantitatively.
tried
to
characterize
points
This enabled him to d e s c r i b e g e o m e t r i c a l
to
point
constructriangles.
in t h e
plane
o b j e c t s and
their
i n t e r r e l a t i o n s n u m e r i c a l l y a n d t o r e p l a c e t h e E u c l i d e a n a r g u m e n t s by s i m p l e calculations.
Later,
were deduced.
from
this
discussed certain precursory In m o d e r n
approach
the
Cartesian coordinate systems
Though they a r e called a f t e r him,
terms,
the
Descartes himself
merely
steps.
introduction
of
Cartesian
coordinates
identifies
p o i n t s in t h e E u c l i d e a n p l a n e w i t h t h e e l e m e n t s of t h e t w o - d i m e n s i o n a l
the (Eu-
2
clidean) vector space IR . of
points
conditions,
coinciding while
Straight
with them,
circles
are
equations (for the details, In m a n y r e s p e c t s , On t h e o n e h a n d ,
cf.
lines
which
are turn
associated chapter
represented out
with
to
by t h e
be s o l u t i o n s
special
kinds
of
collection of
7).
t h i s i d e a h a s b e e n of e m i n e n t
influence to
mathematics.
t h e t r e a t m e n t of g e o m e t r i c p r o b l e m s by c o o r d i n a t e s
facilitates the constructions,
on t h e o t h e r h a n d ,
c e s s i b l e o n l y in t h e e x t e n d e d
theory.
One of t h e m i s t h e q u e s t i o n ,
w h i c h f i g u r e s in t h e p l a n e c a n be d r a w n
ruler
and compasses.
linear
quadratic
often
numerous aspects are
It r e d u c e s t o f i n d i n g out w h e t h e r
ac-
with
the coordinates
of
a l l p o i n t s t o b e c o n s t r u c t e d c a n b e c a l c u l a t e d f r o m t h e g i v e n o n e s by
solv-
ing
basic
of
linear
or
quadratic
equations,
i.e.
by
combining
a r i t h m e t i c a l o p e r a t i o n s w i t h t h e e x t r a c t i o n of s q u a r e
roots.
the
four
1. H i s t o r i c a l Development of the Coordinate Concept
16
This automatically leads to number theoretical problems. We just mention a f e w famous classical examples which have been shown to be insoluble within the framework of
Euclidean constructions:
The Delian problem (to find a
cube whose volume is twice as much as that of a given one), the trisection of an arbitrary angle and squaring the circle. the highlights in this f i e l d is the construction of the regular 17-
One of
gon by Gauß. His investigation exerted a strong impetus on the further development of modern number theory. S t i l l another expansion,
lying closer to the heart of geometry i t s e l f ,
made possible by the introduction of coordinates, (plane algebraic)
nition and classification of
was
namely the precise d e f i -
curves.
We shall
discuss
it
thoroughly in chapter 7. Here we merely state that the curves of second degree are in essence identical with the conic sections which had already been studied systematically e a r l i e r by Apollonius of Perge (ca. 260 - ca. 190 BC). Even much more important than these achievements
is the
epistemological
gain due to Descartes' formulation. The above representation t r a n s f e r s the 2 a f f i n e plane IR
into a model of the Euclidean plane.
In this way the con-
sistency problem of Euclidean geometry reduces to that of the real numbers. We do not want, however, to comment on the dispute which was kindled between
d i f f e r e n t philosophical
intuitionism, Instead,
schools within mathematics
(constructivism,
etc. ) by this question.
we p r e f e r to consider another d i f f i c u l t y which is connected
with
one of Euclid's axioms, namely the (in)famous parallel postulate. It may be stated as f o l l o w s : "For each ( s t r a i g h t ) line g and every point Ρ not coinciding with g,
there
is one and only one line h which contains Ρ and is disjoint to h.
It
is
called the p a r a l l e l to g through P. " This assumption is incomparably more complicated and much less evident than all other axioms of Euclid. This suggested that the statement just mentioned could possibly
be a consequence of the remaining postulates.
then be unnecessary,
or better:
It
would
should be considered as a geometric
theo-
rem. Every attempt to prove it f a i l e d , though in some cases this was by no means clear.
A thorough analysis,
gaps or inconsistencies.
however,
revealed f o r all published
arguments
Most of them silently used an assumption which in
some sense was expressed in simpler terms than the original parallel postulate,
but logically said the same.
1.2. A f f i n e and P r o j e c t i v e Geometry
At any rate,
17
in this way several equivalent f o r m s of the proposition under
discussion were discovered,
as f o r example the well-known theorem on the
angle sum in a planar triangle (180°). Gauß was the f i r s t who concluded the parallel axiom might perhaps be independent of the others.
In order to more clearly present the considerations
to f o l l o w , we shall work with modern terminology. In particular, we denote as absolute geometry those parts of without making use of parallels,
Euclidean theory which can be proved
and thus do not require the incriminated
postulate. The question thus can be expressed b r i e f l y as f o l l o w s : a proper part of Euclidean geometry (and if
so,
Is absolute geometry
which statements does it
include?) or are both identical with each other? Actually, dent.
the f i r s t alternative is true:
Gauß proves
the axiom of parallels is indepen-
this by seeking realizations
of
absolute geometry
for
which the parallel postulate does not hold. Without knowledge of Gauß' unpublished investigations
three decades b e f o r e ,
Nicolai
ky (1792-1856) and Jänos Bblyai (1802-1860) f o l l o w e d path.
Two
versions of
Ivanovic essentially
Lobachevsthe
this non-Euclidean geometry are conceivable
same (there
may be either infinitely many parallels to g passing through Ρ or none; cordingly one speaks of hyperbolic or e l l i p t i c geometry).
ac-
Very simple mo-
dels have been found by Christian Felix Klein (1849-1925) and Jules-Henri Poincarfe (1854-1912).
They represent
the
(hyperbolic)
non-Euclidean
plane
by a part of the Euclidean one and hence v e r i f y the consistency of the l a t ter if that of the f o r m e r is taken f o r granted. Thus the existence of Euclidean geometry implies that of its non-Euclidean counterpart (the e l l i p t i c variant can be constructed similarly). These considerations again show the immense influence of the coordinate approach on our understanding of geometry. that
the s p e c i f i c properties
of
Nevertheless it soon turned out 2 the a f f i n e plane IR involve the necessity
to discuss certain exceptions separately.
For instance,
equation of a straight line g in a Cartesian ( x , y ) - f r a m e ! in various ways,
as e . g .
let us examine the We may express it
the well-known p o i n t - s l o p e - f o r m or the intercept
f o r m ; see chapter 7. Neither is able to describe all straight lines. to the y - a x i s and its parallels; origin
The f i r s t cannot be applied
the second f a i l s if g is incident with the
(0,0).
The complication caused by this is undesirable. the
inevitable
discrimination
between
parallel
The same also applies and
intersecting
pairs
to of
18
1. H i s t o r i c a l Development of the Coordinate Concept
lines in a f f i n e geometry. To avoid both drawbacks,
Jean Victor Poncelet (1788-1867) extends the a f -
f i n e plane by a number of "ideal" of "improper" points. As we shall learn in chapter 7, each of them is incident with all lines of of
parallels,
but not with the others.
a particular
All improper points together
set form
the "infinite" (straight) line. The addition plane,
of
the ideal
and the special
instance,
elements enlarges the a f f i n e to the
cases described
above "vanish
projective
into thin air".
For
two lines which are parallel in the a f f i n e interpretation now in-
tersect at infinity. But we have achieved much more. In the p r o j e c t i v e interpretation, two a r b i t r a r y straight lines always have precisely one point in common,
exactly in
correspondence with the still valid observation that two points determine a unique connecting line. Anyhow, in the geometry of the p r o j e c t i v e plane there is a p e r f e c t symmetry between points and lines,
expressed by the duality principle of
Poncelet
(theorem 7. 1. 1). The price to be paid f o r this improvement is extremely low. sists in the necessity
It mainly con-
to use three homogeneous or projective
coordinates
instead of the two a f f i n e (Cartesian) ones. The parameters themselves are irrelevant; to
the
what counts is their proportion.
so-called
barycentric calculus
of
This idea can be traced back
August
Möbius (1790-
Ferdinand
1868). The description of
points,
lines and incidence relations
in the
projective
plane does not leave anything to be desired. Things change, however,
if we
want to consider curves of higher degree. In the p r o j e c t i v e theory they are characterized as zero sets of homogeneous polynomials ( i . e .
forms). A prob-
lem is that not every algebraic equation with real c o e f f i c i e n t s has a solution in R. To
circumvent
the
closed f i e l d , e . g .
difficulty
one
has
to
replace
R by
an
algebraically
that of the complex numbers (C). In the p r o j e c t i v e plane
over C all considerations can be carried through systematically without e x ceptions. As long as continuity arguments are irrelevant, we could construct a geometry equally well over any other algebraically tic 0,
for
instance over A (set of
addition the existence of
models of
all
closed f i e l d of
algebraic
numbers).
characteris-
This shows in
Euclidean and non-Euclidean
which d i f f e r essentially f r o m the "naive" approach.
theories
1 . 2 . A f f i n e sind P r o j e c t i v e Geometry
Of
course,
dimensional
one
neither
structures
need
nor
(planes).
19
will
restrict
the
investigation
When w r i t t e n in c o o r d i n a t e s ,
to
t h e r e is
s o l u t e l y no d i f f i c u l t y t o t r a n s f e r a l l t e r m s a n d r e l a t i o n s t o h i g h e r sional spaces. In t h e
two-
We s h a l l d i s c u s s t h i s m a t t e r a t t h e end of c h a p t e r 7.
following
period
generalizations
in d i f f e r e n t
directions
have
developed. They gave r i s e to c o m p l e t e l y new and e x t e n s i v e m a t h e m a t i c a l ciplines
ab-
dimen-
(present-day
algebraic
geometry,
braic manifolds, schemes and so on).
theory
of
Unfortunately,
have enough room to account f o r them properly.
varieties,
i. e.
been disalge-
in t h i s book w e do n o t
20
1. H i s t o r i c a l Development of the Coordinate Concept
1.3. D i f f e r e n t i a l Geometry The f i r s t attempts to develop some kind of
infinitesimal calculus go
far
back to classic antiquity. Archimedes (ca. 287-212 BC) succeeds in evaluating volumes and surface a r e as
of
simply-shaped
solid
bodies
like
spheres,
circular
cylinders
and
cones. To that end, he employs the so-called exhaustion method. Clearly these early precursors of
integration theory cannot be viewed
as
systematic procedures. The same holds true f o r the work of Johannes Kepler (1571-1630) or Bonaventura Cavalieri (1598-1647). Both, Kepler's "Faßregel" (invented to determine the volume of wine casks, thus serving a very important practical purpose),
and C a v a l i e r i ' s principle,
proposed in 1635,
are
transformations of the Archimedian contemplation to more general bodies. At about the same time, d i f f e r e n t i a l calculus is developed by several maCrucial intermediate steps are the methods to find the tan-
thematicians.
gent to a curve at a given point by Kepler
(applications to celestial
me-
de Fermat (1601-1665).
But these considerations
also
are based mostly on ad-hoc-considerations though gradually a general
con-
chanics! ) and Pierre
cept begins to emerge. The synthesis of both trends is contained in the contemporary but independent investigations of
G o t t f r i e d Wilhelm Leibniz (1646-1716) and Isaac New-
ton (1643-1727). Newton
gets his f i r s t results concerning
d i f f e r e n t i a l calculus
ver publishes anything about it. His "Principia", mechanics, style.
(or,
as he
the theory of fluxions) around 1671, but during his l i f e t i m e ne-
called it:
in which he founds modern
do not require limiting processes and argue in classical
Later,
as a counterpart,
Greek
the theory of fluentes (= integration)
fol-
lows. Although Leibniz only in 1684 comes to d i f f e r e n t i a t i o n and 1686 to the integral concept,
he quite early brings his achievements to the attention
of
the s c i e n t i f i c community. T h e r e f o r e the extremely unpleasant and disgusting priority
quarrel,
which
the f o l l o w i n g years,
is picked by some people
is groundless.
of
limited
intellect
(who at least in the beginning do not take part in the b a t t l e ) pursue tally d i f f e r e n t paths.
in
This is the more so since both authors to-
1.3.
Differential
21
Geometry
From the t e c h n i c a l
point
of
view
Leibniz'
interpretation
turned out
to
be
f a r b e t t e r because it is much more g e n e r a l l y a p p l i c a b l e than the m o r e physical
ideas
of
Newton.
differentiation
(d)
A large
part
and i n t e g r a t i o n
of
his n o t a t i o n ,
(J"),
is s t i l l
like
the
in use t o d a y .
symbols
for
This
an-
is
o t h e r r e a s o n f o r the success of his i n t e r p r e t a t i o n in comparison t o t h a t
of
his opponent. Only in the l a s t f e w decades N e w t o n ' s a n a l y s i s has seen a p a r t i a l sance"; The
some r e c e n t developments can be v i e w e d as i t s
new
calculus
is a l m o s t
hydrodynamics by Leonhard
field
(1736-1813), bi
are
Euler
Pierre
Simon
applied
(1707-1783),
as to c e l e s t i a l
and o t h e r s as w e l l latter
immediately
descendants.
physics,
especially
Bernoulli
Daniel
mechanics.
Laplace
to
Eminent
(1749-1827),
"renais-
contributors
Joseph
to
(1700-1782) to
the
Lagrange
Louis
W i l l i a m Rowan Hamilton (1805-1865) and Carl Gustav Jacob Jaco-
(1804-1851).
During t h i s p e r i o d tions
of
one r e a l
the or
investigations
complex
are
variable.
in the
justification
rentiable
or
mappings.
explanation
only
restricted
The f u n c t i o n concept
d e v e l o p e d beyond an i n t u i t i v e understanding. out
large
to
func-
itself
is
Most s c i e n t i s t s c o n s i d e r
smooth,
i.e.
infinitely
not
with-
often
diffe-
The c o n f u s i o n caused by the u n r e f l e c t e d usage o f
d e f i n e d t e r m s is c l a r i f i e d only
ill-
gradually.
We o w e the n e c e s s a r y s y s t e m a t i z a t i o n and s c h e m a t i z a t i o n of the r e l e v a n t m a thematical
objects
mainly
to
Cauchy
Augustin-Louis
(1789-1857)
and
Karl
T h e o d o r Wilhelm WeierstraS (1815-1897). It is now most i m p o r t a n t to e s t a b l i s h a t h e o r y of higher dimensional and,
in connection w i t h this,
out in d i f f e r e n t
f u n c t i o n s of
many v a r i a b l e s .
dehnungslehre"
(Graßmann [1844])
spaces in any ( f i n i t e ) On the
other
hand
proposes
in his f a m o u s " L i n e a l e
an
algebraic
Initially assumed. sential
(smooth)
an e x p l i c i t
mathematicians
develop,
which
of
Ausvector
partially
stimulated
1.4),
by
differential
surfaces. embedding
in t h r e e - d i m e n s i o n a l
Gauß is the f i r s t who f r e e s himself
tool
theory
dimension.
many
p r o b l e m s and open questions of t h e o r e t i c a l geodesy (sec. geometry of
carried
directions.
Hermann Günther GraBmann (1809-1877)
Firstly,
spaces
This is
enables
him t o
do so,
are
of
Euclidean
space
this restriction.
intrinsic
IR
The
(Gaussian)
is es-
coordi-
nates. Even if dered
external
important,
q u a n t i t i e s which depend on the embedding a r e s t i l l Gauß points out that inner
properties
have a much
consimore
1. H i s t o r i c a l Development of the Coordinate Concept
22
fundamental meaning.
This is because they can be defined and in principle
also measured within the surface i t s e l f .
(The demand of practical
bility is always clearly visible in Gauß' work).
applica-
The abstraction f r o m sur-
rounding space is a major step f o r w a r d in comparison with older theories. Finally.
Bernhard Georg Friedrich Riemann (1826-1866) puts both threads t o -
gether and builds a unified theory of functions. His definition of the word "manifold", however,
is not exactly identical with today's.
ferences is the assumption of a metric.
By and large,
One of the d i f -
his conception
cor-
responds to what we nowadays call a Riemannian space. One cannot say any more whether Riemann admits manifolds whose atlases (= collections of local coordinate systems) contain more than one chart or if he r e s t r i c t s to those spaces f o r which one global reference f r a m e s u f f i ces. The reason is that he presents his theory to the public in his inauguration lecture,
Riemann [1854].
He has to show consideration f o r the non-
mathematicians in the audience and hence avoids too technical deductions. But this is not the decisive f a c t o r .
All essential assumptions are
already
included, even if several more decades have to go by until the modern manif o l d concept emerges, mainly through contributions f r o m El win Bruno Christ o f f e l (1829-1900),
Eugenio Beltrami (1835-1900),
Ricci-Curbastro
Gregorio
(1853-1925) and Tullio Levi-Civitä (1873-1941). During this phase a tendency to increasing abstraction (e.g.
weaker
attempt
to
differentiability
discuss
all
conditions
ramifications
history up to the beginning of the
20 t h
of
etc.) the
and
is observed. development;
century, c f .
generalization We shall
not
concerning
its
Scholz [1980].
Meanwhile there is a trend towards topology, though already very early a l gebraic aspects play a leading role,
at least occasionally.
We merely men-
tion the absolute calculus of £ l i e Joseph Cartan (1869-1951). with the technical tools f o r many geometric
It endows us
investigations.
In this connection the ideas exhibited by Marius Sophus Lie (1842-1899) in a series of very influential papers on d i f f e r e n t i a l equations are of
utmost
importance to us. The basic innovation is to link up algebraic
and topological
resp.
(diffe-
r e n t i a l ) geometric structure. To this end he considers a smooth manifold on which a product is defined which obeys the group laws.
Multiplication and
inversion are required to be smooth operations. These o b j e c t are now called Lie groups. They very o f t e n come into play naturally
when
dealing
with
mathematical
branch of physics can do without them.
or
physical
problems.
Today
no
1.3. Differential Geometry
23
A l r e a d y Lie [1970] h i m s e l f n o t e s t h a t an a n a l y t i c a t l a s c o m p a t i b l e w i t h g i v e n one c a n a l w a y s be f o u n d ( c f .
chapter
3).
On t h e o t h e r
hand,
the much
w e a k e r d i f f e r e n t i a b i l i t y c o n d i t i o n s s u f f i c e . T h e c u l m i n a t i o n of t h e
accord-
i s t h e p o s i t i v e s o l u t i o n of H u b e r t ' s 5 t h Problem, s e e
section
ing r e s e a r c h 6. 7.
T h e s t a r t i n g p o i n t f o r Lie a n d h i s s u c c e s s o r s is t h e n e a r l y b i j e c t i v e t i o n s h i p b e t w e e n Lie g r o u p s and Lie algebras. As v e c t o r s p a c e s ,
the
relalatter
a r e much e a s i e r a c c e s s i b l e t h a n t h e m a n i f o l d s t h e m s e l v e s . For
example,
groups,
the
classification
w h i c h may be c o n s i d e r e d
o t h e r s c a n be c o n s t r u c t e d , bras.
of
the
as the
finite
dimensional
simple
Lie
"building blocks" f r o m which
i s r e d u c e d t o t h e a n a l o g o u s t a s k f o r Lie
all
alge-
I t s s o l u t i o n w a s a c h i e v e d by Wilhelm K a r l J o s e p h Killing (1847-1923)
and C a r t a n ( K i l l i n g [1888-1890],
Cartan
[1952-1953]).
24
1. H i s t o r i c a l Development of t h e C o o r d i n a t e Concept
1.4. G e o d e s y a n d Cartography
The m a i n p r o b l e m s of geodesy a r e t h e d e t e r m i n a t i o n of f i g u r e and
gravity
f i e l d of t h e E a r t h . After earlier tentative considerations,
t h e E a r t h m i g h t be a disk s w i m m i n g
in t h e o c e a n , Pythagoras and Aristotle ( 3 8 4 - 3 2 2 BC) p r o p o s e d i t s
spherical
s h a p e . Though a c c e p t e d by m o s t s c h o l a r s , t h i s s u g g e s t i o n w a s d i s p u t e d c o n t r o v e r s i a l l y d u r i n g t h e whole of a n t i q u i t y and well i n t o t h e Middle Ages. The f i r s t r e l e v a n t q u a n t i t a t i v e r e s u l t c o n c e r n i n g t h e s i z e of t h e E a r t h obtained
by Eratosthenes of Cyrene (ca.
284 - ca.
2 0 0 BC).
His
was
starting
p o i n t w a s t h e o b s e r v a t i o n t h a t in t h e end of June t h e s u n l i g h t w a s r e f l e c t ed in a deep f o u n t a i n n e a r Syene (Assuan).
The s o l a r
culmination
summer solstice at this place thus occurred close to the zenith, t h e same t i m e in A l e x a n d r i a t h e Sun s t o o d a b o u t 7°12',
i.e.
around
while
at
one f i f t i e t h of
the complete circle to the South. From the
approximately
known d i s t a n c e
between the two
locations
which
a m o u n t s t o c a . 5 0 0 0 s t a d i a , he deduced t h e c i r c u m f e r e n c e of t h e E a r t h t o be nearly 250000 stadia.
Within t h e p r e c i s i o n of m e a s u r e m e n t ,
he l a t e r
cor-
r e c t e d t h i s value t o 2 5 2 0 0 0 s t a d i a , g e t t i n g a r o u n d n u m b e r , namely 7 0 0 s t a dia, f o r t h e l e n g t h of one d e g r e e (Hoppe [1966]). U n f o r t u n a t e l y t h e r e a r e two obstacles to t r a n s l a t i n g this into the system. First,
metric
it is n o t known which of t h e v a r i o u s s t a d i u m s in use a t t h e
t i m e E r a t o s t h e n e s employed. F u r t h e r m o r e , we c a n n o t say any m o r e how he c a l culated the distance.
P r o b a b l y he d e r i v e d it f r o m t r a v e l t i m e s a n d / o r
road
lengths. The a c t u a l d i s t a n c e a s t h e c r o w f l i e s b e c a u s e of t h e u n a v o i d a b l e b e n d s
is
clearly considerably shorter.
If and how E r a t o s t h e n e s t o o k t h i s e f f e c t i n t o
consideration
the
and
whether
distance
of
4325
stadia
proposed
Wolf [1973] ( t h e a u t h o r does n o t i n f o r m us a b o u t t h e b a s i s of h i s t i o n s ) is b e t t e r ,
may be l e f t u n d e c i d e d .
In any c a s e ,
by
calcula-
one g e t s a v a l u e f o r
t h e E a r t h ' s c i r c u m f e r e n c e which is in a c c o r d a n c e w i t h t h e t r u e n u m b e r w i t h in 5. . . 10 %. In view of t h e p r i m i t i v e m e t h o d s which w e r e a t command t h i s is a r e m a r k a b l e
Eratosthenes'
accuracy.
It t o o k a n u m b e r of c e n t u r i e s u n t i l f u r t h e r s t e p s f o r w a r d could be made. Only in 1669 C h r i s t i a a n Huygens (1629-1695) e n r i c h e d geodesy by a f u n d a m e n -
1.4. Geodesy and Cartography
t a l l y new aspect.
25
He noted that the daily rotation of the Earth generates
centrifugal forces.
This leads to an enlargement of
(a ) in comparison with the polar
radius (a p ).
the equatorial
radius
Hence the t e r r e s t r i a l
sur-
f a c e should resemble a flattened e l l i p s o i d much more than a sphere. Newton calculated f r o m the available data an expectation value of the f l a t tening f = (a E -a p )/a E of about 1:230. He assumed the solid Earth to be p e r f e c t l y elastic. This is not a good approximation, as we know today.
There-
f o r e the real f l a t t e n i n g is less than the theoretical. To decide the controversy if the f i g u r e of the Earth is spherical (a p = a E ) or can be described better
by an oblate (a
prolate spheroid
the Paris Academy of
(a p > a^.),
ρ
< a ) or, Ε
on the contrary,
Sciences dispatched
a
two
l a r g e - s c a l e expeditions. They got order to carry through measurements of as large parts of the meridian as possible,
one in the vicinity of the equator
(Peru), the other in a northern t e r r i t o r y
(Lapland).
In the course of this, numerous points along the meridian were marked and the r e l a t i v e positions defined by triangulation.
A direct measurement of an
arbitrary triangle side yields the scale f a c t o r and consequently the length of the arc (which a f t e r w a r d s is transformed to sea-level). A comparison with the astronomically determined latitude d i f f e r e n c e between the northernmost and southernmost grid points e. g. gives the distance c o r responding to one degree. America.
In Norrbottsn this is much more than in Latin
This immediately corroborates the flattening of the Earth.
P a r t i a l l y due to lack of experience and sloppiness, many problems arose during both expeditions. A detailed and readable account of
the d i f f i c u l t i e s they had to
can be found in Wolf [1973], In this book the reduction methods,
surmount intermedi-
ate and f i n a l results are listed also. In
the
17th,
18th
initiated similar
and
early
19th
triangulations.
centuries,
several
European
sovereigns
They were aimed at constructing
of f i r s t order net points with best possible accuracy.
systems
Other less important
points were selected and added later by relative positioning. Among the many land surveyings of this kind, some are of eminent historical significance.
The f i r s t genuine triangulation is the Dutch Survey (meridian
between Alkmaar and Breda) which started in 1614 under the supervision of Willebrord Snell (Snellius,
1591-1626).
He already employed essentially
methods which became standardized later. extremely tedious and laborious, yet.
The numerical calculations
all
were
since logarithm tables were not available
1. H i s t o r i c a l Development of the Coordinate Concept
26
Snell's procedure has been applied on many other occasions as well,
so we
need not comment on them in greater detail. New ideas were brought in only much later. Mainly the Hannover Survey under GauB has to be acknowledged in this respect. are introduced in one stroke: to be mentioned. observations
Here two important
Firstly the invention of
novelties
the heliotrope has
It consists of a pivoted mirror which allows f o r
over
very
long distances by r e f l e c t e d
sunlight.
precise
In this
way
much larger triangles can be measured directly. Furthermore, the instrument serves as an optical telegraph and thus accelerates the transmission of
in-
f o r m a t i o n and orders considerably. Even much more important, however, is the consequent employment of the e x tremely useful method of least squares.
It had been discovered by Rudjer
Josip Boskovic (1711-1787) and was developed further by Adrien-Marie Legendre (1752-1833). Gauß [1823] himself brings the idea to perfection.
In com-
bination with his law of error propagation, it is the rigorous mathematical foundation of the great survey of the Kingdom of Hannover. All f i r s t order triangles
are
subjected
much higher degree of
to
the
analysis
simultaneously,
thus
achieving
a
consistency.
Friedrich Wilhelm Bessel (1784-1846) combines all available results of
sur-
veys and deduces f r o m them the closest spheroidal approximation to the true f i g u r e of the Earth (by a least-square f i t ) . Later investigations, bers.
based on improved data, yield slightly d i f f e r e n t num-
Nevertheless Bessel's spheroid is used up to the present in Western
Europe as the r e f e r e n c e surface f o r the GauB-Kriiger coordinates (see below and chapter 29), while the Eastern European countries p r e f e r the ellipsoid of
Krassowski.
Of the remaining models of the Earth which we shall discuss elsewhere in more detail,
we have to mention the so-called International Ellipsoid,
Hayford [1909]. It is also the fundament of many o f f i c i a l map systems. veral
international
organizations
like the IAU or the IUGG propose
cf. Setheir
own r e f e r e n c e systems. The exact dimensions can be looked up in chapter 29 Since,
caused by topographic peculiarities,
strongly f r o m any spheroid,
the t e r r e s t r i a l
surface d i f f e r s
it is not at all important which of the above-
mentioned surfaces is taken as the basis f o r cartography and positioning. Much smoother are the surfaces of constant total potential U of the Earth. U is the sum of a purely gravitational part U
and a component U
ing
by
9
the
centrifugal
acceleration
Laplace [1799] finds that U
produced
the
diurnal
obeys the potential equation AU
ζ
describrotation.
= 0 and can
27
1 . 4 . Geodesy and C a r t o g r a p h y
be c o n v e n i e n t l y
developed w i t h r e s p e c t
to a suitably
chosen
base
of
the
s p a c e of spherical harmonics ( c h a p t e r 23). The f i r s t 10 or 15 o r d e r s (a f e w h u n d r e d t e r m s ) of t h i s s e r i e s can be d e duced r e l i a b l y quantify the
f r o m t h e o r b i t s of large-scale
structure
artificial of
satellites.
the gravitational
All t o g e t h e r
they
field.
per-
Local
t u r b a t i o n s which depend on i r r e g u l a r i t i e s of t h e m a t t e r d i s t r i b u t i o n in t h e Earth's crust
could be seen only in much h i g h e r o r d e r s .
In l i m i t e d
areas
(most f a m o u s e x a m p l e : t h e n o r t h e r n slope of t h e H a r z Mountains) plumb line deviations of 10" and m o r e f r o m t h e " e x p e c t e d " value a r e q u i t e common. One of
t h e equipotential surfaces U = c o n s t . , namely t h e p a r t i c u l a r
passing through a well-defined point,
is known as t h e geoid. In t h i s
n e c t i o n gravimetric measurements a r e a l s o of g r e a t
one con-
importance.
A l t i t u d e d e t e r m i n a t i o n s a r e c a r r i e d out r e l a t i v e t o t h e geoid a l o n g t h e
lo-
cal U - g r a d i e n t .
fix
U n f o r t u n a t e l y t h e r e is no u n i v e r s a l a g r e e m e n t how t o
the zero height either.
The s t a t e s of W e s t e r n E u r o p e r e f e r t o an
average
w a t e r - l e v e l in t h e p o r t of A m s t e r d a m ( c l e a r e d f r o m t i d a l and m e t e o r o l o g i c a l e f f e c t s ) . On t h e o t h e r hand, E a s t e r n E u r o p e a n c o u n t r i e s u s e t h e
Kronstadt
g a u g e . The l a t t e r is a b o u t 2 dm h i g h e r t h a n t h e f o r m e r . In Germany t h e z e r o level (Normalnull, NN) by c o n v e n t i o n lies e x a c t l y 37 m below a c e r t a i n m a r k n e a r B e r l i n ("Normalhöhenpunkt"), and so on. Until r e c e n t l y t h e t r i g o n o m e t r i c p r o c e d u r e w a s t h e only way t o a c h i e v e a p o s i t i o n i n g a c c u r a c y of c e n t i m e t r e s over d i s t a n c e s of s e v e r a l h u n d r e d metres.
(The t y p i c a l s i d e l e n g t h s of t r i a n g l e s of f i r s t o r d e r a r e c a .
kilo30 to
60 km). In t h e l a s t f e w y e a r s new and t o t a l l y d i f f e r e n t m e t h o d s w o r k i n g t o a s i m i l a r d e g r e e of p r e c i s i o n have been f o u n d . We j u s t m e n t i o n in p a s s i n g t h e d i rect
determination
of
distances
with
laser
light
which is r e f l e c t e d
at
a
p r i s m a t i c m i r r o r . From the precisely measurable travel time f r o m the source t o t h e r e f l e c t o r and back a g a i n ,
one i m m e d i a t e l y g e t s t h e d i s t a n c e of
both
points. Thus i n s t e a d of t r i a n g u l a t i o n s ,
trilaterations can be used a s w e l l . The a c -
c u r a c y is n o t l i m i t e d by t h e clock, but by t h e unknown v a r i a t i o n of t h e r e f r a c t i o n i n d e x of t h e a i r which can only be e s t i m a t e d .
Uncontrollable
ef-
f e c t s of t h e w e a t h e r make t h e c a l c u l a t i o n of refraction and t h e velocity of light w i t h i n t h e a t m o s p h e r e d i f f i c u l t . The s a m e h o l d s t r u e f o r t h e l a s e r r e f l e c t o r s which have been e s t a b l i s h e d on t h e Moon d u r i n g s e v e r a l s p a c e m i s s i o n s . a n c e up t o a f e w cm. Here, h o w e v e r ,
They a l s o a l l o w t o f i n d t h e
beyond t h e a b o v e - m e n t i o n e d
dist-
problems
1. H i s t o r i c a l Development of the Coordinate Concept
28
The classical Lunar theories are at best
t h e r e is one more obstacle.
to p r e d i c t the Moon's position with an e r r o r of about 1 m, w h i l e i n t e g r a t i o n which takes r e l a t i v i s t i c time
intervals)
a little
better.
forces
into account does
Nevertheless
there
is
(over
still
able
numerical limited
a gaping
of one or t w o o r d e r s of magnitude between observational and
hole
calculational
accuracy. Radar ranging b a s i c a l l y
applies the same idea.
system in t w o d i f f e r e n t ways:
It is used in the
planetary
e i t h e r the signal is passively r e f l e c t e d by a
s u f f i c i e n t l y l a r g e body (Moon, Mars, Venus), or one employs a s p a c e c r a f t as a m p l i f i e r and a c t i v e
transponder.
satellite geodesy has undergone a rapid development. A f e w dozen
Recently,
time signal t r a n s m i t t e r s in e x a c t l y known Earth o r b i t s serve as a standard. If f o r an o b s e r v e r on the ground three or more of them are above the h o r i zon at the same time,
it is possible to deduce the exact position f r o m the
a r r i v a l times of the r a d i o
waves.
The method c a p t i v a t e s because of
its s i m p l i c i t y which makes it possible
f i n d o n e ' s position even in rough t e r r a i n (mountains etc. ) or r e g i o n s out f i x e d landmarks (deserts, maintenance
of
a score
it is not inexpensive.
of
Greenland,
satellites
Antarctica).
with
to
with-
Since it r e q u i r e s the
high-precision
(atomic)
clocks,
The best-known network of this kind is probably
the
Global Positioning System (GPS). Another
high-precision
positions cm,
of
large-size
technique, radio
which
telescopes
allows
to
determine
the
on d i f f e r e n t continents
is the s o - c a l l e d Very Long Baseline Interferometry CVLBI).
It consists
in simultaneously
observing
the radiation
g a l a c t i c ) r a d i o source f r o m various sites. on tape,
all
stations numerically
of
The received
t o g e t h e r with accurate time marks.
r e s u l t s of
Later,
a distant signal
is
its p r o j e c t i o n onto the line of
The r e s o l u t i o n
of
all
and calculate the i n t e r f e r e n c e
recorded
effects. or,
ra-
sight.
these procedures
i n e v i t a b l e t o take the temporal
(extra-
one has to compare the
These y i e l d the baseline length between each pair of o b s e r v a t o r i e s ther,
relative
up to ± 1
is e x t r e m e l y
variability
of
high.
It
the t e r r e s t r i a l
is
therefore
surface
into
consideration. The most important periodic e f f e c t s are tidal forces e x e r t e d by Moon and Sun.
They are p e r c e p t i b l e
even in the " s o l i d " Earth,
to t i d e s with amplitudes of a f e w decimetres. tion of the involved heavenly bodies,
where they g i v e
rise
Since they depend on the mo-
they may be eliminated
numerically.
F u r t h e r m o r e , secular e f f e c t s are d i r e c t l y observable nowadays. The most i m -
29
1 . 4 . Geodesy and C a r t o g r a p h y
p r e s s i v e a m o n g t h e m is t h e continental drift p r e d i c t e d by A l f r e d Wegener (1880-1930). theory
Though
his
(Wegener [1915]),
professional in i t s
colleagues
modern
disguise
at
first
ridiculed
this
a s plate tectonics it
is
one of t h e b e s t - e s t a b l i s h e d and m o s t a c t i v e b r a n c h e s of geophysics. The i n d i v i d u a l l i t h o s p h e r i c p l a t e s move w i t h r e s p e c t t o each o t h e r w i t h t y p i c a l v e l o c i t i e s of a f e w c m / y e a r .
Over t i m e i n t e r v a l s of a d e c a d e o r
t h i s s u m s up t o q u i t e l a r g e a m o u n t s .
so,
In c e r t a i n p l a c e s even 10 c m / y e a r
oc-
cur. T h i s is m o r e t h a n one m i g h t t h i n k . At t h e b o r d e r of t w o n e i g h b o u r i n g p l a t e s the
effect
can
occasionally
be
made
visible.
For
instance,
the
western
p a r t s of I c e l a n d b e l o n g t o t h e N o r t h American, t h e e a s t e r n p a r t s t o t h e E u rasian plate.
Both d r i f t a w a y f r o m each o t h e r by ca.
2 cm a y e a r .
Simple
m a r k i n g s on p a r a p e t s of b r i d g e s and s u c h l i k e (shown on s i g h t s e e i n g a r e s u i t a b l e t o make t h e a c c u m u l a t e d m o t i o n e a s i l y
tours)
discernible.
The r e q u i r e m e n t s of geodesy lead Gauß to w r i t e a s e r i e s of p a p e r s on t h e theory of surfaces.
He i n t e n t i o n a l l y
starts
out
from a totally
different
p o i n t t h a n a l l h i s p r e d e c e s s o r s did. 3 it w a s g e n e r a l l y a s s u m e d t h a t a s u r f a c e 3 is embedded in IR .
B e f o r e Gauß,
It is d e f i n e d a s t h e s e t of a l l s o l u t i o n s of an e q u a t i o n like F ( x , y , z) = 0. Here F d e n o t e s an a r b i t r a r y s m o o t h f u n c t i o n of t h e t h r e e C a r t e s i a n
coordi-
nates. In c o n t r a s t t o t h i s , Gauß n o t e s t h a t g e o d e t i c m e a s u r e m e n t s as e. g.
triangu-
l a t i o n s a r e r e s t r i c t e d t o t h e s u r f a c e of t h e E a r t h or i t s i m m e d i a t e ty.
For t h a t
vicini-
r e a s o n he p u t s t h e e m p h a s i s of his s u r f a c e t h e o r y on
p r o p e r t i e s which do n o t depend on t h e s u r r o u n d i n g s p a c e .
He,
unlike
those Rie-
mann, n e v e r t h e l e s s h o l d s t o t h e e x i s t e n c e of an embedding. Gauß' a p p r o a c h s t a r t s out f r o m a p a r a m e t r i z a t i o n of t h e s u r f a c e g w h i c h , r e q u i r e s two inner o r Gaussian co-
corresponding to its two-dimensionality, ordinates u , v .
The i m p l i c i t
relation
F(x,y,z) = 0 thus
f o r the surface points: assumed t h a t rentiable).
χ = x(u,v);
all f u n c t i o n s x , y , z
is r e p l a c e d
y = y(u,v);
are smooth
S o m e t i m e s only p a r t s of 3
can
by e x p l i c i t
ζ = z(u, ν).
(i.e.
formulas
It is
silently
infinitely often
be d e s c r i b e d in t h i s w a y .
diffeThen
m i n o r m o d i f i c a t i o n s a r e n e c e s s a r y , but they do not c a u s e any s e r i o u s d i f f i culty. Just
this
mainly
formulation
initiated
dimensional case.
by
opens
the
Riemann.
road Gauß
to
differential
himself
only
geometry which
considers
the
is
two-
30
1. H i s t o r i c a l Development of the Coordinate Concept
R e s t r i c t i n g the Euclidean m e t r i c of the embedding s p a c e R 3 to g, we g e t an e x p r e s s i o n f o r the s q u a r e of the a r c element, namely d s 2 = d x 2 + dy 2 + dz Z . If we s u b s t i t u t e the defining equations of x , y , z ,
this b o i l s down to a ho-
mogeneous second d e g r e e polynomial in the c o o r d i n a t e d i f f e r e n t i a l s du and St dv, the 1 fundamental form of 3· According to the d e f i n i t i o n , it is p o s i tive d e f i n i t e and inter a l i a a l l o w s to deduce a v a r i a t i o n a l
condition
for
t h e geodesies o n g .
The sectional curvatures at a c e r t a i n point of g lead by a theorem of J e a n Frederic
Frenet
(1816-1900)
to
another
(possible
indefinite)
quadratic
f o r m , which is known a s the 2 n d fundamental form of g. As a m e a s u r e of the deviation of the s u r f a c e f r o m i t s t a n g e n t i a l plane we may use the Gaussian curvature, i. e. the r a t i o of the determinant of
the
second f u n d a m e n t a l f o r m to t h a t of the f i r s t . The importance of this discovery
cannot be o v e r e s t i m a t e d .
For the
first
time we have found an a b s o l u t e invariant of the s u r f a c e (a quantity which is independent of the choice of the Gaussian c o o r d i n a t e s ) . Probably the most i n f l u e n t i a l r e s u l t of s u r f a c e theory is the f a m o u s theorems egregium, see Gauß [1827], surface
can be c a l c u l a t e d
It b a s i c a l l y s t a t e s t h a t the c u r v a t u r e of
f r o m the m e t r i c and i t s f i r s t
two
a
derivatives
with r e s p e c t to the i n t r i n s i c c o o r d i n a t e s u, v. An i n t e r e s t i n g c o r o l l a r y is t h a t two s u r f a c e s g and S can be developed on (smoothly and i s o m e t r i c a l l y t r a n s f o r m e d into) each other only if the G a u s s ian c u r v a t u r e s in c o r r e s p o n d i n g points of g and δ have the same v a l u e s . This l e a d s to an isometric = length-preserving b i j e c t i o n between g and It
is
simultaneously
angle-preserving
(conformal) a n d
area-preserving.
The
r e v e r s e of the l a s t p r o p o s i t i o n is a l s o t r u e , c f . c h a p t e r s 3 and 4. Already one year e a r l i e r , (Gaussian) curvature,
Gauß found a f o r m u l a f o r the i n t e g r a l
of
the
extended over a t r i a n g l e in g which is bounded by g e -
odesies. Following c o n t e m p o r a r y r o m a n t i c Zeitgeist, he c a l l s it, somewhat b o m b a s t i cally,
"theorema elegantissimum". Its a p p l i c a t i o n s go f a r beyond the r e a l m s
of geodesy and d i f f e r e n t i a l geometry. The integral curvature of a c l o s e d s u r f a c e g, namely the i n t e g r a l over g of the G a u s s i a n c u r v a t u r e ,
can be reduced to this by t r i a n g u l a t i o n .
The o u t -
come is an e x p r e s s i o n which contains only topological invariants like the Euler number o r t h e genus.
This f i r s t connection between local and global a s p e c t s has not been published
by Gauß (except
for
some r e m a r k s
in his l e t t e r s
to
several
col-
1.4. Geodesy and C a r t o g r a p h y
leagues),
31
such t h a t it h a s t o be i n d e p e n d e n t l y f o u n d and p r o v e d a g a i n much
l a t e r by P i e r r e Ossian Bonnet (1819-1892). This r e s u l t , differential
since t h e n known a s t h e theorem of GauB-Bonnet, h a s geometry
like n o t h i n g
else.
Many e x t e n s i o n s
and
shaped
generaliza-
t i o n s up t o t h e p r e s e n t day e n r i c h e d algebraic topology e n o r m o u s l y . But l e t us r e t u r n t o t h e main t o p i c of o u r d i s c u s s i o n !
Often a
pictorial
r e p r e s e n t a t i o n w i t h t h e help of maps is m o s t a p p r o p r i a t e t o i l l u s t r a t e
po-
sition-dependent
de-
tails.
data.
In p a r t i c u l a r ,
this holds f o r local
geographic
A chart (in g e o g r a p h y ) is a o n e - t o - o n e mapping f r o m (some p a r t o f ) 2
t h e E a r t h ' s s u r f a c e o n t o a r e g i o n w i t h i n t h e E u c l i d e a n p l a n e R . The l a t t e r may be t h o u g h t of as a s h e e t of p a p e r o r s o m e t h i n g s i m i l a r . We d e m a n d t h e a s s i g n m e n t t o be i n f i n i t e l y o f t e n d i f f e r e n t i a b l e (in b o t h d i r e c t i o n s ) . Obviously,
there
are
countless
possibilities
to
obey
these
T h e r e a r e no g e n e r a l r u l e s which p r o c e d u r e is t h e b e s t .
conditions.
Depending on t h e
p u r p o s e one m u s t a p p l y d i f f e r e n t c r i t e r i a . We t h e r e f o r e have t o be s o m e w h a t vague and r e s t r i c t o u r s e l v e s t o a couple of p r i n c i p a l m a t h e m a t i c a l
remarks.
F o r t o p o l o g i c a l r e a s o n s a s i n g u l a r i t y - f r e e r e p r e s e n t a t i o n of t h e w h o l e t e r r e s t r i a l s u r f a c e by a s i n g l e c h a r t does n o t e x i s t ,
even if we m e r e l y
requi-
r e c o n t i n u i t y of t h e map and i t s i n v e r s e . F o r t h e moment we a s s u m e t h e E a r t h t o be s p h e r i c a l . The ( G a u s s i a n ) c u r v a t u r e of i t s s u r f a c e is e v e r y w h e r e p o s i t i v e ,
but t h a t of t h e p l a n e i s z e r o .
On a c c o u n t of t h e t h e o r e m a e g r e g i u m i t is i m p o s s i b l e t o c o n s t r u c t a c h a r t which m a p s a f i n i t e s e c t i o n of t h e s p h e r e into t h e p l a n e .
Every c h a r t
of
t h e E a r t h t h u s i n e v i t a b l y is s p o i l t w i t h distortion. N e i t h e r can we a c h i e v e c o n f o r m i t y and a r e a - p r e s e r v a t i o n a t t h e same t i m e , s i n c e t h i s would be e q u i v a l e n t t o i s o m e t r y .
On t h e o t h e r hand,
i t is well
p o s s i b l e t o f u l f i l l one of t h e s e t w o c o n d i t i o n s . Of t h e n u m e r o u s c o n f o r m a l m a p s we h e r e j u s t s k e t c h t h e t w o most important;
historically
f o r more i n f o r m a t i o n see Hoschek [19691.
The s i m p l e s t m a p p i n g p r o b a b l y is t h e stereographic projection. It h a s been d e s c r i b e d a l r e a d y by Hipparchus but may be of even o l d e r o r i g i n . by c h o o s i n g t h e
equatorial
p l a n e a s t h e image r e g i o n
and
We g e t i t
projecting
all
p o i n t s on t h e E a r t h in p e r s p e c t i v e ( f r o m t h e S o u t h p o l e ) . We s h a l l d i s c u s s t h i s t r a n s f o r m a t i o n in g r e a t e r d e t a i l in c h a p t e r 7 and a p ply i t a f t e r w a r d s t o a l a r g e number of g e o m e t r i c p r o b l e m s . Much of i t s i m p o r t a n c e s t e m s f r o m t h e f a c t t h a t it maps c i r c l e s t o c i r c l e s . B e t t e r s u i t e d f o r n a v i g a t i o n is a n o t h e r c h a r t which g o e s back t o
Gerardus
Mercator ( G e r h a r d
which
Kremer,
1512-1594).
He f i r s t
uses
a cylinder
is
32
1. H i s t o r i c a l Development of t h e C o o r d i n a t e Concept
t a n g e n t i a l t o t h e E a r t h a t t h e e q u a t o r a s an i n t e r m e d i a t e i m a g e s u r f a c e on w h i c h he p r o j e c t s f r o m t h e c e n t r e of t h e E a r t h . r o l l e d o u t in t h e
T h e r e a f t e r the cylinder
is
plane.
T h i s Mercator projection i s c o n f o r m a l
as well,
but
is r e g a r d e d
higher
s a i l o r s b e c a u s e i t h a s t h e a d v a n t a g e t h a t t h e c i r c l e s of l o n g i t u d e a n d i t u d e a r e r e p r e s e n t e d by o r t h o g o n a l s y s t e m s of s t r a i g h t l i n e s . f i e s to mark out the course.
The two t e r r e s t r i a l poles,
This
however,
by lat-
simpli-
are shift-
ed t o i n f i n i t y ; h e n c e t h e u s e f u l n e s s of M e r c a t o r ' s m a p p i n g in t h e
(ant)arc-
tic sea is limited. As an e x a m p l e of an a r e a - p r e s e r v i n g m a p we only m e n t i o n t h e one p r o p o s e d by J o h a n n H e i n r i c h Lambert (1728-1777) w h i c h p r o j e c t s r a d i a l l y gent plane at the North pole.
L i k e a l l c h a r t s of t h i s k i n d ,
onto
the
tan-
it s u f f e r s f r o m
t h e d r a w b a c k of e x t r e m e l y h e a v y d i s t o r t i o n s in t h e m a r g i n a l r e g i o n s .
It w a s
a l s o k n o w n a l r e a d y in a n t i q u i t y . To r e d u c e t h i s e f f e c t ,
over t h e c e n t u r i e s many s u g g e s t i o n s have been
which we a r e not going to discuss
made
here.
In a d d i t i o n t o t h e a r e a - o r a n g l e - p r e s e r v i n g m a p s t h e r e a r e o t h e r s w h i c h do not
possess
any
of
the
properties
but
minimize
the
total
distortion
(in
some sense). The a s s u m p t i o n t h a t t h e E a r t h is s p h e r i c a l s u f f i c e s f o r most g e o g r a p h i c a p plications.
To t a b u l a t e
the
coordinates
of
the
fundamental
trigonometric
p o i n t s of h i g h - p r e c i s i o n s u r v e y s on t h e o t h e r h a n d r e q u i r e s a c l o s e r x i m a t i o n t o t h e t r u e s h a p e of o u r p l a n e t . t o be a c c o u n t e d f o r . cuss
briefly
some
of
In p a r t i c u l a r ,
appro-
the oblateness
has
At t h e end of t h i s o v e r v i e w w e t h e r e f o r e w a n t t o d i s the
projections
which
are
apt
to
chart
(flattened)
spheroids. T h e m o s t o b v i o u s i d e a of c o u r s e i s t o u s e t w o a n g l e s , ographic coordinates. is r o t a t i o n a l l y
symmetric
around
the geographic
longitude,
b u t on a c c o u n t of
its figure axis,
ans t h e r e are various possibilities f o r the The angle a t plane
and
the centre
the
in a n a l o g y t o t h e g e -
S i n c e t h e E a r t h in t h e p r e s e n t g e n e r a l i z e d c a s e
direction
of t h e E a r t h to
the
point
since it c a n n o t be m e a s u r e d d i r e c t l y .
the
to
keep
meridi-
latitude.
which to
we a r e a l l o w e d
t h e e l l i p t i c i t y of
still
be
is s p a n n e d represented
by t h e is
equatorial
not
suitable
It i s b e t t e r t o t a k e t h e elevation of
the (North) pole. This choice is n o r m a l l y p r e f e r r e d to o t h e r c a n d i d a t e s , t r a t e on i t . nent
s o we s h a l l
A common f e a t u r e w i t h t h e s p h e r i c a l c o o r d i n a t e s is t h e
necessity
to
calculate
trigonometric
functions.
This renders
concenpermathe
ap-
1.4. Geodesy and Cartography
plication laborious.
33
T h e r e f o r e Söldner c o n s t r u c t s a c h a r t which
t h e c i r c l e s of l o n g i t u d e and l a t i t u d e by v e r t i c a l
and h o r i z o n t a l
spectively.
isometrically.
E q u a t o r and all m e r i d i a n s a r e mapped
represents lines,
The Söldner coordinates d e f i n e d t h a t w a y u n f o r t u n a t e l y l a c k c o n f o r m i t y . order to overcome this defect,
Gauß m o d i f i e s t h e idea.
central meridian as a reference basis. rudimentary; scripts.
He d i s t i n g u i s h e s
His i n v e s t i g a t i o n s ,
however,
re-
In a
remain
o n l y K r ü g e r [1912] c l o s e s s o m e g a p s in G a u ß ' u n p u b l i s h e d m a n u -
He e s t a b l i s h e s a c o m p r e h e n s i v e t h e o r y of t h i s transversal Mercator
projection.
Mainly
in t h e
German
literature
one u s u a l l y
speaks
of
Gaue-
Krüger coordinates. T h e y a r e t h e f u n d a m e n t of many o f f i c i a l s u r v e y s . We d e v o t e c h a p t e r 29 of t h i s book t o a c o m p r e h e n s i v e a n a l y s i s of t h i s t o p i c , c l u d i n g a d e t a i l e d a c c o u n t of a l l r e l e v a n t
formulas.
in-
34
1. H i s t o r i c a l Development of t h e C o o r d i n a t e Concept
1 . 5 . Special Coordinates
E v e r y c h a r t of a m a n i f o l d d e f i n e s a c o o r d i n a t e s y s t e m .
The s a m e h o l d s t r u e
f o r a n y d i f f e o m o r p h i s m b e t w e e n t w o r e g i o n s of a E u c l i d e a n s p a c e . T h e v a r i e ty
of
possible
reference
haustive case-by-case
frames thus
investigation
tion seems d i f f i c u l t .
is p o t e n t i a l l y
infinite,
cannot be achieved.
Nevertheless
a few general
and
Even a
rules exist
b e o b e y e d w h e n e v e r one w a n t s t o c h o o s e a s u i t a b l e
an
which
should
parametrization:
(1) T h e invariance group of t h e c o o r d i n a t e s m u s t c o r r e s p o n d t o t h e t r i e s of t h e p r o b l e m . labour,
T h i s u s u a l l y r e s u l t s in a c o n s i d e r a b l e
since the calculation
of u n i n t e r e s t i n g
ex-
classifica-
or needless
symme-
saving
of
information
is avoided. If, f o r example, well
as the
a d i f f e r e n t i a l e q u a t i o n has to be i n t e g r a t e d which,
initial
about some axis,
or boundary
conditions,
is r o t a t i o n a l l y
i t is a l m o s t a l w a y s a d v i s a b l e t o c h o o s e t h e
angle with respect
to the axis
permits a dimensional
a s one of
the coordinates.
as
symmetric azimuthal This
often
reduction.
On a c c o u n t of Noether's theorem a s i m i l a r a r g u m e n t a p p l i e s if t h e r e a Lie g r o u p of s y m m e t r i e s ;
(2) A n o t h e r r u l e of t h u m b i s t o p r e f e r t h e i s an a l t e r n a t i v e .
is
c f . c h a p t e r 6.
The r e l e v a n t
criterion
"simpler" coordinates of
complexity,
p e n d s on t h e k i n d of q u e s t i o n one w a n t s t o a n s w e r .
if
there
however,
de-
T h e r e f o r e we s h a l l
be c o n t e n t w i t h t h i s somewhat vague f o r m u l a t i o n . U s u a l l y orthogonal c o o r d i n a t e s a r e f o u n d m o r e c o n v e n i e n t t h a n
non-or-
b e c a u s e metric and Christoffel symbols ( a n d a
fortiori
thogonal
ones,
the standard differential operators grad, f e w e r nonvanishing components.
rot,
div a n d Δ a l s o )
contain
But in low d i m e n s i o n a l s p a c e s t h e
f e r e n c e is r e l a t i v e l y u n i m p o r t a n t .
It g r o w s o n l y s l o w l y w i t h t h e
dif-
dimen-
sion. T h e a d d i t i o n a l a m o u n t of l a b o u r c a u s e d by t h e l a r g e r n u m b e r of
terms
sometimes
given
problem.
is c o m p e n s a t e d
by a m o r e s u i t a b l e d e s c r i p t i o n
of t h e
In t h i s c a s e n o n - o r t h o g o n a l f r a m e s c a n a l s o b e c o m e u s e f u l in
practise. An i n s t r u c t i v e e x a m p l e a r e t h e Weyl coordinates w h i c h we s h a l l t h o r o u g h l y in p a r t s D a n d G of t h i s book.
They a r e a p p l i c a b l e
discuss success-
1.5. Special
fully
Coordinates
whenever
group.
we
35
are
provided
with
a
(nontrivial)
finite
symmetry
T h e y a r e i n t i m a t e l y l i n k e d w i t h m o d e r n numerical integration me-
thods, e s p e c i a l l y f o r s p h e r i c a l
domains,
w h i c h w i l l o c c u p y u s in
part
F. In s p i t e
of
these
f i c a t i o n of
d i f f i c u l t i e s which
the conceivable
v a r i o u s c a t e g o r i e s can be c a r r i e d A natural above,
are
coordinates,
obstacles at
least
to
a systematic
classi-
a rough subdivision
into
out.
s p l i t - u p h a s been mentioned a l r e a d y .
For the r e a s o n s
o r t h o g o n a l c o o r d i n a t e s y s t e m s (i. e. t h o s e w i t h m u t u a l l y
l a r axes) a r e used a l m o s t exclusively.
discussed
perpendicu-
As w e s h a l l d e m o n s t r a t e ,
this is o f -
t e n a c o n s e q u e n c e of h i s t o r i c a l d e v e l o p m e n t s o r p e r s o n a l t a s t e a n d n o t
al-
ways objectively justified. Besides t h a t ,
o t h e r o r d e r i n g s c h e m e s a r e of i n t e r e s t .
It i s n a t u r a l t o
dis-
c r i m i n a t e b e t w e e n algebraic a n d transcendental s y s t e m s in IRn, a c c o r d i n g w h e t h e r t h e c o o r d i n a t e s a r e a l g e b r a i c f u n c t i o n s of t h e s t a n d a r d components or not.
The f o r m e r a r e n o r m a l l y e a s i e r t o h a n d l e b e c a u s e
need not evaluate complicated expressions. l e r a t i o n of t h e r e q u i r e d tems,
calculations.
This is connected with an
one
acce-
A p a r t f r o m Cartesian a n d Weyl
t h e confocal coordinates and many o t h e r s b e l o n g t o t h i s
to
Cartesian
sys-
class.
F o r t h e a b o v e - m e n t i o n e d r e a s o n s g e n e r a l l y v a l i d r e c o m m e n d a t i o n s c a n n o t be given; way. her
t h u s in t h e f o r m u l a r y ( p a r t H) a l l s y s t e m s a r e p r e s e n t e d in t h e s a m e T h i s i s i n t e n d e d t o m a k e it e a s i e r f o r t h e r e a d e r t o d e c i d e on h i s o r
own.
T h e c r i t e r i o n f o r t h e i n c l u s i o n in t h e l i s t w a s t h e a p p l i c a b i l i t y
to a
lar-
g e r n u m b e r of p r o b l e m s . T h i s a d m i t t e d l y i s a n u n s y s t e m a t i c s e l e c t i o n m e t h o d i n f l u e n c e d by p e r s o n a l
bias.
In t h e l i t e r a t u r e m o s t l y t h e o r t h o g o n a l c o o r d i n a t e s a r e i m p r o p e r l y r e d . In p r a c t i s e ,
prefer-
t h e y a r e much m o r e i m p o r t a n t t h a n t h e o t h e r s . T h i s i s why
t h e y d o m i n a t e t h e t a b l e s in p a r t H. But in a d d i t i o n a f e w m o r e
parametriza-
t i o n s h a v e b e e n i n c l u d e d w h i c h a r e e x p e c t e d t o be u s e f u l in f u t u r e
applica-
tions. As i n d i c a t e d
above,
i t is i m p o s s i b l e t o d i s c u s s
v e c o o r d i n a t e s t o a n y d e g r e e of c o m p l e t i o n .
all potentially
Nevertheless the
constructi-
comparatively
f e w s y s t e m s w h i c h a r e e m p l o y e d f r e q u e n t l y h a v e a n u m b e r of f e a t u r e s in c o m mon. They a r e d i s t i n g u i s h e d m a t h e m a t i c a l l y by c e r t a i n s y m m e t r i e s .
The
majority
of t h e m h a s b e e n d i s c o v e r e d o r i n v e n t e d a l r e a d y a l o n g t i m e a g o , s o m e e v e n in a n t i q u i t y .
Of c o u r s e ,
the argument
is a l s o t r u e
in t h e o p p o s i t e
direc-
36
1. H i s t o r i c a l Development of t h e C o o r d i n a t e Concept
t i o n : One u s u a l l y w a n t s t o m a n a g e w i t h t h e b e t t e r - k n o w n m e t h o d s b e f o r e one t r i e s s o m e t h i n g new. Among t h e o l d e s t n u m e r i c a l d e s c r i p t i o n s of p o i n t p o s i t i o n s a r e t h e Cartesian a s w e l l a s t h e polar coordinates. We p r e s e n t e d t h e i r h i s t o r y a l r e a d y in the
foregoing 3
space R
sections.
into a direct
If
we
(in m o d e r n 1 2
sum R ®R
terms)
split
and use t h e c a n o n i c a l
the
configuration
parameter
on
the
f i r s t summand, b u t p o l a r c o o r d i n a t e s on t h e second, we g e t an i n t e r m e d i a t e f o r m , n a m e l y cylinder coordinates. Other decompositions systems of s i m i l a r
can be used e q u a l l y w e l l ,
shape.
thus getting
m o r e hybrid
T h e i r e x p l i c i t c o n s t r u c t i o n c a u s e s no
problems,
n e i t h e r does t h e c a l c u l a t i o n of t h e m e t r i c c o m p o n e n t s , t h e C h r i s t o f f e l s y m bols,
e t c . We n e v e r t h e l e s s included s e v e r a l of t h e m in t h e t a b l e s .
All m e n t i o n e d c o o r d i n a t e f r a m e s have,
aside f r o m orthogonality,
one m o r e
common f e a t u r e which we should e m p h a s i z e . A l o t of i m p o r t a n t p h y s i c a l p r o b lems lead t o o r d i n a r y
or p a r t i a l
differential equations
of
second
order,
e . g . t h e m o t i o n of p o i n t m a s s e s in N e w t o n i a n m e c h a n i c s o r t h e t e m p o r a l d e v e l o p m e n t of f l u i d s in h y d r o d y n a m i c s (s. c h a p t e r 5). If
the
coefficient
differential
matrix
equation
of
the
is d e f i n i t e ,
highest
order
semidefinite
derivatives
or
of
indefinite,
a
partial
one
respec-
t i v e l y s p e a k s of e l l i p t i c , p a r a b o l i c or h y p e r b o l i c p r o b l e m s . The
simplest
elliptic
differential
equation
of
second
degree
is
the
one
which is obeyed by Newton's gravitational potential U, namely AU = 0. Here Δ d e n o t e s t h e Laplacian. It is d i s c u s s e d a t l e n g t h in c h a p t e r s 3, 22, and 23.
A c a r e f u l a n a l y s i s of t h o s e
(orthogonal)
coordinates
in R
which a d m i t
a
separation of t h e j u s t d e f i n e d potential equation or p o s s i b l y even t h e much m o r e g e n e r a l Hamilton-Jacobi equation ( c f . c h a p t e r 5) by p r o d u c t s o r s u m s g o e s back t o S t ä c k e l .
A d e t a i l e d a c c o u n t of h i s r e s u l t s f o l l o w s in c h a p t e r
27.
E x c e p t f o r t h e a b o v e - m e n t i o n e d C a r t e s i a n , s p h e r i c a l and c y l i n d r i c a l f r a m e s , t h e l a r g e and p r a c t i c a l l y e x t r e m e l y i m p o r t a n t f a m i l y of Stäckel coordinates contains several other well-known types,
among t h e m ellipsoidal, conical
a n d torus coordinates.
A d i f f e r e n t , o c c a s i o n a l l y a p p l i e d method t o c o n s t r u c t new o r t h o g o n a l f r a m e s is t o c h o o s e an a r b i t r a r y s m o o t h r e a l f u n c t i o n on t h e domain one w a n t s t o represent as the f i r s t coordinate. priately. rarely,
A f t e r w a r d s it h a s t o be amended
One of many e x a m p l e s a r e t h e Roche coordinates w h i c h , a r e used in b i n a r y s t a r a s t r o p h y s i c s .
approif
In t h i s c a s e t h e N e w t o n
only po-
1.5. Special
Coordinates
37
tential serves as the initial function. A m o d i f i c a t i o n is t o s t a r t w i t h s e v e r a l f u n c t i o n s w h o s e g r a d i e n t s a r e
eve-
rywhere mutually perpendicular.
rea-
lize this
We s h a l l h a v e m a n y a n o p p o r t u n i t y t o
idea.
All t h e s e s y s t e m s h a v e in common t h a t t h e y a r e direct. T h i s m e a n s : w e may e x p r e s s t h e s t a n d a r d p a r a m e t e r s of t h e E u c l i d e a n s p a c e e x p l i c i t l y in of t h e
terms
coordinates.
T h e s i n g l e e x c e p t i o n a r e t h e Weyl coordinates p r o p o s e d by N e u t s c h & F i c h t n e r [1988] Coxeter
in
order
to
solve
problems
with
finite
invariance
is e x a c t l y the opposite;
though it
t h e Weyl c o o r d i n a t e s a s p o l y n o m i a l s in t h e C a r t e s i a n
is possible to components,
g e n e r a l o u t of t h e q u e s t i o n t o s o l v e f o r t h e m . T h i s , h o w e v e r , t e r much,
write
it is
in
does not m a t -
b e c a u s e of t h e s p e c i a l p r o p e r t i e s of t h e Coxeter groups. We s h a l l
classify them
in p a r t
D of
this
book.
like m e t r i c ,
C h r i s t o f f e l symbols or
div,
are
Δ,
representable
In f a c t ,
the
easily
all
differential
and
explicitly.
essential
information
operators
grad,
For
the
rot,
practically
i m p o r t a n t c a s e s ( d i m e n s i o n s 2 t o 4) w e s h a l l c a r r y t h i s t h r o u g h in 30.
(of
type).
Here the situation
and
groups
chapter
38
2. N o t a t i o n and Conventions
2. Notation and Conventions 2.1. Sets and Topological Spaces
To begin w i t h ,
we have to a g r e e on c e r t a i n pieces of n o t a t i o n ,
since in the
l i t e r a t u r e on our s u b j e c t a l a r g e number of d i f f e r e n t conventions and s y m bols a r e in use.
A l l d e f i n i t i o n s and theorems given in t h i s chapter
will
be
r e t a i n e d throughout the t e x t and used without f u r t h e r comments or e x p l a n a tions. More about them can be found in e v e r y i n t r o d u c t o r y book; frain
from
proofs.
reproducing
the m o s t l y
trivial
and g e n e r a l l y
t h e r e f o r e we known or
Only in a f e w s e l e c t e d cases w e sketch the u n d e r l y i n g
We apply the usual n o t a t i o n scheme f o r sets. by a l i s t of
i t s elements,
A set can e i t h e r be
I x,y,z
described
I
like
χ has the p r o p e r t y
This means:
idea.
e.g.
=
or a c h a r a c t e r i z a t i o n
re-
obvious
An o b j e c t y is contained in Μ if
6(x)
and only
if
it possesses
the
p r o p e r t y S. We then w r i t e y e M; o t h e r w i s e y t M. The Cartesian product o f the s e t s A, B,C, . . .
is
Α χ Β χ C χ
(a,b,c,... ) •
The l o g i c a l plies"),
a € A, b 6 B, c e C,
{
symbols have the usual meanings:
'
N
an a r b i t r a r y
1
set
Ω is
a bijection
of
Ω onto
itself.
p e r m u t a t i o n s o f Ω f o r m the synmetric group S(£i) of Ω. Its isomorphism does not depend on Ω, but only on i t s c a r d i n a l i t y .
All class
F o r that r e a s o n one a l s o
writes S
i f # Ω = η < o°. η The n o t a t i o n (a , a , . . . , a , a ) means the p ermutation K 1 2 k-i k at
to
a
( l s i s
ments o f Ω ( i f length k ( o r cycles are If
(the
π e S(n) tors
shorter:
while
is mapped
a k-cycle).
to
aj
each
remaining
ele-
Such a p e r m u t a t i o n is c a l l e d a cycle Cycles of
length 1 a r e f i x e d points;
of)
Ω
is
finite,
we
can
as a f i n i t e product of t r a n s p o s i t i o n s . uniquely
According
to
determined by π, which
of
the
write
in c o n t r a s t
two
A c y c l e of
every
of 2-
permutation
The r e q u i r e d number of to
possibilities
p e r m u t a t i o n π even or odd. The signum of it, sgn(n), spectively.
and a l l
carries
transpositions.
cardinality
is not
mod 2).
k-1),
any) a r e i n v a r i a n t .
which
fac-
i t s p a r i t y ( = number call
the
is then + 1 or - 1,
occurs,
we
re-
length k thus is an even (odd) p e r m u t a t i o n
whenever
46
2. N o t a t i o n and Conventions
k is odd (even). other! ε
abc...
Be a w a r e that length and signum vary oppositely
to
each
Instead of sgn(a, b, c, . . . ), we sometimes also use one of the symbols abc... or ε
The assignment π ι—> sgn(n) is a homomorphism f r o m c a t i v e group { ± 1).
Its kernel ( = the set of
into the m u l t i p l i -
even permutations)
is the al-
ternating group on Ω. It is denoted Α(Ω) or A π. For η £ 2 is A a normal subgroup of index 2 in S . η η T w o cycles commute with each other if they are d i s j o i n t . Every permutation π of Ω = {1
η ) can be w r i t t e n as a product of mutually d i s j o i n t
cycles.
This r e p r e s e n t a t i o n is unique up to the order of the f a c t o r s and the t r a r y choice of in question,
we
the initial only
element in each cycle.
need to
pick
an element
To f i n d the
a j e Ω,
call
arbi-
partitioning
its
image
a,
whose image is a^, and so on. A f t e r a f i n i t e number of steps we reach the starting
point
= a^ again.
The f i r s t
cycle then is ( a ^ . - . a ^ ) ,
g e t the others we repeat the procedure with any element b ence set Ω Ν ^ } ,
of the
and
to
differ-
as long as not all l e t t e r s of Ω have been considered y e t .
If
in this canonical cycle decomposition the number of i - c y c l e s equals k , k k k we say that π has the (cycle) structure ||l , 2 , 3 . . . . ||. We also o c c a s i o nally use f o r reasons of write
||l2,
2, 3|| instead of
simplicity ||l2,
21,
31,
obvious abbreviations;
for
example
we
4°, 5°, 6°, 7°||, etc.
The a l t e r n a t i n g group A η is simple f o r η a 5. In this case Sπ has no other n o n t r i v i a l normal subgroup than A . As the only a l t e r n a t i n g group of f i n i t e π rank, A possesses a nontrivial normal subgroup, namely the (Klein) four 4
It consists of the unit element and the t h r e e double 2 transpositions and is isomorphic to 2 .
group (Vierergruppe).
The subgroups of S(n) are the permutation groups on Ω. The orbit or block of transitivity of χ 6 Ω under the action of G s ε ( Ω ) is the set χ
G
=
{ x*
I
? * G }
G is transitive on Ω whenever x G = Ω holds. y
The o r b i t s under the conjuga-
tion χ ι—» χ , x , y e G, are the conjugacy classes of G. The group ring K[G] of the group G over a f i e l d Κ is the set of sums of the shape KtG]
=
{
Σ
reG
k^ r
k
j 6
K,
r e G J
all
formal
2.2.
with
Groups
the
47
multiplication
rule
^j^-l^y
= ( k ^ H y ^ )
e K[G].
This
indeed
is a r i n g w i t h 1.
Amongst the w e l l - k n o w n general theorems of
group theory we merely
T h e o r e m 2.2.1 (Homomorphism theorem): L e t φ e H o m ( G , H ) be a g r o u p h o m o m o r p h i s m .
Then the k e r n e l o f
Kern φ
is n o r m a l
in G;
t h e image o f
χ e G
cite:
φ,
?>(x) = 1
φ,
Img φ
is a s u b g r o u p o f
H,
and w e
have
Img φ
Vice versa,
every normal
=
G / Kern φ
s u b g r o u p is the k e r n e l
of
a suitable
homomorphism.
48
2. N o t a t i o n and Conventions
2.3. Matrices
In t h i s s e c t i o n we p r e s e n t a f e w e l e m e n t a r y f a c t s f r o m l i n e a r The
set
of
all
(m, n)-matrices
w i l l be d e n o t e d by K
(m,n)
.
(m
rows,
η columns)
with
Here Κ a l w a y s is a r i n g ,
algebra.
coefficients
i n DC
which u s u a l l y w i l l
be
t h e f i e l d of r e a l (K = IR) or c o m p l e x n u m b e r s (IK = C). The components o r entries of t h e m a t r i x Μ e K ( m , n ) a r e M^ e Κ (1 s i s m, 1 < j < n). The transpose of Μ t h u s is t h e m a t r i x l M e K ( n ' m ) w i t h t
M
= Μ
ij
Jl
We s y m b o l i z e t h e complex-conjugate of a number o r m a t r i x by a b a r : M, t h e a n d Β 6 Κ1-"' by j u x t a p o s i t i o n :
matrix product o f A e DC If m = η, w e s a y Μ
AB e Κ
i s quadratic. T h e zero matrix o f d i m e n s i o n
'
(n, n) i s
de-
or simply 0, t h e identity o r unit matrix by 1 o r 1. The n o t a n η t i o n 0 w i l l a l s o be a p p l i e d t o t h e zero vector. n o t e d by 0
Diag(\i
λ ) is t h e diagonal matrix of dimension ( n , n ) w i t h c o e f f i c i e n t s
λ , n a m e l y t h e m a t r i x Μ e DC(n,n) f o r which λ
ι
if
i
Μ ij otherwise h o l d s . The m a t r i x Μ is monomial if it c o n t a i n s in each r o w and e a c h column e x a c t l y one c o e f f i c i e n t o t h e r t h a n z e r o . equal
If t h e s e n o n z e r o e l e m e n t s a r e
all
t o 1, w e g e t a permutation matrix.
A m o n o m i a l m a t r i x is r e p r e s e n t a b l e in t h e f o r m Μ = D -Ρ = Ρ"Dg w i t h a p e r m u t a t i o n m a t r i x Ρ and d i a ge o n a l m a t r i c e s D and D . The f a c t o r s P, D , and 1 2 ι D a a r e u n i q u e l y d e t e r m i n e d by M. The f u l l monomial group of dimension η over K, which c o n s i s t s of a l l m o n o mial m a t r i c e s , is i s o m o r p h i c t o (K x ) n : S . The n o r m a l s u b g r o u p (DCx)n c o n ti tains the diagonal matrices, permutation
t h e f a c t o r is r e a l i z e d by t h e s u b g r o u p of
all
matrices.
A scalar matrix i s a d i a g o n a l m a t r i x of t h e f o r m λ · 1 = Diag(A
λ),
i.e.
2.3.
Matrices
49
a m u l t i p l e ( w i t h f a c t o r λ e K) of t h e i d e n t i t y m a t r i x . The v e c t o r s p a c e Κ
w i l l n o r m a l l y be a b b r e v i a t e d
as Κ ,
its
elements,
t h e row vectors a r e s i m p l y c a l l e d vectors over Κ (of dimension n). The column vectors o b t a i n e d
l i e i n t h e dual space K < n , l ) .
by t r a n s p o s i t i o n
We w r i t e f ( v ) o r s o m e t i m e s a l s o v f f o r t h e image of t h e v e c t o r ν u n d e r a linear mapping or f u n c t i o n f . The c o m p o s i t i o n of t h e m a p s f , g , h , . . .
(first
f , t h e n g, t h e n h, and so on) gives hogof(v)
=
h(g(f(v)))
vf9h
=
The a d v a n t a g e of t h e l a t t e r n o t a t i o n should be obvious: (1) f , g, h o c c u r in t h e n a t u r a l o r d e r of a p p l i c a t i o n ;
and
(2) we need no b r a c k e t s . F u r t h e r m o r e r o w v e c t o r s a r e e a s i e r t o p r i n t t h a n column v e c t o r s ,
because
t h e y r e q u i r e l e s s r o o m . F o r t h e s e r e a s o n s we have d e s c r i b e d t h e e l e m e n t s of Kn a s r o w v e c t o r s .
E v i d e n t l y t h e use of column v e c t o r s (with t h e
a c t i n g f r o m t h e l e f t ) would lead t o an e q u i v a l e n t q u i t e common,
e s p e c i a l l y in t h e o l d e r
T h e trace (Spur) o f
the
(n.n)-matrix
description.
matrices
It is
also
literature. Μ
is
π 1=1 w h i l e det(M) m e a n s i t s d e t e r m i n a n t .
Μ is singular if det(M) = 0, and regu-
lar (invertible) o t h e r w i s e .
latter
In t h e
case
the 1
c h a r a c t e r i z e d by e a c h of t h e t w o e q u a t i o n s Μ Μ In g e n e r a l ,
inverse is Μ
= 1 and Μ
1
1
.
is
Μ = 1.
Mk is t h e k t h power of Μ (k e Z; w h e n e v e r k is l e s s t h a n
Μ m u s t be r e g u l a r ) .
It
zero,
We have M° = 1 and Μ1 = M.
The dimension of a v e c t o r s p a c e V over Κ (= c a r d i n a l i t y of any K - b a s i s of e.g.,
dim^tK11) = n.
V) w i l l be d e n o t e d by dim^tV),
thus,
m i s u n d e r s t a n d i n g t o be f e a r e d ,
we omit t h e s u b s c r i p t "K" and simply
When t h e r e is no write
dim V. The g r o u p of
all
invertible K-linear
transformations
of
an
n-dimensional
v e c t o r s p a c e V = Kn over t h e f i e l d Κ is GL(V) = GL(K n ). It is i s o m o r p h i c t o t h e g r o u p of r e g u l a r (n, n ) - m a t r i c e s w i t h c o e f f i c i e n t s in K, which we s h a l l call
the
(general) linear group GL (Κ). π
h o w e v e r o f t e n be i r r e l e v a n t .
This
meticulous
distinction
will
50
2. Notation and Conventions
Moreover, group)
are
the normal subgroups SL(V) = SL(K ) and SL (K) (special η determined
by the
additional
restriction
linear
"determinant = 1" in
the corresponding GL. The f a c t o r groups are isomorphic with the multiplicative group of the ground f i e l d :
The λ-eigenspace (λ e C) of a complex ( n . n ) - m a t r i x Μ is the set
ER(X,M)
veC •
ν Μ= λ ν
{
and the multiplicity of λ is the dimension of ER(X, M): mult M U)
dim ER(A,M)
Numbers λ e C with mult^tX) > 0, hence ER(A,M) * 0 (null space) are called eigenvalues of M. When counted with multiplicities,
they f o r m the spectrum
of the matrix M:
EW(M)
λ s C
mult^U) >
The sum of all eigenvalues of Μ (with multiplicities) equals its trace,
the
product of the eigenvalues is identical with the determinant. The i t h row of the matrix Μ e K ( m ' n ' is equal to the image of the i t h basis vector under the linear map φ: Km —> Κ
represented by M. The kernel of Μ
or φ is
Kern Μ
Kern φ
while the image of Κ
=
|veK™
v^ = ν Μ = 0 J
is the subspace Img Μ = Img φ of Κ
ER(0, M)
spanned by the
rows of M. The dimension of the image is the rank of Μ or φ, Rank Μ =
Rank φ
=
dim Img φ
Rank Μ thus is the maximum number of linearly independent rows (or columns) in M. We moreover have
2.3.
Matrices
51
Theorem 2.3.1:
( a ) Assume t h a t U and V a r e v e c t o r s p a c e s over Κ and φ: U —> V is l i n e a r . The f a c t o r s p a c e of U w i t h r e s p e c t t o t h e k e r n e l of φ is
canonically
i s o m o r p h i c w i t h t h e image of φ: U / Kern((p)
a
Img(*>)
(Homomorphism theorem f o r v e c t o r s p a c e s ) .
£
V
For t h e d i m e n s i o n s t h e
equa-
tion dim Kern(p) + dim Img(qp)
=
dim U
holds. (b) The s p a c e (A,B> g e n e r a t e d by t w o s u b s p a c e s A and Β in V h a s t h e d i m e n sion dim κ" 1 )ο(κϊΓ 1 )
holds and the coordinate changes λ « λ
1
as well as κ°κ
1
were assumed to be
smooth. For that reason the above criterion is independent of the charts ic and λ.
Hence it describes an geometric characteristic of the map f
In d i f f e r e n t i a l perties;
geometry,
one is interested
only
statements which require a special choice of
of much less, if any relevance.
itself.
in such "intrinsic" the coordinates
proare
76
3. Manifolds
3.3. Curves and Tangent Spaces The characterization of
all 1-dimensional manifolds does not present any
difficulty:
Theorem 3.3.1: ( a ) Every connected one-dimensional manifold is diffeomorphic to either the real axis IR or the circle S1. (b) Both cases exclude each other: IR and S1 are not even equivalent topologically.
Thus there are only two essentially d i f f e r e n t types of connected manifolds of dimension 1: the set IR of real numbers and the periphery S1 of the unit circle. For reasons of space limitation, we refrain from including the simple proof here. This result suggests
Definition 3.3.1: A curve in the manifold Μ is a smooth mapping r : I —» Μ from a connected 1-dimensional manifold I into M. Depending on whether I is homeomorphic to IR or to S1, we shall call the curve f open or closed. In the former case we shall suppose - except if we explicitly stated otherwise - that I is an open interval in IR.
For the time being Μ always denotes a (smooth real) n-dimensional manifold. The set of all curves passing through a f i x e d point χ € Μ splits up naturally into equivalence classes:
3.3.
C u r v e s and Tangent S p a c e s
77
Definition 3.3.2: Two open c u r v e s y : I
—» Μ a n d
—> Μ a r e tangent t o e a c h o t h e r
I
χ e Μ if f o r s u i t a b l e p a r a m e t e r v a l u e s t
v
v
=
1
w
€ I and t 1 2
e I
in
2
x
•
h o l d s a n d a m a p ic e tt c o n t a i n i n g χ c a n be f o u n d w i t h d
r
dt
The
classes
of
the
( K
ι
° V t=t
equivalence
=
< * rΙΚΟ») J -jT2 lt=t
dt
1
relation
given
by
2
this
condition
are
the
tangent vectors t o Μ w i t h base point x. T h e y f o r m t h e tangent space TM . χ The t a n g e n t
v e c t o r of t h e c u r v e y: I —> Μ a t t h e parameter t 6 I, t h a t
a t t h e place χ = y ( t ) ,
is
i s d e n o t e d by y ' ( t ) 6 TM .
By a p p l i c a t i o n of t h e chain rule i t i s e a s y t o d e m o n s t r a t e t h a t t h i s
parti-
t i o n i n g i n t o t a n g e n t v e c t o r s d o e s n o t d e p e n d on t h e c h a r t κ. S o m e t i m e s a n a l t e r n a t i v e d e s c r i p t i o n of t h e t a n g e n t v e c t o r s i s of
practical
value:
Theorem 3.3.2: Two curves γ
ι
and y
2
in Μ w i t h
V V
=
y
2(V
x
=
a r e t a n g e n t a t t h e p o i n t χ if a n d o n l y if f o r a l l s m o o t h ( o r even a l l functions f:
Μ —> R
d t
is f u l f i l l e d .
1
lt-t
d t
1
2
U-t
2
C1-)
78
3.
Manifolds
T a n g e n t v e c t o r s a r e d i f f e r e n t i a l o p e r a t o r s w h i c h c a n be a p p l i e d t o
smooth
f unctions:
Definition 3.3.3: S u p p o s e χ 6 Μ a n d ν e ΤΜ χ .
The directional derivative of a C ° ° - f u n c t i o n f
w i t h r e s p e c t t o t h e v e c t o r ν is f(r(t+h))-f(r(t)) v(f)
=
11m h-»o
h
w h e r e y i s an o p e n c u r v e in Μ w i t h y ( t ) = χ and y ' ( t ) = v.
I t i s c l e a r t h a t t h e s p e c i a l c h o i c e of y is
irrelevant.
T a n g e n t s p a c e s h a v e a n a t u r a l v e c t o r s p a c e s t r u c t u r e i m p o s e d by t h e
charts:
Theorem 3.3.3: ( a ) C o n s i d e r a ( r e a l ) m a n i f o l d Μ of d i m e n s i o n n. T h e n f o r a l l χ 6 Μ w e h a v e the
isomorphism
TM
X
s
Rn
w h e r e a d d i t i o n of t a n g e n t v e c t o r s a t χ and m u l t i p l i c a t i o n by r e a l
num-
b e r s a r e d e f i n e d in t h e o b v i o u s w a y w i t h t h e h e l p of an a r b i t r a r y
chart
κ containing x. (b) The v e c t o r of
s p a c e s t r u c t u r e t h u s f o u n d (in c o n t r a s t
t h e v e c t o r s ! ) is i n d e p e n d e n t
of κ.
This means:
i n d u c e s a l i n e a r m a p p i n g of ΤΜ χ i n t o i t s e l f . of t h e Jacobi matrix of t h e
to the
It i s g i v e n by t h e
transition.
Proof: Trivial and s t r a i g h t f o r w a r d calculation;
coordinates
Every c h a r t
l e f t to the
reader.
change action
3.3. Curves and Tangent Spaces
79
A chart κ also induces a natural basis on the tangent space at any point in the domain U , f o r we have
Theorem 3.3.4: Let Μ be a manifold of dimension η, κ e k[X)
=
(x 1
a chart. We set f o r all χ € U^ x")
6
Rn
To each i e {1 n) there is an open interval coordinate line through x, that is the curve κ : ,, ν K^lt)
=
κ
-1/ 1 (x
a x 1 , such that the —» Μ with
ith
i-l . i+1 n, X ,t.x ,...,x )
is defined f o r all t e I . Its tangent vector κ ^ ( χ ' ) at the point χ w i l l
be
denoted by d . ( a ) In the κ-coordinates we get the relation a d
=
θχ1 The l a t t e r
has to
be interpreted
as the
directional
derivative
along
the coordinate lines, at a
1
f
=
( b ) The coordinate derivatives (3 lar,
a (f) 1
=
— „1 ax
θ ) f o r m a basis of TM . In particuη χ
l
every tangent vector ν e TM
can be represented uniquely as a l i -
near combination of the a ^ ν
=
v1 a
ι
( c ) The v-action on the set of all smooth functions is a derivation:
v(fg)
=
ι v1 a Cfg)
=
ι v1
9 t f 8 )
—
ax
=
v(f)-g + f-v(g)
80
3.
Manifolds
3.4. Vector Bundles
B e f o r e being able to d e f i n e v e c t o r shall
have to
carry
and,
out an a u x i l i a r y
more g e n e r a l l y ,
construction
which
tensor f i e l d s , is
we
interesting
in
its own r i g h t . It does not make sense to speak of manifold
M,
since
vectors
are
the "set of
meaningful
only
all
tangent
vectors"
in combination
with
of
a
their
base points. Though tangent spaces at two d i f f e r e n t points x, y e Μ are i s o morphic to each other,
a "canonical" b i j e c t i o n between the elements in TM
X
and those in TM y does not e x i s t , Mainly f o r this reason we have to introduce the s o - c a l l e d tangent bundle as a substitute f o r the union of all tangent spaces of M:
D e f i n i t i o n 3.4.1: L e t Μ be a m a n i f o l d , ( a ) The tangent bundle of Μ is
TM
=
I
χ e Μ, ν
e TM
Cb) The projection of the bundle is the map which attaches to each pair in TM i t s f i r s t component,
i.e.
the base point:
π :
TM
(χ,ν)
»Μ I—> χ
The outstanding p r o p e r t y of TM is
Theorem 3.4.1: ( a ) The tangent
bundle TM of
C°°-manifold i t s e l f
an n-dimensional
(smooth)
m a n i f o l d Μ is a
(with the dimension doubled).
Its a t l a s Μ consists of the induced charts κ with κ e 1ί , d e f i n e d via TM Μ
3.4. Vector Bundles
κ(χ,ν)
81
=
(χ
χ ;v
ν )
e
R
χ IR
=
IR
Here
k(x)
=
(x 1 , . . . , χ " )
are the coordinates of the base point and (v 1 the v e c t o r ν with respect to the basis of TM
v n ) the components of which is canonically
giv-
en by the chart κ: ν
=
ν1 a
ι
( b ) The bundle p r o j e c t i o n π: TM —» Μ is i n f i n i t e l y o f t e n d i f f e r e n t i a b l e .
Proof: We consider t w o charts α and β of Μ whose domains are U and U , r e s p e c ~ ~ « Ρ t i v e l y . Then the induced charts α and β are w e l l - d e f i n e d on the p r e - i m a g e s it β°α
(U ) and it α
(U ) of U and U under π. The smoothness of the t r a n s i t i o n β α β f r o m α t o β is t r i v i a l f o r the x - c o o r d i n a t e s . For the v-components
the analogous statement f o l l o w s d i r e c t l y f r o m the d e f i n i t i o n of the
partial
d e r i v a t i v e s 9 . This shows that ι
U TM is
an a t l a s
of
the
manifold
=
TM.
·( ic κ € U 1 (_ I Μ J The
unlimited
differentiability
of
it is
evident.
Structures
similar
with m a n i f o l d s .
to
the
tangent
bundles
frequently
occur
in
connection
It is t h e r e f o r e reasonable to f o r m a l i z e the above c o n s t r u c -
tion in o r d e r to include more general situations as w e l l . This can be done on d i f f e r e n t levels. lowing t e r m i n o l o g y s u f f i c e s .
For our present
intentions the
fol-
3. Manifolds
82
Definition 3.4.2: Let Μ be an n-dimensional manifold. An N-dimensional vector (space) bundle (N e IN) on Μ is a t r i p l e (Μ,Β,π),
composed of
( 1 ) the base space Μ i t s e l f ; (2) a second manifold Β (= "bundle") of dimension n+N; and f i n a l l y ( 3 ) a smooth s u r j e c t i v e mapping π: Β —» Μ, called the bundle projection. We assume in addition that a nonempty family of trivializations Φ^ (i 6 I ) exists with the properties ( a ) Every
is a diffeomorphism f r o m i r ' N l ^ ) to U ( χ RN. Here the U
£ U
are open sets. ( b ) The U n c o v e r M:
U U leu
= Μ
1
( c ) The t r i v i a l i z a t i o n s are compatible with the fibres ττ" 1 (χ):
where ( d ) For
denotes the natural projection f r o m U^xR" onto U^
each two
of Φ »Φ J i
subscripts
i, j 6 0 f u l f i l l i n g
UnU
* 0,
the
restriction
to some f i b r e if'tx)
=
{χ> χ RN
is a (vector space) isomorphism of RN. It thus induces a transformation in GL(R H ).
The tangent bundle is t h e r e f o r e just the simplest special case of this definition. The terminology connected with vector bundles may seem somewhat technical, but its are
importance
applicable
to
lies vector
in its
economy.
spaces,
as
Practically
for
instance
all
operations
constructing
the
which dual
space, tensor products, Graßmann algebras or direct sums etc. , can be c a r ried over without change to vector bundles. der the f i b r e s i T ' t x ) ,
It is only necessary to consi-
χ € Μ, separately.
Since the bundle space Β is a manifold i t s e l f ,
we are able to
introduce
3.4. Vector Bundles
,00
83
C - f i e l d s of vectors,
tensors, and so on. There is nothing more to be done
than to reformulate our f o r m e r approach a l i t t l e . The appropriate tool is
Definition 3.4.3: A (smooth) section of a vector bundle (Μ,Β,π) is a C^-map Β
which, when combined with the projection, gives rise to the identity on the base space M: Id
Μ
It is customary and convenient to w r i t e the image of χ under ν
,s
comes out. This is equivalent with the required e x p r e s s i o n . Similarly,
i (ω)
(b) is found with the intermediary s t e p
=
f
Σ )
dx '
λ
.. .
dx *
λ
1
a
(dx * )
i
λ
dx
1 , 1
λ
. . .
a
dx
k
which is an immediate consequence of the a n t i d e r i v a t i o n p r o p e r t y of i , r e V
o r d e r i n g and a p p l i c a t i o n of i (dx a ) V Statement
=
dxa(v)
=
v(xa)
=
va
(c) at l a s t f o l l o w s f r o m the elementary o b s e r v a t i o n
(proven by
i n s e r t i o n of the d e f i n i t i o n s ) t h a t f o r every d i f f e r e n t i a l f o r m ω, L (du) V holds.
Consequently,
d e r i v a t i v e commute.
=
d(L ω) V
when r e s t r i c t e d to Λ(Μ),
Cartan d i f f e r e n t i a l and Lie
The same is t r u e with d«i
since the C a r t a n r u l e dd = 0 implies d ° ( d o i +i od) V V
=
d ° i d = V
V
+ i «d in the p l a c e of L , V V
( d « i +i od)»d V V
This r e d u c e s the p r o p o s i t i o n in question to the s p e c i a l c a s e of a O - f o r m ω. Under t h e s e c i r c u m s t a n c e s , ( d « i +i od)(u)
=
however, we have
d ( 0 ) + i (du)
=
i (du)
=
du(v)
=
ν(ω)
=
L (ω)
94
3. Manifolds
3.6. Partition of Unity
The i n t e g r a t i o n of a l t e r n a t i n g d i f f e r e n t i a l f o r m s of maximum rank is a d i rect
generalization
of
t i a b l e ) f u n c t i o n s in
the Riemann integral of
(infinitely
often
differen-
IRn.
The underlying idea is quite simple.
We f i r s t consider the special
which the m a n i f o l d is covered by a single chart κ: Μ —* ical point ξ 6 Μ to χ = (χ 1
IRn
case
in
mapping the t y p -
x n ) e IRn.
The f o r m ω 6 Λ η (Μ) we want to i n t e g r a t e may be expressed in c o o r d i n a t e s g i ven by κ as
ω(χ)
where f :
=
fix1
x n ) dx 1 λ . . .
λ dx n
κ(Μ) —» R denotes some suitable smooth f u n c t i o n . We set
J
ω
J...J
=
Μ
fix1
x n ) dx 1 . . .
dx n
Χ6Κ(Μ)
(to be i n t e r p r e t e d as a Riemann i n t e g r a l ) . The general situation,
when one chart alone does not s u f f i c e to map M,
much more complicated. above
case
striction
by
is
Under these circumstances we have to reduce to the
splitting
up the integral
discussed b e f o r e .
If
the
re-
there are i n f i n i t e l y many contributions,
we
are f a c e d with a convergence problem.
into
parts
which
satisfy
We solve it by constructing a " p a r t i -
tion of unity" ( d e f i n i t i o n 3 . 6 . 2 ) . This can be achieved on every manifold, even with some additional properties
(theorem 3 . 6 . 3 )
which
guarantee
that
only
a finite
technical
number
of
terms have to be summed up. These remarks should g i v e the reader an intuitive f e e l i n g f o r the ideas und e r l y i n g the i n t e g r a t i o n theory of The
rest
of
this
these approaches.
section
aims
manifolds.
at
a correct
mathematical
elaboration
of
It has been included f o r the sake of completeness and may
be skipped without hesitation if
one is w i l l i n g
to accept
the existence
of
a p a r t i t i o n of unity as a f a c t . We next have to introduce a f e w pieces of notation f r o m s e t - t h e o r e t i c a l
to-
3.6. Partition of Unity
95
pology.
Definition 3.6.1: Let Μ be a topological space. An (open) covering of Μ is a f a m i l y
U
=
I
i e D I
of open subsets L^ S Μ whose union is the whole of M. U is locally finite if every χ e Μ has an open neighbourhood which intersects only f i n i t e l y many U
nontrivially.
A covering
8
is a refinement of tt, if f o r all j e J there is an i e 0 with V^ £ I K
The
term "partial covering" has the obvious meaning. The topological space Μ is called paracompact if each open covering has an open locally f i n i t e partial covering, a f i n i t e partial covering. The support supp(f) of
and quasicompact if each covering has
A quasicompact Hausdorff space is compact.
a continuous function f : Μ —> IR is the
topological
closure of the set of all χ e Μ with f ( x ) * 0.
We now come to the central notion which w i l l enable us to construct certain global
f i e l d s on manifolds as well
as to integrate
differential
forms
maximal rank:
Definition 3.6.2: ( a ) By a partition of unity on the topological space Μ we mean a set
Ζ
.00 of C -functions φ : Μ —» IR which s a t i s f y the f o l l o w i n g requirements: J
of
96
3. Manifolds
( 1 ) The set of all s u p p t ^ ) is locally f i n i t e ; ( 2 ) we always have
£ 0; and
( 3 ) f o r all x,
Σ vx) =
1
jeJ
( b ) Ζ is subordinate to the open covering
U
=
j Uf
i e 0 j-
if f o r each j e Jl it is possible to find an appropriate i 6 0 such that supp(^) £ U(.
The summation in ( a ) on account of
(1) contains only a f i n i t e number
of
nonvanishing contributions and thus is w e l l - d e f i n e d . Remarkably,
partitions of unity exist f o r all manifolds (and even conside-
rably more general spaces). With their help we shall overcome the d i f f i c u l ties mentioned at the beginning of this section. Except f o r the terminology,
we f o l l o w the argumentation of Warner [1983],
where even a l i t t l e more is proved. Our purposes, however, the f u l l strength of Warner's deductions,
do not require
and we can simplify parts of the
calculations in comparison to the just cited book. We start our considerations with a lemma which is interesting
in its own
right.
Theorem 3.6.1: Every open covering of a manifold Μ has a countable refinement which consists only of open sets with compact closures.
Proof: We set dim(M) = n. To any given χ € Μ we select a chart κ which has χ in its domain of definition, U^. The image of K(X) in the Euclidean space IRn. T h e r e f o r e ,
then is an open neighbourhood we can find a positive
real
3.6.
P a r t i t i o n of Unity
number c,
97
such that the closed ball Κ with radius ε and centre κ(χ) is a
subset of k(U ). κ
Since κ is homeomorphic, the pre-image κ x. The a r b i t r a r i n e s s of the choice of χ € Furthermore, the manifold Μ by definition able basis β = {Β , Β , Β , . . . } . We use the 1 2 3 Β
Β 6 8
(Κ) is a compact neighbourhood of Μ implies that Μ is paracompact. is a Hausdorff space with countnotation
compact
and consider a nonempty open subset U and some point u in U. On account of the paracompactness of Μ there is an open neighbourhood W of u with compact closure W . We may assume without r e s t r i c t i o n t h a t W" is contained in U. As an open s e t ,
W is the union of suitable basic s e t s B ^
which occurs in this representation,
For every
Bt
B t is a (closed) subset of the compact
set W" and thus compact as well. The relevant
are consequently members
of the c o l l e c t i o n S. Repeating this argument f o r all u e U, we are provided with an exhaustion of our neighbourhood U by ε - s e t s . Since U was an a r b i t r a r y (nonempty) open set, 6 i t s e l f is a basis of the topology of Μ (which,
as a subset of 8,
is
countable). We summarize the previous discussion: Μ possesses a countable basis e = {C^, C , C 3 > . . . ) all of whose members have compact closures. We t h e r e a f t e r
construct open subsets D^D^, D^, . . .
of Μ according to the
following instruction: F i r s t , we set m
= 1 and D = C . 1 1 1 Suppose we have already introduced m^ e Ν and D^ S M, where D k
=
C υ ... ν C 1 m k
In this c a s e ,
the compact topological closure D of D is covered by the k k s e t s D^ η C . Because of compactness, there is some μ 6 IN with D~ k
S
l
C υ ... υ C μ
We s e l e c t one such μ > m , call it m , and define k k+1
98
3. M a n i f o l d s
D k+l The s e q u e n c e D , D^, D 3 , . . .
=
C υ .. . υ C 1 μ
f o u n d t h r o u g h t h i s r e c u r s i o n is an open c o v e r i n g
of Μ ( s i n c e e h a s t h e same p r o p e r t y ) . All D^ D have c o m p a c t cclosui l o s u r e s D , and we have t h e i n c l u s i o n D~ £ D~ . Now k k k k+l c o n s i d e r an open c o v e r i n g
U =
I U
1 e D I
of M. F o r a l l k e IN, t h e d i f f e r e n c e D \ D is c o m p a c t and c o n t a i n e d k+2 k+l t h e open s e t D \ D . This i m pr l i e s t h a t K k+3 k
" [Dk+3NDk] I
{
1 6
in
° }
covers D \D k+2 k+l By c o m p a c t n e s s , we may c h o o s e a f i n i t e p a r t i a l c o v e r i n g .
Let us d e n o t e i t
by Bk- In t h e s a m e way we f i n d a f i n i t e p a r t i a l c o v e r i n g in t h e open c o v e r ing
U η Dg
of D z S D 3 ·
1 € II J
Clearly, δ
=
δ
ο
υδ
ι
υ S υ... 2
is a l o c a l l y f i n i t e r e f i n e m e n t of U. As a union of c o u n t a b l y many
finite
f a m i l i e s , Β is c o u n t a b l e i t s e l f , and every member of δ is open w i t h c o m p a c t closure.
Hence, δ i s t h e r e f i n e m e n t we w e r e looking f o r .
Though t h e p r o o f of t h e l a s t t h e o r e m is s o m e w h a t involved, t u r a l and s t r a i g h t f o r w a r d .
it is quite n a -
The n e x t i n t e r m e d i a t e r e s u l t i s much s i m p l e r .
is u s e f u l t o s t a t e i t a l s o a s a s e p a r a t e lemma.
It
3.6. Partition of Unity
99
Theorem 3 . 6 . 2 : For each point χ of a manifold Μ and every open neighbourhood U of x, a compact f:
set
Κ with
χ e Κ c U and
a nowhere
negative
smooth
function
Μ —» R with compact support exist which vanishes outside U and has the
constant value 1 on K.
Proof: It s u f f i c e s to prove the assertion f o r tained in U,
an arbitrary
x-neighbourhood
con-
since the conclusion then immediately carries over to U i t -
self. We choose a chart κ with U
κ
3 χ and replace U by U π U . The action of κ κ
t r a n s f e r s the problem into Rn, and we may r e s t r i c t our investigation to the special case Μ = IRn. Furthermore,
we extend all functions to Rn by setting
them equal to 0 outside the original domains. It is moreover allowed to assume χ = 0 (apply an appropriate translation! ). If necessary, we reduce U further to the open ball
y
y e R"
with radius 2 ε and centre 0. Here,
< 2 ε
ε is a suitably selected positive
real
number. A f t e r these preparations,
we set
y e R"
and seek an arbitrary
y
s ε
nonnegative C -function h: IR —> IR, which is
larger
than zero f o r positive arguments and vanishes everywhere else, f o r instance
exp(-t h(t)
)
(t>0)
= 0
Then the function f : R —> R, defined by
(tsO)
3. Manifolds
100
=
h r c " l y | ] h[2C-|y|] + h[|y|-e]
does the job.
We now come to the key result of this section.
T h e o r e m 3.6.3:
Suppose u
=
{ u
I i
e
0 }
is an open covering of the (smooth) manifold M. Then there e x i s t s a countable partition
Η =
{ « V W · · · )
of unity, which is subordinate to U, such that all functions φ have compact supports.
Proof:
Like in the proof of theorem 3 . 6 . 1 we construct a sequence D^D , D 3 , . . . of open subsets with the properties discussed there:
each D~ is compact and
contained in its successor D . Ali D together cover M. We add to this 6 k+l k family an initial member, namely Dq = e>. Let χ e M. We introduce N(x) as the largest number υ € IN , f o r which χ t D~ ο ν and choose an i(x) e 0 with χ e U . B y theorem 3. 6. 2 we can find a nonl (χ)
negative real-valued C -function f
on Μ whose support is compact and lies
in U η |D~ \D I ι i (x)+2 l (x)J Beyond this,
f
attains in a certain neighbourhood W of χ the constant
3.6.
Partition of Unity
101
value 1. We now f i x k and let χ run through the compact set ing W
D k \D k
· The correspond-
f o r m a covering of D^XD^ of which we pick out a f i n i t e partial c o -
vering. This leads to a f i n i t e collection X of functions f f o r every k. Assembled k χ in their natural order, they provide us with a countable f a m i l y of f u n c tions
X
Χ υ Χ ν Κ υ 1 2 3
=
if we simply w r i t e φ
{ * ΐ ' * 2 · ν · ·
}
instead of f
f o r brevity. The associated neighbour*J hoods f o r m a locally f i n i t e open covering J
·[ W ,W ,W , . . . ]· \ ι a a J of M. The sum
* -
Σ
JeW
is thus w e l l - d e f i n e d and positive on M. The normalized functions
ψJ
=
*y φ
then are a partition of unity with the desired additional
properties.
102
3. Manifolds
3.7. Oriented Manifolds
The t o p o l o g i c a l tions
f r o m IR
group GL(IRn) s GL (IR) of all i n v e r t i b l e linear t r a n s f o r m a n onto i t s e l f s p l i t s into t w o connected components which w e
call GL + (IR n ) ss GL + (R) and GL~(IRn) a GL'(IR). η η the sign of the determinant.
The marks "+" and " - "
indicate
The set GL + (R n ) of orientation preserving maps is a normal subgroup of
in-
while the orientation reversing t r a n s f o r m a t i o n s f o r m
the
dex 2 in GL(R n ),
single coset GL (IRn). This decomposition corresponds with a b i p a r t i t i o n of the ( o r d e r e d ) bases of IRn into
two
orientation classes.
The basis b 1
b η is said
to
posi-
be
tively or negatively oriented, depending on which component of GL(IRn) c o n tains the linear map t r a n s f e r r i n g
( f o r all i ) the i t h unit v e c t o r
e(
to
the
basic element b 1. (The accompanying m a t r i x has the r o w s b I b n ). By this d e f i n i t i o n , the p o s i t i v e class is the one in which the standard
ba-
sis e 1, . . . , e η lies.
The g e n e r a l i z a t i o n of this idea to manifolds is p e r f e c t l y
obvious.
Definition 3.7.1: ( a ) The charts κ and λ of the m a n i f o l d Μ have the same orientation are oriented equally) if the Jacobi determinant of the chart λοκ
1
(i.e.
transition
is e v e r y w h e r e positive.
(b) In case this condition
is f u l f i l l e d f o r
pairs κ,λ of
all
charts
in
the
M - a t l a s 51, we call It (and Μ i t s e l f ) oriented. ( c ) Finally,
the m a n i f o l d (M,11) is orientable if
an oriented
atlas U' of
Μ
e x i s t s which is compatible with U. Remark: The
Euclidean space IRn,
oriented,
the
sphere S n ,
and
the
torus 1Γη = s'x. . . xS 1
while the Möbius strip and the Klein bottle belong
known n o n - o r i e n t a b l e manifolds.
to
the
are
best
We shall not need the l a t t e r t w o m a n i f o l d s
and thus do not discuss them in more d e t a i l . In order to g i v e at least one instructive example,
we mention without p r o o f
3.7. Oriented Manifolds
the real
that
projective
103
space P n (R)
is
orientable
for
odd,
but
not
for
even n.
Every chart κ of
an n-dimensional manifold Μ which represents some point
χ € Μ induces an o» r i e n t a t i o n on TM * s R . The analogous statement f o r the cotangent
space Τ Μ
is true as w e l l .
Charts which are
contained
o r i e n t a t i o n class of κ c l e a r l y lead to the same o r i e n t a t i o n as κ.
in
the
We imme-
d i a t e l y conclude f o r tangent and cotangent bundles:
Theorem 3.7.1: *
The o r i e n t a b i l i t y of one of the three manifolds Μ, TM, and Τ Μ implies the o r i e n t a b i l i t y of the others.
We can also c l a s s i f y the d i f f e r e n t i a l f o r m s in a similar way:
Definition 3.7.2: L e t Μ be an n-dimensional m a n i f o l d and ω € Λ η (Μ) an a l t e r n a t i n g d i f f e r e n t i al f o r m of maximum rank on M. We call ω a volume form if ω ( χ ) * 0 f o r e v e r y χ e M. A l l volume f o r m s can be derived f r o m a p a r t i c u l a r one, e . g . ω, by m u l t i p l y ing with a nowhere vanishing smooth function f :
Μ —> R. All volume f o r m s on
Μ are thus r e p r e s e n t a b l e as f u with suitable f .
We say that ω and f u are in
the same orientation class whenever the quotient function f is p o s i t i v e . Locally,
we choose the o r i e n t a t i o n of d i f f e r e n t i a l f o r m s a l w a y s in c o r r e s -
pondence with that of the 1 - f o r m s . Thus the volume f o r m d x 1 A . . . A d x n is considered as p o s i t i v e ( n e g a t i v e ) if the coordinate d »i f f e r e n t i a l s in this o r d e r d e f i n e a p o s i t i v e ( n e g a t i v e ) basis of Τ Μ.
dx1,...,dxn
A d i r e c t examination whether a given atlas U is oriented can be quite ous. ble
It is even more d i f f i c u l t to f i n d an oriented a t l a s which is with
great
U.
For
interest.
that
reason,
the
following
orientability
tedi-
compati-
criterion
is
of
104
3. Manifolds
Theorem 3.7.2: A m a n i f o l d Μ is o r i e n t a b l e if and only if there is a volume f o r m on M.
Proof: We assume that each chart κ in the atlas I of Μ has a connected domain U . κ This is not a severe l i m i t a t i o n , because every a t l a s can be brought to this special
f o r m if
we replace the charts with their r e s t r i c t i o n s
on the
con-
nected components. A volume f o r m ω g i v e s r i s e to a smooth r e a l - v a l u e d f u n c t i o n which is d e f i n ed on K ( U ) via κ |dx*A. . . Adx Adx j n
Here,
(χ1
it is in the connected region K(U ) either
s t r i c t l y negative. is
f ω
χ η ) is the coordinate v e c t o r of the point χ 6 M. Since f
not vanish,
κ
=
positively
strictly
Depending on which of these a l t e r n a t i v e s occurs, or
negatively
oriented.
It
is
clear
that
this
does
positive
or
we say,
defines
an
o r i e n t a t i o n on M. Vice v e r s a , tion f
f o r each t w o charts κ and λ a p o s i t i v e
if 11 is oriented,
func-
with
(AoKf'i^dxV .
can be found.
This implies,
.Adx"j
=
f dx 1
Λ
...
Λ
dx
if we put
ω
=
*I κ ^dx'ft. . .
ω λ
=
λ ^dx1A. . .
Adxnj
and *
Adx"j
the r e l a t i o n
ω
=
λ and the
quotient
function f ° λ
(λ f ) ω
=
(foA) ω
κ is positive
κ in U
η U^.
We merely
have
to
3.7. Oriented Manifolds
105
construct a partition
{116 °} of unity which is subordinate to the covering
u
=
{ υκ
I KS « }
given by the atlas and calculate the sum ω (We must choose the κ
=
L
)
such that
U
φ κ
11
2 Trg((p ) holds). The v e r i f i c a t i o n i that ω is a everywhere positive volume f o r m may be l e f t to the reader as a trivial
exercise.
By similar procedures one can often build global forms or f i e l d s f r o m cally given
"pieces" with the help of
a suitable partition
of
unity.
method is one of the most important tools of d i f f e r e n t i a l geometry.
loThis
3. Manifolds
106
3.8. Integration of Differential Forms
By f a r the most r e l e v a n t gration
on m a n i f o l d s .
application of
As
indicated
the f o r e g o i n g section is the
above,
the
integral
of
a
inte-
differential
f o r m is a vast extension of the classical Riemann i n t e g r a l on (compact
do-
mains in) IRn. We assume the r e a d e r
to be f a m i l i a r with the d e f i n i t i o n of
the l a t t e r
as
the common l i m i t of upper and l o w e r approximating sums and the most element a r y p r o p e r t i e s required f o r its e x p l i c i t The e x e c u t a b i l i t y iour
of
of
evaluation.
the ideas in question depends s t r o n g l y
the Riemann
integral
under
coordinate
changes.
foundation is the substitution rule of theorem 3.8.1.
Its
on the
behav-
mathematical
As a p r e p a r a t i o n
for
the intended g e n e r a l i z a t i o n we need
Definition 3.8.1: We i n t e r p r e t
the Riemann integral of
Β c r " denotes a compact domain,
a smooth f u n c t i o n f :
Β —» IR,
where
as the integral of the η - f o r m ω e Λ η (Β),
given by
ω(χ)
=
f(x'
x " ) dx 1
A
...
Λ
dx n
and w r i t e
J ω
=
Β
J. . . J f (x 1
xn)
dx^.-dx"
Β
Note that the o r i e n t a t i o n plays an important part here since the
Riemann
volume element dx 1 . . . dx n is independent of the order of the f a c t o r s ,
while
the volume f o r m d x 1 A . . . A d x n must be multiplied by - 1 if we impose an odd permutation on the coordinates. rect
only
for
orientation
T h e r e f o r e the f o l l o w i n g statement is
preserving
sign would have to be inserted.
transformations;
otherwise
a
corminus
3.8.
I n t e g r a t i o n of D i f f e r e n t i a l Forms
107
Theorem 3.8.1: Under
an
orientation
preserving
invertible
continuously
differentiable
transformation τ : Β —> IRn, where Β c IRn is a compact domain, the substitution rule
J
Φ
=J
τ(Β)
τ* φ
Β
holds f o r every n - f o r m 0 e Λ (B ) defined on the image BT = τ(Β).
Proof: We denote the orientation ("signum") of an IRn-basis by sgn(. . . ). the typical
point
tution rule of
χ = (χ 1
real
x n ) to y = (y 1
y n ),
If τ maps
the f a m i l i a r
analysis yields the proposition a f t e r
a short
substicalcula-
tion:
J
φ
=
sgnCdy1
dy n )
τ(Β)
J
fty1
y n ) dy 1 Λ . . .
Λ dy n
yer(B) .
1
=
sgn(dx , . . . , d x n ) χίΒ
=
J
aty1
yn)
^
*">
f o x ( x ) det
f ° x ( x ) df1 Λ ... Λ dfn
=
dx
Λ ...
Λ dx n
J τ*ω
xeB
It
is sensible
to
introduce
the
integration
of
differential
forms
in
two
steps:
Definition 3.8.2: Let Μ be an n-dimensional manifold, II an oriented atlas on M, and ω e Λ η (Μ) an η - f o r m with compact support. ( a ) Consider
f i r s t the special
case that there
is a chart κ covering
the
whole of M, which assigns the coordinates ( x 1 , . . . , x n ) e IRn to the point χ e M. I f ,
furthermore, f : ic(M) —> IR denotes the smooth function cha-
3. Manifolds
108
r a c t e r i z e d by ω(χ)
=
fix1
x n ) dx1
Λ
...
Λ
dx"
we set J ω =
J.. . J f (x 1
Μ ( b ) In general,
xn)dx1...dxn
κ(Μ)
we seek a locally f i n i t e countable partition of unity with
compact supports, f o r instance
•
z
{ v w · · · }
which is subordinate to the covering
Ε
=
{ Uk
I
κ . «
}
of Μ by the domains of the charts. T h e r e a f t e r we choose f o r each k e IN some chart κ κ^ e 1U, whose domain U^ contains the support of φ T h e n k integral of ω is
ίω = Μ
the
Σ Κω
keIN υ k
This looks much uglier than it is. Of course, we have to show that the d e f inition is meaningful. Existence
and finiteness of
the Riemann integral
in (a) f o l l o w
from
the
compactness of K(supp(u)) and the continuity of the integrand. We may replace κ by any other chart λ, say, which meets the requirements, without a f f e c t i n g the integral,
cf.
theorem 3.8.1. The expression in ( a ) is
thus w e l l - d e f i n e d . That a partition exists,
of
unity with the properties mentioned under
f o l l o w s f r o m theorem 3 . 6 . 3 .
(b)
really
The same argument as b e f o r e demons-
trates that the partial integrals have f i n i t e values which do not depend on the employed charts κ .
3.8.
I n t e g r a t i o n of D i f f e r e n t i a l Forms
109
If we had taken a d i f f e r e n t partition of unity, f o r instance
' = {ω = Σ ielN jelN
Σ /
ω
Σ ίψ3ω
=
jelN ielN V }
Hence (b) is also a correct definition.
jslN
Part (a) is a special
case of
(b);
thus we must also show that both statements are in accord with each other. To achieve this,
put II = {κ} in the general formula and use the
partition
of unity which consists only of the single function φ: Μ —» {1>.
Applying theorem 3. 8. 1 to each partial integral separately, get the extremely
we immediately
important
Theorem 3.8.2: Assume τ: Μ —> Ν is a diffeomorphism between two oriented manifolds of dimension n. Then f o r each η - f o r m ω with compact support on N, the substitution rule
ίω = Ιτ*ω Ν holds.
Μ
110
3. Manifolds
3.9. Stokes' Theorem
the
in-
t e g r a l on an oriented m a n i f o l d has s t i l l another p e c u l i a r i t y of eminent
Apart
f r o m the substitution
im-
portance.
rule investigated
in the last
section,
It is known as the "theorem of Stokes" though Stokes himself
proved a very r e s t r i c t e d version of
only
it.
B e f o r e w e can s t a t e the proposition in a f o r m which serves our purposes, need some t e r m i n o l o g y .
Our next d e f i n i t i o n is a s l i g h t and natural
we
exten-
sion of what we had studied e a r l i e r .
Definition 3.9.1: We denote by IR" the positive half-space in Rn, that is the set of
all
vec-
t o r s whose f i r s t coordinate is nonnegative:
R;
=I
(χ 1
€ IR
χη)
A regular domain in an oriented manifold Μ of dimension η is a subset Β ε Μ in which to each boundary point χ e bd(B) a chart κ: U
—» Rn can be found
with κ ( χ ) = 0 and
k(U nB) κ
=
K(U ) η IRn κ
+
The meaning of this condition is of course that the boundary of Β is s u f f i ciently
"tame".
By d i r e c t
insertion
of
the d e f i n i t i o n s
(left
to
the
reader
as an easy e x e r c i s e ) one v e r i f i e s
Theorem 3.9.1: The boundary of a r e g u l a r domain Β £ Μ is an oriented submanifold of
codi-
mension 1 in M. Its o r i e n t a t i o n is obtained f r o m that of Μ by t r a n s f e r r i n g f r o m IRn a c c o r d ing to the f o l l o w i n g rule:
3.9.
Stokes'
Theorem
We c l a s s i f y ,
as
usual,
111
the
standard
basis
o r i e n t e d and do the same f o r the b a s i s e
e1 e
2
en
of
IRn as
positively
of bd(IR ) s IR
n
n
n
+
The r e s t r i c t i o n κ| , of an oriented c h a r t κ of Μ to bd(B) then has 'bd(B)
the
same o r i e n t a t i o n as κ i t s e l f . While bd(B) means the topological
space of the boundary points of B,
we
w r i t e SB f o r the j u s t defined (oriented) manifold with point set bd(B).
S t o k e s ' theorem reads in the required form:
Theorem 3.9.2: Let Μ be an n-dimensional oriented manifold and α e Λη 1 (M) with support.
compact
Then we have f o r each r e g u l a r domain Β
{α
=
{da
3B
Β
E s p e c i a l l y f o r Β = Μ, t h i s yields J da
=
0
Μ
Proof: The argumentation is l a r g e l y p a r a l l e l to t h a t of theorem 3 . 8 . 2 ;
we t h e r e -
f o r e discuss only the main d i f f e r e n c e s . The
proposition
clearly
implies
the
existence
of
all
integrals.
We
shall
obtain t h i s as a b y - r e s u l t of the proof below. First,
we consider the covering of Β induced by the c h a r t s of an o r i e n t e d
atlas
I t s r e s t r i c t i o n to the boundary is a covering of 9B with
gous p r o p e r t i e s .
analo-
T h e r e a f t e r we s e l e c t a subordinate p a r t i t i o n of unity on Β
(getting
a t the same time one of 3B) e x a c t l y
section,
cf.
like t h a t
of
the
foregoing
theorem 3. 8. 2.
Both sides o f the equation then decompose into several
partial
and it will s u f f i c e to e s t a b l i s h the f o r m u l a in question f o r each tion s e p a r a t e l y .
integrals, contribu-
This leads to a reduction to the special case t h a t the
in-
t e g r a t i o n domain is covered by a single c h a r t κ: U —» IRn, say. Without r e -
112
3.
striction,
Manifolds
w e may thus impose the a d d i t i o n a l c o n d i t i o n Β £ U . κ
An a p p l i c a t i o n o f κ shows t h a t it is a l l o w e d to consider Β as a r e g u l a r
do-
main in IP". We e x t e n d the f o r m s to be i n t e g r a t e d t o the w h o l e of R n by s e t ting
them
equal
to
0
outside
the
original
domains.
Then
the
following
s t a t e m e n t r e m a i n s t o be p r o v e d : E v e r y α 6 Λ η 1 ( R n ) w i t h compact support obeys the f o r m u l a
J
=J
α
9R n
da
Rn
+
+
We s p l i t a a f t e r having introduced the standard c o o r d i n a t e s of Rn into tial contributions
Α
_ , 1 f i x k
=
k
Χ
η·.
J
, 1 dx
w h e r e k runs f r o m 1 through n.
Λ
. . . Λ
The f
, k-l dx
. da
simple
Fubini
k
, , \k-l
=
theorem w e
, k+1 dx
Λ
. . . Λ
k . 1 dx „ k
Λ
, η dx
It is
clearly
From
3f
(-1)
Λ
a r e smooth f u n c t i o n s .
s u f f i c i e n t to v e r i f y the t h e o r e m f o r a s i n g l e α^.
and the
par-
like
. . . Λ
j η dx
dx
get
the
first
part
of
the
proposition
after
a
transformation.
The second equation is an immediate consequence of 3M = es.
Remark: Our
proof
of
the
theorem
of
Stokes
can be c a r r i e d
over
literally
to
the
s t a t e m e n t which is obtained by r e p l a c i n g the smoothness of α by the w e a k e r c o n d i t i o n t h a t α is a C 1 - f o r m of
rank (n-1).
is s t i l l continuous ( C ° ) and hence
integrable.
The Cartan d e r i v a t i v e da then
We make use of t h i s c o n s i d e r a b l e e x t e n s i o n in the c h a p t e r s 6 ( c o m p l e x lysis;
Cauchy's theorem) and 22 (potential
much m o r e
powerful
them in the sequel.
generalizations
are
ana-
theory).
In the
literature
even
discussed,
but w e
shall
need
not
4.1. The M e t r i c Tensor
4.
Riemannian
113
Spaces
4.1. The Metric Tensor
As announced at the beginning of chapter 3, we want to r e s t r i c t our investigations
to
manifolds
which
are
endowed
with
an
invariant
It can be r e a l i z e d by an element of arc length ds,
length.
measure
of
like in Euclide-
an space Rn where it may be calculated via
ds 2
=
(dx1)2 + . . . +
(in Cartesian coordinates x 1 ) . 2 The special f o r m of ds is not preserved if
(dxV
1 we r e p l a c e the χ by
another
f r a m e , but a change of charts induces a linear t r a n s f o r m a t i o n of the c o t a n gent space of merely
have
IRn. Thus the d i f f e r e n t i a l s dx 1 to
be
multiplied
with
the
(at some f i x e d point χ e IRn)
invertible
Jacobi
matrix
of
the
transition. This operation c a r r i e s ds 2 over to a quadratic f o r m whose signature (+ n , is also invariant. vanish,
i. e.
In g e n e r a l ,
the o f f - d i a g o n a l
coefficients will
no
longer
in the new coordinates y 1 there w i l l be contributions by
ed t e r m s " dy'-dy^ (i *
"mix-
j). 2
In any case, w e represent the squared length element ds the " m e t r i c "
g.
In contrast
that g is i n d e f i n i t e .
to
Riemann h i m s e l f ,
by a (0, 2 ) - t e n s o r ,
we a l l o w
the
possibility
Only the requirement that the determinant of g is no-
where z e r o is indispensable.
The precise f o r m u l a t i o n thus is
Definition 4.1.1: ( a ) A metric on the n-dimensional (smooth) m a n i f o l d Μ is a nowhere d e g e n e r a t e (0, 2 ) - t e n s o r f i e l d g whose signature is constant on M. Under these conditions,
the pair
(M, g ) is c a l l e d a Riemannian manifold
or a Riemann space. ( b ) If
g
ab
chart,
are
the
coordinates
of
the
metric
we denote the c o e f f i c i e n t s of
also f o r m the components of a tensor,
with
respect
the inverse m a t r i x
to by
a
suitable
gab.
which is of rank ( 2 , 0 ) .
They
We call
114
4. Riernanntan Spaces
the metric associated with the g a b contravariant,
the other type (g
ab
)
covariant. As the example _ab d
Τ indicates,
c
=
g
ap
g
cq
Τ
bqd ρ
we can " s h i f t " tensor components and thus switch f r o m con-
t r a · to covariant expressions and vice versa.
Note the index positions
and that we used the same letter f o r both modifications. ( c ) The (arc) length of an open curve y: I —» Μ between the parameter v a l ues t , t
e I £ R is given by a curve integral
over the length element
ds. It can be evaluated in arbitrary coordinates χ 1 j
ds
2
=
g
. a
ab
dx
,
dx
xn f r o m
b
via the formula t l ( r ; t ,t ) 1
r
=
2
. x°xbdt
2
/g
I
ab
t1 where,
as usual, the dots denote derivatives with respect to the curve
parameter t. The curve y is timelike tive,
2
is positive, spacelike if it is nega2 and null or lightlike whenever ds = 0. In the definition of the if
ds
arc length we only admit timelike curves such that the radicand is a l ways £ 0. ( d ) A geodesic connecting the points x, y 6 Μ is a curve running f r o m χ to y whose length, in comparison with "neighbouring" curves having the same property,
is
stationary.
Equivalently,
the
first
variation
51
(with
f i x e d initial and f i n a l points) of 1 vanishes.
We know f r o m the calculus of variations that extremal problems like the one above defining the geodesies on Μ can be reduced to systems of d i f f e r e n t i a l equations. Without p r o o f , we state
ordinary
4.1. The Metric Tensor
115
Theorem 4.1.1: The variation problem
L ( x ' , x ' , t ) dt
=
0
with the smooth Lagrange function L ( x , x ; t ) and f i x e d boundary conditions x ! ( t ) and x l ( t ) is tantamount to the Euler-Lagrange equations
d dL
dL
dt dx
3x'
f o r all i.
From this we deduce a vector d i f f e r e n t i a l equation f o r the geodesies:
Theorem 4.1.2: The geodesic curves on M, dim Μ = n, are characterized in arbitrary coordinates (x 1
x n ) by the conditions x a + Ta
be
ib
=
0
(with the arc length as parameter). Here, we used the Christoffel symbols _ be
f o r the sake of
1 2
ad 8
Γ [8dc,b
_
+ 8bd,c
1 8bc,dJ
brevity.
Proof: The variational problem which defines the curve h?s the f o r m given in theorem 4.1.1. The Lagrange function is
4. Riemannlan Spaces
116
. L
ds dt
=
/
·a -b
=
X
if we f o r the moment leave the independent variable t undetermined. The d t h E u l e r - L a g r a n g e equation is
Thus, by insertion,
g
•b-c XX be, d
d
2 L
dt
g
_ Lg
3L
d ah
3x d
dt ax d
we get [öbxc + i V l be L d dJ 2 L
dc
"C T x + Lg
d
g
de
dt
Xc + g
X bd
2 L
-b . *b·c _ · b· c x + Lg χ χ + Lg xx bd de, b bd, c
L g dc X
r"
"b
L g bd X
2 L Having c a r r i e d through all d i f f e r e n t i a t i o n s ,
we are allowed to put the pa-
r a m e t e r equal to the arc length s. Then we have L = 1 and L = 0, and the f o r m u l a s i m p l i f i e s considerably:
g
db
" b ^ i r ^ 1-b-c x + g +g - g x x 2 [dc, b bd,c °bc,dj
=
_ 0
Here of course the dot has to be interpreted as d e r i v a t i v e with respect s. Multiplication by g a d yields the proposition.
As long hoods,
as we r e s t r i c t
our
investigation
to
sufficiently
small
neighbour-
the geodesies are the shortest lines connecting the end points.
For
l a r g e regions this statement in general does not remain true (example:
two
antipodal points on S ).
117
4.2. C h r i s t o f f e l Symbols and Covariant Derivat ion
4.2. Christoffel Symbols and Covariant Derivation
Under
coordinate
the Christoffel
transformations,
f o r e g o i n g section 4.1 do not behave like tensors.
symbols d e f i n e d
in
the
They obey the more com-
p l i c a t e d rule
ax 1 ax b ax°
a Γ
Jk
~
Γ bc
a V
a ? ai1 a ?
ax a ax b a i 1 a i *
as can be v e r i f i e d by a simple calculation. (l,2)-tensor Ta
be
ax a ax b
The corresponding
law f o r
any
would be
ax1 ax b ax c T1
=
Ta « » ax a 8xJ a i "
Jk
i.e.
only
the
first
term
of
the
formula
above.
The
comes f r o m the nonlinearity of the chart t r a n s i t i o n ;
second
contribution
it mainly consists
of
the d e r i v a t i v e of the Jacobi m a t r i x . N e v e r t h e l e s s the l e f t hand side of the geodesic equation we found in t h e o rem 4 . 1 . 2 is a v e c t o r = ( 1 , 0 ) - t e n s o r . variational
problem
which
led us to
This was to be expected, the f o r m u l a
in question
since
is
the
invariant
under chart changes. It f o l l o w s that the x a cannot f o r m the components of fact,
a tensor
either.
In
one easily c o r r o b o r a t e s this statement by d i r e c t insertion of the d e -
finitions. We are not going to d w e l l on these t r i v i a l
observations and p r e f e r t o
v e s t i g a t e the more general case of p a r t i a l d e r i v a t i v e s of
in-
tensors.
00
We know a l r e a d y that f o r each C - f u n c t i o n f on Μ the f of a (0, l ) - t e n s o r ,
namely the gradient df
If
we,
however,
are the components
partially
=
f
ιa
differentiate
dx a
a vector
field
ν € 5(M),
we
are
118
4. Riemannian Spaces
l e f t with components ν
which u n d e r t h e t r a n s i t i o n t o new c o o r d i n a t e s χ >b behave a c c o r d i n g t o
in p l a c e of t h e χ
v1
.J
=
θ
J
ax 1
v1
a2x'
ax ,J
ax-
axb.
a x ' a x " ax J
ax®
This r e l a t i o n r e m i n d s us of t h e a n a l o g o u s r u l e f o r t h e C h r i s t o f f e l s y m b o l s . A c a r e f u l i n s p e c t i o n shows t h a t t h e a d d i t i o n a l t e r m s in t h e e x p r e s s i o n ν
;b
=
ν
,b
+ Γ
be
ν
e x a c t l y c o m p e n s a t e each o t h e r . In o t h e r w o r d s : t h e ν
!b
form the components
of a t e n s o r of r a n k (1,1). The p r o c e d u r e j u s t c a r r i e d o u t can obviously be e x t e n d e d u n i q u e l y t o a d e r i v a t i o n on a l l t e n s o r f i e l d s :
Definition 4.2.1: The c o v a r i a n t d e r i v a t i v e of a t e n s o r f i e l d of r a n k ( r , s ) on t h e m a n i f o l d Μ r e l a t i v e t o t h e k^1* c o o r d i n a t e is t h e d e r i v a t i o n w i t h r e s p e c t t o "β" which f o r s c a l a r f u n c t i o n s f reduces to the p a r t i a l coordinate derivative f
and
to
,k
+ Γ
kc
ν
f o r a v e c t o r f i e l d v.
In g e n e r a l , we f i n d
Theorem 4 . 2 . 1 : The c o v a r i a n t
d e r i v a t i v e of a ( r . s ) - t e n s o r
f i e l d Τ by t h e i n d e x k is
t e n s o r of r a n k ( r , s + l ) which r e a d s in c o o r d i n a t e s
the
4.2. Christoffel Symbols and Covariant Derivation
a ... a Γ Τ1
b ...b ;k I s
a ... a Γ Τ1
=
a + Γ 1
b ...b ,k I s
p...a τ
r
Y" p r
119
b
.
a ...a τ
1
Γ
kbi
Here,
in the indicated manner,
f o r every upper (contravariant)
additive term has to be inserted,
and f o r each lower
index,
an
(covariant) index a
negative one.
Proof: It is already known that these formulas yield tensors of the correct ranks which,
when specialized to ( 0 , 0 ) - or (1,0)-tensors,
reduce to the
expres-
sions given in the theorem. Insertion into the Leibniz rule leads us to the desired result.
Covariant derivatives can also be introduced if on some manifold Μ there is 3 no metric, but only an a f f i n e connection. By this we mean a system of η smooth real functions (dim Μ = n) which may be chosen a r b i t r a r i l y
in one
coordinate frame. T h e r e a f t e r they have to be extended to other charts of Μ in correspondence with the transformation formula found earlier. We do not attempt to discuss further this possibility as
well
to
the
geodesic
equation
(because
the
latter
which is only
applicable
contains
the
Γβ
be ). The general theory will not be needed in the sequel. Concerning more detailed information, we r e f e r to the modern standard literature on d i f f e rential geometry.
The so-called Levi-Civitä connection originates f r o m the C h r i s t o f f e l bols, calculated a f t e r section 4.1, J.a be It is torsion free:
_ ~
1 ad Γ 28 [8dc,b
+ + gbd,c
Ί gbc,dJ
sym-
120
4. Riemannian Spaces
Another
peculiarity
is
of
great
importance,
since
it
allows
us to
change index s h i f t s with the help of the metric and covariant tion.
inter-
differentia-
The m e t r i c is namely "covariantly constant":
Theorem 4.2.2: All covariant d e r i v a t i v e s of the metric tensor vanish,
g
g
= 0
h
= η 0
ab
5" b;c
= 0
Proof: By theorem 4 . 2 . 1 and the d e f i n i t i o n of the C h r i s t o f f e l symbols we have
g
=
ab;c
g
ab,c
- r
d
ca
g - r db
d
g cb ad
=
0
The c o n t r a v a r i a n t components of g are (as c o e f f i c i e n t s of trix)
rational
f u n c t i o n s of
c o v a r i a n t l y constant. _a s b, c
=
the g
Finally,
and thus,
the inverse
by the chain rule,
ab the d e r i v a t i o n p r o p e r t y of " ; "
,· ad * (g g,h) db ;c
=
g
ad
.„ ;c
db
.ad g grth.„ db;c
+
=
implies 0
A r e l a t e d problem which also leads to covariant d e r i v a t i v e s is the l i z a t i o n of the f a m i l i a r concept t r y in
"parallel
t r a n s p o r t " of
ma-
likewise
elementary
generageome-
IRn.
On account
of
its
linear
lism of
two
vector
even i f
they
are
u(x)
attached
condition u ' ( x ) = v ' ( y ) .
structure, and v ( y ) to
it
is possible
to
in (Rn i r r e s p e c t i v e
different
positions.
In a m a n i f o l d M,
however,
of
This
the
paralle-
reference
frames,
define
is achieved
by
t h e r e is in general
the no
canonical r e l a t i o n s h i p between the tangent spaces at t w o d i f f e r e n t points χ and y.
121
4.2. Christoffel Symbols and Covarlant Derivation
The natural idea to explain parallelism of vectors through the above equation depends on the chart needed to introduce the coordinates and is thus not an intrinsic concept. Nevertheless,
in Riemann spaces (and even in manifolds with a f f i n e connec-
tion) we can at least define the parallelism of
a vector f i e l d with
itself
along a curve y:
Definition 4.2.2: A
vector
field
ν € 8(M)
on
the
Riemannian
manifold
(M,g)
is
parallel
transported ( i n the sense of Levi-Civitä) along the open curve y: I —> Μ if for
all t e I the directional
derivative of
vector y ' ( t ) is equal to zero. y(t)
v a with respect to the tangent
In suitable coordinates, f o r which =
(x 1 ( t )
xn(t))
holds, this condition hence reads
7
b
a
ν
ib
=
r>0
According to theorem 4.1.2, the geodesies subsume under this definition. We merely have to bring the relevant d i f f e r e n t i a l equation by introducing coordinates
ua
the
of the tangent vector of y to the f o r m 'a „a U + Γ
be
b e U U
= 0
or b fa u u
,b
. _a cΙ +Γ u be J
b a = u u ;b
= 0Λ
We thus obtain
Theorem 4.2.3: A curve y in a Riemann space is a geodesic if and only if its tangent tor y ' i s ( L e v i - C i v i t ä ) parallel transported along y.
vec-
122
4. Riemannian Spaces
It is also possible to express the component version of the formula f o r the Lie derivative with covariant rather than partial derivatives.
We only need
to replace the commas by semi-colons:
Theorem 4.2.4: The L i e derivative of an (r, s)-tensor f i e l d Τ with respect to ν € B(M) is, in coordinates,
Proof: A f t e r substitution
of the covariant d i f f e r e n t i a t i o n s by their defining e x -
pressions one notes that all C h r i s t o f f e l symbols cancel.
We are l e f t
with
the same equation in partial derivatives whose validity has been proved a l ready in theorem 3. 5. 5.
4.3. Normal Coordinates
123
4.3. Normal Coordinates We derive f r o m definition 3 . 3 . 3 that tangent vectors are d i f f e r e n t i a l
ope-
rators. A vector f i e l d on an n-dimensional manifold can t h e r e f o r e be i n t e r preted as a system of η ordinary d i f f e r e n t i a l equations in equally many dependent variables (one only has to w r i t e the condition in coordinates). solution consists of
a curve whose tangent vectors are determined
where by the f i e l d .
In geometric language,
The
every-
the appropriate description
is
given by
Definition 4.3.1: Let ν e S(M) be a vector f i e l d on the manifold M. By an integral curve of ν with i n i t i a l point χ € Μ we mean a curve y which is defined on some open interval I S O ? containing 0 which f u l f i l l s y(0)
=
χ
and
r ' Ct) f o r all parameter values
=
v(r(t))
tel.
If no proper extension of y exists which has the same property,
i. e. no in-
tegral curve y: Γ —> Μ of ν with I c I £ R, we call y maximal.
One of the most fundamental results in real analysis is ( c f .
Kamke [1964])
the observation that each system of ordinary d i f f e r e n t i a l equations whose c o e f f i c i e n t functions are infinitely o f t e n d i f f e r e n t i a b l e possesses f o r set
of
initial
conditions
a unique
solution.
The
integral
curve
is
any also
smooth and can be extended to a suitable neighbourhood. For reasons of space limitation, we do not want to investigate how to prove this proposition (a very simple argument employs the integration method of Picard and Lindelöf).
Instead,
we translate the theorem
in question
into
124
4. Riemannian Spaces
the language of d i f f e r e n t i a l geometry.
Theorem 4.3.1: Every vector f i e l d ν on a manifold Μ has f o r any χ € Μ a unique maximal int e g r a l curve with initial point x.
We are mainly interested in the application to geodesies.
Although the r e -
levant d i f f e r e n t i a l equation is of second order, we may reformulate it with a simple trick such that it attains the shape required f o r the above theorem. We just point.
have to choose a chart κ,
say,
of
Μ which contains the
initial
It attaches to a typical element (x, v) of the tangent bundle TM the
coordinate vector (x a ; v a ). With the help of the map , a a-. (χ ; ν )
ι—»
we define a vector f i e l d on TM which by theorem 4.3.1 has a unique (maximal) integral curve. On account of theorem 4.1.2 the l a t t e r is nothing else than the geodesic starting at χ with the tangent vector v ( x ) .
We hence have
Theorem 4.3.2: To each point χ in the manifold Μ and every ν
e TM
there is a uniquely
determined maximal geodesic line y with initial point y ( 0 ) = χ and tangent vector y ' ( 0 ) = ν . X
Let us f o r the moment denote the geodesic with initial conditions ( x , v ) by y
x,v
The result just found implies that y
x,v
is uniquely
defined in some
open interval around 0. Therefore, f o r all (x, v) e TM a positive real number p(x, v) exists such that the domain of y x,v includes the open
interval
between - p ( x , v ) and + p(x, v). From now on we f i x the point χ e Μ and select an arbitrary chart it £ containing x.
Assume that the induced chart κ ( c f .
ates with a typical ν e TM scalar product on
IRn
the coordinates (v 1
^
theorem 3. 4.1) associv n ) e IRn. The Euclidean
is, as usual, written in the f o r m
4 . 3 . Normal C o o r d i n a t e s
125
=
l
u
v
Due t o t h e c o m p a c t n e s s of t h e s t a n d a r d Sphere Π in TM
p
=
mir in
p(x,v)
= R , the number
= 1 j-
X e x i s t s and is > 0. A f t e r a r e p a r a m e t r i z a t i o n we m o r e o v e r f i n d y
x,sv
(t)
=
y
x,v
The i n t e r p r e t a t i o n of t h i s e q u a t i o n is t h a t ,
(st) w h e n e v e r one side is m e a n i n g -
f u l , t h e o t h e r is a l s o and both e x p r e s s i o n s coincide. In t h e p a r t i c u l a r c a s e t = 1 t h i s shows t h a t f o r any t a n g e n t v e c t o r ν € IRn w i t h s ρ
the point y
(1) e Μ is w e l l - d e f i n e d . We t h e r e f o r e o b t a i n
T h e o r e m 4.3.3:
To each p o i n t χ of t h e n - d i m e n s i o n a l Riemannian s p a c e Μ we can c o n s t r u c t a n e i g h b o u r h o o d V of t h e o r i g i n in TM such t h a t f o r a l l ν e V t h e g e o d e s i c X X X w i t h i n i t i a l c o n d i t i o n s (x, v) can be e x t e n d e d a t l e a s t up t o t h e p a r a m e t e r value 1. The a s s i g n m e n t ν
ι—> exp (v) X
d e f i n e s t h e exponential mapping exp : V X
=
X
y
x,v
(1)
—» Μ a t t h e p l a c e x.
It is
obvi-
are
those
ously C .
Of
primary
importance,
especially
R i e m a n n i a n m a n i f o l d s which f u l f i l l ρ
in a p p l i c a t i o n s X
to
physics,
= °o f o r any x. This m o t i v a t e s t h e i n -
t r o d u c t i o n of t h e f o l l o w i n g n o t a t i o n .
D e f i n i t i o n 4.3.2:
A R i e m a n n i a n s p a c e Μ and i t s m e t r i c g a r e c a l l e d geodesically complete if t o each (x, v) e TM t h e g e o d e s i c w i t h s t a r t i n g point χ and i n i t i a l
direction
126
4. Riemannian Spaces
ν can be extended up to a r b i t r a r i l y large real values of its curve parameter.
Almost all Riemann manifolds considered in this monograph s a t i s f y this assumption. We also note the t r i v i a l l y equivalent formulation
Theorem 4.3.4: A Riemannian manifold (M,g) is geodesically complete if and only if the e x ponential mapping exp
X
f o r each χ € Μ has the whole of TM as its domain of X
definition.
The aforementioned construction enables us to introduce distinguished c o o r dinate frames.
To achieve this,
we start
thanks to the implicit function theorem,
out f r o m the observation the map exp
is locally
that,
inverti-
ble. We need a f e w more conventions:
Definition 4.3.3: Let (M, g ) be a Riemann space of dimension η and χ some point in M. We choose a neighbourhood W £ V X X
of Ο e TM , f o r which exp is l nJj e c X X w
tive. The inverse mapping is f
:
U
» TM
The normal coordinates of the point y e U relative to the basis e v J χ I of TM are the numbers y given by the decomposition
f(y)
=
y1
e
e n
j
We w i l l r e f e r to the point χ itself as the origin of the frame thus characterized.
4 . 3 . Normal
Coordinates
127
T h e r e a s o n f o r t h e i n t r o d u c t i o n of t h i s c o n c e p t i s
Theorem 4.3.5: At t h e o r i g i n of any n o r m a l c o o r d i n a t e s y s t e m a l l C h r i s t o f f e l s y m b o l s
van-
ish.
Proof: F o l l o w s i m m e d i a t e l y f r o m a s i m p l e c a l c u l a t i o n w h i c h we s h a l l n o t b o t h e r t o w r i t e down in f u l l .
4. Riemannian Spaces
128
4.4. Curvature The selection of the charts used to describe a manifold Μ is largely trary.
arbi-
Many d i f f e r e n t approaches lead to equivalent representations.
the local ( d i f f e r e n t i a l ) properties defined by the atlas are of
Only
fundamental
importance.
Definition 4.4.1: Two atlases U and U' on the basic space Μ are compatible with each other if they induce the same d i f f e r e n t i a b l e structure. In this case we usually identify the manifolds (M,U) and ( Μ , Ι Γ ) .
Clearly this is an equivalence relation on the class of all conceivable
at-
lases of M. A more practical rephrasing is
Theorem 4 . 4 . 1 : The atlases Ϊ and II' of Μ are compatible if and only if tions k°k'
1
and κ ' « κ
1
all chart
transi-
with κ Ε II and κ' € Μ' are diffeomorphisms of
the
respective domains. A necessary and s u f f i c i e n t condition is that the union U υ IT
itself
is an
atlas.
It is o f t e n d i f f i c u l t to decide whether two given atlases are compatible or whether
two
different
descriptions
lead
to
the
"same"
manifold
(in
the
sense of definition 4.4.1). In the chapters to f o l l o w we shall consider numerous examples of coordinate frames, mainly of the Euclidean spaces, which give an ample confirmation of this statement. Although all compatible atlases are tantamount in principle,
one frequently
will find certain coordinates more convenient than others.
(This has been
the cause f o r writing the book on hand! ).
4.4. Curvature
129
Normally the best choice depends on the intended purpose; some particular manifolds,
however,
possess characterizations of exceptional
mathematical
simplicity. To give an example, this holds f o r charts of Riemannian spaces relative to which the metric has constant components.
We introduce a special notation
to cover this case which is suggested by the experiences from the theory of surf aces.
Definition 4.4.2: A chart κ of the Riemannian manifold (M,g) is f l a t if the components g the metric are constant in space,
i.e.
Ab
of
all Christoffel symbol ( r e f e r r e d to
κ) vanish identically. Analogously, a f l a t atlas is one which consists only of f l a t charts. Finally
the
space Μ itself
is called
flat
if
the
atlas
U^ is
compatible
with some f l a t atlas IT. Η The opposite to " f l a t " is "curved".
In a f l a t coordinate system the covariant derivatives reduce to the partial. Then f o r any vector f i e l d ν e S(M) ν
=
ν
=
;c;d
ν
=
,c,d
ν
,d,c
,d;c
holds. The implied relation a
V
= V
;c;d
a
;d;c
remains valid even in a curved frame, because as a tensor law it is independent of the coordinates. In curved spaces,
on the other hand, it cannot be true f o r all v.
Instead
we have ν
a
;c;d
=
ν •
a
,
>c,d
„a
+
cb
Γ
b
ν
,d
,
_a
+
Γ
db
b
ν
,c
-
_b
Γ
dc
ν
a
,b
(r a c b , d • r a de r e be - r a eb r e d e1l v b
I
an expression which normally is not invariant under the permutation c * d.
130
4. Riemannian Spaces
This strongly
suggests to consider the d i f f e r e n c e v a
- va
;c;d
its v-independent part as a measure of spatial curvature.
;d;c
or
rather
An easy calcula-
tion yields
Theorem 4.4.2: The second order covariant derivatives of any vector f i e l d ν € 8(M) on the Riemann space Μ s a t i s f y the condition
ν where R a
bed
a
;c;d
- ν
a
=
,d;e
_,a
R
bed
ν
b
are the components of a ( l , 3 ) - t e n s o r f i e l d , the Riemann (curva-
ture) tensor. Written down explicitly, R
bed
=
Γ
_
cb,d
these are .
db,c
ds
.
be
cs
db
The analogous rule f o r a 1 - f o r m ω e A 1 (M) is ω
b;c;d
-ω
=
b;d;e
- R
bed
ω
a
while f o r smooth functions F: Μ —> IR F
;c;d
- F
,d;e
holds.
Proof: The quantities Ra M d i f f e r e n c e va
bed
which have been introduced as an abbreviation in the
- va ;e;d
can be found without problems by a simple evalu;d;e
J
r
ation of the covariant derivatives. The remaining relations are gotten similarly by inserting the Here we have to apply the Leibniz rule of d i f f e r e n t i a l calculus.
definitions. That R a bed
are tensor components is a direct consequence of theorem 4. 2.1, the transformation law, and the mathematical properties of the v a .
The Riemann tensor has a lot of symmetries which can be understood best in
4.4.
Curvature
131
the completely c o v a r i a n t form.
Theorem 4.4.3: ( a ) The Riemann curvature of (M, g) can be c a l c u l a t e d f r o m R abed
=
g - Ζ ~ Κ + Ε 2 [ ac,b,d ad,b,c °bc,a,d bd,a,c 1 ef Γ 47 8 I 8ea, c
g
fb,d ~ 8 ea,d 8 f b , c
+
+
ec,a
fd,b
g
g
~ R
+ β
β
be,Γ
- ΰ Β + σ β -σ Β ea ,c °bd,f eb,c ad,f eb,d ac,f
+ Ä
Ä
ed, a
g
ac,e
g
bc,f
~ Κ
bd,f
8
ec,a
R
bd,f
~~ Ρ
fc,b
fd,b
ea,d
R
ed,a
R
ea ,d e f c , b
R
g
+ Κ
ea, c
R
8
eb,d e f c , a
+ R
ec,b
R
eb,c
~ R
ad,f
s
ed,b
σ
fd,a
R
ac,f
ad,,e g b c , f ]
(b) If we i n t e r c h a n g e a with b or c with d, t h e s e c u r v a t u r e components a r e multiplied by -1,
while R
(a, b) « ( c , d ) : R abed
=
abed
remains invariant
- R bacd
=
- R abdc
=
under the
permutation
+ R cdab
and so on.
Proof: Part
(a)
symbols.
is obtained f r o m theorem
4.4.2
by insertion
Proposition (b) is an immediate consequence,
of
the C h r i s t o f f e i
as an inspection
of
the individual t e r m s r e v e a l s .
We saw a l r e a d y t h a t the Riemann t e n s o r R®b space.
Turning
things
around,
the equation
condition which implies the f l a t n e s s have the very handy c r i t e r i o n
vanishes i d e n t i c a l l y Ra
= 0
is
an
in
flat
integrability
bed of the underlying manifold.
We thus
132
4. Riemannian Spaces
Theorem 4 . 4 . 4 : The Riemannian space (M,g) is f l a t if and only if R"
bed
=
r
- Γ
cb,d
_
db,c
ds
_
be
cs
db
is identically true.
Remarks: (a) All formulas discussed in this section which do not contain the metric tensor explicitly can be generalized to (non-Riemannian) manifolds with affine
connection,
but we shall not use this;
compare the
situation
with Neutsch & Scherer [1992], (b) In Euclidean space the Riemann curvature is identically zero, as the quantities derived from it (see below).
Nevertheless
as well the
just
found r e l a t i o n s are valuable, e . g . in order to check the c o r r e c t n e s s of the r a Ki- .>
Two identities s a t i s f i e d by the R ^
are of special interest:
Theorem 4 . 4 . 5 : The curvature tensor of a Riemann manifold is constrained by R"
bed
+ R"
cdb
+ R"
dbc
=
0
and R"
bcd;e
+ R"
bde;c
+ R"
bec;d
= 0
These equations are known as the 1 s t and 2 n d Bianchi i d e n t i t y ,
respective-
iy-
Proof: It is not reasonable here to carry out the necessary calculations in a r b i t r a r y coordinates since the large number of terms would be tiresome.
4.4. Curvature
133
A much better way to proceed is to f i x some point χ ε Μ in advance and employ Γ*
be
normal
(x)
amount
coordinates
cancel; of
with
only their
labour
origin
derivatives
considerably.
at
x.
Then
survive.
The trivial
all
Christoffel
This reduces the
details
of
the
symbols necessary
deduction
are
l e f t to the diligent reader.
We moreover enumerate several tensors related to R a
bed
Definition 4.4.3: The Ricci tensor of a Riemannian space is the contraction of the Riemann curvature tensor given by R
=
ab
R
acb
while the scalar curvature equals
η
R
=
ab
g
y.c
R
.
=
acb
g
ac
g
bd
_ R . „ abed
The Einstein tensor is the expression G
ab
=
R
ab
- - g 2
ab
R
From the Bianchi identities we derive a corollary which is fundamental f o r general r e l a t i v i t y ,
Theorem 4.4.6: The Einstein tensor is divergence f r e e :
134
4. Ri ernannt an Spaces
and symmetric, G
=
ab
G
ba
The l a t t e r property is (in Riemann spaces) shared by the Ricci tensor, R
=
ab
R
ba
Proof:
The symmetries listed in theorem 4 . 4 . 3 lead to R
=
g
R
bd
=
g
R
abed
=
R
cdab
db
and this in turn implies the analogous formula f o r G . Now the requested ab
d i f f e r e n t i a l equation is a consequence of theorem 4 . 4 . 5 . 0
= =
R ab + R ab + R ab be; a ea;b R a -b ; e2 R"
=
R
;e
We have
- R ba
- Rab
be; a
ae;b
and this yields the proposition without e f f o r t .
"Pure curvature" is provided by the Weyl tensor which is also obtained as a modification of R a
bed
Definition 4.4.4:
The Weyl tensor of an n-dimensional Riemannian space (n 2 3) is U*
bed
=
Ra •
bed
+ —
n-2
— — ( n-l) · (n-2)
[ g [ L
δ
bd
d
Ra g
c
be
- g
be
- δ"
c
Ra g
d
bd
+ 3a
c
R
bd
- 3a
d
R
be
R
It has the same symmetries as the Riemann curvature but, in c o n t r a s t to the latter,
all
its
contractions
vanish.
The
importance
of
the
Weyl
tensor
4.4. Curvature
135
stems f r o m the f a c t that two manifolds of dimension η s 4 with metrics can be mapped by a length-preserving map (i. e. are developable) onto each other if and only if the Weyl tensors in corresponding points are equal. In three dimensions,
Wa
vanishes identically; but we are not going to bed discuss the deeper properties of the Weyl curvature since we have no applications f o r them. We merely want to emphasize that the number of independent among the η components of the Riemann tensor is strongly reduced on account of the symmetry and the f i r s t Bianchi identity. It is exactly 1 - n
2 , 2 , , (n-1)
or, in turn, 1,6,20 f o r η = 2,3,4. In the special case of
the theory of surfaces (n = 2),
ents which do not vanish identically are R
1212
, R
1221
the only c o e f f i c i -
, R
2112
, R
2121
. A
tho-
rough consideration shows that the curvature tensor is exactly the (-21f o l d of the Gaussian curvature K: R Since
Κ
is
Gauß [1827], f actor).
essentially
the
only
=
- 2 Κ
curvature
invariant
of
a
surface,
cf.
such a result could have been expected (save f o r the precise
136
4. Riemannian Spaces
4.5. Volume
For
certain
investigations
concerning
tangent,
cotangent
or
other
vector
b u n d l e s o v e r some m a n i f o l d , t h e c o o r d i n a t e f r a m e s d e f i n e d by t h e c h a r t s do not
suffice.
Frequently,
however,
a generalization
which
suggests
itself
t u r n s o u t t o be h e l p f u l :
D e f i n i t i o n 4.5.1:
(a) A covariant n-Bein on an n - d i m e n s i o n a l m a n i f o l d Μ is a s y s t e m of η v e c tor fields e
e such t h a t t h e l o c a l v e c t o r s e ( x ) , . . . , e (x) a t a l l i n i n p l a c e s χ e Μ a r e l i n e a r l y i n d e p e n d e n t and t h u s f o r m a b a s i s of t h e t a n g e n t s p a c e TM . (b) C o r r e s p o n d i n g l y ,
a contravariant n-Bein is an n - t u p e l 1
f o r m s which yield in e v e r y point χ a b a s i s e ( x ) tangent
e1
e" of
n
e ( x ) of t h e
1co-
space.
(c) Two n - B e i n e e l
e° and e 1
e η a r e dual t o each o t h e r if t h e d u a -
lity condition < e ' ( x ) , e (x)>
(d)
is identically f u l f i l l e d .
Μ i s c a l l e d parallelizable i f t h e r e
=
i'j
i s a global n - B e i n o n M.
It is e v i d e n t t h a t t o every ( c o n t r a - o r c o v a r i a n t ) n - B e i n on Μ t h e r e is a uniquely d e t e r m i n e d dual one. of t h e dual b a s i s .
It can be f o u n d by a p o i n t w i s e
construction
(The s m o o t h dependence of t h e p o s i t i o n f o l l o w s f r o m t h e
i m p l i c i t f u n c t i o n t h e o r e m ) . T h e r e f o r e t h e t y p e of t h e n-Bein needed in p a r t (d) is i m m a t e r i a l . The p a r a l l e l i z a b i l i t y
of t w o m a n i f o l d s Μ and Ν i m p l i e s t h a t of t h e
direct
p r o d u c t Μ χ N. In p a r t i c u l a r ,
it is e a s i l y seen t h a t R and S 1 a r e p a r a l l e l i z a b l e ;
E u c l i d e a n space
thus
the
4 . 5 . Volume
137 ,n R'
and the
R χ ...
χ R
torus
.π Τ' have t h e s a m e p r o p e r t y . A much m o r e i n t e r e s t i n g q u e s t i o n is which spheres can be p a r a l l e l i z e d . leads to extremely
deep c o n n e c t i o n s
with
algebra
and a l g e b r a i c
It
topology
(K-theory). We s h a l l come back t o t h i s p r o b l e m in s e c t i o n 11.5 w h e r e
more
e x p l i c i t i n f o r m a t i o n is given. The f o l l o w i n g t e r m i n o l o g y is a d i r e c t a n a l o g y t o t h e c o r r e s p o n d i n g n o t a t i o n concerning coordinate frames.
D e f i n i t i o n 4.5.2:
A c o v a r i a n t n-Bein e called
e on t h e n - d i m e n s i o n a l R i e m a n n i a n s pr a c e Μ w i t h i n orthogonal if f o r all i * j t h e e q u a t i o n g i e ^ e ^ ) = 0
metric
g is
holds.
If we have in a d d i t i o n g(e , e ) = ± 1,
and c o n s e q u e n t l y t h e
condi-
tion gCe^e ) holds f o r
all
i , j 6 {1
n},
=
±δ
i
t h e n-Bein e , . . . , e is ortho1 η
we say t h a t
normal.
R e l a t i v e t o an o r t h o n o r m a l c o o r d i n a t e s y s t e m , a l l c o m p o n e n t s of t h e m e t r i c t e n s o r a r e e q u a l t o one of 1, -1 or 0 and f o r r e a s o n s of c o n t i n u i t y ant.
Hence a l l
C h r i s t o f f e l s y m b o l s and a f o r t i o r i
tensor vanish identically.
t h e Riemann
const-
curvature
O r t h o n o r m a l c o o r d i n a t e s t h u s e x i s t only in f l a t
manif o l d s . In m a r k e d c o n t r a s t t o t h i s , every given n-Bein can be o r t h o n o r m a l i z e d .
One
p a r t i c u l a r m e t h o d is t o apply t h e w e l l - k n o w n p r o c e d u r e of Gram and S c h m i d t s i m u l t a n e o u s l y in a l l p o i n t s of t h e s p a c e .
Obviously,
normal
of
n-Bein.
An i m p o r t a n t
application
t h i s y i e l d s an
this observation
is t h e
s t r u c t i o n of a d i s t i n g u i s h e d volume f o r m on an o r i e n t e d R i e m a n n i a n f o l d , t o which we come n e x t .
orthoconmani-
138
4. Riemannian Spaces
Theorem 4.5.1:
Let Μ be an oriented n-dimensional Riemann space. For each χ e M, we choose a (positively oriented) orthonormal basis a 1 , . . . , « " of the cotangent space Τ Μ and introduce the abbreviation X ωRt( Χ )\ = α 1 The assignment χ ι—»
Λ
...
Λ
n
α
determines a nowhere vanishing η-form
e Λη(Μ)
w h i c h i s c a l l e d t h e Riemann volume form of M.
The i n t e g r a l vol Μ
J".
of ωR over Μ is the (Riemann) volume of M. We likewise introduce the volume of a domain Β £ Μ via vol Β =
ω J R Β
in case the integral has a meaning.
Remark:
The l a s t definition is a proper extension of the former, since the i n t e g r a l on Β can e x i s t s even if Β is not a manifold. The closed n-dimensional unit ball Kn = I χ e IRn J |x| 5 1 I may serve as an example. It has a f i n i t e volume (theorem 4 . 5 . 3 ) though the boundary points which form the sphere Ωπ do not possess neighbourhoods homeomorphic to Rn, a s is required f o r points of a manifold. In f a c t ,
Κ
i s a manifold with boundary. In the f o l l o w i n g i n v e s t i g a t i o n s w e
4.5.
Volume
139
s h a l l n o t make use of t h i s c o n c e p t . For t h a t r e a s o n we r e f r a i n f r o m d i s c u s s i n g it in g r e a t e r
detail.
Proof:
» Had we s e l e c t e d a n o t h e r b a s i s of Τ Μ , like e . g . β o b t a i n e d t h e same
ω
Η(
χ
I
n β , we would have
)> b e c a u s e , f r o m t h e r u l e s of l i n e a r
β1 Λ ...
algebra,
= d e t ( x ) α 1 Λ ... Λ α η
Λ β"
w h e r e τ d e n o t e s t h e i n v e r t i b l e mapping of t h e c o t a n g e n t s p a c e o n t o
itself
which is u n i q u e l y d e t e r m i n e d by τ ί α 1 ) = β 1 . On a c c o u n t of t h e a s s u m e d o r t h o n o r m a l i t y of t h e a 1 and t h e β 1 , t h e m o d u l u s of d e t ( x ) e q u a l s 1; m o r e o v e r τ is o r i e n t a t i o n - p r e s e r v i n g ,
hence d e t ( x ) > 0.
T h i s c o m b i n e s t o d e t ( x ) = 1, and 1
α
n
λ1
Λ . . . Λ α
β
=
Λη Λ ... Λ
β
t h e b a s i s a 1 and ι z e r o s i n c e t h e r e a r e no n o n t r i v i a l l i n e a r r e l a t i o n s b e t w e e n t h e α . Consequently,
If
ωR (χ) is i n d e p e n d e n t
we i n t e r p r e t
α1,...,a"
as the
of t h e choice of
local
values
of
an
orthonormal
not
n-Bein
which is d e f i n e d in some n e i g h b o u r h o o d of χ ( o b t a i n e d f o r i n s t a n c e w i t h t h e help
of
the
Gram-Schmidt
method
d x 1 , . . . , d x n of a s u i t a b l e c h a r t ) ,
from
the
we see t h a t
U
coordinate
X R ( ) is
a
differentials
C°°-f u n c t i o n of
χ
and h e n c e an n - f o r m . For l a t e r applications,
we a l s o n o t e
Theorem 4 . 5 . 2 : As u s u a l , we d e n o t e t h e s t a n d a r d c o o r d i n a t e s of some p o i n t χ in E u c l i d e a n , 1 η η - s p a c e by χ χ . ( a ) The Riemann volume f o r m of IRn is n o t h i n g e l s e b u t t h e E u c l i d e a n volume element ω (χ)
=
dx 1
Λ
. . . Λ
dx"
140
4. Riemannian Spaces
( b ) On the ( n - l ) - s p h e r e
Ωη
=
{ x
6 R" I |x| = 1 }
x
S-1
u r may be brought to the f o r m
ω3(χ)
=
^
, i-l ,1+1 , π ( - 1 ) 1 x1 dx1 Λ . . . A αχ Λ dx Λ . . . Λ dx
Proof: Proposition ( a ) is t r i v i a l .
We introduce the abbreviation r f o r the Euclid-
ean distance of the point (x 1 , . . . , x n ) e IRn f r o m the origin.
It is calculat-
ed via Γ
2
=
, 1,2 (χ ) + . . .
, , n,2 + (χ )
which by d i f f e r e n t i a t i o n implies the relation . r dr
=
χ
I i i dx
+ ... + χ
The sphere Ω is by definition the set π equation r = 1 in Rn. Therefore
the
restriction
of
an ( n - l ) - f o r m
requested spherical volume f o r m provided dr holds.
Α
ω
of
=
ω = R
all
η , η dx solutions
of
the
algebraic
ω e Λ° 1(IRn) on Ω yields π
dx1 A . . .
the
A dx"
Through direct evaluation of the product we check that this condi-
tion is obeyed by ω = ω .
The total volume of Rn is of course oo. In contrast to this, compact domains always have f i n i t e measures. For the investigations to be resumed in part F (chapters 24 and 25) we need
4.5.
Volume
141
Theorem 4 . 5 . 3 : ( a ) The n - d i m e n s i o n a l b a l l Κ
of radius 1 has the volume
vol Κ 2k
-
k π j-jk!
f o r even dimension η = 2k and _ vol Κ 2k+i if
η = 2k+l is odd.
a l l dimensions,
k! _2k+l k -rrr,—r-rr 2 π (2k+l)!
=
Both f o r m u l a s combine into one which is v a l i d
for
namely
Μ
vol Κ
( b ) The s u r f a c e of Κ
is found f r o m
vol Ω '2k
, „ vol Ω 2k+l
(k-1)!
k! _2k+l k ,, 2 π (2k)!
=
depending on the p a r i t y of n. If w e p r e f e r the more g e n e r a l w i t h the Gamma f u n c t i o n ,
expression
we get
2 πη/2 vol Ω
= Γ,
2
( c ) Volume and s u r f a c e a r e connected by the Archimedian r e l a t i o n
νοΚΩ )
=
n·vol(K )
Proof: F o r the moment we denote the t o p o l o g i c a l l y c l o s e d m - d i m e n s i o n a l b a l l o f
ra-
142
4. Riemannian S p a c e s
dius r a s Κ (r). η The c a l c u l a t i o n of the volume of Κ (r) can, m t h a t of Κ (1) = Κ : m m vol Κ ( r )
=
by r e s c a l i n g ,
be reduced
to
r m vol Κ
In the p r o o f of theorem 4 . 5 . 2 we applied the r e l a t i o n ω^ = drAu g between the volume f o r m s of the Euclidean s p a c e and the sphere. This l e a d s to
vol Ω η i.e.
=
^ vol Κ ( r ) σΓ η
=
vol Κ (γ) ^ η βΓ
η·vol Κ
Archimedes' theorem. This in turn a l l o w s us to deduce the volume f o r -
mula f o r Ω f r o m t h a t of Κ . η π The i n t e r s e c t i o n of the hyperplane χ
= c e [-1,1] and Κ = Κ (1) is a ball η η
i s o m e t r i c to Κ (r) with the r a d i u s n-l
We obtain
vol Κ
=
vol Κ
dc
1(n-l)/2
-
C=-l
[-1
vol Κ dc n-l
C=-l
and thus f o r the r a t i o of the volumes f o r two s u c c e s s i v e n:
vol Κ vol Κ
1
-T
1-c"I i c=-l
(n-l)/2
π
dc
=
s i n α da a=0
Here we used the t r i g o n o m e t r i c s u b s t i t u t i o n c = cos a which s u g g e s t s s e l f . The only remaining problem is the evaluation of the i n t e g r a l
s i n α da a=0
it-
4 . 5 . Volume
143
The transformation η πα
Γ, 2 1 n-2 I I - cos α s i n α
=
(k £ 2) a f t e r partial integration and a slight simplification give ι
=
η
5=1 ι Π n-2
as well as I I η n-1
=
n-2 J n-1 n-2
η
n i l η n-1
χ
n-3
=
(n-2) I
ο
=
π
=
2
=
n-2
η
I
l
n-2
l
n-3
n-3
With the elementary integrals I
I
ι
we get the proposition inductively. The t r a n s l a t i o n to the version c o n t a i n ing the r - f u n c t i o n i s an immediate consequence of i t s functional equation.
The d e f i n i t i o n of the volume form as given in theorem 4 . 5 . 1 i s a b i t
incon-
venient because i t r e q u i r e s the s e l e c t i o n of an orthonormal n-Bein.
There-
f o r e a r e p r e s e n t a t i o n r e l a t i v e to an a r b i t r a r y (oriented) frame w i l l be of interest.
Theorem 4 . 5 . 4 : The volume form ω of an n-dimensional oriented Riemann space can be c a l c u l a t e d in any p o s i t i v e l y oriented frame χ ι—» (χ 1
=
/
det gl
dx
Λ . . . Λ dx
χ η ) through
144
4. Riemannian Spaces
Here g is t h e m a t r i x of t h e c o v a r i a n t m e t r i c c o m p o n e n t s
8,J
=
g (
W
Proof: The f o r m u l a in q u e s t i o n is c l e a r l y c o r r e c t a t χ 6 Μ if we u s e n o r m a l c o o r d i n a t e s w i t h o r i g i n χ ( s e c t i o n 4 . 3 ) . The t h e o r e m f o l l o w s f r o m t h e f a c t t h a t b o t h s i d e s t r a n s f o r m in t h e same way if we apply an o r i e n t a t i o n - p r e s e r v i n g c h a r t t r a n s i t i o n to them.
4.6.
Duality
145
4.Β. Duality Aside f r o m the Cartan derivative,
there is still another linear t r a n s f o r m a -
tion on the Graßmann bundle to an oriented Riemann manifold which is independent of the coordinates,
namely the duality.
In contrast
differential
mapping;
to
the
Cartan
the
latter
is
a purely
algebraic
t h e r e f o r e we want to introduce it f i r s t on the Graßmann algebra
A(V) of a suitable vector space V.
Definition 4.6.1: Let V be an n-dimensional real vector space, endowed with the nondegenerate symmetric scalar product < . , . > . We choose a positively oriented orthonormal basis e , . . . , e . l η This means that = 0 holds f o r i * j ,
while all norms
equal ± 1. The volume form is calculable by ω
e
a
ι
...
λ
e
η
The algebraic duality is the linear map »:
Λ( V) α ι—»
Λ( V) *
α
with *
T(k)
n(k+l)
λ
...
λ
e
π(η)
f o r 0 s k a η and π e S . Π
The Graßmann algebra A(V) is the direct sum of its subspaces A k (V) each of
146
4. Riemannian Spaces
which is spanned by the k - f o r m s e
1(1)
Λ.,.Λβ
. Thus there cannot be more n(k) .
than one linear mapping with the required properties. That
it
really
exists
is
a consequence
of
the
anticommutativity
of
the
wedge product of 1 - f o r m s which is compensated by the f a c t o r sgn(n) and of the nonexistence of further linear relations between the generators. It is as simple to see that the definition is independent basis. An alternative formulation is that the product its dual coincides up to sign with the volume form:
[
e
n(l)
Λ.,.Λβ
1 λ n(k)J
[e Λ...ΛΘ 1 [nil) it(k)J
=
of
the
selected
eπ(ΐ) λ. . . Λβ*(k)
with
lie || . . . 11||e 11II ω n(l)" n(k)
11
This property is retained if we move over to any other orthonormal
refer-
ence f r a m e . We compile some t r i v i a l information about the algebraic duality whose proof f o l l o w s directly f r o m the definition.
Theorem 4.6.1: ( a ) The algebraic duality over the (oriented) vector space V = Rn is a l i near b i j e c t i o n between the k - f o r m s and the ( n - k ) - f o r m s . ( b ) Dualizing
twice
reproduces
α e Λ (V)
except
for
a sign f a c t o r
which
arises in some cases. The precise formula is ##
α
=
(-1)
k(n-k)
sgnidet g ) a
Here sgn(det g ) is the sign of the metric determinant.
If the metric is
positive definite and the dimension odd, "*" is an involution.
The duality operation allows us to introduce a scalar product on the Graßmann algebra.
Definition 4.6.2: Every nondegenerate symmetric scalar product < . , . > on the Graßmann algebra A(V) over V = IRn (with orientation) extends to a product (denoted with the same symbol) which is uniquely determined by bilinearity and the f o l l o w i n g conditions:
4.6.
Duality
147
( a ) For α e A k (V) and β 6 A m (V) we have «χ, β> = 0,
whenever k and m are
distinct. (b) If,
however, the ranks of a and β coincide we get
α Λ
*
=
β
ω
where ω means the (positively oriented) Euclidean volume f o r m ,
i.e.
in
standard coordinates ω
One should notice that
=
e
ι
Λ
...
this definition
Λ
e η
is compatible
with
the
identifica-
tion A ^ V ) V. The η - f o r m dual to the C°°-function (= 0 - f o r m ) f is f ω. It is also possible to calculate the scalar product without explicit
refer-
ence to an orthonormal basis:
Theorem 4.6.2: The scalar product of two special k - f o r m s
Α
=
u 1
Λ
...
Λ
u k
Β
=
Ν
Λ
...
Λ Ν
and
(u , ν 1 1
1
k
e V) can be represented as the determinant of the matrix : ι j =
det
The connection with the duality is *
=
# («Α
# Β)
Proof: Follows immediately f r o m the definition.
•
* (βΛ
Α)
148
4. Riemannian Spaces
We c a r r y the a l g e b r a i c duality just introduced over to m a n i f o l d s .
Definition 4.6.3: Assume Μ is an o r i e n t e d n-dimensional Riemann space with m e t r i c tensor g. *
The duality assigns to each k - f o r m ψ an ( n - k ) - f o r m ψ. » * The local value ^rfx) of ψ at the place χ e Μ can be found as the image of k * lir(x) e • Α (Τ Μx ) under the a l g e b r a i c duality over the cotangent space V = Τ Μ by evaluating the scalar X
=
product
g(a, β)
=
g I J at β} ·
* Clearly
ψ(χ) is a smooth f u n c t i o n of x.
r e n t i a l f o r m ( o f rank n-k). well.
Hence
ψ is an a l t e r n a t i n g
diffe-
On A(M) a scalar product can be constructed as
In c o n t r a s t to its a l g e b r a i c counterpart,
however,
it r e q u i r e s an i n -
tegration:
Definition 4.6.4: The scalar product of
two differential
f o r m s α and β of
rank k over
the
With the understanding that A k (M) and Am(M) are supposed to be mutually
or-
oriented Riemann m a n i f o l d Μ reads
κ
=
J α Λ *β Μ
thogonal if k * m, we extend
l i n e a r l y to A(M).
4.7. C l a s s i c a l Vector Analysis
149
4.7. Classical Vector Analysis
The w o r l d we are living in is three-dimensional and to a remarkable degree of accuracy Euclidean. This is the reason why many results concerning mani3
f o l d s have been found f i r s t f o r the p a r t i c u l a r case of R . So,
for
instance,
i n f i n3i t e s i m a l in R
already
calculus
it
shortly was
after
noticed
the
that
discovery
several
e x i s t which have a g e o m e t r i c meaning,
(or
invention?)
differential
of
operators
independent of coordinate
sys-
tems. Most important among them are three invariant d i f f e r e n t i a t i o n procedures of f i r s t order,
namely gradient (grad), curl or rotation (rot), and divergence
(dlv). They are accompanied by the Laplace operator, normally denoted Δ, which of all d i f f e r e n t i a t o r s of
second order is by f a r the simplest and most
widely
used. Since the Cartan differential d is the only invariant d e r i v a t i o n on Riemann spaces known to us, we suspect that the above-mentioned mappings might be expressible
with
its
help.
This
would
immediately
imply
their
invariance
under chart changes. As concerns the g r a d i e n t ,
this is indeed true,
while f o r
the other
f o r m a t i o n s in addition the duality, which is also independent of dinates, To
begin
trans-
the
coor-
operators
under
is needed. with,
we w r i t e
down e x p l i c i t l y
the f i r s t
discussion in Cartesian coordinates ( x , y , z ) . not employ the index notation but f o l l o w
order
For the sake of
c l a r i t y we do
instead the old t r a d i t i o n
press the components of v e c t o r by a d j a c e n t l e t t e r s of the alphabet. 3 The duality over R is c h a r a c t e r i z e d by the f o r m u l a s *
=
1
* *
dx
Λ
dy
dx dy
= =
dy dz
Λ Λ
dz dx
dz
=
dx
Λ
dy
Λ
dz
to
ex-
150
4. Riemanniaxi Spaces
* #
*
The
application
of
Cartesian
(dyAdz)
=
dx
(dZAdx)
=
dy
(dxAdy)
=
dz
(dXAdyAdz)
=
1
(orthonormal)
coordinates
blurs
the
distinc-
tion between c o n t r a - and covariant v e c t o r f i e l d s because the m e t r i c is g i v en by the unit m a t r i x and thus reads in components either g = δ or 1 1 1 1 J g = δ . T h e r e f o r e , one must be a w a r e of the c o r r e c t i n t e r p r e t a t i o n when one wants t o g e n e r a l i z e to a r b i t r a r y charts.
This task is made easier if
we
t r a n s f e r the r e l e v a n t equations into Cartan calculus. In classical vector analysis,
it
tention to (smooth) functions,
i.e.,
variant)
is common p r a c t i s e
vector fields = 1-forms.
to r e s t r i c t
in f a n c i e r language,
Sections
2
3
at(co3
3
in the bundles A (R ) and A (R ) 1 3
0
3
are r e p l a c e d by t h e i r duals which lie in Λ (R ) or Λ (R ), want to f i n d the o p e r a t o r s grad, r o t ,
one's
0 - f o r m s and
respectively.
and div in components.
We
The most c o n -
venient p r e s e n t a t i o n of the intended investigation s t a r t s out f r o m the C a r tan d e r i v a t i v e d. We begin with a 0 - f o r m f . 1-form.
Because of d(A°) £ A1, the d i f f e r e n t i a l of f is a
By section 3 . 5 , df
=
+
+
w h i l e the gradient r e l a t i v e our basis is .. 8rad
The g r a d i e n t
thus
f
is nothing
= else
fäf äf 8fl [ s* · - äi J but
an a l t e r n a t i v e
way
of
writing
the
d i f f e r e n t i a l of a 0 - f o r m . 1 2 Somewhat more complicated is the situation f o r 1 - f o r m s . Now d(A ) S A , and we have to dualize a f t e r having d i f f e r e n t i a t e d in o r d e r to come back t o a 1-form. Let the given f i e l d be (u, v, w ) in c l a s s i c a l , dern notation.
We f i n d
i.e.
u dx + ν dy + w dz in m o -
4.7. Classical Vector Analysis
d (u dx +
ν
dy + w dz) faw
=
[ay
-
=
dvl
.
alj
d y
151
du
dx + dv
Λ
. Λ
fsu
λ
dy + dw
awl
.
, fav
au| , "
Έ9J
faw 3v1 . fau aw") , ^ dx • ^ dy
+
fav au") , ^ - ^ J dz
+
axj
,
[a*
d z
[ΈΈ -
dz
λ
d z
Λ
d x
+
d x
Λ
d y
and, consequently, *_• / ι d (u dx • V dy
... w dz)
=
. , . rot (u,v,w)
=
+
On the other hand,
and we conclude rot =
*
f 3w
3v
au
3w
3v
3u Ί
3
d f o r 1-forms in IR .
Next we consider the Cartan derivatives of 2-forms.
2
3
Since d(A ) S A , we
dualize before and a f t e r the differentiation. In other words: we calculate * · d (u dx + ν dy + w dz). The 2 - f o r m dual to u dx + ν dy + w dz is * (u dx +
ν
dy + w dz)
=
u dy
λ
dz +
ν
dz
λ
dx + w dx
dy
λ
Differentiation yields d
*
(u dx + ν dy + w dz)
=
Γ 3u 3v 3w 1 I + g y + ^ I dx
λ
dy
λ
dz
and a second transition to the dual form provides us with *
, * . d (u dx
+
. ν dy
+
. , w dz)
=
3u 3v _ + _ +
aw _
This can be identified as . . . , div (u,v,w)
=
3u dx
+
dv 3w + Έdy dz
The only remaining case, namely the application of d to 3-forms, does not lead to anything of interest since the three-dimensionality of the underly3
4
ing space implies d(A ) S A
=0.
It is evident that gradient and divergence due to the descriptions with "*" and "d" can be immediately generalized to all IRn (and even to
arbitrary
oriented manifolds). In contrast to this, the given formula f o r the curl is useful merely
in 3-dimensional
spaces because
dtA1) S An
2
if
η denotes
152
4. Riemannian Spaces
the dimension.
If η £ 3, the Cartan derivative of a 1 - f o r m , whose rank is
only 2 would be handier than the ( n - 2 ) - f o r m discussed above. Thus, one n o r 3 mally r e s t r i c t s the notion "curl" to R or - a bit more general - to 3 dimensional manifolds. For reasons of completeness we also give the 2-dimensional counterpart. Now * d(A ι ) £ Λο , and the curl is defined via the explicit expression rot while,
3v öu ax " ay
(u,v)
on the other hand, d (u dx + ν dy)
£ ^ - j dx λ dy
=
and t j ^ j ^ d (u dx 4. V dy)
av au g j - g^
=
*
hold. Hence in the present case "rot" must also be interpreted as " d". We collect the previous deductions and results of this section.
Theorem 4.7.1: ( a ) The gradient of
a C™-function f : Rn —» R is the 1 - f o r m given by
the
differential, grad f
=
df
( b ) The divergence of a 1 - f o r m α in Rn is the O-form div α 2
( c ) In R
=
*
*
d a
3
or R , the curl of a 1 - f o r m a is calculated by rot α
=
#
da
In Euclidean 2-space, rot α is a O-form, in 3-space, on the other hand, a 1-form.
4 . 7 . C l a s s i c a l Vector Analysis
153
If we want to construct chart-independent d i f f e r e n t i a l operators of second order, we must, on account of the Cartan rule dd = 0, insert the transition to the dual between the d i f f e r e n t i a t i o n s . candidates are combinations of d d, These four mappings transform,
Therefore the only
d d, d d , and
in turn,
conceivable
d d .
k - f o r m s to d i f f e r e n t i a l forms of
ranks n-k, k, k and n-k. The second and third of them leave the rank invariant. They are thus of particular interest.
We introduce some more n o t a -
tion.
Definition 4.7.1:
Let Μ be an oriented Riemann space of dimension n. (a) We define by means of the Cartan d i f f e r e n t i a l
d the co-differential
(the co-derivative) δ through
f o r k - f o r m s and extend this linearly to Λ(Μ) = A°(M) ® . . .
Θ Λη(Μ).
(b) T h e Laplace-Beltrami operator i s
Δ
- d δ - δ d
and in p a r t i c u l a r , when applied to a k-form, Δ
=
(-l)nk
d d + (-1)
( c ) A k - f o r m α is closed if da = 0,
and exact if there is a ( k - l ) - f o r m β
f o r which dß = α holds. (d) Dually, α is co-closed whenever δα = 0,
and co-exact provided a (k+1)-
form β with δβ = α e x i s t s . ( e ) Finally, by a harmonic form we mean an a e A(M) with Δα = 0.
The somewhat strange looking sign convention is explained by the f a c t that in Cartesian coordinates of IRn the Laplace operator reduces to the e x p r e s sion
154
4. Riernanntan Spaces
Δ
=
a2 + . .. 1
as is f a m i l i a r f r o m the c l a s s i c a l
theory.
+ a2 η
For compact m a n i f o l d s ,
S is the
a d j o i n t o p e r a t o r o f d. We combine the last statement and a f e w o t h e r mentary p r o p e r t i e s o f d, δ,
ele-
and Δ to
Theorem 4.7.2: L e t Μ be an o r i e n t e d Riemannian space of dimension n. Then L a p l a c e o p e r a t o r and duality map commute:
* Δ
=
Δ *
=
Δ(*α)
i. e. , f o r a l l a e A(M),
*(Δα)
I t e r a t i o n o f d or δ y i e l d s
If Μ is compact,
zero;
d d
=
0
δ δ
=
0
we f u r t h e r m o r e have
( a ) Cartan d e r i v a t i v e
d and c o - d e r i v a t i v e δ are a d j o i n t mappings w i t h
spect t o the s c a l a r product d e f i n e d by
J α Λ
=
Β
Μ ( t h i s is nondegenerate and even p o s i t i v e
=
definite):
( b ) The L a p l a c e - B e l t r a m i o p e r a t o r Δ is s e l f - a d j o i n t :
=
re-
4.7. C l a s s i c a l Vector A n a l y s i s
155
( c ) A Is n e g a t i v e d e f i n i t e :
s
0
with equality only f o r α = 0. (d) A differential
f o r m α e AIM) is harmonic
and c o - c l o s e d .
It
already
s u f f i c e s to
if
and only
require
the
if
it
is
closed
orthogonality
of
α
and Aa:
= 0
( e ) When a is harmonic,
Aa = 0
(da =
0
δα =
Λ
0)
the same is true f o r the dual f o r m
*
a.
Proof: k Consider
*
a e A (M).
calculation.
The equality
Beyond
this,
the
of
* (Aa) and A( a )
Cartan
rule
is v e r i f i e d
(definition
3.5.5)
by
direct
implies
that
dd vanishes i d e n t i c a l l y and thus *
δ δα
=
±
*
*
*
d
*
da
=
*
± d d a
=
0
(the p r e c i s e signs do not m a t t e r ) . We moreover know that the scalar product < . , . > : and p o s i t i v e d e f i n i t e .
A(M)xA(M) —» IR is symmetric
By d e f i n i t i o n ,
-
=
J |da Λ *β - α Λ *ößj Μ
if a and β are f o r m s of
the ranks m-1 and m,
l e f t hand side is obviously equal to 0). * da
Λ
* β
-
Α Λ
* δβ
=
da
Λ
and Stokes' theorem 3. 9. 2 leads to
-
respectively
m-l β + (-1)
=
(otherwise
The integrand is an exact f o r m : * Α Λ
J d(aA*ß) Μ
*
d β
=
0
=
d (ΟΛ
Β)
the
156
4. Riemannian Spaces
This is p r o p o s i t i o n ( a ) ,
which has (b) as an immediate consequence.
State-
ment ( c ) f o l l o w s f r o m
-
=
=
+
since is p o s i t i v e
definite.
The
last
this holds f o r δα and da simultaneously. part of At l a s t ,
+
expression
2
can only
0
be z e r o
if
We thus have proved the n o n t r i v i a l
(d). ( e ) can be deduced easily f r o m
The compactness of
Μ is essential.
(a).
For example,
a harmonic
function
(0-
f o r m ) f by a s s e r t i o n (d) has a vanishing d i f f e r e n t i a l and hence is constant on each connected component of M. Harmonic f u n c t i o n s on compact m a n i f o l d s a r e t h e r e f o r e quite boring. For noncompact
spaces like Rn the situation
is completely
many harmonic f u n c t i o n s e x i s t which are not constant.
different.
Here
We postpone a t h o r -
ough discussion to p a r t F of this book. There we shall g i v e numerous i m p o r tant applications nections with
(spherical
other
harmonics,
problems
as,
Lam£ f u n c t i o n s ) and discover
e. g. ,
numerical
integration
on
con-
spheres,
etc. By the
Cartan
rules
dd = 0 and δδ = 0 every
exact
differential
form
is
closed (and e v e r y c o - e x a c t f o r m c o - c l o s e d ) . The question,
to which extent the r e v e r s a l of this statement is c o r r e c t
a given m a n i f o l d , Tu [1982],
is a topic of cohomology theory ( c f .
Warner [1983]).
Here
case which w i l l be needed l a t e r .
we only
want
to
for
f o r instance Bott &
consider
a very
special
Assume α is a closed 1 - f o r m on the simply
connected m a n i f o l d M. We f i x a basic point X q € Μ and introduce a smooth f u n c t i o n F e Λ°(Μ) through
F(x)
=
J
y
α
where y denotes an a r b i t r a r y path leading f r o m χ to x. The in f a c t is a 0 » meaningful d e f i n i t i o n . If y w e r e another curve connecting X q with x, we could
on account
of
the homotopy
of
y and y
construct *
a surface
(two-
dimensional submanifold) * Γ £ Μ with boundary y - y ( f i r s t y in the f o r w a r d d i r e c t i o n ; t h e r e a f t e r y backward). But in this case the theorem of Stokes
4.7. Classical Vector Analysis
157
would y i e l d
ί α " ί , α = ί * α = Ι α = Jda = ° r
γ
3Γ
γ-γ
Γ
and hence
ί" • J.· y
r
F i x ) thus is independent of the p a r t i c u l a r curve y we use. By a simple calculation we check without d i f f i c u l t y the exactness of a:
dF
=
grad F
=
α
This is the special case k = 1 of the f o l l o w i n g Poincarfe lemma:
Theorem 4 . 7 . 3 : Closed k - f o r m s (k £ 1) on simply connected m a n i f o l d s are e x a c t .
Proof: Thanks to the simple connection we can r e s t r i c t The natural Cartesian coordinates of
Rn
are,
the discussion to Μ = IRn.
as a l w a y s ,
χ1
χη.
The L i e
d e r i v a t i v e r e l a t i v e to the radial v e c t o r f i e l d
ν
=
χ
a
y i e l d s f o r a k - f o r m ω in components * a ω) ν a...a
= ω
l k
a . . . a , m
l
=
m χ + ω
k
xm 9
m
ω
a ... a
1
m m...a
χ k
,a
+ . . . + ω 1
1
+ k ω
a ... a
k
1
k
and vanishes only if
χ
θ
m
ω
a ... a
1
= k
- k ω
a ... a
1
a...m
k
m χ
.a
k
158
4. Riemannian Spaces
holds. The l a t t e r condition is by the Euler homogeneity theorem 2. 5. 1 t a n t amount to ω being homogeneous of degree ( - k ) in the x™. a .. .a 1 k For k > 0 the r e g u l a r i t y of ω at 0 presupposes it to vanish i d e n t i c a l l y , while f o r O - f o r m s we can merely conclude that ω is constant. We now assume k k Then L : Λ (Μ) —> Λ (Μ) is i n j e c t i v e , and there must be a ( l i n e a r ) V map α of Λ (Μ) into i t s e l f which is a l e f t inverse of L : k ν k a 1.
α ο L k ν
=
id
It is not hard to see that the t r a n s f o r m a t i o n given by
α : k
f dx
1
λ
. . .
f o r all ρ e R
c^oL^jfdx
λ
dx
k
(ρ)
J
ι—>
has the desired property.
1λ.
. . Adx
k
j(p)
dx
1
λ
. . .
dx
λ
k
( p )
This is derived f r o m
o^jkf
=
f t t p ) dt
t"-1
+
x'°a1I1f]dx
V..Adx
kJ
(P)
X J
t*"1
[f + x m S m f ]
dt
dx
Λ ...
1
Λ dx
k
(p)
t =0 1 J
at ( t k
)
f ( t p )
dt
dx
1
Λ . . . Λ dx
k
(ρ)
= 0
a
a
[
fdx V . . A d x
One f i n d s equally
k
Cp)
easily
α »d k
=
d ο α k-1
and the i d e n t i t y
d
α « i (ω) k-l ν J
=
d » α
k-l
We next consider theorem 3 . 5 . 5 which,
« i (ω) ν
=
α
k
° d ° i (ω) ν
if ω is closed,
provides us with
4 . 7 . C l a s s i c a l Vector A n a l y s i s
ω
=
α
k
ο L (ω) ν
α
159
ο
k
d ° i (ω) + i «dw ν
ν
J
=
α
k
° d ° i (ω) ν
This leads to d
and t h e
α °i (ω) (_ k-i ν J
=
(k-l)-Form φ
=
a
k-l
«ι (ω) ν
obeys
dip
=
ω
Hence ω is indeed e x a c t .
T h i s would be w r o n g f o r a r b i t r a r y m a n i f o l d s M; t h e e x a c t k - f o r m s a r e e l e m e n t s of a ( n o r m a l l y p r o p e r ) s u b s p a c e of t h e IR-vector s p a c e of t h e c l o s e d k - f o r m s . The f a c t o r g r o u p , t h e s o - c a l l e d k t h ( d e Rham) cohomology g r o u p of M, t h a t is
Η"(Μ)
= { Α 6 AK(M) I Α c l o s e d > / { Α e AK(M) | Α e x a c t >
frequently (e.g.
f o r c o m p a c t M) is f i n i t e - d i m e n s i o n a l .
t o p o l o g i c a l i n v a r i a n t of M.
It is an
important
160
5. Applications to Physics
5. Applications to Physics 5.1. Mechanics In order to make l i f e easier f o r the readers interested in science or engineering, we compile in this short chapter a number of fundamental formulas f r o m theoretical physics. tions w i l l be continued.
The notational conventions of the f o r e g o i n g
sec-
Of course, we do not attempt to substitute a t e x t -
or r e f e r e n c e book of physics. Rather, the considerations to f o l l o w serve as an illustration of our nomenclature system and demonstrate how to apply the d i f f e r e n t i a l geometric calculus. For that reason we deliberately restrict our developments to the most basic physical laws and abstain f r o m a deeper motivation or a detailed discussion of the underlying notions. Nevertheless,
it should be of value f o r many users of this monograph to be
presented some of the main equations in a directly applicable f o r m instead of having to deduce them independently. It is recommended to transform at least a f e w of the formulas listed below to d i f f e r e n t coordinate systems as found, f o r instance, in parts G and H. A comparison of the results with those of a direct evaluation by means of e l ementary calculus will be instructive. Even f o r the simplest examples, we get a considerable reduction of the necessary amount of labour to be invested and a large gain in clarity when we employ the tensor methods developed in chapters 3 and 4. In the sequel, two versions:
some of the physical laws are described in (not less than) Firstly
in the more elegant c o o r d i n a t e - f r e e
language
which
should be p r e f e r r e d during the process of establishing the relevant f o r m u las; and secondly in components (sometimes even in greater detail than demanded by the purpose of The
latter
alternative
is
the d i f f e r e n t i a l equations,
exposition). unavoidable
if
one
seeks
explicit
solutions
since this in general is only possible in
of
coor-
dinates. We start with a f e w ideas f r o m mechanics of point masses. More information can be found in Arnol'd [1978] or Neutsch & Scherer [1992] where f u r t h e r reaching questions are discussed as well.
161
5.1. Mechanics
The state of point
x,
a mechanical system at some instant t is characterized by a
say,
in an n-dimensional
manifold
M,
called
the
configuration
space. The state can and will vary with time; χ thus moves along some curve r in M, parameterized by t: y:
[R
> Μ
t
ι—> x ( t )
We furthermore assume that all forces are conservative, i . e . , ten as the gradient
of
the potential energy V = V ( x ) .
can be w r i t -
This condition
is
met, f o r instance, by the Newtonian law of gravitation. Under these circumstances, the Lagrange equations of motion govern the evolution of the physical system.
We express them most conveniently in an equivalent f o r m via
the Hamiltonian principle, t δ J 2 L ( x ( t ) , x ( t ) ) dt t
=
0
ι
Here the Lagrange function L = Τ - V is calculable f r o m the potential and the kinetic energy T, which depends on the velocity χ = — (and frequently dt also on the position x ) . The variation of the action integral J L dt has to be evaluated
with f i x e d
initial
and final
states
x ( t ) and x ( t ), 1 2
respec-
tively. By the rules of variational calculus the Hamilton principle is tantamount to the Euler-Lagrange equations
(i = 1 χ1
d
aL
aL
dt
dx
ax1
n) if we use an arbitrary chart on Μ to define the
coordinates
of the point χ of M.
As concerns d i f f e r e n t i a l geometry, the solution of the equations of
motion
comes down to a search f o r an orbital curve y: IR —> Μ to the given Lagrange function L:
TM (χ,χ)
» IR ι—>
L(x,x)
5. Applications t o Physics
162
on the tangent bundle T M (= "velocity space") such that the induced curve y:
R
» TM
t
ι—>
(y(t).r'(t))
makes the integral t J 2 L » y ( t ) dt t stationary.
ι
It is important to note that the Euler-Lagrange equations are
of the same shape in any reference system. This is because they are deduced f r o m the variational problem which is coordinate-invariant
itself.
Another formulation of similar importance is obtained when, instead of the velocity coordinates q 1 ,
we introduce the associated canonically conjugate
momenta p^, defined by 3L P'
=
and with their help eliminate the velocities f r o m the Hamilton function
Η (This
is o f t e n possible;
cf.
=
P l q' - L
Neutsch
& Scherer [1992]).
The p^ are
not,
like the velocities, objects in the tangent space TM , but in the dual co« ι tangent space Τ Μ . If we combine them with the position, the (2n)-tuple q ι * (q ,ρ^) is a set of coordinates in the cotangent bundle Φ = Τ Μ. The l a t t e r is in mechanics called the phase space of the system. *
Thus the Hamilton function Η: Τ Μ —» R has to be used in place of the L a grange function L. Then the motion is described by the canonical equations •ι q
3H ap,
an
5.1. Mechanics
163
In comparison to the Lagrangian approach they have the advantage that they f o r m a system of d i f f e r e n t i a l equations of f i r s t order or, what amounts to the same, a vector f i e l d in Φ. We may also view the action integral
S
=
S(q';t)
J L ( q ' ( τ ) , q ' ( τ ) ; τ ) dx
=
with f i x e d initial conditions as a function of time and the coordinates of the f i n a l state.
The variational problem SS = 0 can be used
alternatively
as a foundation of dynamics (Neutsch & Scherer [1992] discuss the details). The partial derivatives of S with respect to the coordinates yield the momenta: 8S 3q and a f t e r
insertion
and a f e w
.
=
simple
Pi
transformations
we
get
the
time-
dependent Hamilton-Jacobi equation as — + Η at
as • · .q ;-
as =
ο
whose solution gives the action function S = S ( q \ . . . , q n ; t ) and at the same time the dynamical evolution of the system. If
the Hamilton function does not contain the time e x p l i c i t l y ,
a further
reduction can be achieved. The expression S
=
Ε t + Wfq1
qn)
solves the variational equation in question whenever the principal function f u l f i l l s the time-independent Hamilton-Jacobi equation aw .q ;aq'
aw
Ε is the mechanical energy of the physical
=
Ε
system.
The l e f t hand side of
this eigenvalue condition f o r W in essence is a quadratic f o r m in the g r a -
5. Applications to Physics
164
dient of
W,
provided the system is "natural".
Similar f o r m u l a s w i l l
occur
again in the sections 5.5 and 5 . 6 as eikonal and Schrödinger equations,
re-
spectively. These techniques are on account of their f l e x i b i l i t y very w e l l
adapted
not
only to describe phenomena of point mass mechanics as e. g. the N-body problem in c e l e s t i a l mechanics. es of
They also play a leading part in other
But we do not intend to dwell tions.
branch-
science.
Instead,
importance.
on a thorough investigation
of
these
we turn over to a r e l a t e d problem which is also of
It is concerned with the c h a r a c t e r i z a t i o n
ρ
=
Β
Jpdx1
the
gravitational
We describe it by a density dis-
action of continuously distributed matter. 3 tribution p: IR —> R whose i n t e g r a l
J
of
ques-
greatest
dx
A
2
Λ
dx
3
Β
(the l a t t e r in Cartesian c o o r d i n a t e s ) states how much mass is contained in some spatial r e g i o n Β which we assume to be regular in the sense of tion 3. 9. 1. In this case one can s t i l l
defini-
describe the g r a v i t y f i e l d caused by
this d i s t r i b u t i o n by a scalar function, the Newtonian potential U. The p o t e n t i a l energy of a "test body" of mass m moving within the f i e l d is then simply given by the product m U. For the sake of letters to
c l a r i t y we use in this and the f o l l o w i n g
characterize
three-dimensional
vectors.
sections
The p o t e n t i a l
is
German calcu-
l a t e d via the f o r m u l a
U(j)
=
U ρ( J )
*p(l>)
= G Β
J-0
which is obtained f r o m its counterpart f o r point masses through a suitable l i m i t process.
Here G is the gravitational constant. 3
The i n t e g r a t i o n domain Β may also coincide with the whole space IR as long as the i n t e g r a l
exists.
From a mathematical point of v i e w ,
the most intriguing p r o p e r t y of the p o -
t e n t i a l is Poisson's equation Δ U
=
- 4 π G ρ
5.1. Mechanics
165
w h i c h r e d u c e s t o t h e potential equation
Δ U
=
*d *d U =
dlv grad U
=
0
if t h e s p a c e i s void of m a t t e r (p = 0 ) . T h e n - d i m e n s i o n a l v e r s i o n of t h i s f o r m u l a a n d i t s s o l u t i o n s , possess
numerous
t h i s book ( c h a p t e r s
6,22,23).
functions,
peculiarities.
They
will
t h e potential
occupy
us
later
in
A d i f f e r e n t , v e r y i m p o r t a n t p r o b l e m of continuum mechanics, n a m e l y t h e m o tion
of
e x t e n d e d rigid bodies ( m o r e p r e c i s e l y :
t h e spinning top) l e a d
to
t h e m a t h e m a t i c a l t h e o r y of t h e rotation groups a n d t h e i r r e p r e s e n t a t i o n s by orthogonal matrices a n d spinors. T h e y a r e a l s o d i s c u s s e d e l s e w h e r e t e r s 8 , 9 , 1 3 ) s i n c e t h i s i s n o t r e a l l y an a p p l i c a t i o n of t e n s o r F o r t h e m o m e n t we w a n t t o l e a v e i t a t
that.
(chap-
calculus.
166
5. Applications to Physics
5.2. Hydrodynamies In contrast to classical mechanics which is concerned with the motion single point masses under the influence of e x t e r i o r f o r c e s , hydrodynamics are based on statistics. ticles
is
regarded
f l u i d are of
as
The individual behaviour of the p a r -
only
the
collective
properties
of
the
interest.
To reach this goal, the general
irrelevant;
of
the methods of
one defines certain
character
of
the flow.
functions which
reflect
This can be done at d i f f e r e n t
averaged
levels.
Here we only discuss the simplest case, namely the mathematical description of
fields
like
(particle
or
mass)
density,
(bulk)
velocity,
and
tempera-
ture. Concerning the relationship to the substantially more general and f l e x i b l e kinetic gas theory which aims at describing the evolution of the system by distribution functions, we r e f e r the reader to the literature ( e . g .
Landau
& L i f s c h i t z [1962-1967]). The definitions of grals
of
the above-mentioned
distribution
quantities
as velocity
functions multiplied by suitable
factors
space is
inte-
assumed
known. The f a m i l i a r mechanical conservation laws of mass, momentum, and energy are r e f l e c t e d in the basic hydrodynamical equations of motion, as there are: ( 1 ) the continuity equation or mass balance; ( 2 ) the momentum conservation in the f o r m of either Euler's equation or, the presence of f r i c t i o n f o r c e s ,
the Navier-Stokes equation;
in
and f i n a l -
ly ( 3 ) the energy balance whose shape depends on the type of the system to be described. The equation of continuity expresses the f a c t that matter (mass) is neither produced nor destroyed,
but only transported f r o m one place to the other.
This means: We consider a f l u i d c e l l Z,
that is some region in space whose boundary
moves with the local velocity u = u ( x , t ) of
the matter
at position J e IR
and time t. Ζ has to be taken so small that the physical parameters like matter density ρ = p(j, t),
particle
density η = n ( j , t )
or
temperature Τ = T ( j , t )
do
not
5.2. Hydrodynamics
167
vary appreciably within Z. On the other hand, the cell must be s u f f i c i e n t l y large,
f o r otherwise the statistical fluctuations would be too strong.
under these circumstances the physical f i e l d functions η,ρ, T,
Only
and so on,
are w e l l - d e f i n e d . Both (contrasting) conditions are compatible by the fundamental of hydrodynamics;
hypothesis
if they were not, we would have to take resort to more
comprehensive theories. We denote the volume of Ζ by dV. The equation of continuity implies that the mass ρ dV contained in Ζ is constant. tion of
the cell,
This is true by the very d e f i n i -
since the f l u i d particles
cannot
(on the average)
move
across the boundary. Hence, the total derivative of the mass contained in Ζ is zero: iL(pdv)
=
0
which we can express also d i f f e r e n t i a l l y as
+ ρ dlv ii
or,
=
0
in components, dp d£
+
Ρ
a
=
u
„ 0
The Einstein summation of course runs f r o m 1 to 3. A more convenient representation is obtained if we replace
by the p a r t i -
al derivative with respect to the time according to the substitution rule dp π
dp
=
sir+
.
dp =
Λ
är
p;a
Although, as we know f r o m the investigations of chapter 3, partial and covariant derivatives of the scalar density function ρ coincide, rable f o r theoretical reasons to introduce the latter.
it is p r e f e -
In this way,
we ob-
tain a more unified representation when we combine the equation of continuity with the other d i f f e r e n t i a l conditions governing the f l u i d motion. We are provided with the f a m i l i a r
d^p + div(ptt)
=
d t p + (pu )
=
0
168
5. Applications t o Physics
Other than the mass, the sum of the momenta of all particles contained in the cell
Z,
the bulk velocity u equals ρ it dV,
which by definition of
not absolutely constant.
it changes under the influence of a force
Rather,
exerted on the f l o w i n g matter.
is
We have to distinguish between interior and
exterior effects. First,
the thermal motion of the atoms or molecules in Ζ produces,
ing on the local temperature Τ = T(j, t ) ,
the gas pressure p.
Its
dependintensity
is determined by the physical structure of the fluid. We express this r e l a tionship by the equation of state
ρ Clearly, External
=
p(p,T)
we can and w i l l not be more specific here. f o r c e s are described
acting on the cell.
by giving
the total
f o r c e per
unit
volume
We describe it by the components f a as r e f e r r e d to our
coordinate system. It is convenient to w r i t e down the gravitation separately
since
its
contribution
to
the
force
density
f
is
proportional
to
the
mass density: ρ grad U
f.g r a v (U is the Newtonian potential,
see section 5.1). The momentum balance thus
is, if we neglect f r i c t i o n e f f e c t s , ί (p u dV) dt
=
(p grad U - grad p) dV
From this we deduce the d i f f e r e n t i a l variant, the Euler equation du dt
grad
υ -
S^LP
which can be changed f u r t h e r into a , a b Q a U ++ u u 3 u U = t ;b
g
ab L ^; b"| U - — L 'b ΡJ
Note the index positioning and the necessity of introducing the metric g ab . The Euler equation may clearly be brought to the covariant f o r m ,
5 . 2 . Hydrodynami es
ο u 9
t a
169
+ u
a;b
b
u
=
a θ u
, gb c u +
a;b uc
t a
itU
=
;a
ρ - —i ®
p
as w e l l . For
reasons of
completeness,
we f u r t h e r m o r e g i v e the extension
which
is
needed to take into account the influence of friction. c a l l e d the dynamic and
Then t w o more f u n c t i o n s , η = η(ρ, Τ ) and ζ = ζ ( ρ , Τ ) , kinematic viscosity, r e s p e c t i v e l y , of
the f l u i d .
must be employed to describe the inertia
They also depend on the chemical
and physical
matter.
It is t h e r e f o r e impossible to give general
der
conceivable
all
circumstances,
although
state of
the
f o r m u l a s applicable
un-
relatively
comprehensive
gas
kinetic t h e o r i e s of v i s c o s i t y have been developed. The momentum conservation f o r a viscous f l o w thus has to obey the NavierStokes equation,
ρ ίϊ-
=
ρ grad U - grad ρ + η Δτι +
which g e n e r a l i z e s E u l e r ' s . t e r are s p a t i a l l y constant;
[ΐΗ Ό
ζ | grad d i v u
+
Here we assumed that the v i s c o s i t i e s of the m a t this is f r e q u e n t l y a good approximation.
Other-
w i s e f u r t h e r terms containing the gradients of η and ζ would have to be inserted. Expressed in an a r b i t r a r y r e f e r e n c e f r a m e , the momentum balance a t t a i n s the f orm
P K
Ö U + PK U
t a
a;b
b
U
.t ;a
= p U
- p
;a
+
fi (3-JJ
+ SC
J
b ;b;a
U
+i)g 6
be
u
a!b;c
When applied to k - f o r m s in 3-dimensional space, the Laplace-Beltrami operator Δ ( D e f i n i t i o n 4 . 7 . 1 ) s i m p l i f i e s to
Δ
and f o r by g r a d ,
=
ν * * U+1 * * (-1Γ d d + (-1Γ d d
the v e c t o r
f i e l d u in p a r t i c u l a r
=
" (-1)"
(k = 1),
* » * *1 d d - d d d d - d d
[
we get
a
representation
div and r o t : *
#
*
*
A u = d d u - d d u =
which is o f t e n p r e f e r r e d .
grad d i v u - r o t r o t u
The N a v i e r - S t o k e s equation now reads
170
5. A p p l i c a t i o n s t o Physics
ρ ^
=
ρ grad U - grad ρ - ij r o t r o t u + [5
If the equation of
s t a t e is independent of
and the same holds true f o r the v i s c o s i t y ,
Sra(ä d i v
71 +
the temperature,
i.e.
11
ρ = p(p),
the continuity and Euler (or N a -
v i e r - S t o k e s ) equations t o g e t h e r f o r m a complete system of conditions which in p r i n c i p l e could be solved f o r given i n i t i a l values of ρ and u a . This assumption is sometimes j u s t i f i e d ;
normally,
however,
it w i l l be n e c -
essary to add an equation which is apt to determine the t e m p e r a t u r e
field.
This is achieved by the energy balance or heat transport equation. E x a c t l y as the t o t a l momentum in Z,
the energy content of the c e l l can be
changed by several transport phenomena. Of the innumerable p o s s i b i l i t i e s we single out only a f e w instructive and t y p i c a l
examples.
The heat equation basically has the shape (Weizel
p
3t
\β
+
h
"
U]
=
[1962-1966])
τ
where the bracket on the l e f t hand side contains in that order kinetic energy, enthalpy
(e = s p e c i f i c energy,
internal
energy,
cf.
Batchelor [1967])
and
the
gravitational
each r e f e r r e d to unit mass.
On the other hand, expansion
or
the f i r s t contribution represents the work done by an
contraction
of
the volume
of
Z,
while τ
is the
sum of
t r a n s p o r t terms whose precise f o r m s depend on the p r e v a i l i n g physical
all con-
ditions. We only mention the heat conduction which is described by
τ
L
=
d i v ( λ grad T )
=
(gab λ Τ
;a
)
;b
=
gab (λ Τ
;a
)
;b
(λ g r a d Τ is c a l l e d the heat flow) and the frictional heat (energy dissipation) τ , which is most easily expressed with the help of the shear tensor
D
ab
We have
=
- \u
2
I a;b
+ u
I
b;al
5.2.
Hydrodynami e s
171
[ w
0
τΗ
or,
=
2
71
[g
bd 8
n
1
n
D »t Dcd
" i 8
ac
bd
g
;e
_
]
g a b Dodj
somewhat simplified, τ
R
=
2 η D ab ίϋ - i g I ab 3
ab
div
tt| J
=
2 η D ab iü
T h e r e a r e many o t h e r heat t r a n s p o r t p r o c e s s e s , etc;
e
u
ab
- i gs 3
ab
uc
like convection,
1 ;cj radiation
but the e x a m p l e s d i s c u s s e d here should s u f f i c e as an i l l u s t r a t i o n .
172
5. A p p l i c a t i o n s t o Physics
5.3. Relativity
Throughout that a l l
the remainder
Greek super-
of
this chapter
we shall
employ
the
convention
and subscripts run f r o m 1 to 3 and the Latin
ones
f r o m 0 to 3; of course, the same agreement applies f o r the Einstein summation. In Newtonian mechanics,
a physical event 8 is completely described by the
c o o r d i n a t e s x " ( f o r the moment supposed to be Cartesian) 1 2
of
the
position
3
3
J = (χ , χ , x ) where it happens,
t o g e t h e r with the time t.
Here j e R
and
t e IR are independent of each other. The theory of relativity combines both quantities
into a f o u r - d i m e n s i o n a l
event vector: x
(t c ^t ; j )Λ
=
=
(/x ο ; x α .)
(x a .)
=
r
=
,
( x 0 , x1 ,2x 3,x. )
e
_4 IR
The f a c t o r c denotes the speed of light in the vacuum, an absolute
invari-
ant in the theory.
system
As an aside,
its value has been f i x e d in the SI
by convention to
c
=
299792458 m sec" 1
the metre (m) becomes a secondary unit depending
With this d e f i n i t i o n ,
on
the p r i m a r y second ( s e c ) . In the
special
theory
with the points of
of
relativity
all
conceivable
a f l a t 4-dimensional manifold,
events
are
associated
the Minkowski space.
distinguishing f e a t u r e is the existence of a metric which in suitable dinates element,
x°
is
expressed
by the Minkowski form of
the
squared
i. e. . 2
ds
In this f o r m u l a ,
-η
ab
=
i)
ab
. a
dx
, b
dx
are the c o e f f i c i e n t s of the constant
matrix
arc
Its
coorlength
5.3.
Relativity
173
' 1 0
0
0
0 - 1 0 0 η
=
0
0 - 1 0
0
0
0 -1
The c o n t r a v a r i a n t m e t r i c components look e x a c t l y the same since η is an involution (η
= 1). The motion of
a f o r c e - f r e e p a r t i c l e then f o l l o w s a g e o -
desic line which in Minkowski space reduces to a s t r a i g h t line.
We are not
going to d w e l l upon a more d e t a i l e d discussion of mechanics in the f r a m e work of
special
relativity
Minkowski f o r m invariant
or i n v e s t i g a t e which chart t r a n s i t i o n s l e a v e the (these s o - c a l l e d Lorentz transformations are
the
main t o p i c of chapter 14). Instead, we immediately go over to the viewpoint of general
relativity.
Now the configuration space is a 4-dimensional, possibly curved m a n i f o l d M. By the correspondence This means,
principle
the m e t r i c g
it
is assumed to
be l o c a l l y
Minkowskian.
like η
has one p o s i t i v e and three n e g a t i v e ab ab 1 3 genvalues and thus the signature ( + , - , - , - ) = ( + , - ) .
ei-
Gravitational fields by the equivalence principle cause f o r all point masses the same a c c e l e r a t i o n . g l e point
or
They can t h e r e f o r e be t r a n s f o r m e d away in a sin-
even on a timelike
curve
in Μ if
we choose
the
coordinates
cleverly. P a r t i c l e s s u b j e c t to no other f o r c e s except g r a v i t y s t i l l
move along
time-
like geodesies: -a
χ
. —a be
*b ·c
+ Γ
x x
(dot = d e r i v a t i v e with respect to the arc length),
or,
if
we introduce
the
dimensionless v e l o c i t y ua = x a , a
u
b
:b
·a
t
u = u + T
_a
b e be
_
u u = 0
The c o o r d i n a t e - f r e e version is simply
g(u,u)
This expresses the o r t h o g o n a l i t y
of
the equation of motion to the f o r m
velocity
and a c c e l e r a t i o n .
If
we
bring
5. Applications t o Physics
174
••a X
—a b c - Γ U U be
=
we discern the meaning of the Christoffel symbols. They describe the g r a v i tational acceleration in the particular reference system we use. In order to determine the metric we moreover need the Einstein f i e l d equations. They generalize the Poisson equation of section 5. 1. Let us suppose that the f i e l d - g e n e r a t i n g matter has density ρ and gas pressure ρ and that
it moves with the velocity
ua.
Then the Energy-momentum
tensor is given by 2
In coordinates,
Τ
=
(pc +p) u β u - ρ g
this reads _ab Τ
=
, 2, . a b ab (pc +p) u u - ρ g
Obviously, Τ is symmetric. We moreover find that energy and momentum are conserved quantities: T ab
;b
=
0
The O-component of the last expression is the r e l a t i v i s t i c equation of continuity, while the three spatial components group to the Euler equation. At the same time we have translated the basic formulas of hydrodynamics to the four-dimensional theory. In the subsequent section we want to consider what changes if we additionally take the electromagnetic
interaction into account.
shall s u f f i c e to formulate the f i e l d equations.
At the moment it
They connect the Ricci ten-
sor R a b with T a b : 8irG R ab - - sg a b R 2 (G = constant of gravitation).
=
4 C
The f a c t (theorem 4 . 4 . 6 ) that the divergence
of the Einstein tensor vanishes leads us back again to energy-momentum conservation.
5.4.
Electromagnetism
175
5.4. Electromagnetism
Especially
beautiful
applications
of
tensor
calculus
are
provided
by
the
mathematical treatment of electromagnetic phenomena. The
evolution
Β = 8(j,t)
of
the
electric
field
& = 5(j,t)
is governed by the Maxwell equations.
and
magnetic
the
They have the
field
following
nonrelativisic f o r m : div 8
=
0
rot 5 + i g
=
0
c at div £
=
4 it ρ
1 dS rot 8 - c- 9t =
4 π . c I'
in Gaussian units (Landau & L i f s c h i t z [1962-1967]). Here, as e a r l i e r ,
c is the vacuum speed of light; ρ denotes the charge den-
sity and j the e l e c t r i c current density. These f i e l d s exert on some particle of charge e,
moving with the velocity
το, the f o r c e g
=
e 6 + e υ χ 8
which is found as a composition of the electrostatic force e Ε and the Lorentz force e τι χ S. If we d i f f e r e n t i a t e the Maxwell equations,
we obtain via the
intermediary
calculation 4tr ^
=
div 6
=
div
=
jj d i v |c rot 8 - 4ir j|
the conservation of charge §£
+
divi
=
ο
=
- 4π d i v j
5. Applications to Physics
176
which s t r o n g l y reminds us of the mechanical continuity
equation.
The r e l a t i o n s h i p is connected with the observation that moving charges g e n erate
electric
currents.
In
this
case,
the
current
density
j
is,
like
section 5.1, found by summation of the products of charge and v e l o c i t y each individual p a r t i c l e within a f l o w
cell.
The t h e o r y becomes much simpler and more elegant if we r e w r i t e i t , ing
to
sion.
the
philosophy
When doing this,
of
relativity
theory,
in
its
l i k e w i s e the last pair of
accord-
four-dimensional
we have to combine the f i r s t t w o Maxwell
(the "homogeneous" ones);
in for
ver-
relations
("inhomogeneous")
for-
mulas. It s u f f i c e s to c a r r y out this process in any coordinate f r a m e since the
re-
sulting tensor equations are valid universally. We s e l e c t an a r b i t r a r y point Ρ in the f o u r - d i m e n s i o n a l space-time
manifold
("world") Μ and determine a Cartesian r e f e r e n c e f r a m e such that the m e t r i c reduces to the Minkowski f o r m η = D i a g t l , - 1 , - 1 , - 1 ) . lence principle
it
is
always
possible
to
find
Because of the equiva-
an inertial 0
1 2
system of
this
3
kind. The 0-component of the position v e c t o r χ = (χ , χ , χ , χ ) corresponds to the time t via the r e l a t i o n x ° = ct. The f o u r - d i m e n s i o n a l
velocity
of
a test
i t s w o r l d line.
By the above argument,
be
naturally
interpreted
as a
particle
is the tangent
vector
of
the e l e c t r i c current density has to
(contravariant)
vector
field.
We
therefore
w r i t e it as Ja
=
(j°,j\j2,j3)
=
(pc,
and charge conservation reduces to the tensor
Ja if
=
j2, j 3 )
relation
0
we at the same time introduce covariant
in place of
partial
derivatives
to g e t invariant f o r m u l a s . T h e r e are several d i f f e r e n t ways to p e r f o r m the required calculations.
One
procedure is to f i r s t w r i t e down all necessary r e l a t i o n s in components
re-
l a t i v e to the given i n e r t i a l f r a m e and then t r y to combine them into tensor equations. The occurrence
of
electromagnetism,
the o p e r a t o r s however,
div
and r o t
in the fundamental
laws
of
suggests to represent the f i e l d s by d i f f e r e n t i a l
f o r m s on M. This leads to a much easier and more elegant f o r m u l a t i o n than
5.4. Electromagnet ism
177
the tensor
approach. 3 we i n t e r p r e t all 3-dimensional vector f i e l d s as 1 - f o r m s in IR which 1 2 3 0 do not only depend on χ , χ , χ but also on the time χ . We view the l a t t e r Hence,
as a p a r a m e t e r and put
β = β(χ°·,χ1,χζ,χ3) · * ! " • > < > '
=
Β dx" α
=
Β dx 1 + Β dx 2 + Β dx 3 1 2 3
τι
=
tj(x°: χ 1 , χ 2 , χ 3 ) '
=
Ε dx" α
=
Ε
l
=
ι (χ ; χ , χ , χ j
,
One can work
0
1
with
2
3
,
· _ • «
=
j dx α
1
·
=
dx 1 + Ε dx 2 + Ε dx 3 2 3 J
1
j dx 1
Α
I
J
2
+ J dx 2
these parameterized d i f f e r e n t i a l
.
-
+ j 3
J
3
dx
forms e x a c t l y
in
the
same manner as with the usual ones. The only d i f f e r e n c e is that d e r i v a t i v e s w i t h respect to the time must be included as w e l l . To prevent misunderstandings,
we denote by β the three-dimensional 1 2 3
f e r s to the components (x , x , x ) alone,
Cartan d i f f e r e n t i a l which
re-
w h i l e d continues to be used
for
the f o u r - d i m e n s i o n a l d i f f e r e n t i a t i o n (including time x ° ) . The technical advantage lies in the e x p l i c i t consideration of of the spatial As concerns marked
the
symmetry
coordinates.
duality,
we » proceed
by an asterisk
similarly:
( ) as b e f o r e ;
the
4-dimensional
the 3-dimensional
mapping
analogue
will
is be
c h a r a c t e r i z e d by a square Γ ) . 3 We c o l l e c t the results. L e t φ, ω be p a r a m e t e r i z e d 1 - f o r m s on IR and f an a r 1 2 3 bitrary O-form, p a r a m e t e r x°.
i.e.
a smooth f u n c t i o n of the coordinates χ , x , x
and the
The rules f o r dualizing f o r m s of ranks 0 , 1 , 2 , 3 , 4 on Μ are in
turn "f
=
#
[f dx° + 0]
[
, 0 ,
dX° Λ ° f
=
°f +
D 1
dx Λψ +
ω
+
dx° Λ Πψ
•, =
-
°f j =
,0
ψ + dx φ
+
λ ι
f dX°
*[dX°ADf] = - f This covers a l l p o s s i b i l i t i e s
(note that det g is negative! ).
By
comparison
178
5. A p p l i c a t i o n s t o P h y s i c s
w e r e g a i n t h e f o r m u l a of t h e o r e m
4.6.1,
**a
(-l)k+1 a
=
f o r k - f o r m s a in M. We e q u a l l y e a s i l y e v a l u a t e t h e r e d u c t i o n f o r m u l a s of t h e 3 C a r t a n d i f f e r e n t i a l s t o t h o s e in R :
d f
=
d | f dx° + tfij = d |dx 0 A^ + ° ω | d [dX0ADV-
n +
(a f ) dx° + 0f ο - dx° λ df + dx° λ d ß + ϋψ - dx° λ ϋψ + dx° λ
=
f]
=
+
- dX° Λ «°φ + dX° Λ
d [dX°A D f]
=
D
(aof)
0
T h e M a x w e l l e q u a t i o n s a r e in t h e t h r e e - d i m e n s i o n a l the
four-dimensional
version equivalent
with
conditions D
a
aDß
=
ο
» y + dQ β
=
l· β π η β
.
d
0
0
=
4 π ρ
„
=
Uϊ C
a n d t h e c o n s e r v a t i o n of c h a r g e i s e x p r e s s e d
d (pc) + ο
D
ö
D
i
t
as
=
0
T h e t r a n s f e r t o 4 - d i m e n s i o n a l f o r m s in s p a c e - t i m e c a n n o w be a c h i e v e d w i t h out e f f o r t . We s h i f t d o w n t h e s u p e r s c r i p t a n d o b t a i n f r o m t h e ( l . O ) - t e n s o r f i e l d J a t h e 1-f orm
5.4. Electromagnet ism
179
J
=
pc dx° - L
whose dual is *
J
N e x t we d i f f e r e n t i a t e this,
d *J
=
dx° Λ j e
Πι
+
•
=
j
pc - dx
ο
•
Λ
ι
getting
DOopc)j
=
dx° Λ
Dc
+ AQ(pc)j
=
0
Thus we have deduced the theorem of charge invariance,
d
*
J
=
0
*
The 3 - f o r m
J is closed and,
nected regions
of
(theorem 4 . 7 . 3 ) . equal to
the w o r l d
if
we r e s t r i c t
manifold M,
Consequently,
our attention to simply
also
exact
by P o i n c a r £ ' s
conlenaa
a 2 - f o r m exists whose Cartan d e r i v a t i v e
is
J.
It is easy to f i n d the desired quantity.
We introduce the field strength as
the 2 - f o r m defined through F
=
- dx° Λ η +
Dß
It connects e l e c t r i c (rj) and magnetic (β) f i e l d s . The dual 2 - f o r m to F is *
F
=
π
ο η + dx
Λ
β
and the Cartan d i f f e r e n t i a l of the l a t t e r yields with the assistance of inhomogeneous Maxwell
d
*
F
=
the
equations
- dx
ο
ββ + dx
Λ
ο
π π η + « η ο
Λ A
4 IT * — J c
=
On the other hand, we have, as a consequence of the two homogeneous laws of electromagnetism,
d F
=
dx°
Λ
«Τ»
+ dx°
Λ
d ° ß
+
e°ß
=
0
This a l l o w s us to base the theory of electromagnetic phenomena on t w o e x -
180
5. A p p l i c a t i o n s t o
c e e d i n g l y simple and e l e g a n t
Physics
formulas:
d
*
F
4 π * — J c
=
d F
P e r h a p s not q u i t e as b e a u t i f u l , sor
but o f t e n expedient a r e the e q u i v a l e n t
ten-
relations -ab
F
F
bc;a
4 π
=
;b
.a
J
C
+ F + F ca;b ab;c
= 0
which a r e accompanied by c h a r g e c o n s e r v a t i o n in the f o r m
Ja
deduced e a r l i e r
;a
=
0
and the equation of motion of
a charged p a r t i c l e
charge e),
[
· a , _a
u + Γ
b be
u u
cΙ
„ab
J
=
eF
u b
The c o v a r i a n t components of the f i e l d s t r e n g t h t e n s o r
0 Ε 1
-E
-E 1 2
0
Β
Ε -Β 2 3 Ε 3 w h i l e the c o n t r a v a r i a n t
read
-E 3 3
0
Β -Β 2 1
-Β
2
Β 1 0
are
(mass m,
5.4. Electromagnet ism
181
We conclude this section by giving the dual tensor,
Β Β 1 2 -Β
1
0
Ε -Ε 3 2
-Β -Ε 2 3 -Β 3 and the
0
Ε -Ε 2 1
Ε 1 0
contravariant
-Β Β 1
-Β 1 2
0
Β -Ε 2 3 Β 3 version.
Β 3
-Β 3
Ε -Ε 3 2 0
Ε -Ε 2 1
Ε 1 0
both in the covariant
5. Applications t o Physics
182
5.S. Optics Although the Maxwell theory of the f o r e g o i n g section is completely general, it
is useful to modify it to a certain degree b e f o r e we apply
it to
the
evolution of e l e c t r i c or magnetic f i e l d s within regions of space which are not void of electrons,
matter.
The reason is that the charges
of
atomic
nuclei
or
react on varying f i e l d conditions.
The f o r c e s exerted by the electromagnetic phenomena cause the particles to change their state of motion;
and this in turn produces e l e c t r i c
currents
discussion would require to consider all particles
individu-
which influence δ and 8. A microscopic ally,
a clearly unworkable task.
Fortunately,
very o f t e n the situation occurs that the f i e l d s change on an
extremely short spatial and temporal scale. Thus a summarizing treatment of the atomic structure (by averaging) becomes feasible. The magnetic induction 8 and the electric field 6 are accompanied by two other t i m e - and position-dependent quantities,
namely the magnetic field 5
and the displacement current 9. In the
limiting
case
of
strong
field
fluctuations
we
are
interested
these variables are connected with 8 and S through linear relations.
in, They
can be expressed with the help of three functions, namely dielectricity ε, (magnetic) permeability μ, and (electric) conductivity σ of the matter. Their exact f o r m is determined by the chemical and physical composition of the medium. We assume that ε, μ, and σ are constant in time,
but possibly
vary with position: ε = ε(,τ), μ = μ ( ί ) , σ = Generally,
ε, μ, and ο- are tensors,
but we r e s t r i c t our discussion to iso-
tropic media f o r which a s u f f i c i e n t l y good approximation by scalar material functions can be achieved.
This excludes an investigation of optical pheno-
mena in crystals which are mostly anisotropic. Basically,
the f o l l o w i n g considerations are unaffected by this
simplifica-
tion. Under the just described conditions, the material equations are 3
=
ε g
8
=
μ 5
5.5.
Optics
183
w h i l e t h e c u r r e n t d e n s i t y in l i n e a r a p p r o x i m a t i o n is given by Ohm's law, j
=
ο- β
The f i e l d e q u a t i o n s div
8 = 0
1 3Ϊ r o t e + - -5V- = c öt div S
=
0
4 π ρ
1 a» r o t S - - 31- = c at
4 π . J c
d i f f e r f r o m t h e Maxwell t h e o r y f o r t h e vacuum m e r e l y by t h e e
substitutions
S, Β -> 5 in b o t h inhomogeneous f o r m u l a s . If we i n s e r t t h e v a l u e s ε = 1,
μ = 1, valid f o r e m p t y s p a c e , we obviously come back t o t h e o r i g i n a l e q u a tions. (or,
However, somewhat
we a r e now mainly i n t e r e s t e d in t h e p r o p a g a t i o n of light more generally,
electromagnetic waves of
sufficiently
high
frequency). In "transparent" m a t e r i a l s t h e p e r m e a b i l i t y n o r m a l l y does n o t d i f f e r a p p r e c i a b l y f r o m 1, hence in w h a t f o l l o w s we s h a l l put μ e q u a l t o u n i t y .
The
r e a s o n l i e s in t h e lack of m a g n e t i z a b i l i t y of t h e s e s u b s t a n c e s . Furthermore,
we d e f i n e t h e refraction index η = n ( t , j ) via t h e Maxwell re-
lation
η
·Ζ~ε
=
which is a p p l i c a b l e w h e n e v e r t h e o p t i c a l c h a r a c t e r i s t i c s of t h e medium v a r y only s l i g h t l y w i t h t h e f r e q u e n c y , i . e . t h e colour of t h e l i g h t . We a s s u m e in a d d i t i o n t h a t we deal w i t h a p o o r c o n d u c t o r (:
- Δ δ
+
0 e t c .
a r e a l w a y s u n d e r s t o o d t o imply t h a t t h e c o n s i -
d e r e d n u m b e r s a r e in R. An i m m e d i a t e c o n s e q u e n c e of t h e above is t h a t G is t h e extension field of R 2 c o n s t r u c t e d by adjunction of i a s a z e r o of t h e p o l y n o m i a l i + 1. I t s d e g r e e o v e r R is t h u s [C:R] = 2. The ( p o s i t i v e ) s q u a r e r o o t of t h e d e n o m i n a t o r of t h e r e c i p r o c a l z" 1 of
z,
t h a t is |z| is c a l l e d t h e modulus of z.
=
•J zz
=
i / x 2 +y 2
It g e n e r a l i z e s t h e a n a l o g o u s q u a n t i t y f o r
n u m b e r s and is multiplicative:
real
6.1. Elementary P r o p e r t i e s of Complex Numbers
191
The (Euclidean) distance of two points w, ζ € C is the modulus |w-z| of the d i f f e r e n c e . With this definition, € is a metric space. The (open) ε-neighbourhood of w is the set
U (w) c
of all points whose distance f r o m w is less than ε (we shall always assume e > 0).
It is also known as the (open circular) disk of radius ε with cen-
tre w. Closed disks are defined similarly. An application of
standard procedures to C yields a Hausdorff space.
We
take the ε-neighbourhoods of arbitrary ζ e C as a basis ( f o r all ε ) .
It a l -
ready s u f f i c e s to include only rational values of ε > 0 and those ζ
which
have rational
real and imaginary parts.
This particular
basis defines the
same topology, but is countable, in contrast to the f o r m e r . 2 The assignment κ: C —> R , which acts by dint of ζ = χ + y i
ι—>
( " f o r g e t t i n g " the algebraic properties),
,2 ( x , y ) e IR'
is a chart and endows C with
the
structure of a 2-dimensional manifold, i . e . an (oriented) surface. We shall have
ample
opportunity
to
apply
this
observation
later;
it
lies
at
the
heart of Riemann's (interpretation o f ) complex analysis. The metric in C is provided by the arc length element ds which is simply the Euclidean one and can be calculated in standard coordinates by the f o r 2 2 2 mula ds
= dx
+ dy . With this,
C even becomes a 2-dimensional
oriented
Riemannian space. The pictorial representation by means of ( x , y ) as Cartesian coordinates of
z = x + yie(Cis
called the Gaussian (number)
plane
or Argand diagram. It is o f t e n advisable to compactify the surface C through the addition
of
an " i n f i n i t e " or "improper" point oo. This results in the 2-manifold y
=
C υ {oo}
It is denoted as the Riemann number sphere, what unhistoric; see section 1.2. This
though this notation is some-
nomenclature, however, has been es-
6. Complex Analysis
192
t a b l i s h e d so s t r o n g l y t h a t a more suitable.
i t is v i r t u a l l y
In any c a s e ,
i m p o s s i b l e t o r e p l a c e i t now by
it is c l e a r w h a t we mean.
In o r d e r t o make if a t o p o l o g i c a l s p a c e ,
we have t o give a c o m p l e t e
listing
of t h e open s u b s e t s in if. They c o n s i s t of t h o s e of C and t h e n e i g h b o u r h o o d s of oo. The l a t t e r ,
by d e f i n i t i o n , a r e t h e c o m p l e m e n t s (in if) of t h e c o m p a c t
C-subsets. One may v e r i f y w i t h o u t any d i f f i c u l t y t h a t if indeed is h o m e o m o r p h i c t o t h e 2 - s p h e r e S . The a l g e b r a i c p e c u l i a r i t i e s
of if, in p a r t i c u l a r
its
interpre-
t a t i o n a s a 1 - d i m e n s i o n a l projective space PC
over t h e c o m p l e x n u m b e r s ,
( i. e.
as
a projective
line)
w i l l be t h e s u b j e c t of o u r s t u d i e s in t h e c h a p -
t e r s t o f o l l o w ; f o r t h e moment we a r e s a t i s f i e d w i t h t h e i n d i c a t e d i n f o r m a tion.
6.2. Convergence of Function Series
193
6.2. Convergence of Function Series For our later (analytical)
investigations,
functions,
we
as f a r as they are concerned with
shall
need several
number of related definitions and theorems. contact with the coordinate concept, quent sections some of
convergence
criteria
complex and a
Since these are only in loose
we merely list in this and the subse-
the most relevant terms and propositions
without
giving detailed p r o o f s or striving f o r the greatest possible degree of
ge-
nerality. Mainly f o r this reason we furthermore avoid to introduce today's t e r m i n o l o gy and p r e f e r the more vivid methods of the 19th and early 20 t h centuries. The arguments leading to the theorems in question usually consist of simple estimates which can either be supplied by the diligent reader or looked up in any textbook
on complex analysis (= theory of complex functions);
the standard literature,
e.g.
Hurwitz & Courant [1964],
Cartan [1966]
cf. or
Peschl [1967], A sequence of points ζ , ζ , ζ , . . .
in a Hausdorff space X, which need not be
mutually d i f f e r e n t , converges to the limit (point) w, if each neighbourhood U of w contains almost a l l members z^ of the sequence ("almost a l l " means all up to at most f i n i t e l y many exceptions). The Hausdorff property guarantees that the limit is unique if it exists. In C, where the topology is determined by the distance relation, we may a l so express this f o r m a l l y as f o l l o w s :
Definition 6.2.1: ( a ) The complex sequence ζ , ζ , z 3> . . . e C has the limit w if to an a r b i t r a ry ε > 0 a positive integer Ν can be found such that f o r all k > Ν the distances
I ζ - w1 I of the sequence members f r o m w are smaller than ε . 1 k If some w e € with this property exists, one writes
and calls the sequence { ζ
| k 6 IN} convergent;
otherwise divergent.
194
6. Complex A n a l y s i s
( b ) The complex numbers ζ , ζ , ζ , . . . f o r m a Cauchy sequence if f o r each g i ven ε > 0 there is an Ν 6 IN such that f o r
a l l m, η > Ν the
inequality
Ι1 ζ - ζ 1Ι < ε holds. m η ( c ) By the series 00
l·-
k = lpartial sums we a l w a y s mean the limit of the
k=l
for Ν
oo, i . e .
Σ
=
lim Ν-*χ>
Σ
If w e want to consider sequences or series on the Riemann sphere !f instead of the complex f i e l d C, a f e w minor m o d i f i c a t i o n s are necessary. is an o r d i n a r y point in
Firstly,
and one t h e r e f o r e has to permit convergence
the l i m i t co. By the very d e f i n i t i o n of
the neighbourhoods of
whenever the moduli of the members of the sequence g r o w
to,
it
a to
occurs
unboundedly.
It is obvious that e v e r y convergent sequence of complex numbers obeys the Cauchy condition.
The r e v e r s e statement is much more i n t e r e s t i n g .
We p r o v e
it here in o r d e r to present at least one example of the estimation methods which are t y p i c a l f o r complex
analysis.
Theorem 6.2.1:
Every r e a l or complex Cauchy sequence converges.
Proof: We again denote the given sequence by {z^ | k 6 IN} and s e l e c t p o s i t i v e tegers Ν ,N ,N , . . .
which f u l f i l l the inequalities
in-
6.2. Convergence of Function Series
ζ
195
-ζ Ν Ν k+1 k
0 t h e r e is some
simultaneously f o r
all
ζ e Γ,
the
sequence is said to converge uniformly. ( c ) Both d e f i n i t i o n s a r e t r a n s f e r r e d to function series in the obvious manner.
Uniform convergence in a domain is a much s t r o n g e r r e s t r i c t i o n than p o i n t wise convergence;
f o r instance,
the limiting function of a uniformly
verging s e r i e s o f continuous functions is i t s e l f continuous.
The
con-
analogous
s t a t e m e n t f o r pointwise convergence would be f a l s e . A t h i r d v a r i a n t of the same concept is also of i n t e r e s t :
Definition 6.2.3: A series GO
a k=0
k
of complex numbers a^ converges absolutely if the s e r i e s formed by t h e i r moduli, 00
converges (and thus has a f i n i t e l i m i t ) .
6.2. Convergence o f Function S e r i e s
197
A useful convergence criterion is
Theorem 6.2.2: L e t f , f , f . . . . be complex functions in the domain Γ and c , c , c . . . . r l 2 3 ι 2 3 s i t i v e r e a l numbers which majorize the f in Γ:
for
all ζ e Γ and k e IN. If
the sum of
the c^ converges,
Kp o -
this implies
the
absolute and u n i f o r m convergence of 00
(and,
consequently,
Absolutely
the continuity of the l i m i t f u n c t i o n ) .
converging
series
are
especially
convenient
to
handle,
because
they do not change t h e i r values if we permute the summands. We are thus a l lowed to "rearrange" them a r b i t r a r i l y . ticular
when dealing
which we turn next.
This is applied f r e q u e n t l y ,
with the HeierstraB version
of
in
par-
complex analysis,
to
198
6. Complex Analysis
6.3. Power Series
Among the f u n c t i o n sequences, the power s e r i e s combine best of a l l the i d e als
generality,
flexibility,
and
convenience
of
application.
This
is
the
main r e a s o n why they form the s t a r t i n g point of all c o n s i d e r a t i o n s in the Weierstraß
version of complex a n a l y s i s ,
the theory
of complex
(analytic)
f unctions.
Definition 6.3.1:
A power series, c e n t r e d a t the point a e C, of the complex v a r i a b l e ζ is a f u n c t i o n s e r i e s of the special f o r m
k k=0 Here,
the coefficients
a r e a r b i t r a r y complex numbers;
likewise f o r
the
centre of expansion a .
The s e t of a l l ζ € C f o r which the sum converges is the domain of convergence o f the s e r i e s .
The domain o f convergence of course depends on the c o e f f i c i e n t s .
I t s gene-
r a l c h a r a c t e r can be described easily:
Theorem 6.3.1:
(a) For
each power
c i r c l e X(p) with
series centre
ρ = p(z|a) a,
E(p) and diverges outside.
such
there that
exists
a uniquely
J) converges
in the
determined interior
is the convergence circle of ip. I t s
of ra-
dius p(p) is c a l l e d the radius of convergence. It may a t t a i n the values 0 (convergence only at a) or oo (convergence everywhere in C;
in this
c a s e the s e r i e s is permanently convergent). ( b ) Inside the c i r c l e of convergence the s e r i e s converges a b s o l u t e l y and on
6.3. Power Series
199
compact subsets even uniformly, ( c ) The convergence radius p(p) can be obtained from the sequence of c o e f ficients
C 1 ' C 2 ' C 3>
·•·
For
instance,
it
is equal
to
the
reciprocal
of
the largest accumulation point of the sequence
This proposition is known as Cauchy's criterion. (An accumulation point is some w ε C all of whose neighbourhoods contain infinitely many terms of the sequence).
Proof: Elementary estimates.
Note that this does not say anything about the behaviour of ρ on the boundary of the convergence circle, i.e. f o r | z-a | = p(p). There are three d i f ferent possibilities:
convergence nowhere,
boundary points.
is not d i f f i c u l t to find examples
It
three conceivable cases.
everywhere,
or only in certain of
each of
these
The absolute convergence in the interior of K(J))
allows to rearrange the series at will.
Two power series which are both
convergent in the domain Γ may be summed or subtracted from each other and multiplied by complex numbers. Furthermore, to multiply the series formally we have to calculate all products of the terms and afterwards reassemble with respect to the degrees, as the formula
indicates. This series represents F ( z ) - G ( z ) if we denote the two f a c t o r s on the l e f t hand side as F ( z ) and G(z). In analogy to this,
it is also possible to form the quotient F(z)/G(z)
if
the denominator function G does not vanish in Γ. We now come to an extremely f r u i t f u l application of these principles. that end we start out from the series
To
200
6. Complex Analysis
00
F(z)
=
y
Ck ( z - a 0 ) "
k=0 centred at a
o
and select a point a
cle of convergence.
which lies in the interior of the
ι
A f t e r suitable rearrangement of the series,
cir-
which con-
verges absolutely in the vicinity of a , we obtain a new representation 00 F(z)
=
Y^
dk ( z - a ^ "
k=0 say. It holds at least in some neighbourhood and thus has a positive radius of convergence.
While the original series converges within a certain c i r c l e
Κ of some radius ρ > 0, the new sum is defined in the interior of some ο ο circle if with radius ρ and centre a . ι Μ ι In the overlapping region η ί , both developments are valid. It may happen that S^ contains points which are not in main of
If so, we enlarge the do-
definition of the function appropriately.
( d i r e c t ) analytic continuation of
F.
This extension is called
We obviously
may i t e r a t e
this
step,
The domain (of definition) of
the
function is the largest connected open subset of C covered by the just
de-
thus coming to an indirect continuation. scribed process.
It is identical with the union of
all
convergence
circles
obtained through repeated analytic continuation. The result is not always independent of the succession of the centres a , a , a , . . . ,
intermediate
which are only subject to the restriction
that
each
of them lies within the circle of convergence of its predecessor. Occasionally, point
without
a f t e r several continuation steps, getting
the
initial
value
of
one returns to the starting
the
function.
This,
should not be seen as a drawback of the method; on the contrary, decisive
advantage.
We merely must interpret
this possibility
however, it is its
in a d i f f e -
rent way. The extended function can attain more than one value at a single point. The underlying idea may be c l a r i f i e d by a suitable example. Let F ( z ) be the square root of ζ and begin the iteration at aQ = 1 (the radius convergence at this place is exactly value of the r o o t ,
equal to 1).
We choose the
thus F ( l ) = 1 and continue analytically in the
of
positive indicated
manner. If,
a f t e r a number of steps, we come back to the initial point 1 a f t e r hav-
ing circled once around the origin ζ = 0, where the function is not d i f f e -
6.3. Power S e r i e s
rentiable,
201
we w i l l f i n d the value F ( l ) = -1. The skeptical readers are ad-
vised to carry out the necessary calculations explicitly. the other branch of
the square root function.
We have reached
A second encircling
of
the
origin leads us back to the initial situation. It is t h e r e f o r e essential to record the intermediate points used during the extension process.
The method sketched above, namely the insertion of
nitely many auxiliary positions, practical
disadvantages.
unfortunately has certain theoretical
For this reason the so-called analytic
fiand
continua-
tion along a curve is normally more appropriate. The idea is to introduce some curve y: [0,1] —> C which starts at the initial point r ( 0 ) = z Q and ends at the place y ( l ) = ζ u p
to which we want to
extend the function. We must interpolate a number of points w , . . . , w 1
the (compact) path y ( [ 0 , l l ) between w to be given in the natural order. the part of the curve f r o m w^
0
= ζ
0
and w
n-l
on
= ζ . They are assumed
η
1
Each of them has to be chosen such that to w^ is completely contained in the inte-
r i o r of both convergence circles with the centres w^ ^ and w^. One may v e r i f y with the help of
elementary
estimates that
Ffz^) is uniquely
deter-
mined by the power series near z q and the path y, independently of the positioning of the w^. A subdivision
with the required properties
can always be constructed
as
long as y is restricted to the domain of F. Two interpretations suggest themselves. The f i r s t , p r e f e r r e d by Weierstraß, defines an analytic function in the indicated way through maximal analytical continuation of a given power s e r i es.
All series developments which are obtainable f r o m each other by this
process (possibly with intermediate steps) are considered as equivalent
lo-
cal descriptions of the same function. The other, founded by Riemann, does not view the domain of definition of an analytical function as a part of the complex Gaussian plane, but as a (Riemannian) surface 5 covering it. In the latter theory, earlier
we have to permit ramifications like the one we met
when discussing the square root.
p r o j e c t i o n ^ —> € need not be invertible;
In special instead,
(isolated) points,
the
d i f f e r e n t "branches"
of
the surface ? may be attached to each other. We can separate them into single sheets by introducing suitable "cuts" ( i . e .
deleting certain lines in C
and their preimages in 5). This, however, is useful only f o r the purpose of localized investigations; In principle,
the Riemann surface itself is a uniform entity.
both views are equivalent.
They supplement each other
since
202
6. Complex Analysis
they emphasize d i f f e r e n t aspects of the theory. We are now able to
divide the boundary points of
of a power series p ( z | a ) into two classes. gular
if
there
is
an
analytic
continuation
p(z|z Q )
which converges in some open set containing z q , Another e x t r e m e l y
the convergence
circle
We call such a point z q reof
the
given
p(z|a),
and singular o t h e r w i s e .
important t r a n s f o r m a t i o n has not been mentioned up t o
now. Every absolutely convergent power series can be d i f f e r e n t i a t e d t e r m by term.
Hence we can associate with the analytical f u n c t i o n 00 F(z)
=
c
V
L·
k
(z-a )k 0
k=0
i t s derivative 00
F'(ζ)
=
V
L·
c
k (z-a )k_1 k
0
k=0
which is also a n a l y t i c . dity of the f o l l o w i n g
At the same time this construction y i e l d s the
vali-
proposition:
Theorem 6.3.2: Every i.e.
analytical
f u n c t i o n is in its domain i n f i n i t e l y
often
differentiable,
00
a C -function.
The concept of complex d i f f e r e n t i a b i l i t y requires a somewhat more d e t a i l e d explanation.
Definition 6.3.2: A f u n c t i o n F: Γ —> C,
which is defined in the domain Γ £ C,
is
(complex)
differentiable at ζ e Γ if the limit R r ' rI z, i)
exists.
F'(z)
-=
n „ F ( z + h ) τ-- F(z). lim
is the derivative of F at the place z.
D i f f e r e n t i a b i l i t y in Γ
6 . 3 . Power S e r i e s
203
of course means that this is true f o r all points in Γ.
Here, the transition to the limit h
0 may be carried out along any path.
The above definition can also be expressed in a different but equivalent way: Each sequence {z+h^} of arguments which tends to ζ yields a convergent set of f r a c t i o n s Ftz+h) - Ftz) h with the limit F ' ( z ) . We consider F ' ( z ) as a new function of z. In marked contrast to the situation in real analysis, complex d i f f e r e n t i a bility in some open region has very far-reaching consequences.
We shall
come back to this aspect as soon as the necessary tools are available (section 6 . 5 ) . Here, we merely add a few remarks concerning power series of analytical functions. It is easy to see that two power series with common centre Z q , which coincide in some open neighbourhood, are equal. They consequently have the same c o e f f i c i e n t s . This implies
T h e o r e m 6.3.3:
Assume F is an analytical function in the domain Γ. The c o e f f i c i e n t s of the power series around a point a e Γ are obtained from the higher order deriv a t i v e s a t a v i a t h e Taylor formula 00
F(z) k=0 where we used the standard notation F
=
(-u
- i v y
+ i u y
- ν ) dx λ d y x
=
0
x
and the p r o p o s i t i o n reduces f o r any regular domain Β to the t w o - d i m e n s i o n a l g e n e r a l i z e d Stokes theorem ( c f . J f ( z ) dz 3B Therefore, is of
the remark a f t e r theorem 3 . 9 . 2 ) : =
J" ω 3B
=
J du
=
0
Β
the v a l i d i t y of Cauchy's theorem is no surprise.
extraordinary
importance f o r the investigation
of
Theorem
6.5.3
holomorphic
func-
tions. In the f o r e g o i n g p r o o f ,
we do not r e a l l y need the continuous
l i t y of the i n t e g r a t i o n path;
differentiabi-
it would be enough to demand that Β admits a
triangulation, i. e. , can be exhausted by t r i a n g l e s .
This is a l r e a d y true
if
the boundary 9B is piecewise C 1 and thus can be built up f r o m f i n i t e l y many C1-sections.
The argument then goes through without
alteration.
6 . 5 . Holomorphic F u n c t i o n s
The l a s t
observation
217
allows
to formulate a slight
variation
of
Cauchy's
t h e o r e m which u s u a l l y is m o r e convenient t h a n t h e old one. It can be s t a t e d as
Theorem 6 . 5 . 4 : Let 3 r 1 .3 r 2 be t w o h o m o t o p i c p a t h s in a domain Γ £ C and f some h o l o m o r p h i c f u n c t i o n in Γ. Then
ι
2
Proof: Apply t h e o r e m 6 . 5 . 3 t o t h e c u r v e y^ - y g ( f i r s t y i in t h e f o r w a r d
direc-
t i o n , t h e r e a f t e r yg b a c k w a r d s ) .
We s h a l l e m p h a s i z e only very f e w of t h e n u m e r o u s i m p o r t a n t c o r o l l a r i e s of t h e Cauchy i n t e g r a l
theorem.
As a f i r s t i n s t a n c e , we c o n s i d e r a c o n n e c t e d open s u b s e t Γ of C and a r e g u l a r domain Β c o n t a i n e d in Γ. The o r i e n t e d b o u n d a r y SB is a c l o s e d
Jordan
c u r v e y, which is a s s u m e d t o e n c i r c l e ζ once in t h e m a t h e m a t i c a l l y p o s i t i v e s e n s e . Then t h e i n n e r p o i n t ζ € Β l i e s t o t h e l e f t of t h e p a t h y. If,
furthermore,
f e 5(Γ) is a h o l o m o r p h i c f u n c t i o n in Γ,
the same
holds
t h o u g h n o t in Γ, b u t only in t h e p u n c t u r e d domain Γ' = Γ \ {z>. T h e r e ,
how-
f o r t h e map d e f i n e d by
e v e r , t h e o r i e n t e d b o u n d a r y c u r v e y = 3B is obviously h o m o t o p i c t o a ( s u f f i c i e n t l y s m a l l ) c i r c l e κ £ Γ w i t h c e n t r e z. We d e n o t e i t s r a d i u s by p. Cauchy's theorem yields
wey
218
6. Complex A n a l y s i s
We p a r a m e t e r i z e κ w i t h t h e a n g u l a r c o o r d i n a t e φ a n d
w
dw
in t h e l a s t i n t e g r a l .
substitute
ζ + ρ e
=
1 ρ e
dp
This leads to 2π
1
of
) dφ
φ=0
wey independently
f(w+pe
t h e p r e c i s e v a l u e of p,
at
least
as
long as κ is
wholly
c o n t a i n e d in Γ. A s i m p l e e s t i m a t e u s i n g t h e c o n t i n u i t y of f a t t h e p o i n t
ζ
p r o v i d e s u s in t h e l i m i t ρ —» 0 w i t h
1 2π1
f(z)
wer Since
by a s s u m p t i o n
ζ has
positive
distance
f u n c t i o n of w i s b o u n d e d a b o v e ( f o r ζ f i x e d ) . absolutely
and
uniformly.
j u s t like t h e i n t e g r a n d ,
It
is
consequently
f r o m y,
the
integrand
Hence t h e i n t e g r a l an
analytic
a n d c a n be d i f f e r e n t i a t e d a r b i t r a r i l y
as
a
converges
function
of
z,
often.
Theorem 6.5.5: ( a ) L e t f b e a h o l o m o r p h i c f u n c t i o n w i t h d o m a i n of d e f i n i t i o n T S C a n d Β a ( s i m p l y c o n n e c t e d ) c l o s e d s u b s e t of Γ. t i o n can be r e c o n s t r u c t e d
In t h e i n t e r i o r of B,
f r o m its values
on t h e
c u r v e y = 3B:
f(z)
1 2nl wey
T h i s r e l a t i o n i s k n o w n a s Cauchy's integral, (b) Moreover,
f is a n a l y t i c ;
i t s n t h d e r i v a t i v e is
(oriented)
the
func-
boundary
6 . 5 . Holomorphic F u n c t i o n s
219
f"»(z)
=
_ η!
Γ f ( w ) dw
2iri
J (w-z)n+1 wey
( c ) Theorem of Liouville: A b o u n d e d a n a l y t i c f u n c t i o n on C i s c o n s t a n t . ( d ) "Fundamental theorem of algebra" a f t e r numbers
is a l g e b r a i c a l l y
closed.
Gauß:
Every
The f i e l d C of
polynomial
of
complex
positive
degree
w i t h c o m p l e x c o e f f i c i e n t s t h u s h a s a t l e a s t one z e r o in C. (It e v e n d e composes completely into linear
factors).
Proof: S t a t e m e n t s (a) a n d (b) h a v e b e e n p r o v e d
already.
If f i s a f u n c t i o n s u b j e c t t o t h e a s s u m p t i o n s t h e e x p l i c i t bound
of L i o u v i l l e ' s
theorem
| f ( z ) | s Μ f o r a l l ζ € C, we f i n d f r o m C a u c h y ' s
with
integral
f ormula
f'n>(0)
e'nl* fiRe")
— R 2π
ί
η! — Ff 2π
y denotes
R
this expression
the
=
Μ — R"
φ=Ο
φ=0
where
Μ dip
circle
of
radius
R > 0
approaches zero,
and
centre
p r o v i d e d η & 1.
0.
In t h e
limit
Consequently,
all
d e r i v a t i v e s of f v a n i s h a t ζ = 0 , a n d t h e f u n c t i o n m u s t b e c o n s t a n t . Now s u p p o s e ,
Ρ is a polynomial over € with η = deg(P) £ 1 w i t h o u t
Then t h e r e c i p r o c a l
f u n c t i o n ^j—- i s a n a l y t i c a l
in a l l of C.
zeroes.
It p o s s e s s e s
a
l i m i t i n g v a l u e f o r ζ —» co ( n a m e l y 0) and t h u s i s c o n t i n u o u s on t h e R i e m a n n s p h e r e . F r o m t h e c o m p a c t n e s s of f we d e d u c e t h e b o u n d e d n e s s of t h e f u n c t i o n ζ —» j^-j- a n d , After all,
by L i o u v i l l e ,
its constancy;
it f o l l o w s t h a t Ρ has a zero ζ
contradiction! e C, s a y ,
t h e f o r m P ( z ) = ( z - z ) - Q ( z ) w i t h Q e Pol (C). 1 n-l t h e p r o o f of G a u ß ' t h e o r e m .
Consequently, existence
(and
even a n a l y t i c .
and c a n be b r o u g h t
By i n d u c t i o n ,
we
c o m p l e x d i f f e r e n t i a b i l i t y of a f u n c t i o n f a l r e a d y i m p l i e s continuity)
of
all
higher-order
derivatives.
to
complete
In f a c t ,
the f
is
H o l o m o r p h y a n d a n a l y t i c i t y in a n o p e n r e g i o n a r e t h u s d i f f e -
rent ways to express the same.
6. Complex
220
T h i s m o s t r e m a r k a b l e r e s u l t i n d i c a t e s how much s t r o n g e r c o m p l e x ability like
i s in c o m p a r i s o n t o i t s r e a l
counterpart
for
Analysis
differenti-
which we have
nothing
this.
The c a l c u l a t i o n
of c e r t a i n i n t e g r a l s is made e a s i e r
by
T h e o r e m 6.5.6: (a) For
any
ζ € Γ,
holomorphic
and e v e r y
function
in
the
punctured
domain
Γ\{ζ),
( p i e c e w i s e ) C 1 - c u r v e y in Γ w h i c h e n c i r c l e s
n a t e d p o i n t ζ o n c e in t h e p o s i t i v e d i r e c t i o n ,
r e s
z
( f ( w ) d w )
=
( 2 i r i )
the
J
_ 1
the
where elimi-
integral
f ( w )
d w
wsj-
is
independent
of
γ.
The
expression
r e s (ω) = r e s ( f d z ) ζ ζ
residue o f t h e 1 - f o r m ω = f ( z ) d z a t t h e p o i n t z .
called
the
It i s a l s o
unaffected
i f w e r e p l a c e t h e v a r i a b l e w by a n o t h e r l o c a l l y u n i f o r m i z i n g
parameter.
(b) The
residue
vanishes
an a n t i d e r i v a t i v e
whenever
near z.
ω is closed,
i.e.
if
the
It t h e r e f o r e c a n a t t a i n v a l u e s
z e r o only if f is not holomorphic at (c)
is
function
f
different
has from
z.
In c a s e t h e r e g u l a r c o m p l e x d o m a i n Β c o n t a i n s o n l y f i n i t e l y m a n y
points
t h e residue theorem
ζ 1 . . . . ,z η in w h i c h f i s n o t h o l o m o r p h i c ,
η J
f ( z )
d z
=
2 π ϊ
9B
^
r e s ^ t f
d z )
=
2 π 1
^
weB
r e s
1=1
( f
z
d z )
1
applies. ( d ) If f h a s in t h e v i c i n i t y of ζ t h e L a u r e n t
f ( w )
=
) )
a\ .
(
series
w ( w -- z )
k
k=-m
w e g e t r e s (f d z ) = a ζ -ι
Proof: Parts
(a)
and
(b)
are
clear;
(c)
follows
from
the
c u r v e y i s in Γ h o m o t o p i c t o t h e u n i o n o f η d i s t i n c t
observation circles,
that
each of
the
which
221
6.5. Holomorphic Functions
surrounds one singularity,
together with theorem 6.5.4.
Finally,
we obtain
statement (d) through a term-by-term integration.
Next,
we turn to two other very important properties of
some functions,
which we f i r s t want to describe abstractly.
Definition 6.5.2: ( a ) We call a continuous real-valued function u in the domain T S C
harmo-
nic if f o r every compact circular disk Κ = K ( p , z ) with centre ζ and r a dius p, which is completely contained in Γ, the condition
1
utz+pe1*1) d(p
Ζπρ
=
u(z)
φ=0
holds. This means that the average of u over the periphery of Κ equals the function value at the centre, ( b ) A potential function in Γ is a twice continuously d i f f e r e n t i a b l e function u: Γ —> R solving the potential equation Au
In the available literature, rant & Hilbert [1968],
=
u
xx
+ u
yy
c f . f o r example Hurwitz & Courant [1964], Cou-
Cartan [19661 etc. , the nomenclature widely
varies.
The compromise used here has been chosen f o r reasons of vividness. U n f o r t u nately,
there is an inconsistency
compared to the notation
(harmonic f o r m s correspond to potential, This, however,
of
chapter
3
not to harmonic f u n c t i o n s ! ) .
does not really matter since we shall see later ( c f .
6 . 5 . 8 ) that both function classes are identical.
theorem
But b e f o r e can do so,
want to find out the interrelation with the holomorphic functions.
we
To this
end, we need
Theorem 6.5.7: ( a ) Let f = u + i v b e a
holomorphic function in the interior
of
a circle
222
6. Complex A n a l y s i s
( c e n t r e z, r a d i u s R), decomposed i n t o r e a l and i m a g i n a r y p a r t s u and v, respectively. the circle.
Assume f u r t h e r t h a t f is c o n t i n u o u s on t h e p e r i p h e r y Then u is uniquely d e t e r m i n e d by i t s b o u n d a r y v a l u e s
of and
c a n b e f o u n d b y Poisson's integral formula, v a l i d f o r r < R,
R2 - r 2
_1 2ir
uiz+re 1 * 1 )
2
R - 2 R r cosΙφ-ι/ι) + r
2
uiz+Re 1 *) dtp
ψ=ο
(b) The conjugate function ν of u is f i x e d e x c e p t f o r an a d d i t i v e
constant,
which we may s p e c i f y t h r o u g h t h e f u n c t i o n value a t t h e c e n t r e z:
R r Ξϊηίφ-ψ) v(z+re
)
v(z) •
u(z+Re 1|S ) dip
-
2
R - 2 R r cos(φ-ψ) + r 2 ψ=ο
(c) F u r t h e r m o r e ,
u and ν a r e h a r m o n i c p o t e n t i a l f u n c t i o n s .
(d) To each p o t e n t i a l f u n c t i o n u in Γ t h e r e is an a d j o i n t f u n c t i o n v, u n i q u e up t o a c o n s t a n t ,
also
such t h a t u + i ν is h o l o m o r p h i c .
Proof:
We w r i t e t h e Cauchy i n t e g r a l in p o l a r c o o r d i n a t e s :
f (z+re
1(l>
)
=
R e
11
Λ
ftz+Re ")
ο
R e ΙΨ
1(1.
- r e l(p
d^i
Φ=ο
T h e r e a f t e r we f o r m t h e a n a l o g o u s e x p r e s s i o n o b t a i n e d f o r t h e mirror point 2
of z + r e'*1, namely ζ + — e 1(P . It r e a d s
I* 0
=
Ή
f (z+Re'1") r e
1|ί
- 1V - R e
dψ
b e c a u s e t h e i n t e g r a n d is h o l o m o r p h i c i n s i d e t h e c i r c l e . p a r a t i o n i n t o f = u + i ν y i e l d s t h e e q u a t i o n s we seek.
S u b t r a c t i o n and s e -
223
6.5. Holomorphic Functions
In both of them, we can ( f o r r < R) interchange the Laplacian with the integral.
This leads immediately
to the validity
of
the potential
Au = 0 and Δν = 0. Setting r = 0, we get the averaging condition: are harmonic.
equations u and ν
If f i n a l l y Δ u
=
u
xx
+u
yy
=
0
we deduce d (u
dy - u dx)
χ
y
and the 1 - f o r m u
χ
connected,
dy - u
y
=
u
xx
dx
Λ
dx is closed.
dy - u
yy
dy
Λ
dx
Since the circular
=
0
disk is simply
the integrability f o l l o w s f r o m Poincar6's lemma, theorem 4.7.3.
Thus there exists an essentially ( i . e .
up to some constant) unique function
ν with dv
=
ν
χ
dx + ν
y
dy
=
- u dx + u y
χ
dy
A comparison of the c o e f f i c i e n t s leads to the Cauchy-Riemann
differential
equations which by theorem 6. 5.1 are equivalent to f = u + i ν being holomorphic. The proof is complete.
The Poisson integral enables us to show that harmonic and potential f u n c tions are the same:
Theorem 6.5.8:
Every potential function in a domain Γ £ C is harmonic,
and vice versa.
Proof: We f i r s t assume that u is harmonic in Γ. Then the maximum principle holds: In every compact domain Κ. £ Γ, the modulus of u attains its maximum at the boundary. We prove this proposition indirectly. inner point ζ of Κ exists with lar,
Let u be a counter-example;
|u(w)| s |u(z)| f o r all w 6 K.
thus an
In particu-
this is true f o r all w on the periphery of any circular disk D which
224
6. Complex A n a l y s i s
is c o n t a i n e d in K, b u t s h a r e s a t l e a s t one b o u n d a r y p o i n t ρ e Rd(K) w i t h K. The d e f i n i t i o n of t h e w o r d " h a r m o n i c " i m p l i e s t h a t u h a s t h e c o n s t a n t v a l u e u(z) on t h e c i r c u m f e r e n c e of D. p o i n t of Κ w i t h t h e r e q u i r e d
In t h i s c a s e ,
however,
ρ is a
An i m m e d i a t e c o r o l l a r y of t h e maximum p r i n c i p l e is:
A function,
h a r m o n i c w i t h i n some domain and v a n i s h e s on t h e b o u n d a r y , zero.
boundary
property.
is
which is identically
This is b e c a u s e t h e maximum of t h e modulus m u s t be equal t o 0.
More g e n e r a l l y ,
a h a r m o n i c f u n c t i o n on a c o m p a c t s e t Κ c C is uniquely d e -
t e r m i n e d by i t s b o u n d a r y v a l u e s (if t h e r e w e r e t w o such f u n c t i o n s ,
their
d i f f e r e n c e would be z e r o on bd K). We a p p l y he l a s t r e s u l t t o t h e s p e c i a l c a s e Κ = c l o s e d c i r c u l a r disk r a d i u s ρ and c e n t r e z.
The r e s t r i c t i o n u |
is c o n t i n u o u s ,
and by
with theo-
K
J 'bd(K) r e m 6 . 5 . 7 t h e P o i s s o n i n t e g r a l gives t h e h a r m o n i c e x t e n s i o n into t h e i n t e r i o r of Κ (which is unique, a s we j u s t s a w ) . The same a r g u m e n t s h o w s t h a t Au = 0 t h e r e .
Next,
we c o n s i d e r t h e r e v e r s e s t a t e m e n t . Let u now be a p o t e n t i a l f u n c t i o n
on Γ. Moreover,
t a k e an a r b i t r a r y p o i n t ζ e Γ and a c o m p a c t c i r c u l a r
disk
in Γ c e n t r e d a t z. It w i l l be enough t o show t h a t t h e z e r o f u n c t i o n is t h e only s o l u t i o n
of
t h e p o t e n t i a l e q u a t i o n in Κ which v a n i s h e s on t h e b o u n d a r y . This is t r u e b e c a u s e P o i s s o n ' s i n t e g r a l p r o v i d e s us w i t h a c o n t i n u a t i o n
in-
t o t h e i n t e r i o r of t h e disk.
We d e n o t e t h i s f u n c t i o n by u . The d i f f e r e n c e
u - u
is
f u n c t i o n and
u - u
=0,
also
a potential
identically
zero
on Rd(K);
thus
and we a r e done.
In o r d e r t o c l o s e t h e gap in t h e f o r e g o i n g p r o o f , we a s s u m e t h a t u is a p o t e n t i a l f u n c t i o n which v a n i s h e s on bd(K). By t h e o r e m 6 . 5 . 7 (d), we can f i n d a h o l o m o r p h i c f u n c t i o n f : Κ —» C whose r e a l p a r t is i d e n t i c a l w i t h u. c o n s t a n t of i n t e g r a t i o n is chosen such t h a t f ( z ) = 0 h o l d s f o r some
The point
ζ 6 Rd(K). Then f = 0 in t h e e n t i r e domain Γ; t h e same c o n c l u s i o n h o l d s f o r u, and t h e t h e o r e m e v i d e n t l y f o l l o w s .
6.6.
Angle-Preserving Transformations
225
6.6. Angle-Preserving Transformations By theorem 6 . 5 . 2 , a n a l y t i c a l (or holomorphic) mappings of the complex numbers are conformal,
except f o r the zeroes of the derivative.
We may ask
which r e g i o n s can be t r a n s f o r m e d b i j e c t i v e l y into each o t h e r by a n a l y t i c a l f unctions. If we r e s t r i c t our a t t e n t i o n to simply connected domains,
the complete a n -
swer is given by the Riemann mapping theorem 6. 6. 3. To prepare i t s p r o o f , we shall need a number of a u x i l i a r y r e s u l t s :
Theorem 6.6.1: Let f j . f g . fg. · · ·
be
a
uniformly bounded sequence of holomorphic
functions
in a complex domain Γ; i. e. |f k Cz)|
< Μ
where the bound Μ does not depend on k e IN and ζ e Γ. Then on each compact subset of Γ a uniformly convergent subsequence of the f
k
e x i s t s (theorem of Hontel).
Proof: Consider an a r b i t r a r y compact subset Β of Γ. We f i r s t claim: To each e > 0 one can find a δ > 0 which f o r all a, b e Β with f i e s the inequality
| a - b | < δ and all k s a t i s -
| f a . J - f ^ i b ) j < ε.
Suppose this is f a l s e .
,
k
Then t h e r e a r e a positive ε and two sequences {a }
and {b } in Β with la - b I < - , but If (a ) - f (b )|1 £ ε. Since Β is comk ' n n ' n ' n n n n p a c t , we can c o n s t r u c t convergent subsequences < a ' } and { b ' > of the a and k k η b , r e s p e c t i v e l y , which tend to the common limit point w e B. π We f i x
ρ > Ο subject
to the condition
that
the
open c i r c l e
Κ = K(2p, w)
around w of radius 2p is contained in Γ. Then obviously a number Η e x i s t s such t h a t f o r all m > Η the inequalities hold.
|1 am ' - w | < ρ as well as
The Cauchy i n t e g r a l f o r m u l a (theorem 6 . 5 . 5 ) moreover yields
|' ibn' - w |1 < ρ
6. Complex Analysis
226
If ( a ' ) - f ( b1 ' ) I
' m m
mm
- Ι Γ f (z) £ ( J m 3K
a
["(z-a' ) _ 1 - ( z - b ' ) _ 1 ] dzl m m I |
2n π
a'-b' m m — 4πρ Μ 2" min{(z-a')(z-b')}
s
s
2pM p" 2 i f 1
=
2M — np
which in the limit η -» , and so on in the indicated manner, n 2 The "diagonal sequence" F = f evidently converges f o r all ζ and consen n,n k quently f o r all points with rational coordinates in Γ. Finally,
we cover the compact set Β by suitable neighbourhoods of the ζ
( e . g . circular disks with f i x e d radius - ) and choose an arbitrary f i n i t e subcover Κ , Κ , . . . ι 2 Each of the circles Κ contains a rational point ζ : hence f o r all ζ e Β at least one ζ
J
J
J
can be found whose distance from ζ is less than δ. The conver-
gence property of the F^ proved above guarantees the existence of a number N(e),
depending on c, with IF ( z J - F 1 m k
(z ) I η k '
Ν (here, of course, we need the finiteness of the covering).
In particular,
the special z^ constructed earlier obey this
restric-
tion, and we have IF
'm
(z)-F (z) I η '
s
IF
'm
(z)-F (ζ ) m j '
I +
(z )-F ( ζ ) ' m j n j '
IF
I +
(z ) - F ( ζ ) 1I ' n j n
IF
which was to be shown.
It is advisable to formulate the next intermediate step separately.
s
3 ε
6.6.
Angle-Preserving Transformations
227
Theorem 6.B.2: Consider
a s e q u e n c e of h o l o m o r p h i c f u n c t i o n s f , f , f , . . .
in t h e o p e n
do-
m a i n Γ S C w h i c h is u n i f o r m l y c o n v e r g e n t in e a c h c l o s e d s u b s e t of Γ. If a l l f are injective or, k f unction
a s one s a y s in c o m p l e x a n a l y s i s ,
ζ
ι—>
univalent, t h e
limit
F(z) = lim f (z) η n-W>
e i t h e r h a s t h e same p r o p e r t y or is c o n s t a n t .
Proof: We k n o w a l r e a d y f r o m t h e o r e m 6 . 2 . 2 and t h e o r e m 6 . 5 . 5 t h a t t h e l i m i t f u n c t i o n F is h o l o m o r p h i c .
Suppose F t o be n o n c o n s t a n t and s e l e c t a point a e Γ
a n d a c i r c l e κ w i t h c e n t r e a on w h i c h F d o e s n o t a t t a i n t h e v a l u e w = F ( a ) ( t h i s i s p o s s i b l e ! ). T h e
expression „ N
_ "
ι Γ F ' ( z ) dz i^T J F ( z ) - w κ
i s by t h e r e s i d u e t h e o r e m e q u a l t o t h e n u m b e r of z e r o e s ( c o u n t e d w i t h m u l tiplicities)
of F ( z ) - w in t h e r e g i o n e n c l o s e d by κ.
In p a r t i c u l a r ,
Ν is
in-
teger. T h i s r e m a i n s t r u e if w e r e p l a c e F by one of t h e f . But t h e i n t e g r a l d e n p e n d s c o n t i n u o u s l y on F; t h e r e f o r e i t d i f f e r s f o r s u f f i c i e n t l y l a r g e η by an a r b i t r a r i l y s m a l l a m o u n t f r o m N. S i n c e b o t h n u m b e r s a r e i n t e g e r s , must coincide.
By c o n s t r u c t i o n ,
they
e a c h f η- w h a s o n l y o n e ( s i m p l e ) z e r o in t h e
d o m a i n b o u n d e d by κ; h e n c e Ν = 1. We n o w a p p l y o u r a r g u m e n t a t i o n
to a circle which does not s u r r o u n d
the
p o i n t a . T h i s l e a d s t o a s i m i l a r i n t e g r a l w h i c h t u r n s o u t t o b e 0, s i n c e o t h e r w i s e a l l f w i t h l a r g e η w o u l d h a v e a s o l u t i o n of f (z) = w i n s i d e t h e π π circle, contradicting the assumption that f is u n i v a l e n t . T h u s t h e r e i s η e x a c t l y o n e ζ w i t h F ( z ) = w, n a m e l y a .
We a r e n o w p r e p a r e d t o w r i t e down t h e m a i n r e s u l t on c o n f o r m a l
equivalence:
228
6. Complex Analysis
Theorem 6.B.3: Every simply connected domain Γ c C is conformally equivalent to the open unit disk
Κ
Thus,
=
j ζ e C
|z| Γ,
is analytic
Γ —> Κ which,
(Riemann mapping theorem,
as w e l l also
as
called
its the
main theorem of conformal mapping). We may even r e q u i r e in addition that a given point w e Γ is t r a n s f o r m e d to f ( w ) = 0 and that the d e r i v a t i v e f ' ( w ) conditions,
is r e a l and positive.
With these
by-
the f u n c t i o n f is unique.
Remarks: ( 1 ) Γ must be simply connected since the desired mapping is a homeomorphism f r o m Γ onto the unit disk. ( 2 ) The r e s t r i c t i o n to proper subsets of C, i. e. is l i k e w i s e essential,
the exclusion of C i t s e l f ,
because every analytic mapping f
f r o m C into Κ
is bounded by 1 and thus constant by L i o u v i l l e ' s theorem 6 . 5 . 5 This is not compatible with the assumptions on f .
(c).
It is the more a s t o n -
ishing that a l l other Γ are admissible. ( 3 ) As a c o r o l l a r y ,
w e obtain: T w o a r b i t r a r y simply connected complex d o -
mains other than C are d i f f e o m o r p h i c to each other,
but not to C ( a l -
though they are homeomorphic with C! ).
Proof: Since the composition of f i n i t e l y many c o n f o r m a l b i j e c t i o n s is of the same type,
we are a l l o w e d to impose certain
additional
restrictions
on the
do-
main Γ. By assumption,
there is a complex number a which is not contained in Γ. We
consider the l o g a r i t h m l n ( z - a ) , cording to the equation
defined via the inverse function of exp
ac-
6.6. Angle-Preserving Transformations
e
Since Γ is simply c o n n e c t e d ,
ln(z-a)
=
the part
above Γ s p l i t s into i s o l a t e d l e a v e s . ln(z-a)
229
z - a
of t h e Riemann s u r f a c e which
We s e l e c t one of t h e m .
lies
This b r a n c h of
is an a n a l y t i c a l f u n c t i o n g: Γ —> C, b e c a u s e i t s only
singularities
( a t co and a) a r e not in Γ. Furthermore,
g is u n i v a l e n t , ζ
ι
=
f o r f r o m g ( z j ) = gtZg).
a + e x p ( g ( z )) ι
=
a + e x p ( g ( z )) 2
=
ζ
2
immediately follows. We now c h o o s e some w e Γ. It pcissesses a c i r c u l a r n e i g h b o u r h o o d which is c o m p l e t e l y c o n t a i n e d in t h e domain of g. A f t e r s h i f t i n g by 2tri, we g e t o t h e r c i r c u l a r disk which,
t h a n k s t o t h e i n j e c t i v i t y of e x p ° g ,
is
an-
disjoint
w i t h g ( D . The a u x i l i a r y f u n c t i o n
g ( z ) - g(w) + 2iri is c o n s e q u e n t l y and a n a l y t i c a l ,
bounded
in Γ.
Except
for
this,
it
is obviously
We a r e t h u s a l l o w e d t o r e p l a c e Γ by i t s image u n d e r transformation, t o be bounded.
univalent
a s i t is composed f r o m f u n c t i o n s e n j o y i n g t h e s e p r o p e r t i e s . the just
described
or - t o e x p r e s s it d i f f e r e n t l y - we m i g h t have s u p p o s e d Γ Without r e s t r i c t i o n ,
we s h a l l a s s u m e f r o m now on Γ t o be
c o n t a i n e d in t h e open u n i t disk
Κ =
| z e c |
By s h i f t i n g and a s u i t a b l e n o r m a l i z a t i o n ,
| z | < 1 j·
we can beyond t h i s a c h i e v e
t h e d i s t i n g u i s h e d p o i n t w is l o c a t e d a t t h e o r i g i n of t h e n a t u r a l nate system:
that
coordi-
w = 0.
After this reduction,
we c o n s t r u c t t h e f u n c t i o n f a m i l y S c o n s i s t i n g of
all
f u n c t i o n s , which a r e h o l o m o r p h i c in Γ, bounded by 1, u n i v a l e n t , and s u b j e c t to the
restrictions
fCO) = 0 and
f ' ( 0 ) > 0.
In any c a s e ,
3 contains
the
i d e n t i t y and is t h u s not empty. We s e l e c t an a r b i t r a r y n o n z e r o p o i n t a e Γ and keep i t t h r o u g h o u t t h e r e s t of t h e p r o o f . Each f e g omorphism).
maps Γ o n t o a domain f ( D £ Κ ( s i n c e it is a h o m e -
230
Our next
6. Complex A n a l y s i s
aim is to determine f
For the sake of s i m p l i c i t y ,
such that
|f(a)| is as l a r g e
as
possible.
we w r i t e
sup I
and build a sequence f ^ f ^ f 3 > . . .
|f(a) I
f e 3 j-
in 3, f o r which
Ilm f ( a ) π
=
μ
holds. As we saw b e f o r e , the identity is a member of 3. and we may conclude μ £ |a| > 0 .
Montel's theorem 6 . 6 . 1 guarantees f o r e v e r y given compact sub-
set of Γ the existence of
a partial
sequence g j . g , g g
say,
which converges u n i f o r m l y to a holomorphic function F.
of
the f ,
On account of
the
conditions F ( 0 ) = 0 and |F(a)| = μ, the limit function is not constant,
and
theorem 6. 6. 2 thus implies that F is univalent,
since the f
are.
It remains to show that the image domain F ( D completely f i l l s out the int e r i o r of the unit disk in which it is obviously
contained.
Suppose not.
Then there is some point w, lying in K, but not in F ( D . We lw e x p r e s s it in polar coordinates: w = ρ e , where 0 < ρ < 1. Furthermore, we introduce the abbreviation 2 w c
2Vp
=
= (l+p)Vp
e
ίφ
1+p
The f u n c t i o n
w ζ ζ
ι—»
h(z)
(z-c)
-ίφ
= ρ (c ζ - 1)
maps Κ b i j e c t i v e l y onto i t s e l f . C, namely
p1/2
e1*
and ρ
1/2
e 1 *.
z
(z-c)
c ζ - 1
The d e r i v a t i v e h' has e x a c t l y t w o z e r o e s in The associated f u n c t i o n values are w and * -2 w = ρ w. (By the way, the
i t s mirror image r e l a t i v e to the unit c i r c l e ,
arguments are also t r a n s f o r m e d into each other by the inversion at 3K). -ι * The inverse function h has r a m i f i c a t i o n points only in w and w . Both •
points are outside F ( D : The t r a n s f o r m a t i o n h
1
connected domain F ( D .
·
w by d e f i n i t i o n and w , because we have thus is unramified, In other words,
i.e.
univalent,
it is a c o n f o r m a l
|w | > 1.
over the simply equivalence.
231
6.6. Angle-Preserving Transformations
But, as one easily v e r i f i e s , to zero than ζ i t s e l f ; distances:
|h
1(z)|
every ζ e Κ has an image h ( z ) which is closer
the inverse function h
> |z| f o r
all
ζ
with
1
consequently
|z| < 1.
This
is a
increases the contradiction
to the defining maximum condition on F. So we do have F ( D = K. The existence part of the Riemann mapping theorem follows. It remains to prove the uniqueness.
This is a consequence of the f a c t that
there is only one angle-preserving mapping of Κ onto i t s e l f , the origin invariant and has a positive derivative there,
which
leaves
namely the iden-
t i t y function F ( z ) = z. The last statement in turn is an immediate corollary of the c l a s s i f i c a t i o n of all ( c o n f o r m a l ) isomorphisms of the unit disk:
Theorem 6.6.4: The conformal automorphisms of the unit disk are just the linear fractional transformations of the special f o r m
τ(ζ)
=
a ζ + b b ζ + a
with complex numbers a, b, f o r which the by-condition a a - b b
=
1
holds.
Proof: These function are clearly
b i j e c t i v e mappings of
the open unit disk
onto
i t s e l f and holomorphic in both directions. We have to demonstrate that there are no further transformations with the required properties. Suppose, τ is an analytical automorphism of K. We may r e s t r i c t the discussion to the special case τ ( 0 ) = 0 and τ ' ( 0 ) > 0, because this normalization can be achieved a f t e r w a r d s by a suitable rational The function f :
Κ —» C defined by
linear
transformation.
232
6. Complex Analysis
' ζ f(ζ)
1
τ(ζ)
( ζ * 0)
= • τ'(0)
( ζ = 0)
is holomorphic in Κ and has everywhere on the boundary of K, points ζ with |z| = 1 ,
This implies
at all
the constant modulus 1. (This can be corroborated by
a simple limit process). theorem 6 . 5 . 8 ,
i. e.
Thanks to the maximum principle in the proof
of
the modulus |f(z)| is bounded above by 1.
|τ(ζ)| s |z|' in all of
K.
For the inverse transformation τ
1
the same argument applies, and we conclude |τ(ζ)| = ζ . Furthermore, in the limit ζ —» 0, we get τ ' ( 0 ) = 1. Theorem 6 . 3 . 4 says that τ ( ζ ) can be developed into a power series τ(ζ)
=
z + a
2
ζ
2
+ a
3
ζ
3
+ ...
which converges everywhere in K. If one of the c o e f f i c i e n t s a
were d i f f e r -
ent f r o m 0,
|τ(ζ)| = |z|.
this would o f f e n d against our previous equation
Hence we come to the desired conclusion: a f o r all ζ with
= 0, or, equivalently, τ ( ζ ) = ζ
Izl < 1.
Remark: The same considerations even yield a bit more, namely the extremely important lemma of Schwarz,
Theorem 6.6.5: A holomorphic function f which transforms the interior of the unit into i t s e l f
is either a rotation,
circle
that is a linear map of the special
f ( z ) = e i a ζ with some real constant a, or f o r all ζ with 0 < |z| < 1 , inequality
|f(z)| < |z| is true (and then also | f ' ( 0 ) | < 1).
form the
6.7. L i e Groups
233
6.7. Lie Groups Many mathematical and physical questions lead naturally to problems in a l gebra or d i f f e r e n t i a l geometry.
So, f o r instance,
composition of mappings is always associative.
it is well-known that the
This is why the symmetries
of any o b j e c t f o r m a group. On the other hand, just those spaces which o c cur in s c i e n t i f i c investigations are usually endowed with some topological, if
not
even
differentiable
structure.
It
suggests
itself
to
combine
both
aspects. We owe the f i r s t systematic considered
smooth
(i.e.
approach in this direction
infinitely
often
to Lie [1970].
differentiable)
which a product exists which defines a group multiplication. especially
its
interrelation
called Lie algebras, ful.
Later,
with a distinguished
on
This idea, and algebras
(today
a f t e r their discoverer) proved to be extremely
fruit-
much more general
class of
he
manifolds,
objects have been investigated.
Here,
we
Though we only need a f e w elementary concepts and results f r o m this
ex-
merely mention b r i e f l y the topological groups. tremely multilayered f i e l d which has continued to be of topical interest up to now, it seems appropriate to include at least a f e w advanced results. For reasons of space limitation we r e f r a i n f r o m discussing explicit p r o o f s ; almost all statements can be v e r i f i e d easily with the means we have at our disposal. The only two exceptions are theorem 6 . 8 . 3 by Ado, which is not inherently difficult,
and,
above all,
the solution of
so-called 5 t h Hilbert problem (see below).
but somewhat complicated,
The latter
demands very
the
deep-
rooted and specialized methods and can t h e r e f o r e not even be sketched in this book. The f u r t h e r developments during the 20 t h century in d i f f e r e n t as there are algebraic geometry, (algebraic) topology,
directions,
and group theory or
the recent applications in physics (dynamical systems, c e l e s t i a l mechanics, quantum theory,
etc. ) can only be mentioned without detailed
explanation.
The literature on these branches is abundant and easily accessible. But let us put these remarks in more concrete terms!
6. Complex A n a l y s i s
234
Definition 6.7.1: ( a ) A topological group is a t o p o l o g i c a l s p a c e G w i t h a p r o d u c t
• :
s u c h t h a t (G, •) i s a g r o u p ,
G χ G
—>
G
(a,b)
ι—»
a-b
and b o t h ,
a
1
and a - b ,
depend
continuously
on a a n d b. CO
( b ) If G, (a,b)
beyond this,
is even a C - m a n i f o l d a n d t h e g r o u p
t—> a · b a s w e l l a s t h e i n v e r s i o n a
a
1
multiplication
are smooth mappings,
we
c a l l G a Lie group.
It i s e a s i l y s e e n t h a t s u b g r o u p s and f a c t o r g r o u p s of G a r e a l s o cal groups.
E v e r y o p e n s u b g r o u p s is a l s o
topologi-
closed.
T h e r e s u l t w e w a n t t o q u o t e n e x t h a s a l r e a d y b e e n f o u n d by Lie h i m s e l f :
Theorem 6.7.1: F o r a n y L i e g r o u p G w i t h C°°-atlas U, one c a n a l w a y s f i n d a n a n a l y t i c a l l a s 11' w h i c h i s c o m p a t i b l e w i t h U s u c h t h a t m u l t i p l i c a t i o n a n d i n v e r s i o n G are (real) analytic
in
mappings.
Lie a l s o s h o w e d t h a t w e a l r e a d y g e t Lie g r o u p s i f ,
in d e f i n i t i o n 6 . 7 . 1 ,
m e r e l y r e q u i r e G a n d a l l g r o u p o p e r a t i o n s t o be t w i c e c o n t i n u o u s l y rentiable.
at-
T h i s r a i s e d t h e q u e s t i o n w h e t h e r t h i s a s s u m p t i o n c a n be
we
differelaxed
f urther. In a t a l k
delivered to the 2nd I n t e r n a t i o n a l
Mathematical
C o n g r e s s in
r i s , H i l b e r t p o s e d a s one of 2 3 q u e s t i o n s t o f u t u r e g e n e r a t i o n s of
Pa-
mathema-
ticians:
Hilbert's 5 t h Problem: To w h a t e x t e n t i s L i e ' s c o n c e p t of c o n t i n u o u s t r a n s f o r m a t i o n g r o u p s sible to our investigations t y of t h e f u n c t i o n s ?
even w i t h o u t t h e a s s u m p t i o n of
acces-
differentiabili-
6 . 7 . L i e Groups
235
A l i t e r a l q u o t a t i o n c a n be f o u n d e . g .
in H i l b e r t [1900] o r ( w i t h a
histori-
c a l e v a l u a t i o n of t h e a t t e m p t s a r i s i n g f r o m i t ) S k l j a r e n k o [1971]. T h e c o m p l e t e a n s w e r i s d e d u c e d f r o m a c o m b i n a t i o n of t h e p a p e r s by P o n t r j a g i n [1934],
G l e a s o n [1952], M o n t g o m e r y & Z i p p i n [1952].
It
reads:
Theorem 6.7.2: E v e r y t o p o l o g i c a l g r o u p d e f i n e d on a C ° - m a n i f o l d i s a Lie g r o u p .
We s h a l l n o t u s e t h i s t h e o r e m ;
in a l l a p p l i c a t i o n s ,
t h e s m o o t h n e s s of
the
r e l e v a n t m a p s will be given f r o m t h e o u t s e t . The n e x t
d e f i n i t i o n is a l r e a d y f a m i l i a r to us (section
of c l a r i t y ,
however,
2.2).
For t h e
w e r e p e a t t h e e x p l a n a t i o n of t h e t e r m s
sake
"commutator"
and "solvability" once more.
Definition 6.7.2: ( a ) T h e conmutator of t w o e l e m e n t s a, b of a g r o u p G is
[ a , b]
=
a
1
b
1
a b
If U a n d V a r e s u b g r o u p s of G, w e d e n o t e by [U,V] t h e s u b g r o u p g e n e r a t ed by a l l c o m m u t a t o r s [ u , v ] w i t h u e U and ν e V. ( b ) T h e lower central series of G is t h e s e q u e n c e G , G^, G 3
recursively
c a l c u l a b l e via G
= G and G = [G,G ]. ι k+i k If t h e r e i s a n a t u r a l n u m b e r k w i t h G^ = 1, t h e g r o u p G i s nilpotent.
(c) Similarly,
= G a n d G 1 " " ' = [ G ^ ' . G 1 " ], w e o b t a i n t h e de-
setting G
rived series of
G.
If
it
eventually
leads
down t o
the
trivial
group,
t h e n G i s solvable ( o r soluble).
One s e e s w i t h o u t d i f f i c u l t y t h a t n i l p o t e n t g r o u p s a r e s o l v a b l e ; s t a t e m e n t is,
in g e n e r a l ,
the
reverse
false.
We r e t u r n t o t h e Lie g r o u p s and i n t r o d u c e s o m e m o r e n o t a t i o n of f u n d a m e n t a l importance.
236
6. Complex A n a l y s i s
D e f i n i t i o n 6.7.3:
( a ) The g r o u p g e n e r a t e d by a l l c o n n e c t e d s o l v a b l e n o r m a l s u b g r o u p s of a Lie g r o u p G is t h e radical of G. We d e n o t e i t by Rad(G). (b) A c o n n e c t e d Lie g r o u p G w i t h Rad(G) is semisimple.
The r a d i c a l
can a l s o be c h a r a c t e r i z e d
as the maximal
connected
solvable
n o r m a l s u b g r o u p of G. The g r o u p G is s e m i s i m p l e if and only if i t is c o n n e c t e d and does n o t c o n t a i n c o n n e c t e d s o l u b l e n o r m a l s u b g r o u p s e x c e p t 1. Both a s s e r t i o n s a r e b a s e d on t h e f a c t t h a t t h e p r o d u c t of s o l u b l e
normal
subgroups has the same property. The s e m i s i m p l e o n e s a r e in a c e r t a i n s e n s e t h e b u i l d i n g b l o c k s f r o m which a l l Lie g r o u p s can be c o n s t r u c t e d .
T h e i r r e l e v a n c e is a l s o d e r i v e d f r o m
T h e o r e m 6.7.3:
( a ) Every Lie g r o u p G c o n t a i n s a t l e a s t one Levi complement of t h e r a d i c a l . By t h i s we m e a n a m a x i m a l s e m i s i m p l e s u b g r o u p S. It f u l f i l l s t h e c o n d i tions Rad(G)-S
= G
Rad(G) η S
=
and 1
(theorem of Levi).
(b) All Levi c o m p l e m e n t s a r e c o n j u g a t e in G (theorem of Mal'tsev). ( c ) The f a c t o r g r o u p G/Rad(G) of a c o n n e c t e d Lie g r o u p G is s e m i s i m p l e .
These p r o p o s i t i o n s r e d u c e t h e i n v e s t i g a t i o n of c o n t i n u o u s Lie g r o u p s t o t h e d i s c u s s i o n of t h e ( s o l u b l e ) r a d i c a l , t y p e of e x t e n s i o n .
To c a r r y t h i s o u t ,
t h o d s of t h e s u b s e q u e n t
section.
its (semisimple) complements,
and
one mainly a p p l i e s t h e a l g e b r a i c
the me-
6 . 8 . Lie A l g e b r a s
237
Β.8. Lie Algebras
The c e n t r a l i d e a of Lie t h e o r y is t o a s s o c i a t e w i t h each Lie g r o u p G an a l g e b r a w h o s e u n d e r l y i n g s p a c e is t h e t a n g e n t s p a c e TG^ a t t h e t r i v i a l ment of G. R e m a r k a b l y ,
t h i s mapping in many c a s e s is i n v e r t i b l e ,
ele-
and t h e
s t r u c t u r e of t h e Lie a l g e b r a t h u s o b t a i n e d in t h e l a r g e d e t e r m i n e s t h a t of G. T h i s Lie correspondence is of d e c i s i v e r e l e v a n c e f o r an u n d e r s t a n d i n g of t h e Lie g r o u p s . With i t s h e l p ,
f o r instance,
Killing [1888-1890) s u c c e e d e d t o c a r r y
a c o m p l e t e c l a s s i f i c a t i o n of a l l s i m p l e complex Lie g r o u p s .
through
In doing so,
p e r f o r m e d a m a j o r p a r t of t h e p r o g r a m d r a w n up in t h e f o r e g o i n g
he
section
(some i n a c c u r a c i e s which s l i p p e d i n t o Killings work w e r e c o r r e c t e d l a t e r by C a r t a n [1952-1953];
c o n c e r n i n g h i s t o r y and r e l e v a n c e of t h e
classification,
c f . Yaglom [1988], Coleman [1989], Hawkins [1982] u . a . ) . It is t h i s p a r a l l e l i s m b e t w e e n d i f f e r e n t i a l g e o m e t r i c and a l g e b r a i c which m a k e s L i e ' s t h e o r y so v e r s a t i l e ,
aspects
and it h a s such a g r e a t many a p p l i -
c a t i o n s in m a t h e m a t i c s and p h y s i c s t h a t we s h a l l not a t t e m p t t o l i s t
them
t o any d e g r e e of c o m p l e t e n e s s . All we can do is t o give some n o n r e p r e s e n t a tive examples. Chevalley's description
of t h e Lie a l g e b r a s in c o m p a r i s o n
to the
original
c o n s t r u c t i o n by Killing h a s t h e g r e a t a d v a n t a g e t h a t a l l t h e structure constants of t h e a l g e b r a (see below) become i n t e g e r and n o t only r a t i o n a l n u m bers. Recently, trary
L i e ' s c o n s t r u c t i o n s could be t r a n s f e r r e d a l o n g t h i s l i n e t o
fields
of
positive
characteristic
(Chevalley [1955];
the
arbi-
investiga-
t i o n s of t h e 19 t h c e n t u r y a r e s t r i c t l y c o n f i n e d t o t h e g r o u n d f i e l d s IR and C).
So,
e.g.,
it
became
possible
to
classify
"almost
a l l " finite simple
groups w i t h i n a u n i f o r m s c h e m e ( a s "groups of Lie type" = Chevalley groups"
or modifications t h e r e o f ,
t h e s o - c a l l e d "twisted types"; m o r e
can be f o u n d in G o r e n s t e i n [1968]).
information
Only t h e alternating groups A (n £ 5) η
and t h e 26 sporadic groups a r e o u t s i d e t h i s c a t e g o r y .
As soon a s we have
developed
once a g a i n ,
some t e c h n i c a l
tools,
we s h a l l
come back
if
only
b r i e f l y , to t h i s important point (section 21.6). We a r e n o t going t o p r e s e n t t h e c l a s s i f i c a t i o n of s i m p l e c o m p l e x Lie a l g e b r a s in t h i s book,
b e c a u s e we have no a p p l i c a t i o n s f o r t h e r e s u l t .
It
is,
238
6. Complex Analysis
however,
closely related to the Coxeter theory of part D. Especially
the
root systems and the Dynkln diagrams described there will play a fundament a l part.
They are slight modifications of concepts which were also i n t r o -
duced by Killing [1888-1890], During the last decades,
many branches of physics utilized Lie groups and
algebras. We cannot do more than j u s t mention (celestial) mechanics and hydrodynamics. In this theory of fluid motion, f o r example, the KdV- (Korteweg-deVries-)equation has aroused great interest.
This is a completely in-
tegrable dynamical system with infinitely many degrees of freedom.
It is
studied with the help of algebraic methods based on certain generalized Lie algebras,
t h e Kac-Moody algebras.
The l a t t e r are also valuable,
if not essential,
f o r a deeper understanding
of f i n i t e groups; furthermore they have important applications in particle physics (key phrase:
superstrings). One can say without exaggeration
Lie groups and Lie algebras are the ideal means to c h a r a c t e r i z e
that
(continu-
ous) symnetries.
But let us now return to our topic proper! We s t a r t with the a b s t r a c t d e f i nition of the term "Lie algebra".
Definition 6.8.1:
An algebra L over the field Κ is a Lie algebra if the product [.,.]:
L χ L
—»
L
(a,b)
ι—>
[a, b]
(aside from the self-evident bilinearity) f u l f i l l s the following two conditions: (a) For every a e L, we have [a,a] (antisymmetry);
=
0
and
(b) f o r all a, b, c e L, [[a,b],c] + [[b,c],a] + [[c,a],b] (Jacobi-identity).
=
0
6.8. L i e Algebras
239
Remarks: ( 1 ) The antisymmetry implies an identity, by polarization,
namely [a,b] + [b,a]
Vice versa,
which is deduced without e f f o r t
=
0
if the characteristic of the ground f i e l d is d i f f e r e n t f r o m
2, the reverse statement is also true,
but f o r char Κ = 2, the version
used in the definition is stronger. (2) The symbol [ . , . ] f o r the Lie product stems f r o m the f a c t that the mat r i x commutator [Α, Β]
=
Α Β - ΒA
endows the vector space K ( n , n > of the (n, n)-matrices with c o e f f i c i e n t s in Κ with a Lie algebra structure. We even have the much more general, but nevertheless elementary
Theorem 6.8.1: ( a ) Every K-algebra L with associative product ( a , b )
i-» a-b becomes a Lie
algebra with the new multiplication rule [a.b]
=
a-b - b-a
f o r all a, b e L. ( b ) The set of the derivations of an algebra A, i . e . the linear maps D of A into itself with D(a-b) = D(a)-b + a-D(b) f o r all a,b e A, is a Lie
al-
gebra Der(A) with the Lie product [D ,D ] 1 2
=
D «D 1 2
- D °D 2 1
( c ) The adjoint transformations to the elements χ of an arbitrary Lie a l g e bra L, ad(x):
y ι—> [ x , y ]
240
6. Complex Analysis
a r e called
inner derivations
of L. They form a (usually proper) s u b a l -
gebra ad(L)
= I ad(x) | χ e L |
of Der(L). The assignment ad:
L —> Der(L) χ ι—> ad(x)
is the adjoint representation of L. Its image ad(L) is isomorphic with the f a c t o r a l g e b r a L / Z ( L ) of L by i t s centre, that is the subalgebra Z(L) (d) The
Killing f o r m
= \ ζ 6 L
[z, L]
g, defined by
g(x,y)
= tr |ad(x)-ad(y)j
is nondegenerate (det g * 0) if and only if L is semisimple (has no nontrivial solvable ideals). (e) Every simple Lie algebra is semisimple. Α semisimple Lie a l g e b r a decomposes into a direct sum of simple ideals.
(Of course, representations of Lie algebras employ in the image space the product provided by the commutator instead of m a t r i x multiplication; with this single exception, the definitions of section 2 . 4 remain valid). Similar propositions as that of the l a s t theorem can be stated f o r the commutator of vector f i e l d s on manifolds:
T h e o r e m 6.8.2:
( a ) The set SUM) of all vector f i e l d s on a smooth manifold Μ is a Lie a l g e bra with the commutator as product.
6 . 8 . Lie Algebras
(b) In p a r t i c u l a r ,
241
the differential operators
d e f i n e an i n f i n i t e - d i m e n s i o n a l r e a l Lie a l g e b r a on t h e s p a c e of a l l gular
vector
fields
in C = C\{0),
c o m m u t a t i o n r e l a t i o n s of t h e L
called
the
Virasoro algebra.
reThe
are
k =
(J-1}
L.
+J
The m a p p i n g L^ —> k y i e l d s a graduation on t h e V i r a s o r o a l g e b r a .
It i s i n -
t i m a t e l y c o n n e c t e d w i t h Kac-Moody a l g e b r a s and of f u n d a m e n t a l r e l e v a n c e f o r p a r t i c l e p h y s i c s (string theory). We now r e t u r n t o o u r g e n e r a l c o n s i d e r a t i o n s .
The a d j o i n t r e p r e s e n t a t i o n
is
n o t a l w a y s f a i t h f u l , b u t we have t h e i m p o r t a n t
Theorem 6.8.3:
Every f i n i t e - d i m e n s i o n a l Lie a l g e b r a over an a r b i t r a r y f i e l d of istic 0 possesses a faithful representation
character-
whose d i m e n s i o n is a l s o
finite
(theorem of Ado).
N e x t we come t o some t e r m i n o l o g y which is e x a c t l y p a r a l l e l t o t h a t of s e c t i o n 6 . 7 . We d e l i b e r a t e l y choose t h e same f o r m u l a t i o n in o r d e r t o e m p h a s i z e t h i s a s c l e a r l y a s p o s s i b l e . The a n a l o g u e of d e f i n i t i o n 6. 7. 2 is
D e f i n i t i o n 6.8.2:
(a) If U and V a r e s u b a l g e b r a s of some Lie a l g e b r a L, we d e n o t e by [U,V] t h e s u b a l g e b r a g e n e r a t e d by a l l c o m m u t a t o r s [ u , v ] w i t h u € U and ν e V. (b)
T h e lower central series o f
c a l c u l a b l e via L
ι
= L and L
L is t h e s e q u e n c e L , L , L , . . . , k+i
recursively
= [L, L ]. If t h e r e i s a n a t u r a l n u m b e r k k
w i t h L^ = 1, t h e a l g e b r a L is nilpotent. (c) Similarly,
setting L
rived series of L.
0
If
t h e n L i s solvable ( o r
= L and L
k+1
it e v e n t u a l l y soluble).
= [L
k
,L
k)
],
we o b t a i n t h e de-
l e a d s down t o t h e
zero
algebra,
242
6. Complex A n a l y s i s
Obviously,
a l l nilpotent L i e algebras are soluble,
but not vice versa.
We want to pursue the analogy between L i e groups and a l g e b r a s a bit f u r t h e r (cf.
definition
6.7.3):
Definition 6.8.3: ( a ) The s o l v a b l e ideals of a L i e algebra L generate the Jacobson radical of L.
We denote it by J a c ( L ) .
( b ) A L i e a l g e b r a L with J a c ( L ) = 0 is semisimple. ( c ) If L is noncommutative,
i.e.
Z ( L ) * L, and, f u r t h e r m o r e , does not c o n -
tain any ideals excepts f o r 0 and L i t s e l f ,
Hence,
we call L simple.
the Jacobson r a d i c a l is the l a r g e s t soluble ideal in L.
The theorems of Levi and Mal'tsev can also be c a r r i e d over to a l g e b r a s ,
but
only with s l i g h t m o d i f i c a t i o n s and the additional r e s t r i c t i o n char Κ = 0:
Theorem 6.8.4: L e t L be a L i e a l g e b r a over a f i e l d of c h a r a c t e r i s t i c 0. (a)
(Theorem of Levi): cobson r a d i c a l ,
L contains at least
i.e.
a subalgebra
S,
one Levi complement of such that
the
Ja-
L s p l i t s into a
direct
All Levi complements are c o n j u g a t e under
auto-
sum of the f o r m
L
( b ) (Theorem of Mal'tsev): morphisms of
=
Jac(L) ® S
L.
( c ) The f a c t o r a l g e b r a L / J a c ( L ) is semisimple.
We are s l o w l y approaching the central topic of shall be able to w r i t e down the main results, notation.
Lie theory,
but b e f o r e
we
we need a f e w more pieces of
6.8.
243
Lie Algebras
Definition 6.8.4: ( a ) T h e left translation w i t h a n e l e m e n t χ of t h e n - d i m e n s i o n a l L i e g r o u p G is t h e map
L : χ
G
—>
G
y
ι—>
L (y) = x - y X
T h e differential dL p r o v i d e s u s w i t h a n i s o m o r p h i s m b e t w e e n t h e t a n g e n t s p a c e s TG a n d TG . It a l l o w s u s t o c a r r y o v e r an a r b i t r a r y b a s i s y xy of t h e t a n g e n t s p a c e TG a t t h e u n i t e l e m e n t t o TG . i n ι χ (b) A v e c t o r
field
ν ε B(G)
left invariant if
is
the j u s t defined bases or,
is
to say it o t h e r w i s e ,
constant
relative
to
if f o r a l l x , y e G t h e
equation (dL ) v ( y ) X
=
v(L ( y ) ) X
=
v(x-y)
holds.
T h e s p a c e of l e f t i n v a r i a n t v e c t o r f i e l d s on G i s c l o s e d u n d e r f o r m i n g c o m mutators each
and thus
v(l) e TGj
a
a subalgebra well-defined
of
B(G).
vector
Is it
field
also
with
can be f o u n d . This makes t h e f o l l o w i n g c o n s t r u c t i o n
easy to
the
see t h a t
given
initial
for value
possible:
Definition 6.8.5: ( a ) L e t G b e a L i e g r o u p o v e r DC ε {R,C}. T h e associated Lie algebra of G i s the
space
ν e 8(G)
2(G) •
ν is l e f t
invariant
{
of a l l l e f t i n v a r i a n t v e c t o r f i e l d s w i t h t h e c o m m u t a t o r
as product.
It
h a s t h e g r o u n d f i e l d K, a n d i t s d i m e n s i o n e q u a l s t h a t of G. ( b ) A s s u m e L i s a Lie a l g e b r a o v e r Κ e . By A d o ' s t h e o r e m , (k k) b e d L in a s u i t a b l e m a t r i x a l g e b r a Κ ' is the
set
we c a n e m -
. T h e universal Lie group of L
244
6. Complex A n a l y s i s
S(L)
=
< expU)
I λ e L )
g e n e r a t e d by t h e images of a l l L - m a t r i c e s u n d e r t h e e x p o n e n t i a l map.
We a r e now p r e p a r e d t o f o r m u l a t e t h e main p r o p o s i t i o n of t h e w h o l e t h e o r y . The v a r i e t y of i t s a p p l i c a t i o n s is immense.
T h e o r e m 6.8.5:
Let L be a r e a l o r c o m p l e x Lie a l g e b r a . Every Lie g r o u p w i t h £(G) = L is a h o m o m o r p h i c image of i?(L). (This j u s t i f i e s t h e t e r m " u n i v e r s a l Lie g r o u p " ) .
T h e Lie correspondence
G
L = .2(G)
L
G = S(L)
t h u s e s t a b l i s h e d c o n n e c t s t h e o b j e c t s b e a r i n g t h e same n a m e s (e. g. ple, nilpotent,
solvable,
semisim-
. . . Lie g r o u p s and a l g e b r a s ) w i t h each o t h e r .
s u b g r o u p s of G a r e b i j e c t i v e l y a s s o c i a t e d w i t h t h e s u b a l g e b r a s of L, l a r l y f o r n o r m a l s u b g r o u p s and i d e a l s ,
The simi-
etc.
The a u t o m o r p h i s m g r o u p s of G and L a r e c a n o n i c a l l y i s o m o r p h i c , and t h e r a d icals are mutual
counterparts: Rad G
Jac L
relationship
mainly c o n s i s t s
in t h e
reduction
of t h e d i f f i c u l t i e s one e n c o u n t e r s in c l a s s i f i c a t i o n p r o b l e m s of Lie g r o u p s with specified properties.
The c o r r e s p o n d i n g
Lie a l g e b r a s
are
technically
much e a s i e r . The s o - c a l l e d classical Lie groups a r e t h e s p e c i a l l i n e a r , g o n a l o r s y m p l e c t i c ones over t h e r e a l or c o m p l e x n u m b e r s ,
unitary,
ortho-
i. e. t h o s e d e -
s c r i b e d by t h e s y m b o l s SL (C), SL (IR), Sp (C), Sp (R), SU (C) and SO (R). η η 2n 2n η η It is c u s t o m a r y t o d e n o t e t h e a s s o c i a t e d Lie a l g e b r a s by G e r m a n (or G o t h i c ) minuscules,
a s will be a p p a r e n t f r o m t a b l e 6 . 8 . 1 .
Concerning the
theoreti-
6.8.
cal
Lie
Algebras
impact
of
245
the classical
Lie
theory (part D of this book),
algebras
and t h e i r
connection
with
t h e r e a d e r may l i k e t o c o n s u l t t h e
Coxeter
monographs
by H u m p h r e y s [1975] o r J a c o b s o n [1962] o r t h e v o l u m i n o u s o r i g i n a l p a p e r s o f Killing
[1888-1890].
The
latter
source
also
contains
the
classification
the s e m i s i m p l e L i e g r o u p s o v e r C.
Tab.
L i e group
complex:
real:
6.8.1
Lie
algebra
SL ( C ) η
$r co
Sp
sp
2n
(C)
su
SL
sr
Sp
CR)
2n
CR)
SO ( R ) π
n2-l
η
SU C O η
η
dimension
sp
so
2n
η
π
CO
I
CR)
(R)
(2n-1)
Η
CN-1)
n2-l
CR)
2n
η
η
CO
η
- η 2
C2n-1)
Cn-1)
of
246
7. P r o j e c t i v e
Geometry
7. Projective Geometry 7.1. Affine and Projective Coordinates
The h i s t o r i c a l
development towards projective geometry
a t l e n g t h a l r e a d y in s e c t i o n 1 . 2 ; ther
has been
discussed
s o w e s h a l l go in m e d i a s r e s w i t h o u t
fur-
delay.
T h e affine plane o v e r t h e f i e l d Κ is t h e t w o - d i m e n s i o n a l
vector
space Κ ,
w h o s e e l e m e n t s a r e c a l l e d points. A (straight) line g i s t h e s e t of a l l l u t i o n s of a n a l g e b r a i c e q u a t i o n of d e g r e e 1, l i k e u q + u ^
+
so-
= 0.
T h e c o e f f i c i e n t s u 0 , u 1, u2 e Κ a r e t h e Plücker coordinates of g. β All t r i pκ l e s (u . u , u ) a r e p e r m i t t e d , e x c e p t t h o s e w i t h u = u = 0 ; t h e s e w o u l d n o t 0 1 2 1 2 describe
a nontrivial
straight
line
are
not
condition. unique,
On t h e since
other
hand,
multiplication
the of
nonzero f a c t o r λ € Κ merely r e p l a c e s the above polynomial l i n e by a n e q u i v a l e n t •> ^ with λ e Κ
will
one.
The t r a n s i t i o n
be denoted
Plücker
inhomogeneous o r affine coordinates Xj.Xg of
other,
or
m a k e u s e of
we say,
an i l l u s t r a t i n g
l i e s on g" o r "g p a s s e s t h r o u g h P",
a
of
e q u a t i o n of
coordinates
geometric
and so on.
terminology
with
like
"P
T h i s m a n n e r of s p e a k i n g
is
it. t h e r e is
a
are
system
u + u x + u x 0 1 1 2
2
u + u yJ + u y ο 1 1 2 '2
the normalization.
are
a p o i n t Ρ = ( χ , χ ) in
u n i q u e l y d e t e r m i n e d connecting line w h o s e P l ü c k e r c o o r d i n a t e s u o> u ^ u ^
The P l ü c k e r c o o r d i n a t e s uq,
the
(Xu , Au , Xu ) ο 1 2
( d i f f e r e n t ) points Ρ = ( x ^ x ) and Q = ( y ^ y ^ ·
f o u n d by s o l v i n g t h e l i n e a r
a
fixed
Ρ a n d g a r e incident
so w e l l e s t a b l i s h e d t h a t w e n e e d n o t c o m m e n t u p o n F o r e a c h p a i r of
by
rescaling.
t h e p l a n e o b e y t h e l i n e e q u a t i o n of g , each
u(
(u , u , u ) t o o' l ' 2
a s rescaling. H e n c e ,
only d e t e r m i n e d up t o an a r b i t r a r y If t h e
from
coordinates
the
u t
.u2
= 0 = 0
of t h e s t r a i g h t
They s p e c i f y the r a t i o
u : ua>
line g a r e independent
of
w h i c h d e s c r i b e s t h e slope
of g. Two lines g and h w i t h c o o r d i n a t e s u , u , u 0 1 2
a n d ν , ν , ν , r e s p e c t i v e l yJ , ο ι 2 ^
are
247
7 . 1 . A f f i n e and P r o j e c t i v e Coordinates
parallel to each other
(we express this by the symbolic notation g||h),
they have the same slope,
if
i. e.
u : u 1 2
=
ν
: ν 1 2
It makes a principal d i f f e r e n c e whether even
u : u : u 0 1 2 holds or not. are d i s j o i n t .
=
ν
0
: ν
: V 1 2
In the f i r s t case the lines are identical;
tion point in common.
Its coordinates
x
in the second they
they have e x a c t l y one intersec-
If g and h are not p a r a l l e l , j>x2
can
be obtained f r o m the linear
conditions
Whenever t w o s t r a i g h t also
mutually
u + u x + u x 0 1 1 2 2
= 0
v + v x + v x 0 1 1 2 2
= 0
lines h, k are p a r a l l e l
parallel:
to a third one ( g ) ,
g||h and g||k imply h||k.
Parallelism
they
is an
are
equiva-
lence r e l a t i o n . This
and 2
plane IR The
similar
observations
are
very
familiar
to
us;
the
real
affine
of
lines
brings
is the standard model of Euclidean geometry.
distinction
between
parallel
and
nonparallel
along numerous complications and inconveniences.
pairs
Mainly f o r this reason the
t r a n s i t i o n to projective geometry was c a r r i e d out ( f o r a more d e t a i l e d
ac-
count, see section 1.2). The points and lines of the projective plane IP2 Κ over the basic f i e l d DC are 3 the o n e - and two-dimensional subspaces of DC . Incidence of a point Ρ and a line g occurs if and only if Ρ c g. 3 T w o points g e n e r a t e a two-dimensional subspace of DC and thus possess e x T w o points g e n e r a t e a two-dime a c t l y one line connecting them. Since lines are of codimension
1, they
are maximal
3
in DC .
Two
different
by the homomorphism
theorem,
3
lines t h e r e f o r e span the whole of Κ , and, their plane,
intersection
is a subspace of
dimension 2+2-3 = 1: In the
projective
t w o lines a l w a y s have a (unique) point in common. The a f f i n e notion
of p a r a l l e l i s m thus loses its meaning in p r o j e c t i v e g e o m e t r y . A point Ρ is generated (as a subspace) by any nonzero v e c t o r contained in
248
7. P r o j e c t i v e Geometry
it,
f o r i n s t a n c e by χ = (χ , χ , χ ) e Ρ\. I t s c o o r d i n a t e s (which a r e f i x 0 1 2 ed only up t o r e s c a l i n g ) a r e c a l l e d t h e homogeneous o r projective coordinates of
P.
This
leads to the following interpretation
of
the
projective
plane: P2K
P 2 (K)
=
=
(K3\0)/(K\0)
3 As a s u b s p a c e of codimension 1 in Κ , a s t r a i g h t line g can be d e s c r i b e d by a linear equation,
s i m i l a r t o t h e s i t u a t i o n in a f f i n e g e o m e t r y : u x + u x + u x 0
0
1 1
2
2
= 0
It i s s e l f - e v i d e n t t h a t we may c h o o s e t h e s c a l i n g of t h e so d e f i n e d line coordinates a t w i l l ,
b u t t h i s is t h e only f r e e d o m we have.
the a f f i n e coordinates,
now a l l t r i p l e s a r e a d m i s s i b l e ,
In c o n t r a s t
with the sole
to ex-
c e p t i o n of (0, 0, 0). Beyond
this,
it
will
prove
useful
to
interpret
coordinates a s r o w v e c t o r s (= ( 1 , 3 ) - m a t r i c e s ) column v e c t o r s ,
i.e.
the
homogeneous
point
and t h e line c o o r d i n a t e s
as
(3, l ) - m a t r i c e s .
F u r t h e r m o r e , f r o m now on we s h a l l i d e n t i f y p o i n t s and l i n e s in
with the
a s s o c i a t e d coordinate vectors and t h u s speak of t h e "point x '= (XQ, x ^ . x ^ ) " and t h e "line u =
(Uq, u , u )".
The i n v a r i a n c e u n d e r
rescaling has
always
t o be t a k e n i n t o a c c o u n t . I n c i d e n c e of some p o i n t χ and some line u c o n s e q u e n t l y is e x p r e s s e d by t h e linear
condition x-u
=
x u 1 1
=
x u + x u + x u 0 0 1 1 2 2
=
The f i r s t v e r s i o n h a s t o be r e a d as a m a t r i x p r o d u c t , Einstein convention is a p p l i e d ,
which
0 in t h e second
(only in t h e p r e s e n t
e x t e n d e d t o p a i r s of i d e n t i c a l l o w e r i n d i c e s .
chapter
7)
In t h e p a r t i c u l a r c a s e of
the is the
p r o j e c t i v e p l a n e , t h e s u b s c r i p t s r u n f r o m 0 t h r o u g h 2, f o r g e n e r a l projective spaces of dimension η t h e y v a r y b e t w e e n 0 and n. This s i m p l e and e l e g a n t f o r m of t h e incidence r e l a t i o n i n d i c a t e s t h a t t h e p r o j e c t i v e c o o r d i n a t e s a r e highly s u p e r i o r t o t h e a f f i n e ones. T h i s a d v a n t a g e b e c o m e s even m o r e e v i d e n t if we t a k e t h e c o m p l e t e s y m m e t r y b e t w e e n 2
p o i n t s and l i n e s in IP Κ i n t o c o n s i d e r a t i o n , t h e duality principle of Poncelet:
which i s given e x p r e s s i o n t o by
7. 1. A f f i n e and P r o j e c t i v e
Coordinates
249
Theorem 7.1.1: 2 The duality of the p r o j e c t i v e plane IP (K) is the map which associates
with
e v e r yJ point line
(χ , χ , χ ) its dual line * ( x , x , x ) and, r e v e r s e l y , with each 0 1 2 0 1 2 (u , u , u ) i t s dual point (u ,u ,u ). In m a t r i x notation, this simply 0
1
2
0
1
2
corresponds to transposing. Any true
statement
concerning
points and s t r a i g h t
plane and incidence between them yields we dualize
a likewise
lines
in the
correct
projective
proposition
if
it.
Proof: Incidence of the point χ and the line u means that x-u = 0 holds. is ^ - ' x
The Poncelet p r i n c i p l e reduces the amount of most",
The dual
= 0 and thus equivalent.
because
certain
self-dual
interesting). 2 The a f f i n e plane Κ is naturally
labour by almost 50 % ( " a l -
configurations
and theorems
exist;
they
are e s p e c i a l l y
2 embedded in Ρ Κ.
In o r d e r
to v e r i f y
this
assertion,
we only have to map the a f f i n e point (x , x ) to ( l , x ,x ) e p \ . 1 2 1 2 This mapping is c l e a r l y i n j e c t i v e . Its image consists of all points of the p r o j e c t i v e plane whose 0 t h coordinate does not vanish. They are c a l l e d pro2
per or finite points;
the remaining χ 6 IP K, namely those with x q = 0,
are
the improper, infinite or ideal points, sometimes also r e f e r r e d to as the points at infinity. The a f f i n e s t r a i g h t ordinate
vectors
lines correspond with those p r o j e c t i v e lines whose
co-
,u ,u ) are l i n e a r l y independent of l ( l , 0 , 0 ) . We d e 2 note them as proper lines. In Ρ Κ, only one improper line e x i s t s . This line at infinity is
l(u
t(l,0,0).
The t r a n s i t i o n f r o m the a f f i n e to the p r o j e c t i v e point of
view can be d e -
scribed as f o l l o w s : If we are given the a f f i n e plane Κ , we call its points and lines p r o p e r ; the incidence r e l a t i o n s between these o b j e c t s are e x a c t l y 2 the same as in Κ . To each set of
parallel
with a l l of them,
lines we associate an ideal point which is incident
but with no other proper line.
Finally,
w e complete
system thus constructed by adding a single line at i n f i n i t y .
The l a t t e r
the has
250
7. P r o j e c t i v e Geometry
to be incident
with all
ideal,
but no f i n i t e points.
All
points
and
lines
together f o r m the (obviously unique) p r o j e c t i v e plane over the f i e l d K. The reversal
is even simpler.
We delete an arbitrary
incident with it f r o m the p r o j e c t i v e plane
line and all
points
What remains is the a f f i n e
plane. The points Ρ ,P , Ρ , . . . are said to be collinear if a line exists which 1 2 3 passes through all P ^ analogously we call the lines g , g 2> g ,... copunctal if there is a point lying on each of them. All points on a line f o r m a point row; the t o t a l i t y
of
all
straight
lines
through a f i x e d point is a line bundle. The connecting line of two points A, Β has the line coordinates [AB]
=
' ( Α Β -Α Β , Α Β - Α Β , Α Β - Α Β ) 1 2
2 1 2 0
0 2
0 1
1 0
and all its points are of the f o r m α A + β Β with (α, β) e K 2 \{(0,0)>.
The
intersection point of the line a, b is [ab]
=
(a b - a b ,a b - a b ,a b - a b ) 1 2
2
1 2 0
0 2
0 1
1 0
and the lines passing through it have a representation α a + β b with the same r e s t r i c t i o n on α and β as before. A triangle Δ = Δ(Α, Β, C; a, b, c)
consists
of
three
points
A,B,C
and
three
lines a, b, c, such that A is incident with b and c, Β with c and a, C with a and b, while there is no incidence between A and a, Β and b, or C and c. It is clear that a triangle is determined by either
its vertices A,B,C
or
its sides a, b, c. T h e r e f o r e we sometimes p r e f e r to w r i t e simply Δ = A ( A , B , C ) or Δ = A ( a , b , c ) .
The vertices or sides can be chosen with the single
striction that they must not be collinear or copunctal, The r o l e
of
the coordinate axes in a f f i n e geometry
re-
respectively. is in the
projective
plane taken over by the coordinate or fundamental triangle. It has the v e r tices
(1,0,0);(0,1,0);(0,0,1)
and
the
sides
t(l,
0, 0); 1 ( 0 , 1 , 0); l ( 0 , 0,1).
shall see in the next section how to construct point and line
We
coordinates
by using it. If we w r i t e down the coordinate vectors of three points A,B,C one underneath the other, we get the vertex matrix
7 . 1 . A f f i n e and P r o j e c t i v e
251
Coordinates
E(A, B,C)
In p e r f e c t a n a l o g y ,
w e d e f i n e f o r t h r e e l i n e s a , b , c by j u x t a p o s i t i o n of
the
c o o r d i n a t e v e c t o r s t h e side matrix
S(a,b,c)
T h e f r e e d o m t o r e s c a l e i s e x p r e s s e d by t h e f a c t t h a t i t i s a l l o w e d t o tiply E(A,B,C) f r o m the l e f t and S ( a , b , c ) vertible diagonal The c o l l i n e a r i t y
mul-
f r o m t h e r i g h t by a r b i t r a r y
in-
matrices. of
three
p o i n t s A , B , C in t h e p r o j e c t i v e
l e n t t o t h e v a n i s h i n g of d e t E ( A , B , C ) ;
plane is
t h e t h r e e l i n e s a, b, c a r e
equiva-
copunctal
if a n d o n l y if d e t S ( a , b , c ) is z e r o . W h e n e v e r E(A, B, C)· S ( a , b, c) is a r e g u l a r vertices valid.
a n d a , b, c t h e
s i d e s of
diagonal matrix,
a triangle;
the reverse
A, B, C f o r m statement
is
This gives us a simple means t o c o n s t r u c t t h e sides when t h e
ces a r e given or vice versa.
also verti-
We o n l y n e e d t o c h o o s e t h e m a t r i c e s E ( A , B , C)
and S ( a , b , c ) inverse to each o t h e r .
The c o o r d i n a t e s c a n t h e n i m m e d i a t e l y be
r e a d o f f . I n s t e a d of t h e r e c i p r o c a l m a t r i x , adjoint,
the
we could use, f o r i n s t a n c e ,
w h i c h d i f f e r s f r o m i t o n l y by a n u m e r i c a l f a c t o r
T h e f o l l o w i n g theorem of Desargues is of e m i n e n t
the
(determinant).
importance to
projective
geometry:
Theorem 7.1.2: L e t Δ = Δ(Α, Β, C; a, b, c) and Δ' = Δ( A ' , Β ' , C ' ; a ' , b ' , c ' ) h a v i n g no V e r t e x e s o r s i d e s in c o m m o n . below are
triangles
in
stated
equivalent:
( a ) The l i n e s [AA'], triangles are (b) The
be t w o
Then t h e t w o a s s e r t i o n s
intersection
a n d Δ' a r e
[BB'],
ICC'J c o n n e c t i n g c o r r e s p o n d i n g v e r t i c e s of
both
copunctal. points
collinear.
[aa'l,
[bb'l,
[cc']
of
associated
sides
in
Δ
252
7. P r o j e c t i v e Geometry
Proof: We a r e allowed to choose the normalization such t h a t each of the two v e r t e x m a t r i c e s Ε = E(A, Β, C) and E ' = E ( A ' , B ' , C ' ) S = S ( a , b , c ) or S ' = S ( a ' , b ' , c ' ) , E-S
is inverse to i t s side
matrix
respectively, =
E'-S'
=
1
which is possible by the preliminary remark. The p r o p o s i t i o n s
(a)
and
(b)
are
mutually
dual;
it
will
thus
suffice
to
prove the implication (a) =» (b). We f i r s t consider the special c a s e in which the common point Ρ of the t h r e e connecting lines [AA' ], [BB' ], [ C C ] coincides with one of the s i x For symmetry r e a s o n s , If t h i s condition
vertices.
we may assume Ρ = C.
holds,
lie on the
straight
line [PA] = [CA] = b. But A' is incident with the side b ' as well;
the points P,
thus we
have A' = [bb' ]. S i m i l a r l y , is t h e r e f o r e reason,
[A'B'J = c ' .
B' = [aa'].
A,
and A'
all
The line connecting [ a a ' l and [ b b ' ]
It evidently also passes through [ c c ' ] .
For
we may c o n c e n t r a t e on the generic case t h a t Ρ is d i f f e r e n t
this from
each o f the Vertexes A, Β, C; a, b . c . As we saw b e f o r e , t h i s implies the v a l i d ity of r e l a t i o n s of the form Ρ
=
a A + α' A'
=
β Β + β' Β'
=
y C + y ' C'
with a p p r o p r i a t e nonzero c o e f f i c i e n t s α, α ' ; β, β ' ; y, y ' in the ground f i e l d K. I f we w r i t e t h e s e t h r e e r e p r e s e n t a t i o n s as the rows of a ( 3 , 3 ) - m a t r i x ,
we
get E(P,P,P)
=
D-E + D ' - E '
with D = Diagta, ß, y) and D' = D i a g l a ' , β', y ' ) . The rank of E ( P , P , P ) is equal to 1, as is the rank o f DE + D ' E ' . If,
on the other hand,
one can find i n v e r t i b l e diagonal m a t r i c e s D, D'
with
r a n k ( D E + D ' E ' ) = 1, then all rows of DE + D ' E ' are proportional and d e t e r mine a common point on [AA'],
[ B B ' ] and [CC'].
P r o p o s i t i o n (a) thus is tantamount to the e x i s t e n c e of r e g u l a r diagonal m a t r i c e s D , D ' f o r which Rank(DE+D'E')
=
1
7 . 1 . A f f i n e and P r o j e c t i v e C o o r d i n a t e s
is t r u e .
A s s e r t i o n (b) is dual t o t h i s ,
253
and we c o n c l u d e w i t h o u t
difficulty
t h a t it e q u i v a l e n t t o Rank(SM+S'M')
=
1
f o r some d i a g o n a l m a t r i c e s Μ and M' w i t h n o n z e r o d e t e r m i n a n t s . S D"1 + S' D'" 1 This immediately
=
E - 1 D"1 + E'
1
D'" 1
implies the theorem,
=
Moreover,
E"1 D"1 (D'E'+DE) E ' " 1 D'" 1
since t h e m u l t i p l i c a t i o n
by
regular
m a t r i c e s d o e s not c h a n g e t h e r a n k .
We d e l i b e r a t e l y r e s t r i c t e d our i n v e s t i g a t i o n t o t h e c o n s t r u c t i o n of t h e f i n e and p r o j e c t i v e p l a n e s over that,
a given
basic
f i e l d K.
af-
The a d v a n t a g e
in c o r r e s p o n d e n c e w i t h t h e t o p i c of t h e p r e s e n t book,
is
we have f r o m
t h e o u t s e t c o o r d i n a t e s a t our d i s p o s a l . The c o m m u t a t i v i t y of K, h o w e v e r , not essential;
w i t h a pinch of s a l t ,
all our previous findings remain
is
valid
if Κ is a skew f i e l d . In t h i s m o r e g e n e r a l c a s e , we only have t o be c a r e f u l w i t h t h e o r d e r of t h e f a c t o r s in p r o d u c t s and, e q u a t i o n s which c o n t a i n d e t e r m i n a n t s
in p a r t i c u l a r ,
modify the
slightly.
As o u r a p p l i c a t i o n s w i l l mainly c o n c e n t r a t e on t h e f i e l d s Κ = 18 o r t h e m o r e i m p o r t a n t c a s e Κ = €, we r e f r a i n f r o m d i s c u s s i n g t h e i n d i c a t e d tion explicitly. tigated.
generaliza-
In g e o m e t r y , much more a b s t r a c t t y p e s of p l a n e s a r e i n v e s -
It seems a p p r o p r i a t e t o say a t l e a s t a f e w w o r d s a b o u t t h e m .
One s t a r t s o u t f r o m a d e f i n i t i o n which is not t i e d t o t h e e x i s t e n c e of g r o u n d (skew) f i e l d . I n s t e a d , void
sets
f
and
a
a projective plane is b u i l t up f r o m t w o n o n -
Ϊ. ("points"
and
"lines") b e t w e e n
which
a
relation
("incidence") is d e f i n e d , such t h a t t h e f o l l o w i n g f o u r a x i o m s h o l d s : (1) F o r each t w o p o i n t s P, Q e f t h e r e is a uniquely d e t e r m i n e d connecting line [PQ] e ΐ which is i n c i d e n t w i t h both of t h e m . (2) F o r each t w o l i n e s g, h e if t h e r e is a uniquely d e t e r m i n e d intersection point [gh] e Τ which is i n c i d e n t w i t h b o t h of t h e m . (3) The p l a n e c o n t a i n s a t l e a s t f o u r p o i n t s , no t h r e e of which a r e c o l linear. (4) The p l a n e c o n t a i n s a t l e a s t f o u r l i n e s , punctal.
no t h r e e of which a r e
co-
254
7. P r o j e c t i v e Geometry
The f o r m u l a t i o n chosen h e r e l e a d s d i r e c t l y back t o P o n c e l e t ' s · p r i n c i p l e of duality,
b e c a u s e t h e c o n d i t i o n s (1) and (2) a r e dual t o each o t h e r ,
and t h e
couple (3) and (4) h a s t h e same p r o p e r t y . With r e g a r d t o economy,
t h i s s y s t e m of a x i o m s is n o t o p t i m i z e d ,
however.
The f o u r p o s t u l a t e s a r e not i n d e p e n d e n t . We could, f o r e x a m p l e , o m i t one of the last
two,
since
(3) is a c o n s e q u e n c e
of
(1), (2), (4),
while dually
(4)
can be deduced f r o m (1), (2), (3). A f f i n e p l a n e s a r e c o n s t r u c t e d in t h e same way as we did b e f o r e , namely by d e c l a r i n g one line a s " i m p r o p e r " and d e l e t i n g it f r o m t h e p r o j e c t i v e p l a n e . It is t h e r e f o r e s u p e r f l u o u s t o r e p e a t o u r e a r l i e r
discussion.
As a l a s t r e m a r k t o t h i s t o p i c , we m e n t i o n w i t h o u t p r o o f t h a t a p r o j e c t i v e p l a n e is i s o m o r p h i c t o some IP (DO over a s u i t a b l e skew f i e l d Κ if and only if i s D e s a r g u i a n . T h i s of c o u r s e m e a n s t h a t it obeys t h e o r e m 7. 1. 2. We do n o t i n t e n d t o p u r s u e t h i s any f u r t h e r and r e f e r t h e i n t e r e s t e d to the l i t e r a t u r e ,
a s , e. g . ,
P i c k e r t [1955],
reader
7.2.
Projectivities
255
7.2. Projectivities
Many a g e o m e t r i c p r o b l e m c a n be s i m p l i f i e d c o n s i d e r a b l y by a p r o p e r of c o o r d i n a t e s . are admissible, Let us give an
choice
Which r e f e r e n c e f r a m e s a n d w h i c h t r a n s i t i o n s b e t w e e n d e p e n d s on t h e s t r u c t u r e one w a n t s t o k e e p
them
invariant.
example: 2
T h e E u c l i d e a n p l a n e IR i s e n d o w e d w i t h a n a t u r a l m e t r i c w h i c h a l l o w s u s t o d e f i n e t h e distance of t w o p o i n t s a n d angles b e t w e e n t w o s t r a i g h t l i n e s .
If
w e d e m a n d t h e s e m e a s u r e s t o be i n v a r i a n t d u r i n g t h e t r a n s f o r m a t i o n , w e a r e r e s t r i c t e d t o m a p p i n g s of t h e f o r m χ
ι—>
χ D + a
2 (2 2) translation vector a e IR and a m a t r i x D 6 IR ' which
with an a r b i t r a r y
i s s u b j e c t t o t h e s u p p l e m e n t a r y c o n d i t i o n D l D = 1 (orthogonality). Such t r a n s f o r m a t i o n s a r e called Euclidean; contains
the
set
of
t h e y f o r m a g r o u p g(IR ),
translations (D = 1) a s a n o r m a l
all
which
subgroup.
It
2
i s o m o r p h i c w i t h IR , w h i l e t h e f a c t o r g r o u p c o r r e s p o n d s t o t h e g r o u p of orthogonal (2,2)-matrices,
is
real
2
i.e.
GO(IR ):
g(IR2)
£
IRZ : G0(IR2) 2
(The
extension
splits;
a
complement
of
IR
is
provided
by
the
Euclidean
t r a n s f o r m a t i o n s with a = 0). An s eor be lt ya i nreedq uifi r e wet h ado c o n sael rt ve er nd ,a t i bv ue t i m t tnhoet iwnacni dt e nt hc ee lr ee nl agtt ihosn a nind IRa2n gilse s i nt ov a rbe i2 a n t u n d e r t h e t r a n s f o r m a t i o n (in o t h e r w o r d s : w e i n t e r p r e t IR a s t h e a f f i n e real plane). includes
T h e g r o u p of p e r m i t t e d c h a r t c h a n g e s t h e n b e c o m e s l a r g e r
and
all χ (2 2)
w i t h a r e g u l a r m a t r i x Μ e IR '
ι—»
, or,
χ Μ+ a
more precisely,
2
Μ e GL(IR ). 2
T h e l a s t c o n s i d e r a t i o n c a n be e x t e n d e d i m m e d i a t e l y t o a n y a f f i n e p l a n e Κ . 2
An affine transformation of Κ
2
i s s i m p l y a b i j e c t i o n of Κ
onto
itself
by
256
7. P r o j e c t i v e Geometry
2 which χ e Κ g o e s o v e r
to
χ
where a e Κ
2
2 and Μ e GL(K ).
*
=
χ Μ+ a
The set of these maps is the affine group
A{K2)
K 2 : GLCK2)
st
2 E s p e c i a l l y f o r the j u s t discussed case Κ = IR, we f i n d that ) is a m a n i 2 f o l d of dimension 6, w h i l e the Euclidean group £(IR ) is only 5 - d i m e n s i o n a l . The r e n u n c i a t i o n of
2
length and angular measurements in IR
is a l s o
combined
w i t h a c o n s i d e r a b l e gain of f r e e d o m in the choice of c o o r d i n a t e s . This even 2 becomes m o r e e v i d e n t when w e i n v e s t i g a t e the p r o j e c t i v e plane Ρ (Κ). As w e saw in s e c t i o n 7. 1, the point r o w termined
by
a
linear
condition;
incident w i t h
therefore
in
the
a fixed
present
p o l y n o m i a l t r a n s f o r m a t i o n s of d e g r e e 1 a r e admissible.
line is
situation,
Beyond t h i s ,
It i m p l i e s t h a t t w o
ar
same
coordinate
vectors,
describing
the
mapped t o v e c t o r s which a r e l i k e w i s e p r o p o r t i o n a l stricts
the p o s s i b i l i t i e s
for
point,
have
t o each o t h e r .
the point coordinates of
the
image
only
we also
have t o t a k e the s c a l i n g i n v a r i a n c e into account. dependent
de-
lineto
This to
be re-
linear
f unctions: χ
w h e r e Κ is an a r b i t r a r y w i t h det Κ * 0,
or,
ι—>
regular
*
χ
= χ Κ
(3,3)-matrix
more c o n c i s e l y ,
w i t h e n t r i e s in Κ:
Κ e GL (Κ). 3
( 3 3)
Κ 6 IK
'
2
On account of the s y m m e t r y between points and lines in IP Κ imposed by d u a l ity,
the
line coordinates a r e s u b j e c t to an analogous t r a n s f o r m a t i o n ,
for
example, u
Here L e GL 3 (K), cidence
condition *
equation χ
but this m a t r i x χ u = 0 for
*
ι—»
u
= L u
is not independent of
a point
χ and a line
K.
Instead,
the
u must have the
indual
* u
= 0 as a consequence and v i c e v e r s a . From χ u = 0 we t h e r e -
f o r e deduce 0
=
*
x u
*
= x K L u
7.2. P r o j e c t 1vi t i e s
This is only true if
257
Κ and L are,
up to a scalar f a c t o r ,
inverse to
each
other:
Κ L
=
λ·1
with λ e K* = DC\{0>. On the other hand, we may r e s c a l e the coordinates.
This permits us to
im-
pose the n o r m a l i z a t i o n λ = 1, which w i l l be done f r o m now on. We conclude:
Definition 7.2.1:
2
The projectivity of IP (K) given by the regular in Κ is the t r a n s f o r m a t i o n of
(3,3)-matrix
Κ with
the point and line coordinates
entries
according
to
the rules χ
ι—>
χ
*
= χ Κ
* u
ι—»
u
-1 = Κ
u
A l l p r o j e c t i v i t i e s f o r m a group isomorphic with
GL (K)/Z(GL (HO) 3
=
PGL CK)
3
3
the projective group on Ρ 2DC. The last a s s e r t i o n r e f l e c t s the f a c t that the scalar matrices, ers, map a l l v e c t o r s to multiples and consequently 2 the sets of points or lines in IP OC.
and no o t h -
induce the identity
The group PGL3(D0 of pro j e c t i v i t i e s is 8-dimensional and thus even
on
strictly
l a r g e r than the a f f i n e group. Aside f r o m the i n t e r p r e t a t i o n chart t r a n s i t i o n s ,
of
a f f i n e and p r o j e c t i v e t r a n s f o r m a t i o n s
as
which we p r e f e r here, we could equally w e l l have viewed
the maps in question as automorphisms of IK
and
If we want to do so, however, we have to observe that not every t r a n s f o r m a tion which respects the incidence r e l a t i o n in the plane is contained group
*i(K ) or
PGL (HO.
For
example,
for
certain
ground
fields
in the
(like
C:
258
7. P r o j e c t i v e Geometry
complex conjugation) there might exist additional symmetries produced by nontrivial field automorphisms of DC. Such a τ € Aut OC simply has to be a p plied to all coordinates of points and lines. If we are only interested in the incidence structure,
but not in the c a t e -
gorization of the geometric o b j e c t s as points or lines, we can also enlarge the automorphism group by the duality operation,
etc.
The term "automor-
phism of an a f f i n e or projective plane" f o r this reason requires a precise definition. We are not going to discuss these problems to any extent and leave the subj e c t as it stands. In what follows, the only automorphisms we need are the ( a f f i n i t i e s and) p r o j e c t i v i t i e s .
2
Having fixed a r e f e r e n c e frame, in the a f f i n e plane Κ the point Ο = ( 0 , 0 ) , called the origin of the coordinates, is distinguished, together with two s t r a i g h t lines a j and a.^, the axes, which intersect in 0. The Plilcker c o o r dinates of a and a are ( 0 , 0 , 1 ) and ( 0 , 1 , 0 ) : the associated point rows 1 2 contain all ( x ^ O ) and (0, x^l,
respectively.
The configuration j u s t described is c h a r a c t e r i s t i c f o r the r e f e r e n c e system in use and is known as its coordinate cross. The transition to a new set of coordinates is simply e f f e2c t e d by the definition of another pair of axes. In the p r o j e c t i v e plane Ρ DC, this construction is replaced by the fundamental triangle introduced in section 7.1, points
(1, 0, 0); ( 0 , 1 , 0); (0, 0 , 1 )
of
whose vertices are situated at the
our coordinate
system,
while the
sides
are the lines t ( l , 0. 0); ι ( 0 , 1 , 0); *·(0, 0 , 1 ) .
2 It is obtained from the basic cross in Κ by adding the improper line as the third side,
together with the three points of intersection.
the situation in the a f f i n e plane,
Similar to
a projectivity is tantamount to
select-
ing an ( a r b i t r a r y ) triangle as fundamental. The p r o j e c t i v e group t h e r e f o r e a c t s s t r i c t l y transitive on the set of all 2 2 t r i a n g l e s in IP (DC), j u s t as the affine group does on the crosses in DC . This implies the relation i4(DC2)
χ
= χ Μ + a
2 where Μ e GL (DC) and a 6 Κ amounts to the projectivity with the matrix 1 -
0 0
in point coordinates;
a
a 1 2 Μ Μ 11 12 Μ Μ 21 22
the corresponding (inverse) matrix for the line coor-
dinates is 1
-b
0
Ν 11 Ν 21
0
-b
1
2 Ν 12 Ν 22
-1
and
Ν = Μ
That all
matrices of the form 1 * * ' 0 * « 0 » * in GL3
τ
τ
Α
β+δτ = α,+γτ
The a r i t h m e t i c of the point at i n f i n i t y (oo) hasto be obtained f r o m the proj e c t i v e version:
If the denominator in the last f o r m u l a vanishes,
r a t o r is a l w a y s nonzero;
in this case we have τ
the nume-
= oo. The image of the ide-
al point is in the same way found f r o m
(0, 1) · A
=
(y,6)
as α
CO
Since
it
is permitted
to
rescale
δ — 7
=
the homogeneous
coordinates
of
the
ele-
ments of p ' k ,
t w o m a t r i c e s which d i f f e r only by a nonzero constant f a c t o r 2 λ e K, give r i s e to the same t r a n s f o r m a t i o n of the subspaces of Κ , i . e . , the same permutation of the points in P I . The set of all projectivities of the line IP1(K) t h e r e f o r e is a f a c t o r of GL 2 (K),
group
namely GL (DC) / K* 2
=
GL (Κ) / Z(GL ( Κ ) ) 2
=
2
PGL (DC) 2
the projective group. This p a r a l l e l s the t h e o r y of the f o r e g o i n g section;
the most important
pro-
p e r t y of PGL (K) which has no equivalent in higher dimensions is
Theorem 7.3.1: The
group
PGL 2 (K)
of
projectivities
is
strictly
3-fold
transitive
on
the
line p V k ) .
Proof: We f i r s t
show that every t r i p l e
a,b,c
of
pairwise
different
is c o n j u g a t e under the action of P G L ^ K ) to a p a r t i c u l a r one,
points
in p'lK
say oo, 0, 1 (in
this o r d e r ) . If a 6 K, we apply the t r a n s f o r m a t i o n τ ty.
We are
therefore
a l l o w e d to
ι-* ( τ - a )
assume without
1
which moves a to i n f i n i restriction
of
generality
262
that τ
7. P r o j e c t i v e Geometry
a =
In
the
next
step,
we
reduce
the
ι-> τ - b t o t h e s p e c i a l c a s e a = oo, b = 0.
field element p r o o f of t h e
other
than
zero,
and
task
by
the
translation
The t h i r d p o i n t , c, t h e n i s a -1 by c completes the
a multiplication
3-transitivity.
It r e m a i n s t o s h o w t h a t e v e r y
invertible α
β
7
δ
in GL 2 (K) w h i c h h a s oo, 0 , 1 a s f i x e d p o i n t s i s a s c a l a r m a t r i x and c o r r e s p o n d s t o t h e i d e n t i t y in t h e f a c t o r g r o u p P G L ^ K ) .
(0,1) A
implying γ = 0, while 0 = 0 namely Diagta, δ),
(r,3)
=
therefore
The i n v a r i a n c e
of
(0,*)
l e a d s t o β = 0. T h u s ,
A is a diagonal
a n d f r o m 1* = 1 we d e d u c e α = δ,
matrix,
a s s t a t e d in t h e
theo-
rem.
Doubly
transitive
groups
are
not
very
frequent;
threefold
transitivity
an e x t r e m e l y r a r e o c c u r r e n c e .
More i n f o r m a t i o n c o n c e r n i n g m u l t i p l y
tive permutation
be p r o v i d e d
groups
will
in s e c t i o n
20.6
is
transiHathieu
on t h e
groups. T h e o r e m 7 . 3 . 1 h a s q u i t e a n u m b e r of c o n s e q u e n c e s w h i c h a r e of f u n d a m e n t a l importance. prescribed
While w e c a n a l w a y s f i n d a p r o j e c t i v i t y of P1K w i t h images
of
three
points,
this
is
impossible
for
arbitrarily
four
or
more
points. In o r d e r
to give an i l l u s t r a t i o n ,
a , b, c, d 6 Ku{to>.
The
first
we s e l e c t f o u r m u t u a l l y d i f f e r e n t
three
of
them
fix
a
unique
points
projectivity
σ e PGL 2 (K) w h i c h t r a n s p o r t s t h e m in t u r n t o oo, 0, 1. T h e n t h e i m a g e d0^ of d is
determined
as
well.
It
only
depends
on t h e
write
CR(a, b, c , d)
In p a r t i c u l a r ,
=
d0"
given
points
a, b , c , d .
We
7.3. The Cross Ratio
263
CR(co, 0,1, d)
=
d
f o r all d € K\{0,1}. "CR" is an abbreviation of "cross ratio", a term whose meaning w i l l become apparent soon. By its definition,
CR(a,b, c , d ) is an invariant,
i.e.
it does
not change if we subject the points a , b , c , d to any p r o j e c t i v e t r a n s f o r m a tion. Up to now, the cross ratio is only defined f o r the case that no two of the arguments coincide.
Its value is also a point on the p r o j e c t i v e line P 1 (K).
The direct calculation of CR(a, b,c, d) does not present any d i f f i c u l t y .
Let
us assume f o r the moment that a , b , c , d are all contained in K. The p r o j e c tive transformation which brings the quadruple (a,b, c , d ) in the correct o r der to (co, 0, 1, * ) is represented,
among others, by the matrix a(c-b)
b(c-a)
Μ = b-c
a-c
and turns out to be Μ_
(c-a)b+(a-c)T (c-b)a+tb-c)r
as can be seen by inserting τ = a, b.c. We thus obtain the cross ratio if we put τ = d, CR(a,b,c,d)
=
(c-a)b+(a-c)d (c-b)a+(b-c)d
(a-c)(b-d) (a-d)(b-c)
In f o r m e r times, one p r e f e r r e d to write the last expression as CR(a, b, c, d) This alternative
=
-c b-c a-d ' b-d
has its origin in the theory of proportions and
explains
the name "cross ratio". The formula f o r C R ( a , b , c , d) requires a f e w supplementary conventions in the case that one of the points lies at infinity. stead of τ
ι-» τ
K
the transformation τ-b c-b
If,
for
instance,
a = co,
in-
264
7. P r o j e c t i v e Geometry
s u f f i c e s , f r o m which we deduce CR(n>,b,c, d) = ( b - d ) / ( b - c ) . The o t h e r s p e c i a lizations are similar. Furthermore,
it
is a b s o l u t e l y
evident
that
the cross
ratio
t h e only projective invariant of f o u r p o i n t s in p'tK).
is
essentially
This is t o say
that
a l l o t h e r i n v a r i a n t s can be e x p r e s s e d in t e r m s of CR. Each of t h e a r g u m e n t s a , b , c , d can be r e g a i n e d f r o m t h e o t h e r t h r e e and C R ( a , b , c , d). We s u m m a r i z e t h e p r o p e r t i e s f o u n d up t o now:
Theorem 7.3.2:
(a) The c r o s s r a t i o of a, b, c, d e IP1 CK) is c a l c u l a t e d
w i t h t h e h e l p of
the
It is d i r e c t l y a p p l i c a b l e if none of t h e a r g u m e n t s is equal t o t h e
in-
f ormula ,
f i n i t e point.
Otherwise,
(a-c)(b-d) (a-d)(b-c)
t h e f o l l o w i n g r e l a t i o n s hold: CR(a>, b, c, d)
b-d b-c
CR(a, oo, c, d)
a-c a-d
CR(a, b, oo, d)
b-d a-d
CR(a, b, c, to)
a-c tPi
They a r e f o r m a l l y d e r i v e d f r o m t h e g e n e r i c ones by o m i t t i n g t h e
two
f a c t o r s c o n t a i n i n g "co" o r f r o m t h e s u b s t i t u t i o n r u l e oo-x oo-y
_ y-oc
if x , y e Κ and t h e e q u a t i o n
a s long a s χ * 0. The same a g r e e m e n t a l s o a l l o w s t o d e f i n e t h e r a t i o w h e n e v e r only t h r e e of t h e f o u r p o i n t s a r e d i s t i n c t , (b) The c r o s s r a t i o is i n v a r i a n t u n d e r p r o j e c t i v i t i e s :
cross
7 . 3 . The C r o s s R a t i o
265
CRta", b " , c11, d " )
=
CR(a,b,c,d)
f o r all π ε PGL (DO. 2
( c ) Two o r d e r e d q u a d r u p l e s ( a , b , c , d ) and ( a ' , b ' , c ' , d ' ) f r o m Ρ (Κ) a r e e q u i valent r e l a t i v e to the group PGL^OO if and only if CR(a,b,c,d)
=
CR(a', b ' , c ' , d ' )
holds.
We now r e t u r n to the p r o j e c t i v e plane.
Every s t r a i g h t
2 line g in IP (DC) is
c a n o n i c a l l y isomorphic to p'flK), a s can be e a s i l y d e m o n s t r a t e d d i r e c t l y in homogeneous
coordinates
(χ , χ , χ ) of
the
points
χ incident
with g.
At
l e a s t one of the t h r e e components is redundant when we r e s t r i c t to g; if we omit it, we a r e l e f t with p r o j e c t i v e c o o r d i n a t e s on g = P1K. The c r o s s r a t i o n of f o u r points on g is then defined a s well.
It does not
depend on the r e f e r e n c e f r a m e , b e c a u s e a c h a r t change is e f f e c t e d by a l i n e a r c o o r d i n a t e t r a n s f o r m a t i o n which induces a p r o j e c t i v i t y on g. The i n v a r iance g r o u p of g in PGL^tDC) c o n s i s t s of the maps which have (the line c o o r dinates of) g as fixed vectors.
They all t o g e t h e r f o r m a subgroup
isomor-
phic with PGL 2(DC). At the s a m e time we have c o n s t r u c t e d the c r o s s r a t i o n of f o u r p o i n t s in the plane.
Dually,
collinear
we a l s o have the p o s s i b i l i t y to i n t r o d u c e the
c r o s s r a t i o of f o u r copunctal lines g , g , g , g . This is done by choosing a line h which is not contained in the same bundle a s the given l i n e s .
The
c r o s s r a t i o of the g^ then is understood a s the a n a l o g o u s quantity of
the
i n t e r s e c t i o n points with h, CR(g1,g2,g3,gt) CR(a,b, c,d)
depends
on
the
=
CR([hgiJ,[hg2],[hg3],[hg4])
order
of
its
arguments.
Nevertheless, 24
4! = 24 p e r m u t a t i o n s of the 4 v a r i a b l e s yield only at most — = 6 4
the
different
v a l u e s since the c r o s s r a t i o does not change if we apply an element of the f o u r group
266
7. Projective Geometry
V4
=
I 1, ( a b M c d ) , ( a c ) ( b d ) , ( a d ) ( b c )
J
as can be seen by d i r e c t evaluation of the r e l e v a n t expression.
In g e n e r a l ,
the 6 remaining quotients are mutually d i f f e r e n t . We compile them here e x plicitly
f o r the convenience
of
the r e a d e r .
The cross r a t i o
of
a, b, c, d is
denoted by λ. Then we f i n d
CR(a,b, c , d )
=
CR(b, a, d, c )
=
CR(c, d, a, b )
=
CR(d,c,, b, a )
CR(a, b, d, c )
=
CR(b, a, c , d )
=
CR(c, d, b , a )
=
CR(d, c,, a, b )
CR(a, c, b, d )
=
CR(b,d, a, c )
=
CR(c,a,d,b)
=
CR(d,b,, c, a )
CR(a,c, d , b )
=
CR(b, d, c, a )
=
CR(c,a,b,d)
=
CR(d,b,, a, c )
CR(a, d, b, c )
=
CR(b,c, a, d )
=
CR(c,b,d,a)
=
CR(d, a,, c, b )
CR(a,d, c, b )
=
CRCb.c, d, a )
=
CR(c, b, a, d )
=
CR(d, a,, b , c )
Though,
the cross r a t i o
exceptional
λ·
1_λ·
b
are usually
configurations
projective ( 1 ) If
1 λ =
1-λ 1 1-λ λ-1 λ λ
Τ=λ· V " ' λ^Τ
different,
which
for
deserve
our
}
certain
special
attention
λ we
because
obtain
of
their
invariance:
λ e {oo, 0 , 1 } ,
each of
these three
we speak of a singular quadruple.
values occurs
(2) Similarly, again
if
λ attains one of all
of
them
are
twice.
In this
case
The condition is equivalent with the
coincidence of t w o of the f o u r points a, b , c , d ;
ther,
λ
as stated above, the six values
{
of
=
hence the name.
the numbers - 1 , 2 , i found
twice.
Then
which the
belong
set
toge-
{ a , b , c , d)
is
c a l l e d harmonic. Whenever the points are ordered such that even C R ( a , b , c , d) = -1, we say that the quadruple ( a , b , c , d ) is harmonic. the fourth harmonic point to (3) Finally, ratic
If this holds,
d is known as
a,b.c.
it is possible that λ equals one of the solutions of
equation λ2 - λ + 1
=
0
the quad-
7 . 3 . The C r o s s R a t i o
267
Then a, b , c , d a r e equianharmonic,
2 and b o t h z e r o e s of λ - λ + l a r e
s e n t e d t h r i c e by p e r m u t a t i o n s of t h e a r g u m e n t s . condition
c a n n o t be f u l f i l l e d in every f i e l d .
However,
repre-
the relevant
If t h e c h a r a c t e r i s t i c
of
Κ is e q u a l t o 2, i t can never be t r u e ; w h i l e f o r c h a r Κ = 3, we would g e t λ = -1, i. e. t h e h a r m o n i c c a s e .
In a l l c h a r a c t e r i s t i c s o t h e r t h a n 2
and 3, we have λ * -1, and b e c a u s e of
(λ+l)
(λ 2 -λ+1)
( - λ ) m u s t be a p r i m i t i v e t h i r d r o o t of u n i t y .
These (and hence e q u i a n -
h a r m o n i c s e t s ) e x i s t w h e n e v e r - 3 is a s q u a r e in K, f o r t h e s o l u t i o n of the quadratic equation f o r λ reads
λ
2
2
Of t h e s e t h r e e s p e c i a l c a s e s , t h e h a r m o n i c is by f a r t h e m o s t i m p o r t a n t in projective geometry.
7. P r o j e c t i v e Geometry
268
7.4. B6zout's Theorem We have learnt to characterize
the straight
lines in the p r o j e c t i v e
plane
IP (IK) over the ground f i e l d Κ by linear equations in point coordinates. al to this representation
is the description
of
the points in line
Du-
coordi-
nates. Both are particular examples of algebraic curves, which are defined as the sets of zeroes of homogeneous polynomials in three variables (χ , χ , χ ).
In
doing so, it is immaterial whether we interpret the X ( as point or line coordinates. A f u r t h e r generalization suggests i t s e l f .
The investigation can be extended
to higher dimensional spaces, f o r instance to IPn(K). We only have to consider vectors over Κ with (n+1) components (χ
χ ), η
0
and algebraic
equa-
tions now define submanifolds of codimension 1. The r e s t r i c t i o n
to
homogeneous polynomials
is essential,
since
only
then
the invariance under rescaling is retained which has proved to be so useful in the f o r e g o i n g sections. On the other hand, the choice of the underlying f i e l d Κ is much less important;
to avoid unwanted complications,
however, we shall always assume it
to be algebraically closed. As a particular example, the reader is urged to keep in mind the set Κ = C of the complex numbers (and r e f e r to Gauß' fundamental theorem of algebra; theorem 6 . 5 . 5 ) .
Definition 7.4.1: ( a ) An n-ary form over a f i e l d Κ is a homogeneous polynomial in η variables taken f r o m K. I f ,
in particular,
ternary, quaternary forms,
η = 2,3,4
one speaks of
binary,
etc.
( b ) All zeroes of a f o r m F: Kn —> K, i . e .
the vectors χ = (χ
ι
χ ) e Kn η
with F(x)
=
F(x
ι
χ ) π
=
0
are the points of the hypersurface defined by F; or 3 this is also called a point,
in the cases η = 1, 2
(algebraic) curve or surface,
respec-
7.4.
E t e z o u t ' s Theorem
269
tively. ( c ) If χ * 0 i s a n o n t r i v i a l z e r o of F, on a c c o u n t of t h e h o m o g e n e i t y of F a l l m u l t i p l e s of χ a r e l i k e w i s e s o l u t i o n s of t h e
FUx)
=
F(Ax
λχ ) π
1
equation
=
0
=
κ
The s e t
DC-x
is
called
=
j λ χ
λ € κ |
ray (of solutions) of
the
F.
We u s u a l l y
interpret
it
as
a
p o i n t in s o m e p r o j e c t i v e s p a c e . T h e n i t s c o m p o n e n t s x t a r e t h e h o m o g e n e o u s c o o r d i n a t e s of x. (d) We do n o t m a k e a d i f f e r e n c e b e t w e e n v e c t o r s of t h e s a m e r a y .
Therefore,
t h e number of s o l u t i o n s of an a r b i t r a r y s y s t e m of f i n i t e l y many F^: Kn —> Κ is a l w a y s t h e c a r d i n a l i t y of t h e s e t of t h e
F
forms
simulta-
n e o u s s o l u t i o n r a y s of t h e F . In o t h e r w o r d s ,
i t i s t h e n u m b e r of intersection points of t h e
hyper-
s u r f a c e s belonging to t h e given f o r m s .
Any i n f o r m a t i o n on t h e c o m m o n p o i n t s of h y p e r s u r f a c e s i s of g r e a t to geometry.
Under very general
yields their number.
conditions,
the theorem
of
interest
B^zout
We s h a l l f o r m u l a t e it in a v e r s i o n s u i t i n g o u r
e s a t t h e e n d of t h i s
[1769]
purpos-
section.
But b e f o r e , we p r e s e n t s o m e h e u r i s t i c a r g u m e n t s w h i c h c a n b e t u r n e d i n t o a n e x a c t m a t h e m a t i c a l t h e o r y w i t h t h e m e a n s of m o d e r n algebraic geometry ( s e e Hartshorne
[1977]).
Generically,
t h e z e r o s e t of a f o r m o v e r a n a l g e b r a i c a l l y c l o s e d f i e l d
has
codimension
1; w e t h e r e f o r e e x p e c t
the
i n t e r s e c t i o n of t w o , t h r e e , ...
under
"normal" circumstances
h y p e r s u r f a c e s s h o u l d h a v e c o d i m e n s i o n 2,
If s o m e of t h e f o r m s p o s s e s s common f a c t o r s of p o s i t i v e d e g r e e ,
conclusion
i s of
b o u n d of t h e In
...
that
the
nates,
course
invalid;
it
nonetheless
provides
us with
a
3, this
lower
dimension.
n-dimensional
projective
space
P n (K),
endowed
with
(n+1)
coordi-
we t h u s need a t l e a s t η a l g e b r a i c a l l y i n d e p e n d e n t f o r m s t o g e t
l a t e d p o i n t s = r a y s of
solutions.
iso-
270
7. P r o j e c t i v e Geometry
To get acquainted with Bezout's theorem,
we f i r s t consider several
simpler
special cases which are important f o r themselves.
Theorem 7.4.1: Every f o r m in 2 variables over
an algebraically
closed f i e l d splits
com-
pletely into linear f a c t o r s .
Proof: We w r i t e down the f o r m in question - let it be of degree η - e x p l i c i t l y , F(x ,χ )
=
0 1
f
xn + f
0 0
1
χ"" 1 χ 0
1
+ ...
+ f
n-l
χ
x n_1 + f
0 1
x"
n l
If f
vanishes, then χ is a f a c t o r of F; otherwise a f t e r having t r a n s f e r n ο red to the inhomogeneous coordinate χ ζ
=
1 X 0
we obtain a polynomial of degree n, namely
f(z)
=
F(l.z)
=
f
0
+f
1
z + . . . + f
n-l
z"" 1 + f
η
zn
which, on account of the assumption that the underlying f i e l d Κ is a l g e b r a ically (ζ χ0 -
closed,
has a zero ζ ,
say.
Then F ( l , ζ ) = 0,
and F is divisible
by
V ·
Thus, we may separate a linear f a c t o r f r o m F. By induction with respect to the degree, the theorem f o l l o w s .
Already f o r ternary forms, the above statement is f a l s e , as simple examples show. Up to now we only considered one f o r m at a time; now we ask f o r s i multaneous solutions of two binary forms.
Theorem 7.4.2: Let
7 . 4 . B 6 z o u t ' s Theorem
A(x , x ) ο ι
271
a
0
x
k k-1 + a χ χ 0 1 0 1
+ a
k-1
k-1 . k x x + a χ 0 1 k 1
and TW \ B(x , x ) 0 1
1_ no— 1 . IB—1 . m b x + b x x + . . . + b x x + b x 0 0 1 0 1 , m-101 m l m
=
be t w o b i n a r y f o r m s of d e g r e e s k and m o v e r t h e a l g e b r a i c a l l y c l o s e d f i e l d K. Then t h e f o l l o w i n g a s s e r t i o n s a r e e q u i v a l e n t : ( a ) The s y s t e m A(x) = 0; B(x) = 0 h a s a n o n t r i v i a l
solution;
(b) Both f o r m s have a n o n c o n s t a n t common d i v i s o r ; ( c ) T h e r e a r e f o r m s F and Η of d e g r e e s m-1 and k-1, obeying t h e i d e n t i t y A(x) F(x) + B(x) H(x)
=
0
f o r a l l χ e K2; (d)
T h e resultant o f
A and
B,
that
i s Sylvester's determinant
constructed
f r o m t h e c o e f f i c i e n t s of t h e f o r m s a c c o r d i n g t o t h e s c h e m e
Res(A,B)
=
det
(m r o w s w i t h a, k r o w s w i t h b), is i d e n t i c a l l y z e r o . The r e s u l t a n t is homogeneous of d e g r e e m in t h e a ' s and of d e g r e e k in the b's.
272
7. P r o j e c t i v e Geometry
Proof: ( a ) => ( b ) :
If
( ξ , ξ ) * ( 0 , 0 ) is a common z e r o of both f o r m s A and B, then 0
1
A ( x , x ) and B(x , x ) are multiples of ( ξ χ - ξ χ ). r ο ι ο ι η ο ^o ι ( b ) => ( c ) :
Let Κ be a f a c t o r of A and Β with d e g ( K ) > 0. By theorem 7.4.1,
we can f i n d a linear f a c t o r L ( c ) =* ( a ) :
We decompose,
again with the help of theorem 7.4.1,
A , B , F , H into linear f a c t o r s . which
does
either
| K. Now set F = 5- and Η = -
divide
Η not
at
all
or
with
lower
multiplicity
(consider the degrees! ). We may thus, e . g . , assume that Then L "
| BH and L
(c) β (d):
=
forms
L*1
than
| A, but i f
A
f H.
| B.
If the f o r m s F and Η are,
F ( x ,χ ) 0 1
all
Among those of A, there must be at least one
f
0
explicitly,
x™"1 + f x m " 2 χ + . . . 0 1 0 1
+ f
m-2
χ
χ*" 2 + f x"" 1 1 m-1 1
0
and
H(x , x ) = h x 0 1 0
k
+ h x k 1 0
l
0
x + . . . + h χ xk 2 + h xkl 1 k-2 0 1 k-1 1
2
a comparison of the c o e f f i c i e n t s of x k+m ral numbers i = 0 , 1 , 2 , . . .
a
0
f
1 1
x j in AF +BH = 0 f o r all
natu-
in turn yields the conditions
+ b
0
0
h
= 0
0
a f + a f 1 0 0 1
+ b
+ b h 0 0 1
= 0
a f + a f + a f 2 0 1 1 0 2
+ b h + b h + b h 2 0 1 1 0 2
= 0
a f + a f + a f + k m-3 k-1 m-2 k-2 m-1 a
k
f
m-2
+ a
k-1 a
k
f f
m-1 m-1
h 1
b h +b h +b h m k-3 m-1 k-2 m-2 k-1
= 0
b h + m k-2
= 0
+
b
+
h m-1 b
k-1
h m k-1
=
0
which a l l t o g e t h e r are a system of m+n linear equations in just as many unknowns,
namelyJ
;h m-1 0
h k-1
algebra,
this has a nontrivial
solution
zero.
f
0
f
But the c o e f f i c i e n t m a t r i x
in the d e f i n i t i o n of the resultant.
According if
to
and only
is the transpose of
the if
the
rules
of
linear
determinant
the m a t r i x
is
occurring
7.4.
B 6 z o u t ' s Theorem
Next,
we
multiple
give
273
a necessary
and
sufficient
criterion
for
the
existence
of
zeros:
Theorem 7 . 4 . 3 : The b i n a r y f o r m F = F ( x q , X j ) o f d e g r e e η o v e r an a l g e b r a i c a l l y c l o s e d f i e l d possesses
a solution
of
multiplicity
at
least
2 if
its
discri-
2 n - 2 in the
coeffi-
and only
if
minant,
Dsc(F)
vanishes.
The d i s c r i m i n a n t
c i e n t s of
F.
itself
=
Res(9 F,a F ) ο ι
is a f o r m
of
degree
Proof: If of As
F has the double or ( sξ χ - ξ χ i ο ^o ι an e l e m e n t a r y
shows,
higher
order
zero
ξ^) * ( 0 , 0 ) ,
it
is a
multiple
differential
calculus
)2.
in t h i s
d i v i s i b l e by
application
case
both
of
partial K
the product derivatives
of
are
(^χ^ξ^^.
On the o t h e r hand,
on account of the
identity
χ 9 F ( x , x ) + χ 3 F ( x ,x ) 0 0 0 1 1 1 0 1 which i m m e d i a t e l y
=
η F(x ,x ) 0 1
f o l l o w s f r o m E u l e r ' s homogeneity theorem,
s o l u t i o n of 9 q F ( 5 o , ζ ^ ) =
F ( x ,χ ) ο ι
ξ ^ ) = 0 is also a z e r o of F.
=
every
common
We w r i t e
( ξ χ - ξ χ ) G(x ,χ ) η ο ^ο ι ο ι
w i t h some f o r m G of d e g r e e n-1 and f i n d , stitution of
rule
d F ( x , x ) and d F ( x , x ) 0 0 1 1 0 1
again by d i f f e r e n t i a t i o n and sub-
ξ,
0 = 9/CVV = 0 = V^o-V =
+
G
depending on the parameter κ. Using a slightly
different
notation than in theorem 7. 4. 2, we develop A and Β into f i n i t e power series of the variables χ
ο
and χ : ι
A(x , χ ) ο ι
=
V
F (χ ,χ , κ·χ ) ι ο ι ο
L i =0 g„
a
r ϊ 1 i(K)xo
X
i
Σ,V9 b (κ) χ 1 0
χ 1
where we expressly emphasize the dependence of the c o e f f i c i e n t s on κ. cording to the relation x g = κ·Χ 0> degrees
s i.
i s m holds,
The
(i,j)-entry
of
the a j and b Sylvester's
a κ-polynomial of degree s j - i ;
one of degree s j-i+m.
are polynomials
determinant
is,
as
Ac-
in κ of long
as
f o r i > m, on the contrary,
(A reduction of the degree could be caused by can-
cellation of the highest order contributions). All terms the determinant is composed of get f r o m each row and each column one f a c t o r . This leads to an upper bound f o r the degree of the resultant as a polynomial in κ, deg
κ
Res(A,B)
s
m-n
In f a c t , we have equality, since the summands of maximum degree combine to the resultant of G (χ , ι ι
F (0,
V
and G (x ,x ) 2 1 2 gotten by setting x q = 0 in F
=
F (0,χ , χ ) 2 1 2
and F .
Clearly R e s f G ^ G ^ cannot be identically 0, because the G^ by assumption do not have nontrivial solutions in common.
7 . 4 . B 6 z o u t ' s Theorem
277
The d e g r e e of Rest A, B) as a f u n c t i o n of κ t h u s is g -g , and t h e r e a r e e x a c t l y g j ' g 2 v a l u e s of κ f o r which Res(A,B) v a n i s h e s , to our initial problem:
Ν = g · g^.
We can c o n t i n u e t h i s a r g u m e n t a t i o n well.
and a s many s o l u t i o n s
in t h e i n d i c a t e d
m a n n e r f o r η £ 3 as
The d e t a i l e d e l a b o r a t i o n can be l e f t c o n f i d e n t i a l l y t o t h e
reader.
interested
The r e s u l t of our e f f o r t s is
Theorem 7.4.5:
The n u m b e r of s i m u l t a n e o u s s o l u t i o n r a y s of η f o r m s F , . . . , F in (n+1) i n 1 π d e t e r m i n a t e s w i t h c o e f f i c i e n t s in an a l g e b r a i c a l l y c l o s e d f i e l d and r e s p e c tive degrees gI
n g
is e i t h e r i n f i n i t e or equal to t h e p r o d u c t of t h e g1
(theorem of B6zout).
Remark:
Of c o u r s e ,
h e r e as a l w a y s ,
we have t o t a k e t h e algebraic multiplicities of
the linear f a c t o r s into account;
some or all of t h e s o l u t i o n s m i g h t
coin-
cide. In e a c h p a r t i c u l a r c a s e t h e c o r r e c t c o u n t i n g can be f o u n d f r o m a s i m p l e c o n s i d e r a t i o n of t h e individual f a c t o r s . Bezout [1769] h i m s e l f
p r e f e r s t h e a f f i n e i n t e r p r e t a t i o n and s t u d i e s t h e
r o s e t s of p o s s i b l y inhomogeneous p o l y n o m i a l s in η v a r i a b l e s χ 1
ze-
χ η in-
s t e a d of f o r m s in homogeneous c o o r d i n a t e s , a s we did. This l e a d s t o t h e f a m i l i a r c o m p l i c a t i o n s which a r e c o n n e c t e d w i t h t h e ( p r o j e c t i v e ) s o l u t i o n s on t h e hyperplane at infinity.
F o r t h i s r e a s o n t h e o r i g i n a l v e r s i o n of B e z o u t ' s t h e o r e m is much m o r e a w k w a r d t h a n t h e one given in t h i s s e c t i o n :
there are numerous exceptions
to
be c o n s i d e r e d which u n d e r c e r t a i n c i r c u m s t a n c e s may r e d u c e t h e n u m b e r of solutions
in t h e
s e n s e of
a f f i n e geometry.
We a r e
not
going t o
discuss
t h e s e q u e s t i o n s any f u r t h e r ; t h e s i t u a t i o n is much like t h e one e n c o u n t e r e d in s e c t i o n 7. 1. The above f o r m u l a t i o n of B e z o u t ' s t h e o r e m is by no m e a n s t h e m o s t g e n e r a l . In m o d e r n a l g e b r a i c g e o m e t r y , cepts are investigated.
f a r - r e a c h i n g e x t e n s i o n s of t h e r e l e v a n t
For e x a m p l e ,
t h e p r o j e c t i v e p o i n t s obeying
con-
certain
s y s t e m s of a l g e b r a i c e q u a t i o n s a r e r e p l a c e d by t h e irreducible components of
the
solution
s e t s whose
d i m e n s i o n s may be p o s i t i v e .
If
one
evaluates
278
7. P r o j e c t i v e Geometry
s u m s o v e r s o - c a l l e d i n t e r s e c t i o n n u m b e r s of t h e t h u s c o n s t r u c t e d algebraic varieties, one g e t s s i m i l a r , b u t even m o r e f u n d a m e n t a l r e s u l t s t h a n B£zout; c f . t h e book of H a r t s h o r n e [1977], For our purposes,
t h e t h e o r y developed h e r e is s u f f i c i e n t . We s h a l l
f o r e n o t e x t e n d t h e d i s c u s s i o n beyond t h i s p o i n t . priate
as the presentation
This is the more
is only i n t e n d e d a s an i l l u s t r a t i o n
of
thereapprothe
d e r l y i n g i d e a s (and t o s e r v e as a p r e p a r a t i o n f o r t h e i n v e s t i g a t i o n of nar complex algebraic curves).
unpla-
7.5. Planar Algebraic Curves
279
7.5. Planar Algebraic Curves
The realm of applications of the two types of projective coordinates in the plane
extends
far
beyond
the
description
of
points
or
straight
lines
as
discussed in section 7.1. Among other things,
they also f a c i l i t a t e the study of the zeroes of
higher
degree ternary forms, the so-called algebraic curves. In this book, we usually r e s t r i c t to curves in IP (C), which, f r o m a practical viewpoint,
are an
uttermost important special case. All definitions and the m a j o r i t y of the theorems can be t r a n s f e r r e d without d i f f i c u l t y to more general p r o j e c t i v e planes, braically the
closed
principal
scription
fields.
properties
in point
and
This of line
s u f f i c e s to algebraic
curves
coordinates.
at least to those over
serve At
our
intentions
to
algesketch
and their
mathematical
any
it
time,
will
be
declear
which generalizations are admissible.
Definition 7.5.1: By a curve we always mean a complex planar algebraic curve if not e x p l i c i t ly stated otherwise. Expressed in point coordinates,
a curve of order η e IN is the set of all 2 solution rays of a not identically vanishing f o r m F: P C —> C of degree n. If
2
the point χ with homogeneous coordinates (X q , x j . x 2 ) e P C
satisfies
the
equation F(x , χ , χ ) = 0, we say that χ lies on the curve F = 0 (and, 0 1 2
the
other way round, that the curve passes through the point). Dual to this,
we have a representation in line coordinates. The
solutions
of a f o r m G * 0 of degree k define a curve of class k. When G(u 0, u ,u 1 2) = 0 holds, the straight line u = (u ,u ,u ) is called a tangent of the curve 0 1 2 G = 0.
Remarks: ( 1 ) The necessity from
the
wish
(section 7. 1).
to consider two mutually to
utilize
the
extremely
dual
kinds of
fruitful
curves
Poncelet
derives
principle
280
7. P r o j e c t i v e Geometry
( 2 ) It is common practise not to distinguish systematically between a curve and its
equations
or
the associated
solution
sets.
We have
followed st
this custom already when we investigated points = curves of
1
class
st
and straight lines = curves of 1 The defining f o r m s of sense.
order.
course have to be understood in the p r o j e c t i v e
A curve thus is l e f t invariant if we multiply its equation
with
a nonzero complex number. ( 3 ) It is very important to c l a s s i f y the curves in decomposable = reducible ones, i . e .
those which can be f a c t o r i z e d into curves of lower order or
class,
and those which do not admit such a split-up.
curve,
be it given in point or line coordinates,
Evidently,
each
is representable in a
unique way as the product of irreducible components. In doing so, we have to acknowledge c a r e f u l l y the multiplicities with which the single components divide the equation of the curve. For example, we must distinguish between the straight line u χ +u χ +u χ = 0 0
r
and the curve of 2
0
1
order with the equation (u χ +u χ +u χ ) 0 0 1 1 2 2
Though the latter has the same zeroes as the f o r m e r ,
1
2
2
=0.
they have to be
counted twice. The quadratic f o r m t h e r e f o r e describes a so-called double line. ( 4 ) Whenever we speak of
a curve of
order η or class k,
this
implicitly
contains a statement on the type of coordinates (point or line) we have to use.
Before we shall discuss the interrelations and connections between the two categories of curves, which greatly surpass the duality proper, we r e f o r m u late a special case of B6zout's theorem in order to adapt it to the present situation:
Theorem 7.5.1: ( a ) Two curves of orders m and η intersect in m-n points. ( b ) Each two curves of classes k and 1 have k-1 tangents in common.
Proof: Theorem 7. 4. 5.
7.5.
281
P l a n a r A l g e b r a i c Curves
We remind the r e a d e r f o r the last time that the c o r r e c t a l g e b r a i c cities
of
all
factors
must
be
taken
into
account;
otherwise
multipli-
the
theorem
would be f a l s e . Under c e r t a i n c i r c u m s t a n c e s , tion
or
ficance,
tangents
coincide.
it occurs t h a t t w o or more p o i n t s o f
This
because the c o n d i t i o n
w e choose.
possibility
is
of
great
does not depend on the special
As a f i r s t a p p l i c a t i o n ,
intersec-
geometrical
signi-
coordinates
we shall use it to go o v e r f r o m one t y p e
of c u r v e s t o the dual. Assume
a curve
straight
line g
line
shall
of
(we
F ( x ) = 0 of
order
in η · 1 = η points. soon g i v e
η to With
an e x a m p l e ) ,
d e g r e e n. M u l t i p l e s o l u t i o n s e x i s t s ,
and only i f gree n(n-l).
i t s d i s c r i m i n a n t vanishes. T h i s i m p l i e s (one half
be given.
It
is
intersected
a suitable
parametrization
t h i s leads to
an a l g e b r a i c
by
of
equation
and g is a tangent t o the c u r v e , The l a t t e r
a
the
by t h e o r e m 7. 4 . 3 has
if de-
of)
Theorem 7 . 5 . 2 : A curve o f o r d e r η has c l a s s at most n ( n - l ) ; s
curves of c l a s s k a r e o f
order
k(k-l).
Here,
a r e d u c t i o n f r e q u e n t l y occurs,
a cubic
(curve),
η = 3.
Its
as is seen a l r e a d y f r o m the e x a m p l e of
equation
in line
coordinates
j u s t d e s c r i b e d method has the d e g r e e 3 - 2 = 6. r e p r e s e n t a t i o n w i t h point c o o r d i n a t e s , It o b v i o u s l y cannot be i d e n t i c a l
If
as found by
we r e t u r n to the
the
original
we obtain a curve o f o r d e r 6 - 5 = 30.
w i t h the curve w e s t a r t e d f r o m ,
but
con-
t a i n s it as a f a c t o r . The t r a n s i t i o n f r o m one curve v e r s i o n to the dual and back again thus p r o duces a d d i t i o n a l branches,
caused by the f o r m a l i n c r e a s e of the d e g r e e .
The
r e a s o n is t h a t a c u r v e which is d e f i n e d in point c o o r d i n a t e s a t t a i n s a v e r y special f o r m i f t r a n s f e r r e d to line c o o r d i n a t e s , We do not ther,
intend to
and v i c e v e r s a .
discuss the questions r a i s e d by this o b s e r v a t i o n
but o n l y consider a f e w types of curves (Clebsch
L e t us begin w i t h the simplest and f r o m a p r a c t i c a l point o f v i e w most p o r t a n t nonlinear
curves.
fur-
[1968]). im-
7. P r o j e c t i v e Geometry
282
Definition 7 . S . 2 : A conic section is a curve of 2 nd order and 2 n d class.
Since 2-1 = 2;
the degree
in this case does not g e t
larger
if
we
switch
over f r o m one version to the other. To i l l u s t r a t e the methods,
we c a r r y through the calculation e x p l i c i t l y .
The
general equation of a conic section in point coordinates is
Κ χ χ 1J ι J if we apply the Einstein convention.
=
0
The subscripts run f r o m 0 through 2.
We may demand in addition that K ^ = K ^ .
Κ
00
Of
the
χ® + Κ 0
six
11
x2 + K 1
22
χ2 + 2 Κ
coefficients
2
x x + 2 K
12
Κ
00
Thus,
1
Κ
12
,
2
only
20
the
explicitly,
x x + 2 K 2
0
five
01
ratios
multiplying the f o r m u l a as a whole by some constant
x x
= 0
are
essential;
0 1
f a c t o r * 0 does
not
change anything. When considering conic sections, tion of
section 7.1.
class than 2).
it is useful to remember the m a t r i x
nota-
(This is no longer true f o r curves of higher o r d e r
According to our e a r l i e r
agreement,
nates are represented by r o w and column v e c t o r s ,
point and line respectively.
or
coordi-
The
above
The rank of Κ can be equal to 1, 2 or 3; Rank(K) = 0 is not a l l o w e d
since
r e l a t i o n then reads =
with the symmetrical
0
matrix Κ Κ Κ 00 Ol 02 Κ Κ Κ 10 11 12 Κ Κ Κ 20 21 22
it corresponds to the z e r o f o r m . sentially d i f f e r e n t ,
i.e.,
simple c o o r d i n a t e t r a n s f o r m a t i o n , der i n v e r t i b l e linear maps.
The three remaining p o s s i b i l i t i e s are
es-
they cannot be t r a n s f e r r e d into each other by a as the rank of a m a t r i x is invariant un-
7.5. Planar Algebraic Curves
283
The case rank(K) = 3 is equivalent with det Κ * 0 (irreducible conic section).
We shall later treat these curves explicitly.
or proper
If Κ is not
regular, we obtain a degenerate conic section. For rank 2, the kernel of Κ is one-dimensional. We assume it to be generated by a vector χ * 0; thus χ · Κ = 0. It corresponds to a certain point on the conic section.
If y is a d i f f e r e n t point on the curve,
the same holds
true f o r all λ χ + μ y with arbitrary complex numbers λ, μ. The straight line connecting χ and y then is contained as a whole in the conic section which therefore decomposes into two straight lines. The same conclusion r e mains valid if rank(K) = 1, but then the two f a c t o r s of degree 1 are identical. In conclusion,
a degenerate conic section in point coordinates is simply a
(straight) line pair (intersecting at x),
if
the rank of Κ is equal to 2,
and a double line f o r rank 1. Going
over
to
line coordinates,
we get
the dual configurations,
namely
point pairs and double points, depending on the rank. From now on we restrict
our investigations to irreducible conic
x-K-tx
sections
= 0 and consider f i r s t the line coordinates u = ,u ,u ) of the 0 1 2 tangent at some curve point χ = (XQ, Χ , χ ). TO find it, we select a second point
l(u
y * χ on u.
Then χ + Xy is
incident
with
the
line u = 0 f o r
all
λ € C. The point of tangency of u is x; therefore the quadratic equation obtained by insertion, 0 or,
=
(x+Xy) Κ ^x+Xy)
=
λ |x Κ *y + y Κ Scj + λ 2 y Κ ' y
simplified, 2 λ y Κ 4x + λ 2 y Κ V
=
0
must have the twice counting solution λ = 0. Consequently,
the c o e f f i c i e n t
of λ vanishes: y Κ *x
=
0
This linear relation in y must be equivalent with y-u = 0. Up to an i r r e l e vant numerical f a c t o r it is allowed to set
284
7. P r o j e c t i v e Geometry
u
=
Κ 'χ
By assumption, Κ is symmetric and invertible, χ
t
=
so we can solve f o r x:
„-1 u Κ
and insert the result into the basic equation.
This yields the formula f o r
our conic section as a curve of class 2, namely 4u
K _1 u
=
0
It is not d i f f i c u l t either to find the two tangents which can be drawn f r o m an ( e x t e r n a l ) point χ e Ρ C to the curve.
Let y be the contact points
of
one of them. We get the condition
χ Κ V
=
y Κ Sc
=
0
A comparison with the equation of the conic section itself provides us with two solutions: y and the point of contact of the other tangent. We c o l l e c t the results just found,
together with their dual
counterparts,
and take the opportunity to introduce at the same time some more important notions.
Theorem 7 . 5 . 3 : ( a ) Let X be an irreducible
conic
section.
Its equation
in point
coordi-
nates Χ = (XQ, X I> X 2 ) can be brought to the f o r m χ Κ Sc
=
0
while in line coordinates u = Su , u , u ) it reads 0 1 2 tu
L u
=
0
Here Κ and L are regular symmetric matrices which, except f o r an unimportant constant multiplier,
are inverse (or,
by the nonzero determinant,
a d j o i n t ) to each another.
if we p r e f e r to
multiply
( b ) If the point χ and the line u (up to some f a c t o r ) f u l f i l l the relation
7.5. Planar Algebraic Curves
285
u
=
Κ lx
χ
=
t . u L
or t h e e q u i v a l e n t dual one,
w e c a l l u t h e polar (line) of x , and χ t h e pole of u w i t h r e s p e c t t o X. (c) Pole and polar determine each other uniquely.
Furthermore,
a l l p o l e s of
l i n e s p a s s i n g t h r o u g h χ l i e on t h e p o l a r of u, and v i c e v e r s a . A pole-polar
pair
x , u is i n c i d e n t
a n d χ i t s p o i n t of ( d ) More g e n e r a l l y ,
if
and
o n l y if
u is a t a n g e n t
to X
contact.
t h e p o l a r of s o m e p o i n t χ i s t h e c o n n e c t i n g l i n e of
p o i n t s of t a n g e n c y of b o t h t a n g e n t s f r o m χ t o X. A n a l o g o u s l y ,
the
the pole
of a s t r a i g h t l i n e u is t h e i n t e r s e c t i n g p o i n t of t h e t w o c o m m o n
points
of u a n d X.
T h e r e a r e n u m e r o u s f u r t h e r p r o p e r t i e s of p o l e s and p o l a r s , be d e f i n e d f o r o n e of t h e m ,
higher
degree
curves,
cf.
C l e b s c h [19681.
which can
also
We d i s c u s s
only
s i n c e i t p r o v i d e s d e e p e r i n s i g h t in t h e s t r u c t u r e of t h e
s e c t i o n s and t h e i r In t h i s r e s p e c t ,
conic
equations.
p a r t (d) of t h e f o r e g o i n g t h e o r e m p l a y s a m a j o r r o l e .
We
c o n s t r u c t w i t h i t s h e l p t o a g i v e n p o i n t E, w h i c h i s s u p p o s e d n o t t o l i e on X but is otherwise a r b i t r a r y , p o i n t F on e,
the polar
b u t n o t on X. I t s p o l a r ,
l i n e e. f,
T h e r e a f t e r we choose
any
i n t e r s e c t s e in a t h i r d p o i n t G.
In t h e s a m e vein w e d e n o t e t h e l i n e c o n n e c t i n g Ε and F by g. We t h u s o b t a i n a polar triangle w i t h r e s p e c t t o t h e i r r e d u c i b l e c o n i c
sec-
t i o n X. E a c h of e,f,g.
its vertices E,F,G
is t h e p o l e of t h e c o r r e s p o n d i n g o p p o s i t e
It i s t o be e x p e c t e d t h a t
the equation
of X a t t a i n s
a
side
particularly
s i m p l e f o r m if w e t a k e Δ(Ε, F, G; e, f , g) a s t h e fundamental triangle of a n e w coordinate frame. This is indeed the case. we
deduce
proposition:
(after
having
From the f a m i l i a r r e l a t i o n between pole and defined
the
multipliers
suitably)
the
polar
following
7. P r o j e c t i v e Geometry
286
Theorem 7.S.4: Relative to a polar triangle,
the equation of a proper conic section is in
point coordinates 1
x x
2
^
2
^
x +
2
=
x + 0
x 1 2
=
2 ^ 2 ^ 2 u + u + u 0 1 2
=
η
0
and in line coordinates 1
Hence,
uu
η = 0
all irreducible conic section can be transformed into each other by
pro j e c t i v i t i e s ,
i. e. chart changes.
For curves of higher order (or class) than 2 the analogous statement dently is f a l s e ,
since a f o r m of degree η contains exactly i(n+l)(n+2)
evico-
e f f i c i e n t s , one of which can be divided out. The remaining -n(n+3) parameters cannot be transformed away by an application of the 8-dimensional p r o j e c t i v e group PGL^C). This is already the case f o r η = 3, where we have 9 essential
parameters.
For f o r m s of
higher
transformations,
degree (n £ 4),
the possible simplification by
namely elimination of (at most) 8 c o e f f i c i e n t s ,
linear
is compa-
rably i n e f f i c i e n t , and we shall not pursue these approaches here. On the other hand, we can t r y to reduce the cubic curves (3 r d order or 3 r d class) to a one-dimensional family of standard curves.
This is indeed f e a -
sible. We shall come back to this question later the e l l i p t i c functions.
(chapter 18) in connection
with
7.6.
Stereographic
Projection
287
7.6. Stereographic P r o j e c t i o n
2
I t is known t h a t the 2 - s p h e r e S
2
the Euclidean plane IR . F o r , 2
of IR
existed,
consequently (χ
then Μ would be, isomorphic
= 1), h o w e v e r ,
Thus t h e r e
disk.
The Euler
number
of
and
the
disk
d i f f e r s f r o m that of the sphere (χ = 2).
2
S
which consists of
only
one c h a r t .
e v e r y c h a r t of the sphere has a s i n g u l a r i t y .
To
This f a c t ,
c o m p l i c a t e s the p o s i t i o n i n g
considerably.
On the o t h e r hand, face
into
onto a subset Μ
2
t o c a r t o g r a p h e r s f o r many c e n t u r i e s , on the sphere
2
like S , compact and simply connected
to the c l o s e d
is no a t l a s of
another way,
cannot be embedded h o m e o m o r p h i c a l l y
if a t o p o l o g i c a l mapping of S
of
put
known points
2
i f we d e l e t e only one point f r o m S , the r e m a i n i n g
is homeomorphic
and even
(analytically)
plane. We want to e s t a b l i s h such a b i j e c t i o n , model S2 by the set
{
it
the s i m p l e s t in f a c t .
3
(x,y,z)
diffeomorphic
2
e IR
2
2
x +y +z
to
the
surentire
To do so,
we
I
= 1 1-
3
and choose the south pole
of a l l unit v e c t o r s in IR
Σ as the e x c e p t i o n a l p o i n t . Σ with
the
south
pole
=
(0,0,-1)
€
Ω
=
3
S2
We connect an a r b i t r a r y v e c t o r ( x , y , z ) o t h e r than
by a s t r a i g h t
line
in space.
It
hits
the
plane ζ = Ο in a w e l l d e f i n e d point w i t h c o o r d i n a t e s ( u , v , 0 ) , the i r r e l e v a n t t h i r d component, The
assignment
Clearly,
other
described
normalizations
is
are
standard f o r m .
known
as
possible,
the
Omitting
stereographic
mapping.
but
we
+ (1-λ)
shall
Our next task w i l l
citly. 2 The l i n e through Σ and ( x , y , z ) e S Μ Σ ) c o n s i s t s of
λ (x,y,z)
say.
w e g e t ( u . v ) e IR .
( x , y , z ) «=» (u, v )
the j u s t
equatorial
(0,0,-1)
=
always
stick
be to f i n d i t
all
(λχ,\y,λζ+λ-1)
to
expli-
288
7. Projective Geometry
w i t h r e a l λ.
One of them must be ( u , v , 0 ) .
U
=
V
=
This i m p l i e s λ = (1+z)" 1 and thus
χ 1+z JL 1+z
The i n v e r s e map is obtained s i m i l a r l y .
The line through Σ and ( u , v , 0 )
con-
t a i n s the p o i n t s
λ (u,ν,0)
The second
+ (1-λ)
intersection with 2 2 2
f r o m the c o n d i t i o n χ +y +z
f o r λ,
one of
(0,0,-1)
the
sphere,,
=
(Au,λν,λ-1)
aside f r o m Σ,
= 1. It y i e l d s the q u a d r a t i c
can be
calculated
equation
λ2 (uZ+v2+l) - 2 λ + 1 = 1 whose s o l u t i o n s (λ = 0 ) has to be d i s c a r d e d since it 2
ponds to the south p o l e ,
2
corres-
—1
w h i l e the o t h e r leads to λ = 2(u +v +1)
and
the
(x,y, z)
are
point 2 u χ
=
2
1+u +v
2
2 ν y
Ζ
The p r o j e c t i o n it:
(x,y,z)
~
=
i-» ( u , v ) ,
~
2
1+u +v
2
2 2 1-U -V 1+U2+V2 and its inverse
w = u+iv 6 C
c a n o n i c a l l y w i t h the f i e l d C of the complex numbers.
This procedure
results
in a c o n s i d e r a b l e gain of mathematical s t r u c t u r e . F u r t h e r m o r e w e a r e p r o v i ded w i t h a v e r y n a t u r a l projective
space a l r e a d y f a m i l i a r to us,
namely the
complex
line
C υ { 3 is only u s e f u l f o r
representation.
description
numerical
of
the
calculations.
investiga-
rotation
groups
The e v a l u a t i o n
of
e x p ( S ) is m o s t e f f i c i e n t l y done n o t w i t h t h e h e l p of t h e d e f i n i t i o n f o r m u la,
b u t by a t r a n s f o r m a t i o n of t h e skew s y m m e t r i c a l m a t r i x S t o
principal
a x e s ( t h i s i s a l w a y s p o s s i b l e ! ). To t h i s end we have t o seek a s u i t a b l e τ e GL(n,R) such t h a t t h e t r a n s f o r m ed m a t r i x S T = τ 1 Sx is d i a g o n a l , S'
=
ST
=
e.g.
τ" 1 S τ
=
DiagU
1
λ ) π
The λ^ t h e n a r e t h e e i g e n v a l u e s of S, and t h e e x p o n e n t i a l mapping p r o v i d e s us w i t h
D'
=
exp(S')
=
, λ Diagle 1
λ •> e "
and t h u s
D =
τ D' τ
=
λ f λ 1 τ Diag e , . . . , e n τ "
N u m e r i c a l p r o c e d u r e s f o r c a r r y i n g out t h e t r a n s f o r m a t i o n t o p r i n c i p a l
axes
o r t h e d e t e r m i n a t i o n of t h e e i g e n v a l u e s can be f o u n d in n u m e r o u s c o m p u t e r libraries;
we a r e n o t going t o give any advice.
At t h e end of our i n v e s t i g a t i o n i n t o t h e e x p o n e n t i a l r e p r e s e n t a t i o n of rotation groups,
the
we should come back once m o r e t o t h e a m b i g u i t y which h a s
310
8. Orthogonal Groups
been already discussed f o r the plane. Since all of
(see chapter
2) the identity
is the only
whose eigenvalues are equal to unity,
= - S yields
D = 1 if
its
eigenvalues
orthogonal
transformation
every matrix S € R t n , n ) are
integer
multiples
with
of
2ni.
Thanks to theorem 8.2. 1, this means in particular that all skew symmetric (3, 3 ) - m a t r i c e s S(v) with
ρ
=
jν I
e
2nl
(and only these) describe the identical rotation. T h e r e f o r e all elements of SO(IR ) are already included in the formula with
I ν I < 2ir, which gives each of them, except f o r D = 1, exactly
D(v)
twice.
This observation would allow us to determine the topological structure
of
the Lie group SO(3,IR), but we shall postpone the necessary considerations. In the next section, the desired result w i l l be handed us on a plate.
8.3.
Rational
Parametrization
311
8.3. Rational Parametrization 1 The r e p r e s e n t a t i o n
of
the
standard
1 - s p h e r e Ω^ = S
2 in R
as
the
set
of
c o m p l e x n u m b e r s w i t h m o d u l u s 1 c a n be e x p l o i t e d f o r a n o t h e r d e s c r i p t i o n of t h e r o t a t i o n m a t r i c e s w h i c h c a n be c a r r i e d o v e r t o o r t h o g o n a l
transforma-
n
t i o n s in IR . All w e n e e d i s t o g e n e r a l i z e t h e s i m p l e s t c o n c e i v a b l e c o n s t r u c t i o n of
num-
b e r s of m o d u l u s 1 in C, n a m e l y f o r m i n g t h e q u o t i e n t of t w o c o n j u g a t e c o m plex
quantities + b i
and
a - b i
(a, b € IR). T h u s , b e g i n n i n g w i t h t h e c a s e of t h e E u c l i d e a n p l a n e , we h a v e t h e n e w c h a r a c t e r i z a t i o n of t h e s p h e r e S 1 ,
s1
It h a s
to
be t a k e n
for instance,
η
s
into
2
=
/
account
j
(
I a-bi
e
r2n{0}
that
Ί J
mutually proportional vectors 2 ( a , b ) a n d A - ( a , b ) = (Aa.Xb) in IR \{0> l e a d t o t h e s a m e
like, point
in Ωζ- T h i s y i e l d s a n i s o m o r p h i s m of Sl^ and t h e o n e - d i m e n s i o n a l real projective space, i. e. t h e projective line:
(compare t h i s with section 7.1). F o r m a n y a n a p p l i c a t i o n i t w i l l be c o n v e n i e n t t o r e p l a c e t h e c o u p l e of projective coordinates a a n d b by t h e i r
t
=
a
€
ratio
IP1 R
=
IR υ {•*>}
312
8. Orthogonal Groups
with the r e s u l t
•
{ PIT I " " · ' » }
As we explained a l r e a d y in chapter 7 about p r o j e c t i v e g e o m e t r y , the " n o r m a l " f r a c t i o n s
(for
which a * 0 ) ,
aside f r o m
we also must admit the
special
Thus we are f o r c e d to include the infinite or
case a = 0 (implying b * 0).
improper point t = oo, subject to the rule 1+ooi 1-ooi It
corresponds
real
values of
to
the point
the parameter
+1 -1
-1
( - 1 , 0 ) e Ώ^ c R t.
which
The t o p o l o g y
of
is not
IP1(IR),
obtainable
d e f i n e d in
with
section
8. 1, goes through the assignment
1+tl 1-ti
over into the natural t o p o l o g y of the c i r c u l a r line S . We return to the m a t r i x notation and apply the f a m i l i a r 2-dimensional gebra)
(al-
representation 1
0 =
0
0
1
-1
0
1
=
1
I
of C. We have discussed it already at length in the last section. ber t i e C ( w i t h t € R) is associated with the skew symmetric
t
0
t
•t
0
The num-
(2,2)-matrix
I
For the g e n e r a l case of a r b i t r a r y ( f i n i t e ) dimension η a 2 we are in an abs o l u t e l y analogous situation.
In conclusion,
we g e t
Theorem 8.3.1: Suppose S € R ( n , n > to be skew symmetric:
lS
= -S.
Then
8.3. Rational
Parametrization
D =
D(S)
is an n - d i m e n s i o n a l
=
313
=
rotation
1
(1+S) · (1-S)
matrix.
=
Moreover,
1
(1-S)
· (1+S)
every e l e m e n t
of
SO(n,IR)
which d o e s n o t p o s s e s s t h e e i g e n v a l u e -1 i s uniquely r e p r e s e n t a b l e in t h i s m a n n e r . The i n v e r s i o n f o r m u l a r e a d s D-l D+l
Proof: The e i g e n v a l u e s of S a r e p u r e l y i m a g i n a r y ,
such t h a t in t h e given e q u a t i o n
t h e d i v i s i o n i s well d e f i n e d . Beyond t h i s , auxiliary
D is an e l e m e n t of SO(n,IR),
as can be seen f r o m t h e
simple
calculation
*D
=
His l^S
and t h e o b s e r v a t i o n
that
=
l·^
=
D -1
of
1+S and
1+S
the determinants
l+lS are
identical
with each other. The t r i v i a l p r o o f of t h e l a s t e q u a t i o n f o l l o w s by solving f o r S.
Remark: ( a ) E x a c t l y like f o r η = 2,
in t h r e e - d i m e n s i o n a l
s p a c e it is a l s o
possible
t o avoid t h e e x c e p t i o n caused by t h e r o t a t i o n s w i t h e i g e n v a l u e -1. We only have t o s e t , s l i g h t l y m o r e g e n e r a l l y t h a n b e f o r e ,
w i t h d e IR and 1,S e£ R ( 3 , 3 ) ; S + l S = 0. C l e a r l y , t h e p a i r s (Xd.XS), w h e r e λ e IR , give t h e same t r a n s f o r m a t i o n , (b) T h i s a r g u m e n t a t i o n e s t a b l i s h e s , going section,
the isomorphism
(d,S)
p a r a l l e l t o t h a t s k e t c h e d in t h e
and
fore-
314
8. Orthogonal Groups
S0(R 3 ) (as topological
IP3 IR
s
spaces).
E s p e c i a l l y f o r η = 2 we g e t t h e r e p r e s e n t a t i o n
D =
™
-
1-t
2t
-2t
1-t
(l+tV>
By c o m p a r i s o n w i t h t h e e x p o n e n t i a l f o r m of D, cos a
sin a
-sin α
cos α
D =
we o b t a i n
cos a
=
sin α
=
1-t 1+t 2 2t 1+t 2
and u n d e r s t a n d t h e g e o m e t r i c a l meaning of t h e p a r a m e t e r t , t As an a s i d e ,
=
namely
tg
e x c e p t f o r t h e t r i g o n o m e t r i c f u n c t i o n s of a,
the d i f f e r e n t i a l
of t h e a n g l e a,
doc
2 dt l+t2
is a l s o r a t i o n a l
in t and d t .
This is o f t e n used t o r e d u c e
i n t e g r a t i o n s t o e a s i e r ones w i t h r a t i o n a l At l a s t ,
trigonometric
integrands.
we give t h e e x p l i c i t f o r m u l a s f o r t h e r a t i o n a l r e p r e s e n t a t i o n of a
t y p i c a l D e SO(3,IR). It is n o t a p p r o p r i a t e rectly.
Instead,
to c a r r y
out t h e i n v e r s i o n of t h e m a t r i x
we f i r s t e x c l u d e ,
(1-S)
as in s e c t i o n 8 . 2 in t h e d i s c u s s i o n
diof
8 . 3 . Rational Parameterization
315
the exponential mapping, the trivial case S = 0, D = 1 and note that D com2 mutes with S and thus is a linear combination of the matrices 1 , S , S :
D =
1+S 1-S
=
A·1 + B-S + C-S
Multiplication by the common denominator yields 1+S
=
A·1 + (B-Α) S + (C-B) S 2 - C S 3
We insert 0 S
=
S(o,ß,r)
-r
jr ο
β -a
and compare the c o e f f i c i e n t s ( 1 , S , S
=
/
aZ+ß2+rZ
and the minimal equation S3
=
- p2 S
A =
1
getting the solution
Β
= C 1+P*
and consequently D =
Hence we have proven:
a 0
are linearly independent!),
the shorthand notation
ρ
-β
1 + — (S + S 2 ) 1+P2
again with
316
8. Orthogonal Groups
Theorem 8.3.2: 3
( a ) The rotation D £ SO(R ) belonging to the parameter matrix
Sict.ß.y)
0
r
-y
0
1
β -a
D
=
1 +
α 0
(s + sn i+a 2 +ß 2 +r 2
( b ) All rotations in IR , except those which possess the eigenvalue -1,
can
be (uniquely) written in this way.
In order to include the hitherto neglected orthogonal transformations with eigenvalue -1, i. e. rotation angle 180°, we go by setting
α
=
a d
. :
=
ß
b d
c :
7
=
d
3 over to p r o j e c t i v e coordinates ( a , b , c , d ) in the parameter space P R .
Theorem 8.3.3: ( a ) Each rotation matrix D e SO (IR) is of the f o r m 3
' a 2 -b 2 -c 2 +d 2 D(a, b , c , d )
2
,2
a +b +c +d
2(ac-bd)
2(ba-cd)
2(bc+ad)
2(ca+bd)
2(cb-ad)
c 2 -a 2 -b 2 +d 2
( b ) If a , b , c are not simultaneously zero, f r o m the identity,
2(ab+cd) b 2 -c 2 -a 2 +d 2
or,
in other words,
'
if D d i f f e r s
the axis of D ( a , b , c , d ) is parallel to ( a , b , c ) e R3.
( c ) The rotation angle φ is determined by . 2 (VI tg W
-
a 2 +b 2 +c 2
—p-
8.3. Rational
Parametrization
317
Proof: The Statement under ( a ) has been confirmed e a r l i e r .
Furthermore,
one f i n d s
f r o m theorem 8 . 3 . 2 and
(a,b,c)-S
=
0
or by d i r e c t calculation that
(a,b,c)-D
and thus ( b ) holds true.
=
(a, b, c )
The eigenvalues of
D are l . e ' ^ . e
of their sum
1 + e 1 * + e" 1(P
with the t r a c e of D g i v e s
(c).
=
1+2
cos φ
a comparison
318
9. Linear Transformations of Complex Spaces
9. Linear Transformations of Complex Spaces 9.1. Pauli Matrices
We a s s o c i a t e with an a r b i t r a r y complex ( n . n ) - m a t r i x Μ via
a pair Η, Κ of Hermitian matrices, f r o m which we can g e t Μ back by the f o r mula
Μ =
Η + i Κ
For η = 1 this s p l i t - u p reduces to the decomposition of into i t s r e a l and imaginary parts.
a complex
number
We analogously denote Η as the Hermitian
and iK as the skew-Hermitian part of M. Real linear combinations of Hermitian m a t r i c e s e n j o y the same p r o p e r t y well.
as
T h e r e f o r e all Hermitian (n, n ) - m a t r i c e s f o r m an IR-vector space (of 2, mension η ).
di-
The simplest
of
the
importance;
for
nontrivial
case
(n = 2) concerns
the
linear
mappings
2
complex plane C
into i t s e l f .
It is of p a r t i c u l a r p r a c t i c a l
this reason we are going to discuss it here at some length. The m a t r i x a
b
c
d
Η
( a , b , c , d e C) is Hermitian if
and only if
are complex c o n j u g a t e to each other.
a and d are r e a l ,
ansatz
b
=
χ - 1 y
w h i l e b and c
It is t h e r e f o r e reasonable to make the
c
χ + 1 y
9.1. Pauli Matrices
319
with real numbers x, y. We can restore the symmetry of the formulas by substituting d
with likewise real w, z,
Η
=
w - z
getting
=
H(w,x,y,z)
w+z
x-iy
x+iy
w-z
=
If we drop the condition that the parameters lie in IR and allow
arbitrary
(2 2)
complex numbers instead, we obtain all elements of C '
; but the s i g n i f i -
cance of the representation in question of course is founded on the especially simple and elegant description of the Hermitian matrices.
We conse-
quently r e s t r i c t our considerations to ( w , x , y , z ) € IR . The invariants of Η are easy to evaluate: the trace is Tr Η =
2 w
while the determinant det Η
=
(w+z) (w-z) - ( x + i y ) ( x - i y )
=
w2 - χ 2 - y 2 - z 2
represents the Lorentz form in the (Cartesian) coordinates ( w , x , y , z ) . indicates an intimate relationship
This
with the special theory of relativity,
which we shall attend to in chapter 14. But b e f o r e , we consider the spectrum of H. It is immediately found f r o m the last two equations:
EW(H)
=
I w ± / x 2 +y Z +z 2
The Hermitian ( 2 , 2 ) - m a t r i x Η can alternatively be written as Η
=
wo- + x
twenty
conjugation
M(a,-b,-c,-d)
discussed.
But G a u ß w a s n o t t h e f i r s t m a t h e m a t i c i a n e i t h e r w h o c a m e u p o n t h i s a l g e b r a ic
concept;
E u l e r [1911-
it
had
been
discovered
and
investigated
much
earlier
by
].
Aside f r o m t h i s i n s t r u c t i v e c o m p a r i s o n ,
we w a n t t o e m p h a s i z e a l s o a s o m e -
338
9. L i n e a r T r a n s f o r m a t i o n s of Complex Spaces
what different parameterization
of t h e m u t a t i o n g r o u p which h a s
likewise
been f o u n d in G a u ß ' s n o t e s . The a n s a t z +a +d - c Κ =
- d +a +b
+ f
(b,c,d)
(b,c,d)
+c - b +a
+a+b f Κ =
leads,
+d+bcf 2
-c+bdf
-d+bcf
+a+c f
+b+cdf
+c+bdf
-b+cdf
+a+d f
a s t h e m u l t i p l i c a t i o n by t h e t r a n s p o s e d m a t r i x s h o w s , t o a m u t a t i o n
if we s u b s t i t u t e f
=
(n+a)" 1
Here η a g a i n h a s t h e a b o v e - m e n t i o n e d i n t e r p r e t a t i o n a s a s c a l i n g f a c t o r . We t h u s have g o t t e n a new r e p r e s e n t a t i o n of t h e g r o u p GM^IR) of m u t a t i o n s ; t y p i c a l e l e m e n t is
Κ =
By d i r e c t c a l c u l a t i o n ,
b +a(n+a)
bc+d(n+a)
bd-c(n+a)
bc-d(n+a)
c +a(n+a)
cd+b(n+a)
bd+c(n+a)
cd-b(n+a)
d2+a(n+a)
one indeed f i n d s Κ lK
=
Κ =
η2· 1
which i m p l i e s t h e m u t a t i o n p r o p e r t y (with s c a l e n).
its
9 . 5 . Euler Angles
339
9.5. Euler Angles 3 The r o t a t i o n s of IR f o r m a three-dimensional seek a p a r a m e t e r i z a t i o n
of
L i e group.
SO^ilR) by three angles,
description of SC>2(IR) with the r o t a t i o n
natural
similar to the f a m i l i a r
To understand
we f i r s t need the special r o t a t i o n s around the three axes of
Cartesian r e f e r e n c e f r a m e .
For s i m p l i c i t y ,
we want to avoid
f i n d the m a t r i c e s
of
the r o t a t i o n s
with
p o s i t i v e d i r e c t i o n ( " l e f t t u r n " ) about the x - ,
D (?)
=
1
0
0
cos
0
cos
sin
D (?)
φ
0
=
φ
the
we
mathematically
and z - a x e s .
They are
0 sin
φ
-sin φ
Dy ( φ )
angle φ in
y-,
his
sub-
s c r i p t s and c a l l the coordinates x, y, z. With the results of section 8 . 2 , easily
to
angle.
A v i a b l e method of this kind goes back to Euler [1758]. approach,
This suggests
φ
cos φ
0 -sin 1
0
0
cos
φ
φ
cos φ
sin φ
0
-sin φ
cos φ
0
0
0
1
How can w e build up the general r o t a t i o n of IR f r o m these? Following
Euler,
we f i r s t c a r r y out a z - r o t a t i o n with the angle a,
y i e l d s new coordinates x ' , y ' , z ' , tion;
say.
T h e r e a f t e r we apply a second
which rota-
this time about the x ' - a x i s with angle β. This d e f i n e s a new f r a m e ,
x",y",z".
Finally,
t a t i o n angle y ) .
we only have to r o t a t e once more around the z " - a x i s
The result of the procedure is the orthogonal
tion with the m a t r i x
E(a,ß,y)
When multiplied out,
it reads
=
D (a) D (ß) D ( y ) ζ χ ζ
(ro-
transforma-
340
9. Linear Transformations of Complex Spaces
sina cosß siny + cosa cosy
sina cosß cosy + cosa siny
sina sinß
cosa cosß siny - sina cosy
cosa cosß cosy - sina siny
cosa sinß
sinß siny
- sinß cosy
cosß
In spite of the unsymmetrical way in which the partial operations are applied, the Euler angles α, ß, y are in widespread use, mainly in the physical literature. The reason is that the "symmetry breaking" by the decomposition into "elementary rotations" is thoroughly welcome and meaningful,
particu-
larly f o r the investigation of the motion of rigid bodies (spinning tops). An extremely, and perhaps overly comprehensive evaluation of the many mathematical
facets of
the Euler angles and their
utilization
for
physical
problems can be found in the handbook of Klein & Sommerfeld [1897-1910]. A presentation of the theory of the gyroscope, which is more apt to practical
aspects,
is
the book
Schneider [1992-1993]
by Whittaker [1937],
mainly
consider
while
astronomical
Bucerius [1967]
applications
and
(rotation
of the Earth, precession, nutation, etc. ). As concerns history and the impact of the theory of the spinning top,
especially the work of Euler him-
self and about later developments in the prize-winning paper of Kowalewskaja [1890] and other achievements, cf. Cooke [1984], The orthogonality of the product matrix given above is manifest, is composed of three rotations. tirely
On the other hand,
since it
it is perhaps not en-
obvious that the Euler matrices Ε(α, β, y) form a group
isomorphic
with SO (IR). 3
In order to corroborate this, we want to have a closer look at the transformations in question. Trivially,
the identical map is equal to E ( 0 , 0 , 0 ) ,
while the inverse matrix
of E(a, ß,y) is
[E(a,ß,y)j
=
D (-y) D (-ß) D (-a)
E(-y.-ß.-a)
Note the order of the angular parameters! The product of two Ε-matrices is by no means easy to calculate; we theref o r e shall not attempt to do so,
but instead v e r i f y the assertion by di-
rectly comparing the Euler approach with the Cayley-Klein formulas of section 9. 2. To begin with, we find without much e f f o r t the (2,2)-matrices belonging to the rotations D («>), D (a>) and D («ρ), χ y ζ
9.5. Euler Angles
341
Δ (φ)
=
Δy (φ)
= -sin 2
Δ (φ)
cos,
ΗΝ
=
exp
by
explicitly
solving
the
2
correspondence
(?)
equations.
The
(2,2)-version
of
Ε ( α , β , γ ) turns out to be
=
Α ( α , β , γ )
e
Δ
i(+a+y)/2
(α)
cos
Δ
(β)
ρ I
Δ
(y)
i e1(+"-*)/2
sinl^
Δ(α,β,τ) ι
s i n
l ( - a -
ι
5
Because of that, the Cayley-Klein parameters of A
eK+a+jr)/2
Β
=
, i (+a-y)/2 , \β i e sIn I
c
=
ι
=
e1 e ' * w w i t h φ e R, and the inversion
Since the above substitutions g e n e r a t e the e n t i r e Möbius g r o u p P G L ^ C ) , proof
the
o f the t h e o r e m is c o m p l e t e .
We a r e now capable of t r e a t i n g our f i r s t g e o m e t r i c
problem:
Theorem 9 . 6 . 3 : The antipode o f
a e P 1 (C) is
Proof: By d e f i n i t i o n ,
a
*
#
=
π(ο- ( a ) )
*
=
π(α )
=
π(-α)
A f t e r a f e w t r i v i a l i n t e r m e d i a t e steps, the t r a n s f o r m a t i o n rules o f
theorem
7. 6. 1 y i e l d the c o r r e c t n e s s of the claim.
The most i m p o r t a n t s u r f a c e of
curves on the sphere a r e the c i r c l e s . 2
second o r d e r ,
three
different
them.
Its
points
intersection
S
algebraic
does not contain c o l l i n e a r point t r i p l e s ;
a, 6, c uniquely with S
t i o n can be f o u n d e a s i l y .
As an
define
a
plane
is the ( o n l y ) c i r c l e
Ε
passing
Κ 3 α, b, c,
We s t a t e it in a s y m m e t r i c a l
i.e.,
through
whose
equa-
f o r m as a c r i t e r i o n
f o r the c o c i r c u l a r i t y of ex, b, c and a f o u r t h point 6.
Theorem 9 . 6 . 4 : The f o u r
2 p o i n t s a, b,c,6 e S r e s p .
their
images
a,b,c,d
under
the
stereo-
348
9. Linear Transformations of Complex Spaces
graphic p r o j e c t i o n l i e on a common c i r c l e if and only if the cross r a t i o
is
r e a l or i n f i n i t e :
CR(a, b, c, 6)
=
CR(a,b,c,d)
e
R υ {»}
=
IP1 (IR)
Proof: The points a, b, c, d have the required p r o p e r t y whenever the same condition holds f o r
M(a), μ(1>), μ(ο), μ( q is an a n t i a u t o m o r p h i s m :
definite
(Eu-
10.1.
The Skew F i e l d o f
Quaternions
(d) The i n v e r s e of q i s (qq)
1
357
q+u
=
q + u
qu
=
u q
-q.
( e ) The c e n t r e Z(IH) of IH i s (R-1 £ R. ( f ) The c e n t r a l i z e r o f a q u a t e r n i o n q i s t h e l i n e a r span of 1 and q, q is real.
unless
In t h i s c a s e t h e c e n t r a l i z e r i s IH.
In o r d e r t o p r e p a r e t h e f o r m u l a t i o n of t h e n e x t t h e o r e m ,
we o b s e r v e
t h e n a t u r a l d e c o m p o s i t i o n of q i n t o t h e r e a l p a r t a = Re(q) and t h e
that
imagi-
n a r y p a r t ν = Im(q) = bi+cj+dk l e a d s t o q2
[a 2
=
+
ν2]
+
[ 2 a v]
The s q u a r e b r a c k e t s in t h i s e q u a t i o n i n d i c a t e t h e s e p a r a t i o n i n t o r e a l imaginary
and
parts. 2
One n o t i c e s t h a t f r o m q
e R at l e a s t one of t h e e q u a t i o n s a = 0 and ν = 0
follows.
The f i r s t c a s e y i e l d s p u r e l y i m a g i n a r y q u a t e r n i o n s ,
is
negative
then
(if
and o n l y if q i s r e a l
q * 0),
while
the
alternative
itself.
occurs
if
2
The c o m p l e t e s o l u t i o n s e t of t h e s p e c i a l e q u a t i o n q s i s t s of all pure
and t h e s q u a r e
possibility
= -1 c o n s e q u e n t l y
con-
norm-l-quaternions.
It i s n o w no l o n g e r d i f f i c u l t t o d e t e r m i n e a l l s u b a l g e b r a s o f IH:
Theorem 10.1.3: The s u b a l g e b r a s U of IH a r e ( 1 ) t h e z e r o a l g e b r a 0 (dim U = 0); ( 2 ) t h e a l g e b r a of r e a l q u a t e r n i o n s IR-1 as IR (dim U = 1); (3) those
s u b a l g e b r a s w h i c h a r e s p a n n e d ( a s v e c t o r s p a c e s ) by 1 and any 2 = -1. They a r e a l l i s o m o r p h i c t o C (dim U = 2); and
element ι with ι ( 4 ) IH i t s e l f
(dim U = 4 ) .
Proof: Suppose that U is a subalgebra
of IH w h i c h i s n o t c o n t a i n e d
in t h e
above
358
list.
10. Quaternions
Then dim U e { 1 , 2 , 3 } ,
because the dimensions'0 and 4 correspond only
to 0 and IH, respectively.
I If dim U = 1, we select some q e U = U\{0>. Its square is, as an element 2 of U, linearly dependent of q and both, q and q , are unequal to 0. Like in the introductory remark to the present theorem, we split up q = a + ν with a = Re(q) and ν = Im(q). If q is not real (a * 0), the above formula implies a
(compare
the real
2
+ ν a
and imaginary
2
_
parts
2 a 1 separately).
After
simplification,
this is equivalent to
But a
2
is positive,
while ν
2
is not.
Hence,
in contrast to the assumption,
q is real, and U = IR. On the other hand, if dim U is 2 or 3, there are nonzero purely imaginary quaternions in U. The square of such an element is, as we saw b e f o r e ,
real
and negative. We are t h e r e f o r e allowed to replace the given quaternion by a 2 suitable real multiple ι with the additional constraint ι = -1. If the dimension of U equals 2, we have case (3) of theorem and are done. Thus, only the possibility dim U = 3 must be discussed. We determine an arbitrary element κ * 0 in U which is orthogonal to 1 and t. Then iK is also perpendicular to 1 and i . This brings about that LK and 2 (LK)
are
linearly
dependent,
cause of IK l 1, however,
which
is only conceivable
this leads to LK = 0,
for
contradiction.
real
LK.
Be-
10.2. Left and Right Multiplication
359
10.2. Left and Right Multiplication
T h e p r o d u c t of t w o
quaternions
q
=
a + b i + c j
+ dk
u
=
a + ß i + y j
+ Sk
and
is (see Tab.
qu
10. 1.1)
=
( a a - b ß - c y - d S )
l + ( a ß + b a + c S - d y )
i
+ ( a y - b S + c a + d ß ) j + ( a ö + b y - c ß + d a ) k
The m u l t i p l i c a t i o n with a q u a t e r n i o n u * 0 f r o m the l e f t or f r o m t h e i n d u c e s a l i n e a r t r a n s f o r m a t i o n of IH. We d e s c r i b e i t w i t h t h e h e l p of m a t r i c e s L(u) a n d R(u),
respectively.
If w e ( a s we s h a l l do t h r o u g h o u t
book) i n t e r p r e t q u a t e r n i o n s as r o w v e c t o r s ,
q
=
i. e.
set
(a,b,c,d)
we i m m e d i a t e l y g e t f r o m t h e m u l t i p l i c a t i o n
table
α
β
y
δ
α
δ
-y
-y -δ
α
β
y -β
a
-β
L
=
L(u)
=
L(a,β,y,δ)
=
-δ
and
right the this
360
10. Quaternions
α
R(u)
=
ϋ(α,β,γ,δ)
β
γ
-β
α -δ
-7
δ
δ r
=
-δ
-jr
α
-β
β
α
The associativity of ΙΗ implies several useful relations;
among others,
we
derive f r o m (uq)u' = u ( q u ' ) the commutativity of l e f t and right multiplications: L(u) R ( u ' ) and vice versa.
This identity
=
R(u') L(u)
can as well
be obtained by direct
calcula-
tion, but this much more tiresome and not obvious at all. Two more important relations can also be considered as direct consequences o f ' the associative law, now written in either of the f o r m s q ( u u ' ) = (qu)u' or (uu')q = u ( u ' q ) .
They are
R(uu')
=
R(u) R ( u ' )
and L(uu':
L(u')
L(u)
One has to be careful with the order of the f a c t o r s ; since we use row vectors, we must multiply f r o m l e f t to right, when we apply the matrix notation.
Clearly,
all
equations
could
(except
for
the
technical
difficulties
caused by the clumsiness of vertical arrays and the waste of space) equally well be written as column vectors. Then we would get the transposed m a t r i ces and equations.
For the reasons expounded in chapter 2, we stick to the
row representation. The determinants of L ( q ) and R(q) are det L ( q )
=
det R(q)
=
2
They d i f f e r f r o m zero whenever q * 0. The last statement is obvious because the invertibility of q implies that of L ( q ) and R(q). The multiplication matrices associated with a real quaternion α are identi-
10.2. L e f t and Right M u l t i p l i c a t i o n
361
cal w i t h t h e s c a l a r m a t r i x a. From t h i s and t h e p r o d u c t f o r m u l a , we d e d u c e the orthogonality
conditions L(q) L(q)
=
L(q) L(q)
=
-l
R(q) R(q)
=
R(q) R(q)
=
-l
a s well a s
In p a r t i c u l a r ,
t h i s gives L ( q ) , R ( q ) e SO^tlR), if q h a s t h e n o r m 1.
362
10. Quaternions
10.3. Rotations of the Quaternion Algebra The norm in Η is by section 10.1 Euclidean with 1, i, j , k as an o r t h o normal basis.
The makes it possible to characterize the 2 - and 3-spheres
very elegantly with the help of
quaternions.
To this end, we only need to compile some results of the f o r e g o i n g considerations:
Theorem 10.3.1: ( a ) The standard-3-sphere in IH s R4 is the set
I q e IH
= 1 J
of all quaternions with norm 1. ( b ) The subset
q € IH
Re(q) = 0, = 1
of the pure quaternions in !f
Μ
q 6 and the 4
associated covering
2 a S 0
We shall
see
later
that
4
(IR)
this result
ss
S
3
Χ
S
3
can be generalized
considerably:
all
SO (IR) with η ϊ 3 possess double covers, the spin groups. We shall have to π deal with them and their elements, the spinors, much more thoroughly in chapter 13. The special f o r m of the extension provided by theorem 10.3.3 and its de3
scription with the sphere, mensions,
however,
is limited to IR
nothing like that remains true.
peculiar nature of these two spaces.
4
and IR . In higher
di-
This emphasizes once again the
10.4. R e p r e s e n t a t i o n by Complex M a t r i c e s
367
10.4. Representation b y Complex Matrices
The n a t u r a l i s o m o r p h i s m of C w i t h t h e s u b a l g e b r a R·1 ® R·1
c m
s p a n n e d by 1 and i s u g g e s t s t o w r i t e t h e t y p i c a l q
=
a + b l + c j
quaternion
+ dk
(a, b , c , d e R) in t h e f o r m q
=
w + ζ j
with w =
a + bi
e C
z
c + di
e C
=
Since j w =
j (a+bl)
=
a j - b k
=
(a-bl) j
=
w j
however, the product formula (W+Zj) (w+zj)
=
(Ww + ZJzJ) + (WzJ + Zjw)
=
(Ww - Zz) + (Wz+ Zw) J
which i s v a l i d f o r a l l W,Z, w , z e C, c o n t a i n s t h e c o m p l e x c o n j u g a t i o n . map q i-» w + z j t h e r e f o r e is n o t a r e p r e s e n t a t i o n
of IH. It is
The
nevertheless
u s e f u l on a c c o u n t of i t s e x t r e m e l y s i m p l e a r i t h m e t i c . F o r e x a m p l e , t h e c o n j u g a t e of q is q
=
a - b l - c j - d k
=
(w+zj)
=
w - j z
=
w - z j
368
10. Quaternions
The norm of q reduces to
q q
=
(w+zj) (w+zj)
=
w w + ζ ζ
and is consequently identical with the sum of the norms of the two
(com-
p l e x ) constituents of q. A genuine representation of H as a complex matrix group is also easy to find. The Pauli matrices i and η e { 2 , 3 } . 2 4 2 Let us consider the case η = η = 2 f i r s t ! We obtain Ν = 2 η with η a 3. 1 2 3 3 The subgroup U g has index 2 in G and is, of course, cyclic. Every element of G outside l>3 is contained in a subgroup conjugate to either U^ or U^ and because of η
= n^ = 2 an involution. We choose two of them, f o r example a
and b, such that the product ab generates
Then G = is a dihedral
group. We are now l e f t with η
ι
= 2 and η = 3. The fundamental relation in this 2
case implies 12 η
»
=
W 3
This is admissible only f o r η = 3 , 4 , 5 . 3 The f i r s t alternative, (n ,n ,n ) = ( 2 , 3 , 3 ) , 12
contains precisely — = 6
12
12
yields
Ν = 12.
The
t w o f o l d and — + — - 8 t h r e e f o l d poles,
group
G
lying on
10.5. F i n i t e Groups o f
Quaternions
half as many axes (3 + 4).
375
G t h e r e f o r e possesses e x a c t l y 8 elements of
der 3, each of them f i x i n g only one axis and the two associated poles.
orThis
implies that the permutation representation of G on the 4 t h r e e f o l d axes is f a i t h f u l , and G isomorphic to a subgroup of the symmetric group S . The o r 4
der of G immediately leads to G s A^. For
(n^, η , n3> = ( 2 , 3 , 4 ) ,
the argument runs analogously.
i n j e c t i o n into the S4 acting on the 4 t h r e e f o l d axes. tion,
however,
We again f i n d an
In the present
situa-
we have Ν = 24 and G = S . 4
At last we come to more involved.
(η , n^, n3> = ( 2 , 3 , 5 ) ;
Ν = 60.
We have — = 30 t w o f o l d poles,
Now things are
i.e.
slightly
15 corresponding
axes
and just as many elements of order 2 in G. They are all c o n j u g a t e to each other,
since the same is true f o r the 30 poles. The c e n t r a l i z e r of an i n v o -
lution consequently
has order
4 and cannot
be anything
else
than a
2
group (isomorphic to 2 ), as there are no elements of order 4 in G,
four
because
2
the n (
are not divisible
by 4.
Each 2
contains
three p a i r w i s e
commuting
involutions. Hence,
in t o t a l ,
there are e x a c t l y 5 f o u r groups which are mutually c o n j u -
g a t e within G. (This assertion f o l l o w s easier f r o m Sylow's theorem, c f . renstein [1968], of
commuting
involutions
In the
same manner
Go-
b e f o r e , 2 we conclude that the representation of G as a permutation group on the 2 or the t r i p l e s
Huppert [1967] etc. ).
is f a i t h f u l .
This
as
induces an
isomor-
phism of G with a subgroup of order 60 in S s> which can only be A s We have h i t h e r t o shown that the groups compiled in the theorem are the only conceivable
ones.
The
existence
of
finite
subgroups
of
SOgtR) and their uniqueness up to conjugacy is also easily For the c y c l i c groups, hedral group D
these
types
we have proved this already in theorem 10. 5. 1. A d i -
is obtained f r o m the cyclic group U
2n a x i s perpendicular to the axis of U . π The positioning of the U - p o l e s is a r b i t r a r y η f o r the o r t h o g o n a l i t y condition,
by adding a t w o f o l d
π and the second
can also be chosen at w i l l .
axis,
except
All such c o n -
f i g u r a t i o n s are equivalent to each other under the action of the e n t i r e t a t i o n group S0 3 (IR),
in
demonstrated.
SO, f o r each n,
there is essentially only one
ro-
possibi-
lity. The t e t r a h e d r a l group G s A^ has 4 t h r e e f o l d poles a, b, c, d, say, which t o g e t h e r f o r m an o r b i t of a = (0,0,1).
G.
We introduce Cartesian coordinates in IR3 with
The sum a + b + c + d then is G - i n v a r i a n t .
If
it w e r e
diffe-
r e n t f r o m 0, G would act i n j e c t i v e l y on the 2-dimensional orthogonal space.
Due to theorem 10. 5. 1, this cannot happen since G is not c y c l i c .
sub-
376
10. Quaternions
The 120"-potation about the axis through a permutes the points b, c, d. this and the o b s e r v a t i o n that we may bring b by an a p p r o p r i a t e t r a n s f o r m a t i o n to the f o r m ( 0 , ? , ? ) ,
From
orthogonal
uniqueness up to the action of
SC^tR)
is immediate. The S - and A - g r o u p s can be t r e a t e d by r e c u r r i n g to the A 4
5
contained
in
4
each of them. Since the l a t t e r can be oriented in space only in one way, except f o r a r o t a t i o n , the analogous conclusion f o r the l a r g e r groups is c l e a r l y true as w e l l .
2
While the c l a s s i f i c a t i o n of
the f i n i t e r o t a t i o n groups in the plane R does 3 not present any problems and R as the next case - as seen - is also quite easy to handle, sion. ral
the d i f f i c u l t i e s g r o w
very quickly with
increasing
dimen-
Even with r e f i n e d group t h e o r e t i c a l methods one soon reaches a natu-
limit.
This is mainly caused by the already in spaces of moderately
large
sion immense v a r i e t y of groups arising through the procedure.
For this r e a -
dimen-
son, only some s c a t t e r e d results concerning a f e w of the spaces in the s e 4
5
quence R ,R , . . .
have been published,
which we shall not describe,
however,
since w e have no applications f o r them. Clearly, arbitrary
t h e r e is no hope at all dimensions,
presentations, momorphisms
e.g. into
tensor products,
to c l a s s i f y all
finite rotation
since every f i n i t e group has f a i t h f u l orthogonal
the permutation representations as w e l l suitable
etc.
groups
SOη(R)
constructed
by
reduction,
as f u r t h e r direct
of reho-
sums,
The problem under discussion thus would as a f i r s t
step r e q u i r e an e x p l i c i t table of all f i n i t e groups. But this is completely out of the question:
even the groups of prime
power
o r d e r p f have w i d e l y varying structures already if the exponent f is of
the
o r d e r of 10. Much more i n t e r e s t i n g to us is a c o r o l l a r y of theorem 10.5.2. The q u a t e r n i onic m u l t i p l i c a t i o n is a s s o c i a t i v e ;
hence the i n v e r t i b l e elements of
H form
a group H x . I t s f i n i t e subgroups can be c l a s s i f i e d easily with our methods. The bulk of the work has indeed already been done.
Theorem 10.5.3: ( a ) Every f i n i t e subgroup G of IHX = H\{0} is isomorphic to one of the lowing:
fol-
10.5. F i n i t e Groups of
Quaternions
377
(1) a c y c l i c group of a r b i t r a r y order Ν € IN; ( 2 ) a double cover of a dihedral group D
2n
(3) the Schur extension 2AA
4
;
of the t e t r a h e d r a l
group;
( 4 ) the Schur extension 2aS^ of the octahedral group; (5) the Schur extension 2AA g of the icosahedral In all of
these cases,
the c y c l i c normal
or
group.
subgroup of
order
2 consists
of the t w o quaternions 1 and - 1. 3 ( b ) G l i e s in the standard-3-sphere if . ( c ) Isomorphic f i n i t e subgroups of Hx are c o n j u g a t e .
Remark: The G mentioned in part tetra-,
octa-,
( a ) under
(2), . . . . (5) are also called binary
and icosahedral groups,
respectively.
are known about them, f o r example representations, rings,
Many important presentations,
di-,
details
invariant
etc. ; see C o x e t e r & Moser [1965].
Proof: If
u e G is of
consequently tion
with
order n,
|u| = 1 ,
we obtain f o r its norm
that is proposition (b).
norm-l-quaternions
induces
a
|u|n = |un| = |l| = 1 and
By theorem 10.3.2,
representation
of
the
conjugastandard
sphere S = f 3 with kernel {±1} and image SO^ilR). We use the symbol " - " to indicate this.
Then G is a f i n i t e subgroup of S s SO^flR) and t h e r e f o r e c o n -
tained in the l i s t given in theorem 10.5.2.
Beyond this,
we have shown in 2
section
10.3 that ± 1 are the only solutions of
the equation
q
= 1 in H.
This implies that there is only one involution in (H, namely - 1. Every quaternion of the even f i n i t e order 2n thus obeys u° = - 1, which in turn y i e l d s un = ( - 1 )
= 1 and o r d ( u ) = n.
We discuss the single p o s s i b i l i t i e s f o r the image group one a f t e r the er.
If
G = is c y c l i c
of
even o r d e r ,
then # ( G )
is also
divisible
othby 2,
and G contains an involution which by what has been said above must be - 1. The o r d e r of a preimage u of u is then t w i c e as l a r g e as that of u i t s e l f . This immediately leads to G = . If G is c y c l i c of odd o r d e r ,
we must distinguish between # ( G ) even or odd.
In the f o r m e r case the g e n e r a t o r u has t w o inverse images u and - u, one of which has o r d e r n. The other,
in c o n t r a s t ,
say,
has t w i c e that o r d e r and
g e n e r a t e s G. If the order of G is odd, the kernel of the c o n j u g a t i o n r e p r e -
378
10. Q u a t e r n i o n s
s e n t a t i o n G —> G i s t r i v i a l ,
i. e. G ss G.
Anyway, t h e c y c l i c i t y of G i m p l i e s t h e same p r o p e r t y of G. There
remain
the groups
listed
under
(2), (3), (4), (5) in t h e
compilation,
n a m e l y3 D , A , S , and A , a s c a n d i d a t e s f o r G. 2n 4 ^ 4 S In t h e s e c a s e s G and c o n s e q u e n t l y also G have even o r d e r s .
The k e r n e l of
t h e h o m o m o r p h i s m G —» G now is {1, -1} = 2, and G a d o u b l e cover of G. To finish the proof,
we m e r e l y have t o v e r i f y t h a t t h e e x t e n s i o n i s of
Schur
t y p e f o r G e {A4> S 4> Ag}. All t h r e e of t h e s e g r o u p s c o n t a i n a s u b g r o u p U i s o m o r p h i c t o t h e Klein f o u r 2
g r o u p 2 . T h a n k s t o t h e o r e m 1 0 . 5 . 2 we can, by c o n j u g a t i n g w i t h a s u i t a b l e invertible quaternion,
choose t h e l a t t e r such t h a t i t c o n t a i n s ,
besides
the
identity, the three 180°-rotations around the coordinate axes. The i n v e r s e image of Ü t h e n is U = {±1, ±i, ± j , ±k}, [ 1, J ]
=
I"1 j " 1 i j
=
(-i) (-j) i j
=
and (Ij)2
=
k2
l i e s in t h e c o m m u t a t o r g r o u p U' of U, and hence a l s o in G ' .
=
- 1
Since - 1 is
t h e only i n v o l u t i o n in G, it m u s t e v i d e n t l y be a l s o c o n t a i n e d in t h e
centre
Z(G). T h i s s h o w s t h a t G is a S c h u r e x t e n s i o n of ( - 1 ) - 2 w i t h G, a s s t a t e d in (a). The only c l a i m n o t y e t e s t a b l i s h e d ,
(c),
follows directly from the
analo-
g o u s p r o p o s i t i o n in t h e o r e m 10. 3. 2 by going back t o t h e p r e i m a g e u n d e r o u r representation.
Several
a p p l i c a t i o n s of t h i s r e s u l t
s h a l l play an i m p o r t a n t
r o l e when
we
i n v e s t i g a t e e f f i c i e n t n u m e r i c a l i n t e g r a t i o n m e t h o d s on t h e 2 - s p h e r e in p a r t F of t h i s book.
11.1. Doubling Method of Cayley and Dickson
379
11. Octaves 11.1. Doubling Method of Cayley and Dickson The geometric relevance of the complex numbers and the quaternions, 2
cially with regard to the parameterizations of S
espe-
3
and S , is caused
basi-
cally by the existence of a multiplicative norm (or modulus) and the associated Euclidean scalar product. For this reason, we want to investigate both of them more thoroughly. particular
interest
is the question whether similar
Of
structures can be de-
fined also in other algebras. To begin with, we formulate the problem more precisely,
at the same introducing some notation:
Definition 11.1.1: Let U * 0 be an algebra over the ground f i e l d K. ( a ) A norm on 11 is a positive definite quadratic f o r m N: 11 —> K. The induced scalar product is, as always, defined via
Clearly,
=
this is possible
not essential,
| j^N(x+y) - N ( x )
only if
-
N(y)j
char(OC) * 2,
but this restriction
is
since we shall later consider only the f i e l d s DC e {R, C}.
In particular,
N(x)
=
( b ) The conjugation associated with Ν is the K-linear map κ: 11 —» 11, which leaves 1 invariant and in the orthogonal complement
l1
=
{ χ € 11
< x , l > = 0 ]·
of 1 corresponds to the multiplication by - 1. The image of χ under κ is
380
11. Octaves
χ
=
κ(χ)
=
2 ·1 - χ
( c ) The n o r m is c a l l e d multiplicative (and V a normed algebra) if it is p o s i t i v e d e f i n i t e , and f o r a l l a, b e U t h e e q u a t i o n
=
holds true.
We l i s t a f e w of t h e m o s t i m p o r t a n t p r o p e r t i e s of t h e c o n j u g a t i o n :
Theorem 11.1.1: Let κ be a c o n j u g a t i o n mapping on some a l g e b r a o v e r K. 2 = (1) κ is i n v o l u t o r y : κ = id, and hence χ = x; (2) The f i x e d s p a c e of κ c o n s i s t s of t h e m u l t i p l e s of 1: x=x
β
xeK-1
(3) Each a l g e b r a e l e m e n t c o m m u t e s w i t h i t s c o n j u g a t e : — χ χ =
— χ x
=
2 χ - χ
2
Proof: Check t h e d e f i n i t i o n s .
It is o b v i o u s t h a t t h e c o n j u g a t i o n s and n o r m s in C and H f u l f i l l t h e s e m a n d s and,
beyond t h i s ,
are multiplicative.
The l a t t e r p r o p e r t y is
de-
excee-
dingly r a r e and a l m o s t c h a r a c t e r i z e s t h e IR-algebras known t o us, n a m e l y IR, C, and IH; e x c e p t f o r t h e s e t h r e e ,
t h e r e is only one f u r t h e r
finite-dimen-
s i o n a l n o r m e d a l g e b r a w i t h u n i t e l e m e n t over IR. Our n e x t g o a l w i l l be t o c o n s t r u c t t h i s o c t a v e a l g e b r a 0 = 0(IR).
Its
is d e r i v e d f r o m t h e f a c t t h a t dim (0) = 8. We u t i l i z e a p r o c e d u r e IK by Cayley and Dickson, t h e s o - c a l l e d duplication method:
proposed
name
381
11.1. Doubling Method of Cayley and Dickson
D e f i n i t i o n 11.1.2:
(a) To any algebra II of dimension η over Κ with conjugation κ, norm N, and s c a l a r product we construct by the adjunction of some Ω the "dou2 U with underlying space tl ® U-Ω. The additional element
bled" algebra
Ω is orthogonal to all of 11 and subject to the normalization 2 = 1. The multiplication in
II is determined by the rules
(a Ω) b
=
(a b) Ω
a (b Π)
=
(b a) Ω
(a Ω) (b Ω)
=
- b a
valid f o r all a, b € B. When written in full, elements in 2I is (a +a Ω) (b +b Ω) 1 2 1 2
=
condition
a 1 b1 -
L
b~a 2 2
J
+
the product of two typical
L
b2 a1
+ a 2 b~ Ω 1 J
(with a r b i t r a r y3 a , a , b , b e U). 1 2 1 2 2 (b) We extend the conjugation to Ϊ by the formula a +a fi = 1 2
The set
a
- a Ω 1 2
2 U = U © β·Π obviously is an algebra of twice the dimension of V.
It is also easy to see by a simple calculation that the duplication of R by adjunction of Ω = i leads to C, from which in turn we get IH by repeating the procedure with j .
If we go one step further,
we arrive at
D e f i n i t i o n 11.1.3:
2 The octave algebra 0 = IH is obtained from the quaternion algebra IH by duplication.
We now follow common practise, call the additional element Ε r a t h e r than Π
382
11.
Octaves
a n d i n t r o d u c e t h e s y m b o l s I, J, Κ f o r t h e p r o d u c t s iE, j E , kE. T h e n w e h a v e
Theorem 11.1.2: The m u l t i p l i c a t i o n
r u l e s of
the basic octaves
1, i, j , k. Ε, I, J, K,
a n o r t h o n o r m a l s y s t e m w i t h r e s p e c t t o , a r e g i v e n in t a b l e
Tab.
1
J
k
Ε
I
J
Κ
1
-1
k
"j
I
-E
-K
J
J
-k
-1
i
J
Κ
-E
-I
k
J
-1
-1
Κ
-J
I
-E
Ε
-I
-J
-K
-1
i
j
k
I
Ε
-K
J
-1
-1
-k
j
J
Κ
Ε
-I
-j
k
-1
-1
-k
"j
i
-1
-J
In t h e n e x t f e w s e c t i o n s ,
I
Ε
we s h a l l c o n s i d e r s o m e of t h e m o s t
l a w s f u l f i l l e d by t h e o c t a v e s in g r e a t e r d e t a i l ,
11.1.1.
fundamental
b u t b e f o r e w e do s o ,
u s e f u l t o g e t r i d of t h e i n c o n v e n i e n c e of t h e C a y l e y - D i c k s o n b a s i s , p r o d u c t s m a y n o t be e a s y t o
form
11.1.1
1
Κ
which
it
is
whose
remember.
T h e s o l e o b v i o u s s y m m e t r y of t h e r e p r e s e n t a t i o n u s e d up t o n o w i s t h e triality a u t o m o r p h i s m τ w h i c h i n d u c e s s i m u l t a n e o u s c y c l i c p e r m u t a t i o n s of triples
(i,j,k)
and
(I,J,K)
in t h a t
order,
the
w h i l e f i x i n g Ε (and n a t u r a l l y
1
also). We g e t a much n i c e r b a s i s if w e r e n a m e a c c o r d i n g t o t a b l e 11.1. 2. T h e s u b -
11. 1. Doubling Method of Cayley and Dickson
383
s c r i p t s of the new b a s i s vectors are to be read mod 7 or, even better,
in-
terpreted as elements in F . The symbol e^ f o r the unit element is j u s t a name without deeper meaning; we shall not use the f a c t that the index set { 0 0 : 0 , 1 , 2 , 3 , 4 , 5 , 6 } is the proj e c t i v e l i n e P 1 (F 7 ) = F7u{oo}.
Tab. 11.1.2 1 e to
1
j
-e
-e
1
k -e
4
2
Ε
I
-e
-e
0
J 3
-e
Κ 5
-e
6
The arithmetic in 0 in the new nomenclature is governed by the following description:
Theorem 1 1 . 1 . 3 : (a) The eight elements e 00 = 1 and e L (L € F7 ) form an orthonormal b a s i s of 0 = 0(R). (b) The conjugation is the linear mapping from 0 onto itself which is determined by e00 = e CO and e L = - e L f o r L € F 7. The norm obeys '
=
xx
=
xx
e
IR-1
and the s c a l a r product is calculated via
=
^ x y
+ y x j
=
| | χ y + y χ j
(c) The products of any two basic elements are (L e F ): e e
2
2 = L
=
, = 1
e
- e 00
=
- 1
384
11. Octaves
g
e
6
L+l L+2
C L+2
— — g Q L+2 L+l
— Q L+4
g
— —
— g L+l
L+4
= —
G
q
e
L+4 L+l
q
L+4
L+l
L+2
6
=
L+4
g
L+2
Note t h a t 1 = I 2 , 2 = 3 2 , 4 = 2 2 a r e t h e s q u a r e s in F* = J^MO). All
formulas
above
are
invariant
under
multiplication
of
the
subscripts
w i t h one of t h e t h r e e s q u a r e s 1 , 2 , 4 a n d / o r a d d i t i o n of F ^ - e l e m e n t s . ther,
Toge-
t h e s e t r a n s f o r m a t i o n s g e n e r a t e a p e r m u t a t i o n g r o u p of o r d e r 21 (and
s t r u c t u r e 7: 3) which c o n t a i n s only a u t o m o r p h i s m s of 0. With t h e s e s y m m e t r i e s , identification
e
=1
00
t h e e n t i r e a l g e b r a can be r e c o n s t r u c t e d
and
a
single
nontrivial
pair
e -e = - e · e = e . 1 2 2 1 4 A s i m p l e r · d e s c r i p t i o n could s c a r c e l y be imagined.
of
products
from
the
like,
for
e x a m pr l e ,
The c o m p l e t e
multiplica-
t i o n r u l e s r e l a t i v e t o t h e e - b a s i s a r e a t t a c h e d as t a b l e 11.1.3 b e l o w .
Tab.
e e e e e e e e
00 0 1 2 3 4 s 6
e -e -e -e e -e e e
0 00
3 β 1 s 4 2
e e -e -e -e e -e e
1 3 00
4 0 2 6 Β
e e e -e -e -e e
-e
2 6 4
00
s 1 3 0
11.1.3
e -e e e -e -e -e e
3 1 0 s 00 6 2 4
e e -e e e -e -e -e
4 5 2 1 6 CO 0 3
e -e e -e e e -e
-e
s 4 β 3 2 0 00
1
e -e -e e -e e e
-e
6 2 s 0 4 3 1 00
11.1. Doubling Method of Cayley and Dickson
385
We c o l l e c t t h e t w o m o s t f u n d a m e n t a l a l g e b r a i c p r o p e r t i e s of t h e Cayley o c t a v e s in
Theorem 11.1.4:
The o c t a v e a l g e b r a 0 = O(IR) is n o r m e d and s t r i c t l y
alternative.
Proof:
Here,
"strictly
alternative"
means t h a t
t h e alternative law h o l d s , 2
c o n s i s t s of t w o p a r t s ,
which 2
namely t h e i d e n t i t i e s a ( a b ) = a b and ( a b ) b = a b ,
w h i l e t h e m o r e r e s t r i c t i v e associative law, i. e. a(bc) = ( a b ) c ,
is n o t
true
f o r a l l a, b, c e 0. If we d e v e l o p a and b w i t h r e s p e c t t o t h e b a s i s
{»e ,0e ,1e ,2 e ,3 e 4, e , e S, e 6} , we see i m m e d i a t e l y t h a t t h e a l t e r n a t i v i t y of 0 is t a n t a m o u n t w i t h t h e a s sertions (e e )e L Μ Ν
(e e )e = Μ L Η
e (e e ) + e (e e ) L Μ Ν Μ L Ν
and (e e )e + (e e )e LMM LNM
=
e (e e ) + e (e e ) LMN L NM
w h e r e t h e s u b s c r i p t s L, Μ, Ν r u n t h r o u g h p'tlF^). Whenever one of t h e m e q u a l s oo, i . e . ,
if 1 i s among t h e f a c t o r s ,
t h i s is t r i v i a l ;
c o n s i d e r t h e t r i p l e s L, Μ, Ν e IF . 7 Then e L > e M a n < l e „ a n t i c o m m u t e p a i r w i s e ,
we t h e r e f o r e need only
and by c o n j u g a t i n g and a f t e r w a r d s
i n t e r c h a n g i n g L «—» Ν b o t h r e l a t i o n s t r a n s f o r m i n t o each o t h e r . it
is s u f f i c i e n t to e s t a b l i s h
first.
the validity
of
only one of
them,
Therefore e. g.
the
Because of t h e c y c l i c a u t o m o r p h i s m e g —> e s + j and t h e i n v a r i a n c e of
a l l e q u a t i o n s in 0 u n d e r t h e s u b s t i t u t i o n e s —> e
, we may s u p p o s e in a d -
d i t i o n t h a t L = 0 and Μ is one of t h e n u m b e r s 0 , 1 , - 1 . t h e r e m a i n i n g 21 c a s e s w i t h t h e h e l p of
The v e r i f i c a t i o n of
t a b l e 11. 1.3 is l e f t t o t h e
dili-
gent r e a d e r . Thus,
t h e o c t a v e a l g e b r a 0 is a l t e r n a t i v e ;
duced f r o m
t h e n o n - a s s o c i a t i v i t y can be d e -
386
11. Octaves
k Κ =
(I J) Κ
I 1
I ( J K) and Ο is s t r i c t l y
Ε - Ε
alternative.
The u n i t e l e m e n t by d e f i n i t i o n 11. 1.2 is i n v a r i a n t u n d e r c o n j u g a t i o n , a l l l i n e a r c o m b i n a t i o n s of i, j , k, Ε, I, J, Κ change t h e i r s i g n s .
while
Consequently,
χ —» χ is a l i n e a r i n v o l u t i o n in 0. The p r o d u c t of t w o t y p i c a l o c t a v e s a = a +a Ε and b = b +b Ε y i e l d s ,
when
we a p p l y t h e c o n j u g a t i o n ,
which t u r n s o u t t o be i d e n t i c a l w i t h t h e p r o d u c t of t h e c o n j u g a t e s in r e verse order,
This i m p l i e s t h a t x x is r e a l .
A c o m b i n a t i o n of b o t h r e s u l t s w i t h t h e
thus
verified formula ab
b a
s h o w s t h a t χ —> χ indeed is t h e c o n j u g a t i o n which b e l o n g s t o t h e n o r m N(a) in t h e s e n s e of
d e f i n i t i o n 11. 1.1.
a a
The s c a l a r p r o d u c t
f o r e . We t h e n have t h e i d e n t i t i e s
and
a a
N(a +a E) 1 2
=
( a +a E) (a - a E) 1 2 1 2
N(b +b E) 1 2
=
(b +b E) (b - b E) 1 2 1 2
is e x p l a i n e d
as
be-
1 1 . 1 . D o u b l i n g Method of C a y l e y and Dickson
N(ab)
=
Tab - b a b a I 1 2 2 j [ l l
=
+ a 1 1 +
=
a
a b b a 1 2 2 1
I I +
is i d e n t i c a l
b b 1 1
- a b + a b + b a b a + a b [ 2Zj [ 1 2 1 2_] 2 1 2lJ
-
a b a b 1 1 2 2
-
b a b a 2 2 1 1
+
b
a a b 2 2 2 2
+
a b a b 1 2 2 1
+
b a b a 1 2 2 1
+
b
a a b 1 2 2 1
+ 1 2
a b a b 1 2 2 1
+
387
+ 2 1
b a b a 1 2 2 1
+ 2 2 -
1
a
1
b
2
a
b
2
-
b
2
a
2
b
1
a
1
with
N(a) N(b)
=
| < a , a > + [ 1 1 2 2 J
=
1 1
+ 1 2
K b , b > + . ι ι 2 2 J
+ 2 1
+ 2 2
s i n c e f o r a l l q u a t e r n i o n s a, b, c , d t h e e q u a t i o n
a b c d
holds.
To p r o v e t h i s ,
+ d c b a - a d c b - b c d a
=
0
we r e m a r k t h a t t h e l e f t - h a n d s i d e e q u a l s 2 X,
X
=
-
e
where
OM
For homogeneity reasons,
we may r e s t r i c t w i t h o u t loss to t h e s p e c i a l
vec-
tors
Of t h e
ele-
a, b, c, d e { l , i , j , k } .
4
= 256 w a y s t o c o m b i n e t h e b a s i s
m e n t s , h o w e v e r , t h a n k s t o t h e n u m e r o u s s y m m e t r i e s of X, o n l y v e r y f e w h a v e t o be e v a l u a t e d . As a l l s c a l a r p r o d u c t s l i e in R - l , abed
is
real.
This
simple
to essentially three cases,
b o t h t e r m s in X v a n i s h i n d i v i d u a l l y
observation
reduces
the
remaining
lest
possibilities
namely
(1) all f o u r v a r i a b l e s a r e d i f f e r e n t ; (2) a, b, c, d s p l i t u p i n t o t w o p a i r s of e q u a l e l e m e n t s ;
and
(3) a = b = c = d. If (1) o c c u r s ,
w e m a y n o r m a l i z e t o a = 1, b e c a u s e a s i m u l t a n e o u s
multipli-
c a t i o n of a , b, c, d w i t h a common f a c t o r i, j , k d o e s n o t c h a n g e t h e v a l u e of X. If w e t h e r e a f t e r a p p l y a s u i t a b l e p o w e r of t h e c y c l i c a u t o m o r p h i s m τ of H (triality),
i t i s even a l l o w e d t o s e t b = i in a d d i t i o n .
So w e n e e d
only
388
11. Octaves
consider
( a , b , c, d) = (1, i , j , k )
or
(l,i,k, j).
Both
quadruples
result
in
X = 0. In c a s e ( 2 ) ,
we may by the same argument assume a = 1 and,
t h a t the second pair equals i. ed,
as
there
are
( a , b , c , d ) = (1,1, i, i)
them again yield X = 0 . F i n a l l y ,
furthermore,
This gives t h r e e p o s s i b i l i t i e s to be or
(l,i,l,i)
or
a l t e r n a t i v e (3) is t r i v i a l .
(l,i,i,l).
checkAll
of
11.2. A l t e r n a t i v e D i v i s i o n Algebras
389
11.2. Alternative Division Algebras
The o u t s t a n d i n g p r o p e r t i e s of t h e Cayley-Dickson o c t a v e a l g e b r a c o n s t r u c t e d in
section
11. 1 a r e
the
nonexistence
of zero divisors
and
the
validity
of
the a l t e r n a t i v e law. We w a n t t o show n e x t t h a t , e x c e p t f o r t h e u n i n t e r e s t i n g O - d i m e n s i o n a l a l g e b r a , t h e e x a m p l e s known t o us, t h e a l g e b r a s IR,C,IH, 0 of t h e r e a l and c o m p l e x n u m b e r s , t h e q u a t e r n i o n s and t h e o c t a v e s , r e s p e c t i v e l y , ed by t h e a b o v e - m e n t i o n e d
can be c h a r a c t e r i z -
peculiarities.
To a c h i e v e t h i s aim, we u t i l i z e some s t a n d a r d m e t h o d s and t e r m s of m o d e r n a l g e b r a which we r e c a p i t u l a t e b r i e f l y . An identity in an a l g e b r a II is an e q u a t i o n which depends on c e r t a i n terminates a , b , c , . . . ,
say,
and y i e l d s a l w a y s t r u e s t a t e m e n t s if we
indeinsert
a r b i t r a r y e l e m e n t s of U f o r t h e s e v a r i a b l e s . S i m i l a r t o t h e conmutator [a, b]
=
a b - b a
of t w o e l e m e n t s a, b € U, d e f i n e d e a r l i e r ,
we i n t r o d u c e a r e l a t e d
expres-
sion, which in a c e r t a i n s e n s e m e a s u r e s "how l a r g e " t h e d e v i a t i o n s f r o m t h e associative law a r e .
Definition 11.2.1:
(a) The associator of t h r e e e l e m e n t s a, b, c of an a l g e b r a U is t h e m a n i f e s t ly t r i l i n e a r
expression [a, b, c ]
( a b) c - a (b c)
(b) A s u b s e t Μ £ 11 is c a l l e d associative if f o r a l l m , m , m € Μ t h e a s s o 1 2 3 c i a t o r [m , m , m ] v a n i s h e s . V 1 2 3
One of t h e m o s t f e r t i l e m e t h o d s t o deduce new a l g e b r a i c
identities
known ones is by polarization. It is a p p l i c a b l e w h e n e v e r t h e given
from
identity
390
11. Octaves
contains
a variable,
e. g.
a,
to
an order
higher
than the f i r s t .
In
this
case we have to replace the variable a by a linear combination of the general shape a + Xb, where b is a new indeterminate and λ an element of
the
underlying f i e l d DC. Then we develop this relation into powers of λ. If Κ contains " s u f f i c i e n t l y many" elements, such that the λ-power occurring in the derived
identity
(as functions on K) are linearly
independent,
this
implies that the c o e f f i c i e n t s of all λ 1 are zero. (The f i e l d s R and C, which shall be the only ones considered in the sequel, are large enough since f i n i t e l y many d i f f e r e n t power functions are always linearly independent over them). An example may serve to illustrate this. If in the R-algebra U the identity [a,a,a]
=
0
holds, we f i r s t get by the above prescription [a+Ab, a+Xb, a+Xb]
=
0
(λ e ER) and f r o m this the expansion 0
=
j
[a,a,a] + λ ^ [b,a,a] + [a,b,a] + [a,a,b]
+ λ 2 ^ [a, b, b] + [b, a, b] + [ b , b , a ] j + λ 3
[b,b,b]
Aside of the original equation [a, a, a] = 0 as the term independent of λ, we find the new identity [b, a, a ] + [ a , b , a ] + [ a , a , b ]
=
0
( c o e f f i c i e n t of λ ) as well as (belonging to λ ) the same with a β b and the identity [ b , b , b ] = 0, which is equivalent to the one we started with. In the
derived
relation,
a still
occurs to
second order.
It
is
therefore
possible to repeat the procedure. This leads to the identity [a, b , c ] + [a, c, b] + [ b , a , c ] + [ b , c , a ] + [ c , a , b ] + [ c , b , a ] We now come to the central topic of our investigations.
=
0
Many of the f o l -
lowing arguments are adapted f r o m Braun & Koecher [1966], where even consi-
11.2. A l t e r n a t i v e Division Algebras
391
d e r a b l y m o r e g e n e r a l r e s u l t s can be f o u n d a l s o .
Definition 11.2.2:
An a l g e b r a V i s alternative if t h e
[ a , a , b]
i s t r u e in II. More e x p l i c i t l y ,
identity
=
[ a , b, b ]
=
the two relations 2 a ( a b)
=
( a b) b
=
0
are
a b
and
ab
2
The a l t e r n a t i n g law h a s numerous consequences, f e w very i m p o r t a n t
of w h i c h w e o n l y c o m p i l e a
examples:
Theorem 11.2.1: L e t U be a n a l t e r n a t i v e a l g e b r a o v e r IR o r €. (a) The a s s o c i a t o r
of t h r e e e l e m e n t s in V d o e s n o t c h a n g e i t s v a l u e
cyclic permutations we a p p l y a n odd ( b ) In U, t h e
of t h e a r g u m e n t s ,
permutation.
identity
[a,b,a]
or,
while it is m u l t i p l i e d
=
0
equivalently,
( a b) a
holds.
=
a (b a)
An a l g e b r a f o r w h i c h t h i s is t r u e ,
i s c a l l e d flexible.
( c ) F o r a l l a , b, c e II, w e h a v e
[a,b,ca]
=
a
[a,b,c]
under
by - 1 if
11. Octaves
392
[ca,b,c]
=
[a,b,c] c
and [a,b2,c] ( d ) Furthermore,
=
[a,b,bc] + [ab,b,c]
the Moufang i d e n t i t i e s (a b a) c
=
a (b (a c ) )
c (a b a )
=
( ( c a) b) a
as well as (a b) (c a) are valid.
Here,
=
a (b c ) a
we used the f l e x i b i l i t y to save the brackets in
ex-
pressions like (ab)a = a(ba). ( e ) The subalgebra generated by any associative set Μ £ U is itself
associ-
ative. ( f ) The same is true f o r all subalgebras of U which are generated by any two elements.
Proof: Polarization of the defining relations [a, a, b] = 0 and [a, b, b] = 0 provides us with [a,c,b] + [c,a,b]
=
0
[a,b,c] + [a,c,b]
=
0
The associator thus goes over into its negative if we interchange the f i r s t two or the last two arguments. This implies (a). Statement (b) is then obtained by an application of ( a ) to one of the initial
identities.
The remaining relations under (c) are found through a short
calculation,
11.2. Alternative Division Algebras
393
d u r i n g which we make r e p e a t e d use of the a l r e a d y deduced s y m m e t r i e s of t h e associator. [a,b,ca]
The f i r s t equation f o l l o w s f r o m
= =
[ca,a,b] = ((ca)a)b - (ca)(ab) = (caZ)b - (ca)(ab) 2 2 2 ( c a )b - c ( a b) + c ( a ( a b ) ) - ( c a ) ( a b ) = [ c , a , b ] - [ c , a , a b ]
=
[a2,b,c] - [a,ab,c]
=
(a2b)c - a2(bc) - (a(ab))c + a((ab)c)
=
a((ab)c) - a(a(bc))
=
a
[a,b,c]
w h i l e the second is proved a n a l o g o u s l y [ca,b,c]
= = = =
[b,c,ca]
=
via
(bc)(ca) - b(c(ca)) 2
=
(bc)(ca) - b(c a)
2
2
( b c H c a ) - ( ( b c ) c ) a + (be )a - b ( c a ) = [ b , c , a ] - [ b c , c , a ] 2 2 2 [a,b,c ] - [a,bc,c] = (able - a(bc ) - (a(bc))c + a ( ( b c ) c ) ((ab)c)c - (a(bc))c = [a,b,c] c
The f l e x i b i l i t y implies beyond t h i s [a,bz,c]
=
(ab2)c - a(bzc)
=
( ( a b ) b ) c - (ab) (be) + ( a b M b c ) - a ( b ( b c ) )
=
((ab)b)c - a(b(bc))
=
[ab,b,c] + [a,b,bc]
w h i c h is the t h i r d of the r e l a t i o n s we need. The p r o o f of t h e Moufang i d e n t i t i e s is now e a s y .
In the f i r s t ,
we f o r m the
d i f f e r e n c e of both s i d e s and s i m p l i f y w i t h the help of the a l r e a d y
verified
s y m m e t r y p r o p e r t i e s of the a s s o c i a t o r : ((ab)a)c - a(b(ac))
Here,
=
((ab)a)c - (ab)(ac) + (ab)(ac) - a(b(ac)
=
[ab,a,c] + [a,b,ac]
=
- [b,c,a] a - [c,b,a] a
(c) has been employed.
c(a(ba)) - ((ca)b)a
and a t l a s t
=
- [ab,c,a] =
[ac,b,a]
0
We g e t the second r e l a t i o n
similarly:
=
c(a(ba)) - (ca)(ba) + (ca)(ba) -
=
- [c,a,ba] - [ca,b,a]
=
[a,c,ba] + [a,b,ca]
((ca)b)a
=
a [a,c,b] + a [a,b,c]
=
0
394
11. Octaves
(ab)(ca) - a((bc)a)
=
(ab)(ca) - a(b(ca)) + a(b(ca)) - a((bc)a)
=
[a,b,ca] - a [b,c,a]
= Ο
again utilizing (c). This is part (d) of the theorem. To establish (e), we start out from the four identity [ab, c, d] - [a, be, d] + [ a , b , c d ]
=
[a,b,c] d + a [b,c,d]
which is true in every algebra and can be corroborated without d i f f i c u l t y by direct calculation. Polarization of the relations in (c) shows that an associator which depends linearly on four elements changes its sign if we permute any two of the a r guments. This implies, among others, - [a,bc,d]
=
[bc,a,d]
=
- [cb,a,d]
=
[ab,c,d]
and analogously [a,b,cd]
=
[cd,a,b]
=
- [ad,c,b]
=
[ab,c,d]
The three terms on the left-hand side of the four identity are therefore all equal. We obtain the new identity 3 [ab,c,d]
=
[a,b,c] d + a[b,c,d]
Restricting now to a, b, c, d e M, the associators [ a , b , c ] and [ b , c , d ] vanish, since Μ by definition is associative. There only remains [ab,c,d]
=
0
and the cyclically permuted equations. Together, they say that Μ υ (ab) is associative as well. In conclusion, we can add to an associative set the products of two arbitrary
elements without
losing associativity.
This implies
(e)
and -
since
all two-element sets are obviously associative - also ( f ) , and the proof the theorem is complete.
of
11.2. A l t e r n a t i v e D i v i s i o n Algebras
By t h e way,
395
t h e r e v e r s a l of ( f ) is a l s o c o r r e c t :
An R - a l g e b r a is
alterna-
t i v e if and only if a l l s u b a l g e b r a s g e n e r a t e d by t w o e l e m e n t s a, b e U a r e associative.
This is a t r i v i a l c o n s e q u e n c e of t h e f a c t t h a t t h e
alternative
l a w s depend on no m o r e t h a n t w o a r g u m e n t s . We f u r t h e r m o r e need
Definition 11.2.3:
(a) If
the
elements
a, b e 11\{0)
a b = 0, w e c a l l t h e m
of
an
algebra
II f u l f i l l
(left a n d right) zero divisors.
the
Clearly,
condition 11 h a s
z e r o d i v i s o r s if and only if ab = 0 implies a t l e a s t one of t h e
no
equa-
t i o n s a = 0 o r b = 0. (b) II is a division algebra if f o r a l l a, b e 11 w i t h a * 0, t h e e q u a t i o n s χ a
=
b
a y
=
b
and
have u n i q u e s o l u t i o n s χ and y,
respectively.
F o r t h e i n t e n d e d c l a s s i f i c a t i o n of t h e a l g e b r a s IR, C,H, 0 t h r o u g h t h e i r
es-
sential properties,
re-
n o n e x i s t e n c e of z e r o d i v i s o r s and a l t e r n a t i v i t y ,
quire another elementary information,
we
namely
T h e o r e m 11.2.2:
A d i v i s i o n a l g e b r a does n o t c o n t a i n z e r o d i v i s o r s .
If t h e dimension is f i -
n i t e , t h e r e v e r s e s t a t e m e n t is a l s o t r u e .
Proof:
The n o n e x i s t e n c e of z e r o d i v i s o r s in division a l g e b r a s is t r i v i a l , f o r a * 0, t h e e q u a t i o n ab = 0 h a s m e r e l y one s o l u t i o n ,
because
which is c l e a r l y 0. £
Vice v e r s a ,
a s s u m e tl h a s no z e r o d i v i s o r s and choose a e U = U\.
Left
and r i g h t m u l t i p l i c a t i o n s w i t h a a r e t h e n 1 i n e a r maps of V i n t o i t s e l f
with
396
11.
trivial kernels.
Octaves
From the homomorphism theorem we deduce t h a t the
images
a r e i s o m o r p h i c t o a n d t h u s ( f o r d i m e n s i o n a l r e a s o n s ) i d e n t i c a l w i t h II.
Be a w a r e t h a t t h e l a s t s t a t e m e n t i s i n d e p e n d e n t of t h e u n d e r l y i n g f i e l d ! We n o w h a v e t h e m e a n s a t o u r d i s p o s a l t o f o r m u l a t e t h e c e n t r a l r e s u l t this
of
section:
Theorem 11.2.3: Let U be an a l t e r n a t i v e
d i v i s i o n a l g e b r a of
IR. T h e n V i s i s o m o r p h i c t o
either
( 1 ) t h e f i e l d R of t h e r e a l
numbers,
( 2 ) t h e f i e l d C of t h e c o m p l e x
positive f i n i t e dimension
over
numbers,
( 3 ) t h e n o n c o m m u t a t i v e s k e w f i e l d IH of t h e q u a t e r n i o n s ( 4 ) t h e a l t e r n a t i v e r i n g 0 of t h e o c t a v e s ,
or
neither commutative nor
associa-
tive.
Proof: We s e t η = dim II a n d s e l e c t a n a r b i t r a r y n o n z e r o e l e m e n t a e U. T h e t i o n e of t h e
solu-
equation
e a
a
i s u n i q u e l y f i x e d . By t h e a l t e r n a t i v e l a w ,
we g e t
e (e a )
e a
2 and e is i d e m p o t e n t :
e
= e.
Moreover,
o b v i o u s l y e * 0. F o r a l l b 6 11 w e
also have
and
2 be'
(b e ) e
=
e(eb)
2 = e b
b e
similarly
=
eb
11.2. A l t e r n a t i v e D i v i s i o n Algebras
After cancellation,
397
the two last relations reduce to b e
=
e b
=
b
Hence U c o n t a i n s ( e x a c t l y ) one u n i t e l e m e n t , namely e, which we f r o m now on call
1.
It c o m m u t e s w i t h a l l e l e m e n t s of 11. The same h o l d s f o r a l l
real
m u l t i p l e s of 1, which f o r m a o n e - d i m e n s i o n a l s u b a l g e b r a IR-1 i s o m o r p h i c t o R. As u s u a l , we i d e n t i f y it w i t h t h e f i e l d IR. If η = 1, we have U = (R and a r e done. We may t h e r e f o r e a s s u m e w i t h o u t r e s t r i c t i o n t h a t η £ 2. Every n o n r e a l χ e 1I\R g e n e r a t e s by t h e o r e m 11.2.1, p a r t (e) or ( f ) , an a s s o c i a t i v e s u b a l g e b r a of U: t h e p o w e r s χ
do n o t depend on t h e o r d e r of m u l -
t i p l i c a t i o n (k e IN). In o t h e r w o r d s : ΐ is power-associative. The (n+1) v e c t o r s
Ι,χ,χ2
χ" are
linearly
dependent.
Thus,
there
is
a
n o t i d e n t i c a l l y v a n i s h i n g p o l y n o m i a l Ρ of d e g r e e s η w i t h P(x) = 0. Ρ has real
orem 6 . 5 . 5 ;
c o e f f i c i e n t s a n d d u e t o t h e fundamental theorem of a l g e b r a
see Gauß [1799]),
(the-
it s p l i t s into i r r e d u c i b l e p o l y n o m i a l s of
de-
g r e e 1 o r 2. Since Μ h a s no z e r o d i v i s o r s , χ must be a z e r o of one of t h e s e f a c t o r s . The d e g r e e of t h i s p o l y n o m i a l is 2, s i n c e we a s s u m e d χ t IR, and χ does n o t obey a l i n e a r e q u a t i o n w i t h c o e f f i c i e n t s in IR. We t h u s have a r e l a t i o n of t h e f o r m χ* + Μ χ + Ν =
0
w i t h s u i t a b l e x - d e p e n d e n t r e a l n u m b e r s Μ = M(x) and Ν = N(x). The irreducibility of
t h i s minimal
polynomial
of
χ
is e q u i v a l e n t
to
its
discriminant being n e g a t i v e : M2 - 4 Ν
,
where Μ+ 2 χ 1
In p a r t i c u l a r ,
U = IR -1 © [R-x = R-l © IR - i = 11^,
and
after
having
inserted
t h e d e f i n i t i o n i n t o t h e minimal e q u a t i o n of x, we o b t a i n
and U = U 3 (C. χ If η = 2, we even have U ξ €, and t h e s t a t e m e n t of t h e t h e o r e m is t r u e . We t h e r e f o r e a s s u m e η > 2 and f i x t h e e l e m e n t i. Then tt is a p r o p e r
subalge-
b r a of U i s o m o r p h i c t o t h e a l g e b r a C of complex n u m b e r s : IR ® IR · i c II. We c o n s i d e r an a r b i t r a r y y 6 II \ 1 . Lying o u t side U , it c a n n o t be r e a l , and by t h e p r o c e d u r e we a l r e a d y a p p l i e d b e f o r e t o c o n s t r u c t i f r o m x,
we
g e t a new e l e m e n t t e U = IR-1 © R-y obeying
The s u b a l g e b r a g e n e r a t e d by χ and y (or, w h a t a m o u n t s t o t h e same t h i n g , by i and t ) ,
w i l l be c a l l e d 11. Due t o t h e o r e m 11.2.1, U is a s s o c i a t i v e .
The minimal p o l y n o m i a l s of i ± t a r e q u a d r a t i c : ( 1 + ι ) 2 + Μ+· t i + t ) + Ν
0
(i-t)2 + Μ ·(i-t) + Ν
0
w h e r e Μ+ = M(i+t), Μ- = M ( i - i ) , Ν+ = N ( i + i ) , Ν With t h e a b b r e v i a t i o n
= N(i-i) are real numbers.
11.2. A l t e r n a t i v e D i v i s i o n Algebras
λ
399
=
i ι +
i
L
we d e r i v e ( - 2 + λ + Ν+ ) + M + i + M + ι
=
0
( - 2 - λ + Ν ) +M
=
0
i - M
ι
Summing up y i e l d s ( - 4 + N+ + N ) + (M+ + M ) i
+ (M+ - Μ ) ι
= 0
The t r i p l e 1, i, ι is l i n e a r l y i n d e p e n d e n t over R; hence M+ and Μ zero.
must
be
The d i f f e r e n c e of t h e two minimal e q u a t i o n s of i ± ι s h o w s t h a t 2 λ
=
Ν - Ν
+
and A € R. We d e f i n e a new e l e m e n t j t h r o u g h
j
Ai+2t — —
=
getting J2
=
- 1
and i j + j i
=
0
If we now d e f i n e in a d d i t i o n k = i j and t a k e t h e a s s o c i a t i v i t y account,
of U i n t o
t h e n a l l a r i t h m e t i c r u l e s of t h e q u a t e r n i o n a l g e b r a a r e s a t i s f i e d ,
as the examples k2
= j k
(ij)(lj) =
j(ij)
= =
i(ji)j
=
- i(ij)j
(Ji)j
=
- (ij)j
= =
- iZ jZ - i j2
= =
- 1 i
400
11. Octaves
indicate. U is thus a homomorphic image of IH. The conjugation χ —> χ corresponds to that in IH. This implies the relation
x+y which clearly holds as well f o r
=
χ + y
any pair
( x , y ) of
elements in U (the
as-
sumption about χ and y, namely that they are not both contained in a subalgebra
isomorphic
to C,
is irrelevant,
since the conclusion
evidently
re-
mains true f o r IR c IH and C c IH). For every λ e R, we moreover find
λχ
λ χ
such that the conjugation has been shown to be IR-linear. Except f o r this,
it is involutory:
χ = χ and has the (precise) f i x e d space
IR. The norm thus becomes a positive definite quadratic f o r m on U whose associated scalar product we, as usual, denote by
=
i
N(x+y) - N(x) - N ( y )
A direct v e r i f i c a t i o n proves that the elements l , i , j , k all have norm 1, are mutually -orthogonal,
and linearly independent.
T h e r e f o r e , ft is 4 - d i -
mensional and isomorphic to IH. In case η = 4 again the theorem is valid, since then II = Μ. So, f r o m now on, we suppose η > 4 and choose a norm-l-element Ε which is perpendicular to 1, i , j , k . names I , J , K .
Then,
For the products iE, jE, kE we introduce the alias
by our prior results,
(k, E , K ) obey the same rules as ( i , j , k ) ,
the triples ( i , E , I);
(j,E,J);
and
in the correct order.
The remaining products of two basis elements are easily found by the Moufang identities,
e.g.
f o l l o w i n g the prescription below:
If both f a c t o r s are in { I , J , K } , J Κ
=
(jE)(kE)
The products,
=
- (Ej)(kE)
we calculate ä la =
- E(jk)E
=
- E i E
in which one f a c t o r is equal to E,
=
- E I
=
- i
while the other is not,
have been treated already. A typical example of the f e w formulas which are still missing is
11.2. A l t e r n a t i v e D i v i s i o n Algebras
J Κ
=
(ΚΙ)Κ
=
401
- (ΙΚ)Κ
=
- I Κ2
=
I
The only case not yet considered is EZ
=
(Ii Η Ii)
Putting things together,
=
- (IiHil)
=
- I i2I
=
I2
=
- 1
we have seen that {1, i, j, k, Ε, I, J, K} is an
ortho-
normal basis of a subalgebra in 11 isomorphic with 0, and particularly that η a 8. If η = 8, we get U = 0, and the theorem holds. The only possibility
left
is η > 8.
From what we said b e f o r e it
that Β contains an octave algebra.
follows
We p r e f e r to express this not in terms
of 1, i, j , k, I, J, K, but relative to the cyclic standard basis
According to section 11.1, the subscripts are taken f r o m the set P*(F )
=
7
F
7
U {oo>
Like in the last step, we seek an element which is orthogonal to all e^ and has the norm 1. Let us call it f . The products f
=
L
e -f L
00
(L 6 F ) complete a system
=
{eL,fL
j Lep'(F7l
}
of 16 pairwise orthogonal vectors of norm 1, whose squares are all equal to - 1, with the sole exception of e
2
00
= e
00
=1.
The 15 remaining e's and f ' s
anticommute with each other. Aside f r o m the alternative law, since
theorem
11.2.1
applies.
we have to s a t i s f y the Moufang identities, This
uniquely
determines
all
arithmetical
rules in the subalgebra Β £ U generated by 6. The reason is that f o r each pair L, Μ Ε
the product e^ e^ is contained in 0 and of the f o r m ε e^ with
suitable e = ± 1 and Ν s P 1 (F ). This yields
11. Octaves
402
f f
L M
=
(e f
=
- ε f
L
Me f ) CO Μ 03 00
e
f
Ν
=
- (f e He f ) Co L Μ C O
=
- f
= ε ί ( ί β ) = ε ί CO 00 Ν
CD
2
(e e )f L Μ C O
00
β CO
Ν
=
- ε β
Ν
and a l l o w s us to evaluate
f
Μ
e
Ν
=
- ε f
=
- ε f Ν
L
and
f
without ambiguity.
Finally,
e L H
the anticommutativity
implies a l l equations
of
the f o r m
e
The missing rules
follow
L
f
Μ
=
- ε f
Ν
f r o m the observation
by the v e r y construction of
the f
that
the t r i p l e s
(e , f L
are imaginary units in quaternion
bras which s a t i s f y the same conditions as ( i , j , k ) do (in this But the 16-dimensional a l g e b r a 8 is not a l t e r n a t i v e .
,f
00
L
)
alge-
order!).
To v e r i f y this,
we on-
ly need to compare the a s s o c i a t o r s [ e , e , f ] 0
1
2
=
( e e ) f - e ( e f ) = e f + e f 0 1 2
0 1 2
3 2
0 4
= f - f S
= 0
S
and
[e , f ,e ] 0
2 1
=
(e f )e 0 2 1
- e (f e ) 0 2 1
=
- f
6
e
1
- e
0
f
4
= f
S
+f
= 2 f
S
The sum of both does not vanish as it should do in an a l t e r n a t i v e where the identity
[ a , b, c ] + [a, c, b ]
is true.
=
[a,b+c,b+c] - [ a , b , b ] - [ a , c , c ]
=
0
5
algebra,
11.2. A l t e r n a t i v e D i v i s i o n Algebras
The p a r t
of
theorem
1 1 . 2 . 3 which is c o n c e r n e d w i t h a s s o c i a t i v e
403
algebras
d a t e s back t o F r o b e n i u s [1878], A s i m i l a r c h a r a c t e r i z a t i o n of a l g e b r a s w i t h a m u l t i p l i c a t i v e n o r m h a s been given by H u r w i t z . We s h a l l d i s c u s s it in t h e next section.
404
11.
Octaves
11.3. The Theorem of Hurwltz
The g e n e r a l i z e d t h e o r e m of F r o b e n i u s i n d i c a t e s t h e s p e c i a l p o s i t i o n of
the
a l g e b r a s R, C, H, 0. However,
s i n c e i t i s b a s e d on an a l g e b r a i c a l p r o p e r t y ,
tive law,
w h i l e o u r m a i n c o n c e r n is c o n n e c t e d w i t h t h e g e o m e t r i c a l
tions,
we shall discuss also a d i f f e r e n t ,
r i z a t i o n f o u n d by H u r w i t z in 1898,
namely the
though closely r e l a t e d
alternaapplicacharacte-
namely
Theorem 11.3.1: The only
finite-dimensional
normed
algebras
with
unit
elements
over
the
f i e l d of r e a l n u m b e r s a r e IR, C, IH, a n d 0.
Proof: L e t U b e an n - d i m e n s i o n a l
algebra over R which s a t i s f i e s the
w h e r e η e IN. We c o n t i n u e t o u s e o u r s t a n d a r d
nomenclature.
In II, t h e r e i s s o m e e l e m e n t a d i f f e r e n t f r o m 0 , norm
a n d t h e e x i s t e n c e of
implies
=
< a · 1, a · 1>
=
f r o m w h i c h a n d t h e d e f i n i t e n e s s of w e m a y d e d u c e t h a t
Polarizing the
=
1
identity
=
< a , a >
we o b t a i n
+
=
assumption,
< a , c > + < c , a >
the
11.3. The Theorem of Hurwitz
405
By the symmetry of the scalar product, this simplifies to
=
A second polarization provides
+
=
2
Another identity in U is (a b) b To prove it,
=
a
we decompose b = ß-1 + bQ with
< l.t> Q >
= 0 and assume f i r s t
that the corresponding claim with bQ in place of b is correct. (a b) b
=
(ab +ßa) TF+βϊΓ 0 Ο
=
Then
(ab )ΪΓ + ßaET + ßab + ß2a 0 0 0 0
while a
coinciding with (ab)b,
=
a + a β 2 o o
=
(ab ) b~ + a β 2 o o
because b
= -b . o o Thus, we have to establish the formula (ab)b = a only under the additional hypothesis = 0. We introduce the abbreviation = λ·1 with λ e R and calculate the norm of χ
=
( a b) b - λ a
It turns out to be
=
- 2 λ + λ2
The f i r s t term can be shortened considerably:
=
=
=
λ2
since II is normed by assumption. We transform the second summand with the help of the above identity.
Substituting
406
11. Octaves
(a,b,c,d)
ι—»
(ab,b,a, 1)
we obtain the formula
=
2 - < ( a b ) l , a b >
=
=
λ
The norm of χ now reduces to
=
λ 2 - 2 λ 2 + λ 2
=
0
and we have χ = 0, as desired. If
we
insert
the
special
value
a = 1
in
the
just
verified
equation
(ab)b = a, we are led to · 1
=
b b
and t h e r e f o r e also (a b) b
=
a
=
a (b b)
Since b is a linear combination of 1 and b, we get the f i r s t variant of the alternative law, (a b) b
=
ab2
b (b a )
=
b2 a
The second,
is proved in exactly the same manner; we merely have to w r i t e down all p r o ducts in reverse order. Hence U is an alternative algebra. of zero,
Moreover,
it does not contain
f o r ab = 0 implies
and consequently a = 0 or b = 0.
=
=
0
divisors
11.3. The Theorem of Hurwitz
407
The generalized Frobenius theorem 11.2.3 then immediately yields the r e quired result.
It is natural to ask whether the conditions of Hurwitz' theorem can be r e laxed. In f a c t , there are also normed algebras without 1. They are all constructible f r o m R, C,IH, 0 by the following simple procedure:
Definition 11.3.1: We choose two orthogonal transformations A and Β of the normed algebra II. The (A, Β)-modification of 11 is the algebra with the same elements, but the new product x*y
=
x A yB
=
A(x) B(y)
The essential properties of the modification are
Theorem 11.3.2: ( a ) All modifications of a normed algebra 11 are also normed and have the same dimension as 11. (b) Every normed algebra is a modification of IR,C,IH or 0 and in particular of dimension 1,2,4 or 8.
Proof: That the (A, B)-modification of 11 is normed can be seen f r o m |x*yI *
=
|A(x)B(y)|
=
|A(x)| |B(y)|
=
|x| |y| *
Now let U
be any normed algebra. We write the multiplication in II in the » form ( x*, y ) i-» x*y and select an arbitrary a e 11 with a * 0. Since II has no zero divisors, we derive the existence and uniqueness of some e e 11 which satisfies a*e = a. Then |e| = 1, and l e f t and right multiplication by e are orthogonal transformations of tl . Call them A and B, respectively:
11. O c t a v e s
408
A(x)
=
BCx)
=
χ » e e * χ *
We n e x t i n t r o d u c e a new p r o d u c t on II χ y
by
A _ 1 (x) * B 1 ( y ) *
=
which d e f i n e s an a l g e b r a U. E v i d e n t l y , To f i n i s h t h e p r o o f ,
U
is t h e (A, B ) - m o d i f i c a t i o n of U.
we t h e r e f o r e have t o d e m o n s t r a t e t h a t U c o n t a i n s
a
u n i t e l e m e n t , s i n c e in t h a t c a s e , by t h e t h e o r e m 11.3.1 of H u r w i t z , we m u s t have 11 e induced by the quadratic form μ without comment. An immediate consequence of the definition is λ(1)
=
μ(1)
=
=
1
and, more generally, f o r all χ € 11, λ(χ) Moreover,
=
it is again possible to introduce a conjugation mapping in U. It
is given with the help of λ by χ
=
2 λ(χ) · 1 - χ
=
2 < 1, x>· 1 - χ
The conjugation obeys the rule λ(χ) and is thus an involution:
=
2 λ(χ) λ(1) - λ(χ)
=
λ(χ)
416
11.
χ
=
We n o w c o m e t o t h e a l r e a d y
2 λ(χ)·1 - χ
=
Octaves
χ
announced
Theorem 11.4.3: L e t II b e a q u a d r a t i c a l g e b r a o v e r t h e f i e l d K. T h e n t h e d o u b l e d a l g e b r a is q u a d r a t i c as well.
2
If w e d e n o t e t h e a s s o c i a t e d f o r m s on U a s λ and μ, t h o s e on
11 a s A a n d M,
we d e d u c e f r o m t h e d e f i n i n g e q u a t i o n s ( a £2) b
=
( a b) Ω
a (b £J)
=
(b a ) Ω
( a Ω) (b Ω)
f o r a l l a , b £ II t h e
=
- b a
relations:
Λ(a+bΩ)
M(a+bil)
=
=
ACa)
μ(Ά) + μ ^ )
while the conjugation is given by
a + b Ω
=
ä - b Ω
Proof: We s t a r t
bb
with
=
2 X(b)-b - b2
U
=
2 X(b)-b - 2 *(b)-b + μ ^ ) · 1
and c a l c u l a t e t h e s q u a r e of a+bΩ;
=
μ(^·1
11.4. Quadratic
(a+bfi)2
417
=
a Z + a ( b f i ) + (b£2)a + (bfi)(bi2)
=
a2 - M(b)-1 + 2 M a ) b Ω
A comparison
2 *(a)
Algebras
=
j a 2 - b b j + |b a + b a j Ω
with
(a+bΩ) -
j/i(a)
+
·1
=
2 A(a)-a + 2 *(a) b Ω -
^(a)
+
M(b)J-l
leads to the f o r m u l a s f o r Λ and M. The r e s t is obvious.
The doubling method can thus be i t e r a t e d . H o w e v e r , ly l o s e e s s e n t i a l a l g e b r a i c
in doing so, w e g r a d u a l -
properties:
T h e o r e m 11.4.4: Let I (a)
be a q u a d r a t i c a l g e b r a o v e r the f i e l d K. Then we have
21I
a l t e r n a t i v e «=» II a s s o c i a t i v e . 2 ( b ) U a s s o c i a t i v e ] · 1
1 1 . 4 . Quadrat i c A l g e b r a s
419
We s u b s t i t u t e t h e f o r m e r e x p r e s s i o n and o b t a i n äb - b ä or,
=
2 jx(ab) - X ( a ° b ) j - 1
shorter, äb - b ä
=
(ab) - A ( b a ) j - 1
On t h e o t h e r h a n d , one e a s i l y deduces f r o m t h e g e n e r a l l y v a l i d e x p a n s i o n a(ba) - (ab)a
=
a»(ba) - a»(ab) + a(a»b) -
(a°b)a
( v e r i f i e d by e x p l i c i t l y m u l t i p l y i n g o u t ) t h e i d e n t i t y [a,b,a]
=
(ab)a - a(ba)
=
Since 11 by t h e o r e m 11.2.1 is f l e x i b l e , tion vanishes.
A(ab-ba)-a - -l t h e l e f t hand s i d e of t h e l a s t
equa-
Hence, f o r a t Κ·1 we f i n d by c o m p a r i s o n of c o e f f i c i e n t s A(ab-ba)
=
0
which n a t u r a l l y r e m a i n s t r u e if a is a m u l t i p l e of 1. C o l l e c t i n g t h e mediate results,
inter-
we a r r i v e a t X(ab)
=
A(ba)
and ab - b a
=
0
t h a t means ab
=
b a
T h e s e r e l a t i o n s a l l o w us t o s i m p l i f y t h e above a s s o c i a t o r s t o a c o n s i d e r a ble e x t e n t :
420
11. Octaves
[Α,Α,Β]
j\aj8)a - ß(aa)
=
+ ^ ( a a ) b + (aä)b [B, A, A]
=
- ß(aa)
+ a(ßa)j
- ( a b ) a - ,
the e q u i v a l e n c e
of
( b ) and
(c)
is
ob-
vious. We again i n t r o d u c e the Jordan product a°b = - ( a b + b a ) . Then in e v e r y a l g e b r a
422
11. Octaves
the relation
(a b) a - a (b a )
=
a « ( a b) - a » ( b a) + (a»b) a - a (a»b)
h o l d s , which can be p r o v e n by d i r e c t e v a l u a t i o n . [a,b,a]
=
( a b) a - a (b a )
=
An a l t e r n a t i v e w r i t i n g i s
[a»b,a] - a°[b,a]
The e q u a t i o n a«b which, bra,
=
b»a
=
X(b) a + A(a) b - < a , b > - l
a s we have v e r i f i e d b e f o r e ( t h e o r e m 1 1 . 4 . 4 ) ,
is valid in any
alge-
then yields [a, b , a ]
=
X(a) [ b . a ] - \ ( [ b , a ] ) a - X(a) [ b . a ] + < a , [ b , a ] > - l [a, b, a ]
=
- λ( [b, a ] ) a + < a , [ b , a ] > - l
The l a s t f o r m u l a is s t i l l t o t a l l y g e n e r a l .
Under t h e c o n d i t i o n s (b),
it
re-
duces to [a,b,a] In c o n t r a s t t o t h i s ,
=
-l
we have
=
2 X(a) A(b) - X(a«b)
By a s i m p l e t r a n s f o r m a t i o n we o b t a i n [ a , b, a ]
=
^2 X(a) A ( [ b , a ] ) - A(a» [b, a ] ) j • 1
=
- X(ao[b,a])-l
and w i t h - 2 X(a»[b,a])
=
X(a(ab)-a(ba))
=
λ((ab)a-a(ba))
=
X([a,b,a])
=
0
we s e e t h a t [ a , b , a] = 0. This p r o v e s t h e i m p l i c a t i o n (b) => (a). We now a s s u m e t h a t U is f l e x i b l e . The above i d e n t i t y t h e n a t t a i n s t h e s i m -
11.4. Quadratic Algebras
423
pie f o r m
- λ ( [b, a ] ) a + • 1
=
[a,b,a]
f r o m which A ( [ b , a ] ) = Ο immediately f o l l o w s ( i f
=
0
a is linearly dependent on
1, by comparing c o e f f i c i e n t s ; otherwise the claim is t r i v i a l ) . The linear f o r m λ thus vanishes on all
A(ab)
=
commutators:
A(ba)
This has
A(ab)
=
as a d i r e c t consequence.
( a ° b ) » c - ao(boc)
=
A(a°b)
Moreover,
=
2 A ( a ) A ( b ) -
we get f o r the Jordan product
A ( a ° b ) c + A ( c ) a»b -
-l
- A ( a ) b°c - A ( b « c ) a + < a , b » c > - l =
^ - A ( b ) A ( c ) j a + ^X(a) A ( b ) - j c + ^A(c) - A ( a )
We apply λ to this equation.
The result is
A(a«(b°c))
=
Using the already obtained identities,
or,
j-l
A((a»b)oc)
this can also be w r i t t e n as
A(a(b°c))
=
λ((a°b)c)
A(a(bc)) + A(a(cb))
=
A((ab)c) + A((ba)c)
=
A((ab)c) + A((cb)a)
expanded,
A comparison with the r e l a t i o n
A(a(bc)) + A(c(ba))
424
11. Octaves
which is derived f r o m the f l e x i b i l i t y law by p o l a r i z a t i o n ,
and
substitution
into λ y i e l d s
A(c(ba))
and f i n a l l y ( a ) => (b).
=
A(a(cb))
=
A((cb)a)
Hence the f i r s t three statements are equivalent.
It remains to be shown that ( b ) implies the corresponding 2 linear f o r m Λ belonging to
p r o p e r t y of
the
11.
For the g e n e r i c commutator, the calculation runs as f o l l o w s . We again w r i t e A = a + αΏ, Β = b + βΩ, C = c + y£2 and take the d e f i n i t i o n of A as w e l l as the mutual o r t h o g o n a l i t y of ΙΙ·β and 1 into account and get Λ( [Α, Β ] )
=
λ(ab-ßa-ba+öß)
=
A(ab-ba) + λίΰβ-βα)
=
0
because λ is z e r o when applied to the commutator [a, bl = ab - ba, and βα is c o n j u g a t e to αβ. The argument f o r the a s s o c i a t o r of A, B, C is s i m i l a r :
A([A, B,C])
=
λ((ab)c-a(bc)) + A(a(yß)-y(ßa)) + X ( ( b F ) a - y ( o c b ) ) + λ( ( c ß ) a - ( ß a ) c )
The f i r s t contribution vanishes by assumption, terchange the f a c t o r s in the others;
A([A,B,C])
=
while we are a l l o w e d to i n -
this leads to
λ((yß)a-y(ßa)) + A(a(by)-(ab)r)
+ Μ(cß)a-c(ßa))
=
0
We do not want to c a r r y on with these considerations and only remark as a c o r o l l a r y that all algebras in the sequence
R — > C — > H - * 0 — >
are f l e x i b l e .
20
—
This is true in p a r t i c u l a r f o r the doubled octave algebra
2 0.
11.5.
P a r a l l e l i z a b i l i t y and R e g u l a r V i e l b e i n e
425
11.5. Parallelizability and Regular Vielbeine
In
this
section,
division algebras
are
always
understood
to
be
finite-
d i m e n s i o n a l o v e r t h e f i e l d IR. Our a l g e b r a i c
i n v e s t i g a t i o n s s u g g e s t t h e q u e s t i o n f o r w h i c h η e IN d i v i s i o n
a l g e b r a s in R n e x i s t .
All e x a m p l e s w e k n o w h a v e d i m e n s i o n s η € { 1 , 2 , 4 , 8 } ;
b u t w e c o u l d p r o v e t h e c o m p l e t e n e s s of o u r l i s t o n l y u n d e r c e r t a i n nal a s s u m p t i o n s like,
additio-
e. g. , normedness o r alternativity.
In f a c t , t h e s t a t e m e n t t h a t e v e r y d i v i s i o n a l g e b r a h a s one of t h e f o u r m e n tioned
dimensions
d a y no p r o o f argument Instead,
is
true
without
has been found,
any
restriction.
Nevertheless,
w h i c h i s e n t i r e l y o r a t l e a s t in i t s
to
this
essential
algebraic. o n e h a s t o r e s o r t t o e x t r e m e l y d e e p and d i f f i c u l t t o p o l o g i c a l
a s , w h o s e d i s c u s s i o n by f a r w o u l d e x c e e d t h e s c o p e of t h i s b o o k .
we shall only c i t e the r e l e v a n t topological t h e o r e m s and deduce f r o m t h e g e o m e t r i c and a l g e b r a i c consequences we a r e i n t e r e s t e d
ide-
Therefore, them
in.
More i n f o r m a t i o n c a n be f o u n d in t h e o r i g i n a l p a p e r s r e f e r r e d t o b e l o w a n d t h e a r t i c l e by H i r z e b r u c h [1983] w h i c h i s v e r y w o r t h We r e c a l l
definition
"parallelizability",
4.5.1,
where
we
introduced
reading.
the
terms
"n-Bein"
as well as section 4 . 7 ("cohomology groups"),
and
and sup-
p l e m e n t t h e s e c o n c e p t s by
Definition 11.5.1: ( a ) T h e Euler characteristic ( o r E u l e r number) of a c o m p a c t r e a l manifold Μ is the a l t e r n a t i n g
sum
π
of t h e d i m e n s i o n s of t h e c o h o m o l o g y g r o u p s of M.
Β (Μ) k
=
dim Η*(Μ)
n-dimensional
426
11. Octaves
is also called the k t h Bettl number of M.
The Euler numbers of the spheres are found directly f r o m the Poincar^ lemma. They are
zcs 2 "- 1 )
Z(S 2 n )
=
=
ο
2
Hopf [1935] succeeded in showing that it is possible to construct a nowhere vanishing vector f i e l d on a manifold Μ if and only if χ(Μ) = 0.
Thus the
even-dimensional spheres (except f o r the t r i v i a l case of S ) are not p a r a l lelizable. For spheres of odd dimension k, the existence of a regular k-bein is a very r a r e occurrence.
Complete information is provided by the f o l l o w i n g result,
which has been discovered almost simultaneously and independently by K e r vaire [1958] and Milnor [1958]:
Theorem 11.5.1: The
(n-l)-sphere
Ω s Sn 1 is only η (theorem of Kervalre and Milnor).
parallelizable
if
η e {1,2,4,8}
holds
The original p r o o f s , as well as all other which have meanwhile been found, are based on the periodicity theorem of Bott, a deep and very important r e sult of K-theory. An immediate corollary is
Theorem 11.5.2: Division algebras on IRn exist only f o r η = 1,2,4,8.
Proof: Let U be an n-dimensional division algebra over the f i e l d IR. We choose a basis b
b of II whose f i r s t vector is the unit element b = 1. I n ι For χ e Ω a S , the vectors ν = b ·χ are a basis of IR , because U conn 1 1
11.5. P a r a l l e l i z a b i l i t y and Regular V i e l b e i n e
tains no z e r o divisors.
The f i r s t of these elements,
consequently perpendicular to the sphere.
w ι
=
427
ν
1
- < v , v > v 1 1 1
ν , is equal to χ and
The (n-1) v e c t o r s
=
ν
-x ι
ι
with 2 s i s η are t h e r e f o r e tangential to Ω and, considered as f u n c t i o n s η of x, f o r m a r e g u l a r n-bein. Hence Ω is p a r a l l e l i z a b l e , and theorem 11.5.1 η implies the assertion.
Hopf [1941] had
shown
earlier
(also
with
methods
of
algebraic
topology)
that η must be a power of 2. We shall
encounter
the next chapter, citly,
another
application
of
the K e r v a i r e - M i l n o r
but b e f o r e we want to l i s t
the regular
theorem
vielbeine
in
expli-
which are derived f r o m the division a l g e b r a s C, H, and 0, since they
have many useful g e o m e t r i c
consequences.
We begin with the 1-sphere. The nowhere vanishing v e c t o r f i e l d w^ we have to f i n d is obtained by the described procedure with respect to the standard basis (b , b ) = ( l . i ) of C. At the place (x , x ) it is
Almost as simple is the situation f o r the 3-sphere. usual
basic
in Η y i e l d
quaternions
(b , b , b , b ) = ( l , i , j , k ) . 0 1 2 3
We again r e f e r t o
The
without e f f o r t the required p a r a l l e l i z a t i o n
the
multiplication
rules
by the r e g u l a r
vec-
tor fields w 1
=
w
=
2
II
c-v
+x , -X , + 0 3
V
( -* 2 ·
+x , +X , 3 0
X i
'"'S"
- x , +X , 2 1
)
L e t us now come to the 7-sphere and the Cayley-Dickson octaves! By choosing the
standard
basis
( b , b , b , b , b , b , b , b ) = (e , e , e , e , e , e , e , e ) , O 1 2 3 4 S 6 7 00 0 1 2 3 4 S 6 we obtain with table 11.3.1 at the point ( x , x , x , x , x , x , x , x ) € S the r 00 ο 1 2 3 4 s β tangent v e c t o r s
428
11. Octaves
w 1
-
(
- V +xCO,
w 2
=
(
-V
W 3
=
+X , +x , +x ,1 -X ., "Χ , +X 00 5 6 4 3 1
W 4
=
~x 1. + x o · +x 5 ,
W s
=
(
w β
=
(
w 7
=
(
, +X x ) 3 1 ~Xβ .1 +X1 :. "X5I 4 > + 2
+x , +X ,1 - X 1. ~X 1,+X ,, -X 3 00 4 2 6 0
"V
-X
2 >+
x4)
+X , +X , +X , - x 00 0 •" X 3 ) V ~x 2> 1 β -X , +Χ . - x , +x , +X . +x 4 00• - ν -V 6 3 2 0 -V
+
+X , - χ . +x , +x .,+x ) GO - V -X 2 . -X 5 . 0 4 3 1
Since t h e a l g e b r a s €, H, and Ο a r e n o r m e d , orthonormal
s y s t e m s of t a n g e n t
t h e ν in t h e s e t h r e e c a s e s a r e 1 3 7 v e c t o r s t o S , S , and S . They c o i n c i d e
w i t h t h e w . We t h u s have even c o n s t r u c t e d e v e r y w h e r e o r t h o n o r m a l ( r e g u l a r ) p o l y n o m i a l v i e l b e i n f i e l d s on t h e s p h e r e s in q u e s t i o n .
429
12. 1. Homotopy Groups of Spheres
12. Hopf Mappings 12.1. Homotopy Groups of Spheres
T h i s and t h e s u b s e q u e n t s e c t i o n s e r v e a s a m o t i v a t i o n of w h a t f o l l o w s and t o i n f o r m t h e r e a d e r a b o u t s e v e r a l deep r e s u l t s of a l g e b r a i c t o p o l o g y ;
we
s h a l l need t h e m l a t e r only o c c a s i o n a l l y .
the
For this reason,
we r e s t r i c t
p r e s e n t d i s c u s s i o n t o a m e r e d e s c r i p t i o n of t h e m o s t r e l e v a n t f a c t s w i t h o u t explaining the (sometimes extremely d i f f i c u l t ) proofs. We have a l r e a d y r e c o g n i z e d in s e c t i o n 2. 1, t h a t t h e h o m o t o p y of c u r v e s in t o p o l o g i c a l s p a c e s is a very n a t u r a l c o n c e p t . We t h e r e f o r e e x t e n d i t s r e a l m of a p p l i c a b i l i t y in t h e obvious way:
D e f i n i t i o n 12.1.1:
Two c o n t i n u o u s m a p p i n g s f : X —» Y and f : X —> Y b e t w e e n a r b i t r a r y t o p o l o g i c a l s p a c e s X and Y a r e homotopic t o each o t h e r if we can f i n d a c o n t i n u ous m a p p i n g F:
[0,1] χ X
Y
w i t h F ( 0 , x ) = f Q ( x ) and F ( l , x ) = f ^ x ) f o r all x. by t h e s y m b o l i c n o t a t i o n f
briefly
~ f^.
Homotopy is an e q u i v a l e n c e r e l a t i o n . f:
We e x p r e s s t h i s
The s e t of a l l m a p s h o m o t o p i c t o some
X —) Y i s t h e homotopy class
[f]
of f .
If X and Y a r e ( s m o o t h ) m a n i f o l d s , every h o m o t o p y c l a s s c o n t a i n s a t one C ^ - m a p .
Hence,
" s m o o t h homotopy".
t h e r e is no need t o d i s t i n g u i s h b e t w e e n h o m o t o p y
least and
430
12. Hopf Mappings
The set of all homotopy classes of mappings f r o m X to Y in general has no IT
f u r t h e r structure.
If,
however,
X is a sphere, e . g .
X = S , we can i n t r o -
duce an invariant product on it.
Definition 12.1.2: ( a ) A pointed (topological) space is a pair ( X , x o ) , gical space X and a base point χ
consisting of a t o p o l o -
contained in it.
( b ) The set of all homotopy classes of continuous maps f r o m S point Xq into a connected pointed topological
space ( Y . y Q )
with base is
denoted
by 7tk(Y). ( c ) We describe the k-sphere in the standard f o r m as the set Π of all unit vectors
ζ - (ξ , ξ ξ , ξ ) in IR1+n and choose as its base ^ k-i k the "north pole" x q = (1,0, . . . , 0 ) . The "northern hemisphere"
Ω+
=
I ξ
6
point r
Ω I ξο > 0 }
together with the "equator"
Ä
=
{ ξ e Ω I ξο = 0 }
is mapped by
V = with the scaling f a c t o r ρ obtained f r o m
continuously onto Ω. The image of Ä is the base point Xq, while the r e striction onto Ω+ is b i j e c t i v e l y transformed onto Ω \ { x Q } · ly define φ ( ξ ,ξ
ξ )
=
(1+2ξ ,ρξ
ρξ )
We s i m i l a r -
1 2 . 1 . Homotopy Groups o f Spheres
431
on the "southern hemisphere"
=
{ ξ € Ω
€
0
< θ | k
plus e q u a t o r A. To the continuous mapping f and g f r o m Ω s S a s s o c i a t e a new one, by h(£) = g»(p
h,
which is given in Ω+ by h(£) = f°
y ( [ f ] ) = y , is a homomorphism f r o m π (S n ) f η into the additive group of the integer numbers.
Thus,
r([f]°[g]) = r([f]) + r([g]) Two C^-maps of the η-sphere Sn into itself are homotopic if and only if they have the same degree (this implies IT (S n ) = Z). π is integer-valued and multiplicative: For smooth
( e ) The degree y
12.2. Homotopy Invariants
435
f:
Μ
g:
Ν
and
the formula
Τ,fg
=
7fr 7 g
holds.
Remark: The second invariant in question, here denoted by i ,
was f i r s t
constructed
by Hopf [1935] in order to find the simplest of yet unknown homotopy group 2 of a sphere, namely ii 3 (S ). It turned out that Hopf's definition can be transferred without change to mappings f r o m S 2 " Furthermore, of
t.
1
to S n f o r all η £ 2 (what we shall do f r o m the outset).
there are various d i f f e r e n t , but synonymous characterizations
We shall
describe only three of
the better-known
formulas apt
to
evaluate i .
Now we are prepared f o r
Theorem 12.2.3: Let
f : S2"
1
—> Sn be smooth
(n £ 2).
The f o l l o w i n g
three
definitions
of
i ( f ) are equivalent: ( a ) We choose some ω e A n (S n ) with
J « = 1 sn •
*
The, because of the obvious relation df ω = f du = 0 closed and by the Polncar6
lemma
also
exact
form
f ω e Λ (S
)
has
a
representation
436
12. Hopf Mappings
f ω = da with a e Λη 1 (S 2 n 1 ) . We w r i t e
i(f)
=
J s
α
Λ
da
2n-l
This expression has the same value f o r all forms ω and α s a t i s f y i n g the above conditions. ( b ) With each two d i f f e r e n t fibres F = f _ 1 ( x ) and G = f _ 1 ( y ) ,
where χ * y
are points in S , we associate the linking number V ( F , G ) . To obtain it, we select a submanifold submani U £ S 2n tersects G transversally,
1
with SU = F, which in-
and set
V(F, G)
=
γ^
(± 1)
UnG Here the positive or negative sign applies according to the orientation of TU ®TG (p € UnG). Then ρ Ρ i(f)
=
V(F,G)
irrespective of the f i b r e s chosen. ( c ) If we delete a point P, which lies on neither of the two f i b r e s F ,F , χ
and i d e n t i f y the remaining space, i . e . S
Kf)
=
\ S 2n 2 , given by χ y scx.y)
=
^
The thus defined (integer) number t ( f ) is called the Hopf invariant of f.
The
interchangeability
of
these
three
descriptions
is
investigated
greater detail by Bott & Tu [1982]. The authors also discuss several cations.
We r e s t r i c t our attention ( c f .
in
appli-
Hirzebruch [1983]) to the problem,
which i ( f ) occur in any given dimension. The complete answer is
12.2. Homotopy I n v a r i a n t s
437
Theorem 12.2.4:
( a ) The Hopf i n v a r i a n t i ( f ) of any ( c o n t i n u o u s ) mapping f f r o m s 2 n _ 1 t o S n v a n i s h e s if η is odd. (b) In c o n t r a s t t o t h i s , we can f i n d f o r a l l even η and every m 6 Ζ a map f which h a s i ( f ) = 2m. ( c ) Only f o r η = 1 , 2 , 4 , 8 , cases,
t h e r e a r e f : S 2 n _ 1 —» S n w i t h 2 { i ( f ) .
t h e Hopf i n v a r i a n t a t t a i n s all i n t e g e r v a l u e s .
In t h e s e
In p a r t i c u l a r ,
we
can c o n s t r u c t m a p p i n g s h w i t h i ( h ) = 1 (theorem of Adams).
The l a s t
observation
is i n t i m a t e l y
c o n n e c t e d w i t h t h e o r e m 11.5. 1 by
Ker-
v a i r e and Milnor. The n e x t t w o s e c t i o n s a r e devoted t o t h e e x p l i c i t c o n s t r u c t i o n of t h e s i m p l e s t h : S 2 n 1 —> Sn w i t h η = 2 , 4 , 8 and t ( h ) = 1. F o r t h e s a k e of c o m n 11 p l e t e n e s s , we amend t h i s l i s t by a f o u r t h map h : S —» S which i s l a r g e l y analogous.
This a d d i t i o n u n i f i e s t h e d i s c u s s i o n
considerably.
The f o u r t r a n s f o r m a t i o n s h , h , h^, h ß have unique g e o m e t r i c p r o p e r t i e s which w i l l p e r m i t us t o solve s e v e r a l i m p o r t a n t p r o b l e m s most e l e g a n t l y .
438
12. Hopf Mappings
12.3. Duplication of Angles and Classical Hopf Fibration As a practise lap, Hopf mappings,
so to say, to prepare our considerations of the proper
we consider f i r s t the angle doubling transformation
h^ on
the 1-sphere. Like its counterparts h2> h^, and h8> it is a homogeneous polynomial map of second degree, i. e. , the (Cartesian) components of the image vector are quadratic f o r m s in those of the argument. Apart of the normalization,
as e . g . the choice of the base point f r o m which
we reckon the angle, the just mentioned transformations are uniquely f i x e d and in each case the simplest members in their respective homotopy classes. They are t h e r e f o r e best suited f o r studying the topological questions connected with' the Hopf mappings. Because of their homogeneous-quadratic character, all four h can be extenΠ ded to the entire embedding space R of the spherical domain S . This proves especially useful. The extensions will also be denoted hΠ. The danof confusion between the maps h : S —> S and h : IR —> IR , η η
ger
which coincide on S
, is not probable (and would not cause any trouble
at a l l ) . For the moment, we consider the angle duplication h
only in the r e s t r i c t e d
variant on S1 and w r i t e it f r o m the outset in a f o r m which enables us to t r a n s f e r it to the higher-dimensional Hopf mappings. We find with the Cartesian
coordinates
circle S
(XQ, X^
of
point Χ on the
a typical
periphery
of
the
by section 8.2 the associated central angle (= azimuth) φ via the
identification (cos xi)
φ r
χ . 1 arc tg — χ ο
It is
=
The image point h ^ x ) is calculated with the help of the addition theorems of the sine and cosine functions,
439
12.3. Duplication of Angles and Classical Hopf Fibration
(cos 2φ , s i n 2φ)
(cos2(p - sin 2 ?) , 2 s i n φ cos φ)
-
=
(x 2 -x^,2x χ )
The same result comes out if we consider χ as the image of (1,0) under the rotation with angle φ and repeat this procedure:
h^x)
χ 0
(1,0)
-χ The
duplication
X X 0 1
χ 1 X
1
=
exactly
transformation
two
0
0 h : S 1 —» S1 has the
which equals the algebraic degree, are
ι 2 2 _ . (χ -χ ,2x x )
=
W
(
preimages,
and
1
(mapping)
0 1
degree
f o r to each point on the circle the
differential
is
locally
preserving.
2,
there
orientation-
2
It is s e l f - e v i d e n t that the extension of h j to the domain R
is defined by
the assignment h (χ ,x ) ι ο ί The natural identification of R
=
χ 2 -χ 2 , 2x 0χ1 ο ι J
as the f i e l d C of the complex numbers ad-
mits another interpretation of h : We decompose the arbitrary point ζ by z
=
into real and imaginary parts.
V
z )
=
x + ο
ix
ι
ε
IC
Then
[x 2 - X 2 ]
+
[2 x 0 x j
i
=
z2
is simply the map which associates to each ζ its square. With a pinch of salt,
all mentioned properties remain true if we replace h
by the multiplication of the angle by a f i x e d integer k (or, algebraic version,
in the complex
the squaring of ζ by raising to the k t h power).
The de-
gree then is k, the addition formulas of the angles have to be iterated k times, etc. We are not going to hold f o r t h on this subject, however, we are interested in the transfer to higher dimensions,
since
and this is possi-
ble merely f o r k = 2 (and in special spaces). We can, according to Hopf [1935], construct the classical Hopf f i b r a t i o n h 2 in complete analogy to h . To this end, we parameterize the three-dimensional sphere with two complex coordinates:
440
12. Hopf Mappings
(ζ ,Ζ ) 6 C 0 1 and map t h e p o i n t ζ = (zq, Zj) t o t h e r a t i o
ζ
ζ J:
=
IP1 (C)
€
T h i s c o r r e s p o n d s t o t h e t r a n s i t i o n f r o m two homogeneous c o o r d i n a t e s t o a s i n g l e i n h o m o g e n e o u s one in P 1 (C). By m e a n s of a stereographic projection, 1 2 e s t a b l i s h i n g t h e a l r e a d y f a m i l i a r i s o m o r p h i s m IP (C) s= S , we t h e r e a f t e r go over t o t h e 2 - s p h e r e . 1 2 Depending on t h e m e t h o d used t o c a r r y o v e r f r o m Ρ (C) t o S , i . e . ,
which
p r o j e c t i o n c e n t r e we c h o o s e , t h e r e a r e d i f f e r e n t , but h o m o t o p i c and t h u s in every r e s p e c t equivalent modifications. We a g a i n decide f o r t h e s t a n d a r d p r o j e c t i o n of s e c t i o n 7 . 6 w i t h t h e p o l e a s t h e c e n t r e and s p l i t t h e t w o c o m p l e x p a r a m e t e r s up i n t o t h e i r and i m a g i n a r y
south real
parts: z z
ο ι
=
χ
=
χ
ο
+ 1 Jy o
ι
+ ι yJ i
The q u o t i e n t is x χ +y y ο ι 'ο'ι χ 2 + y2 0 *0
+ i x
2
U + i V
2
+ y *o
0
and we c a n r e a d o f f t h e C a r t e s i a n c o m p o n e n t s ( Χ , Υ , Ζ ) of t h e image v e c t o r : 2U 1 + U2 + V2 2V 2
Z
1 + U + V
2
(x
! + y0 yi> 2 2 2 . 2 x + y + x + y 0 °0 1 2
(X
o
x
0 χ2 + y 2 + 0 'o
y
o
x
i'
1 - U2 - V2 1 + u2 + v2
x
2
ο
+ y
2 ,
"ο
+ χ
2
ι
+ yJ
2
i
12.3. D u p l i c a t i o n of Angles and C l a s s i c a l Hopf F i b r a t i o n
441
This r e p r e s e n t a t i o n , f i r s t f o u n d by Hopf [1935], is f o r our p u r p o s e s a bit inconvenient, mainly because of the d e n o m i n a t o r . We omit i t ( t h i s makes no d i f f e r e n c e on t h e s p h e r e , since t h e 3 - s p h e r e is c h a r a c t e r i z e d as t h e s e t of all p o i n t s w i t h d e n o m i n a t o r = 1). This l e a d s to one of t h e c o u n t l e s s m o d i f i c a t i o n s of h : 2
[wvyi]
~
(
2 ( χ
οχι+νι'
' 2(
VfVi'
'
x
o+VVyi ]
The g r e a t t h e o r e t i c a l and p r a c t i c a l i m p o r t a n c e of h z r e q u i r e s t h a t we e x a m ine t h e e x p l i c i t f o r m u l a s more closely. We u n i f y t h e n o t a t i o n of the independent v a r i a b l e s by c a l l i n g them α , β , ζ , δ . Moreover,
we change t h e o r d e r , which leads to somewhat n i c e r
and a l s o p r o v i d e s an a l t e r n a t i v e
expressions
interpretation.
The q u a d r a t i c dependence of t h e image v e c t o r , f r o m now on denoted (ξ, rj, ζ), on (α,β,ζ,δ) e IR s t r o n g l y s u g g e s t s a r e p r e s e n t a t i o n w i t h q u a t e r n i o n s .
We
assume x
=
ξ 1 + υ j + ζ k
t o be a pure quaternion, while i t s preimage u
=
a + ß l + y j
+ 5k
may be any element of H. This s p e c i f i c a t i o n would in our old t e r m i n o l o g y (see t h e o r e m 10.3.1) simply r e a d u € if3 and χ = h^tu) e f 2 . The only p o s s i b i l i t i e s to c o n s t r u c t a q u a t e r n i o n i c f u n c t i o n h 2 (u) which
is
a homogeneous polynomial of degree 2 (as a r e a l map) a r e e x p r e s s i o n s like uqu, u q u , uqu or uqu with some f i x e d q u a t e r n i o n q * 0. The f i r s t two c a n d i d a t e s a r e r u l e d out since they also a t t a i n values o u t s i d e f 2 ; t h e r e m a i n i n g two a r e e s s e n t i a l l y e q u i v a l e n t . We may (and s h a l l ) t h e r e f o r e use t h e f o r m h 2 (u) Not all q e Η a r e s u i t a b l e ;
=
uqu
a t r a n s i t i o n to the norm shows t h a t t h e
c o n d i t i o n | q | = 1 must be f u l f i l l e d such t h a t But even t h i s
does not
s u f f i c e yet;
must vanish. This l e a d s to
by-
| u | = 1 implies |h ( u ) | = 1.
in a d d i t i o n ,
the t r a c e
of
χ = h^tu)
442
12. Hopf Mappings
0
=
2 Re h 2 (u)
=
u q u + u q u
=
u (q+q) u
and f o r invertible u amounts to the same thing as Re(q) = 0. The parameter 2
quaternion q thus is restricted to if . We express the dependence on q by the symbolic notation Ηq(u) = uqu. The reader is urged to compare this with theorem 10.3.2. All Η , q e f 2 , are Hopf mappings of the same type as h . It should hardly q
2
be necessary to include an explicit proof of the t o t a l l y t r i v i a l
Theorem 12.3.1: The (classical) Hopf f i b r a t i o n s Ηq: IH —» H defined by Ηq (u) = uqu (q e y 2 ),
are all homotopic to each other. They map IH onto the set of pure
quaternions,
3
2
while Η (if ) = if . q
The f i b r e s of Η
q
are,
as we shall see later-on,
great circles on if
3
!
(sec-
tion 12.5). This yields a representation of the 3-sphere as a S - f i b r e bun2 die over S , which plays an important part
in (algebraic) topology.
This
observation more or less initiated Hopf's own contributions to the present topic. Among the infinitely many conceivable parameter values, the choice q = i is particularly clearly
convenient.
is no essential
On account of
the just
restriction
of
generality.
H^u)
=
u i ü
special mapping is
In real coordinates it reads H(a,ß,y,5) f r o m which we derive
=
(ξ,τ},ζ)
described homotopy, In explicit
form,
this this
12.3. D u p l i c a t i o n o f Angles and C l a s s i c a l Hopf
ξ
i + η j
+ ζ k
=
(a2+ß2-y2-62)
Fibration
i + 2 (αδ+ßy) j
443
+ 2 (-ay+βδ) k
and thus
(ζ,η,ζ)
or,
=
Η^α,β,τ,δ)
=
^a 2 +ß 2 -y 2 -ö 2 ,2(αδ+ßy), 2 ( β δ - a y ) j
componentwise, ^ ξ
=
2 α+
-2 β -
2 γ -
V
=
2 (α δ + β y)
ζ
=
2 (β δ - α τ )
.2 δ
We shall p r e f e r t h i s basis in our l a t e r a p p l i c a t i o n s ( c h a p t e r The c l a s s i c a l
Hopf
in the
utilization
cessity
to
additional
fibration of
d i f f e r s f r o m the a n g l e d u p l i c a t i o n
quaternions
be c a r e f u l
with
24).
rather
the o r d e r
of
than c o m p l e x the f a c t o r s
only
quantities,
the
ne-
in products,
and
the
conjugation.
E s s e n t i a l l y the same can be said about the t w o g e n e r a l i z e d Hopf mappings h and h
t o which w e turn next.
4
444
12. Hopf Mappings
12.4. Generalizations
E x c e p t f o r t h e c o n s t r u c t i o n of t h e c l a s s i c a l Hopf f i b r a t i o n h z w i t h nions,
a s d i s c u s s e d in t h e f o r e g o i n g s e c t i o n ,
n i t i o n of h 2>
which is s o m e t i m e s u s e f u l ,
r a l i z a t i o n to the r e l a t e d mappings h
quater-
t h e r e i s an a l t e r n a t i v e
defi-
especially since it a l l o w s a g e n e -
and h .
Definition 12.4.1:
L e t X b e a normed algebra w i t h o r t h o n o r m a l b a s i s S = (e
e ). We d e c o m π
1
p o s e a t y p i c a l e l e m e n t a 6 51 a c c o r d i n g t o
a
into its real components a
=
a
e 1 1
η
e
η
a . We g e t t h e Hopf fibration h η
1
w i t h 51 by i d e n t i f y i n g t h e d o m a i n IR (a,b)
+ . . . + a
associated
w i t h 5t χ it a n d d e f i n i n g t h e i m a g e
of
via h(a,b)
The n a t u r a l
=
£|a|2-|b|z;2(ab)it . . . ,2(ab)J
€
R χ Jt
=
Rn+1
abbreviation
h ( a , b)
w i l l a l s o be u s e d o c c a s i o n a l l y .
=
|ja|2-|b|2;2abj
The standard versions of t h e Hopf m a p f o r
η 6 { 1 , 2 , 4 , 8 } a r e o b t a i n e d w i t h t h e s p e c i f i c a l g e b r a s IR,C,IH,ID.
Theorems
11.3.1 and
11.3.2 mean
that
normed
algebras
exist
only
in
the
a f o r e m e n t i o n e d d i m e n s i o n s a n d t h e n a r e modifications of R, C,IH, a n d 0.
Hopf
f i b r a t i o n s d e d u c e d f r o m m o d i f i e d a l g e b r a s by d e f i n i t i o n o n l y d i f f e r by
ad-
ditional rotations inserted before and/or In p a r t i c u l a r ,
a f t e r the transformation
s t r i c t the investigation without severe loss to the f o u r standard In f a c t ,
h
proper.
they a r e homotopic t o each o t h e r . For t h i s r e a s o n we may r e -
maps the sphere S
2n 1
n
s u r j e c t i v e l y onto S ,
cases.
as demanded at
the
12.4.
Generalizations
445
v e r y b e g i n n i n g of t h i s c h a p t e r .
T h i s i s a t r i v i a l c o r o l l a r y of
Theorem 12.4.1: T h e ( E u c l i d e a n ) n o r m of h (x) is e q u a l t o t h e s q u a r e of t h e n o r m of x .
Proof: With t h e n o t a t i o n j u s t i n t r o d u c e d ,
|hn(a,b)|2
[I| a |12 2--|Ibb|122]
=
b e c a u s e of Jl b e i n g n o r m e d ,
+ 12ab1 2
and the binomial t h e o r e m implies the
For easier
reference,
4 +
Ibl4
+
we have
2 lal2
|b|2
proposition.
w e give a l l h
e x p l i c i t l y in c o o r d i n a t e s . We a l w a y s π r e f e r t o t h e s t a n d a r d b a s i s of t h e u n d e r l y i n g a l g e b r a a n d w r i t e t h e a r g u m e n t v e c t o r in t h e f o r m u
thus setting u
(u
u
1
2n
)
=
(a
a;b π 1
1
= b . In c o n t r a s t , n+i ι of t h e i m a g e χ = h (u) f r o m 0 t h r o u g h n: i
= a
=
i
and u
x
=
0
x( (1 s i < n).
W =
2
b) η
we c o u n t t h e
coefficients
lbl2
-
(2ab)
In t h e f o u r p o s s i b l e c a s e s we g e t t h e f o l l o w i n g l i s t ,
a m e n d by t h e J a c o b i
matrices:
( a ) η = 1: Jt = IR; Β = ( 1 ) ; h : IR2
Dh^u)
IR2: 2
2
X 0
=
U - U 1 2
x
=
2 u
1
=
2
u 1 2
w h i c h we
446
(b)
12.
η
=
2:
J1 =
C;
(1, i );
«
h :
1
< 2
R3:
2 , 2 U + U 1 2
=
0
κ
R
2
2
2
u u 3 L 1
-
u
u
=
2
u L 1
+
u 2
u
u
4
U
1
η
=
4:
Jt = IH;
X
x
x
x
X
=
=
1
2
= 2
2
= 2
3
= 2
4
U
+
L
1
l
I
1
L
1
^ 1
u
u
u
u
U
h :
+
2
=
-
u
+
u 2
u
u 8
6
7
Jt =
0;
3
=
u
U
4
U
B
U
6
-
U
-
u u 3 7
-
u
u
u 5
+
u u 3 8
-
u
u
-
u u 2 8
+
u 3
+
u 4
u
+
u u 2 7
-
u u 3 6
+
u
u
1
6
u
u 5
u
Β
β
4
4
4
u
U
-u
U
-u
U
-u
U
U
2
4
-u U
U
U
-U
-U
U
U
7
-U
3
-u
-U
( 1 , 1 , j , k , Ε, I , J, K ) ;
7
8J
7J
6j
5J
7
-u
8
-U
-U
8:
U 1
2
u
2
-U
η =
+ 3
u
s
U
(d)
3
R°
U
u
Dh ( u ) 4
3j
(u)
( l . l . j . k ) ;
Ο
2 4
4J
2
-U
(c)
U
=
U
Dh
2 U 3
-
4
U
hg:
-u
-u
R18
—»
R9:
-
U
S
Hopf
Mappings
12.4.
Generalizations
2
2
2
2
2
2
2
2
2
2
2
2
2
2
=
2
=
2
=
2
=
2
[
[
U U -U U -U U -U U -U U -U U -U U -U U 1 9 2 10 3 11 4 12 5 13 β 14 7 15 8 1
u u
=
2
=
2
+u u +u u
1 10
1
2
9
-u u
11
2
3
12
-u u 4
+u u +u u
12
3
9
4
11
10
+u u 5
+u u 5
-u u
14
15
β
+u u 6
-u u
13
16
7
-u u 7
16
13
1
8
-u u 8
15
14
U U +U U -U U +U U +U U -U U +U U -U U 1 12 2 11 3 10 4 9 5 16 6 15 7 14 8 13 U
U
h
u u 1
[
13
14
15
-U
2
U
-U
14
3
U
15
-U
4
U
16
+U U +U U +U U +U U 5 9 6 10 7 11 8 12
+U U -U U +U U -U U +U U - U U +U U 2 13 3 16 4 15 5 10 6 9 7 12 8 11
+u u 2
+u u
16
3
13
-u u 4
14
-u u 5
11
+u u
12
+u u - u u 7
9
12
U
10
13
13 11
U
12
U
8
U U -U U +U U +U U -U U -U U +U U +U U 1 16 2 15 3 14 4 13 5 12 6 11 7 10 BE
-U
6
-U
14
-U
12
10
-U
16
15
11
-U
13
u =
+u u
h
U
8
2
U +U + U + U + U + U +U +U - U - U -U -U -U -U -U -U 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
1
Dh ( u )
2
=
1
X
447
U
U
16
-u
10
U
9
u
12
-u
11
-u
2
U -U
-U -U
12
14 15
-U
-U -U
4
6 7
U
3
-U
2
U
8
-U
8
7
-U U
5 6
-U -U
6 5
448
12. Hopf Mappings
12.5. Geometric Peculiarities
As we noticed a l r e a d y ,
all f o u r Hopf mappings h , η e { 1 , 2 , 4 , 8 } , are p o l y Π nomial and homogeneous of second degree and the Cartesian components of the image point thus quadratic f o r m s in those of
the argument
vector.
These
p r o p e r t i e s do not depend on the r e f e r e n c e f r a m e we use; they are shared by the v a r i a n t s of
the hη which we would have obtained if
we had based our
construction on a r b i t r a r y normed algebras without unit element r a t h e r than R.C.W.O. However,
the
algebraic
Hopf flbratlons found in this
manner
are
distin-
guished g e o m e t r i c a l l y f r o m their t o p o l o g i c a l r e l a t i v e s which can be derived by continuous d e f o r m a t i o n s ;
they are in many respects the simplest
s e n t a t i v e s in their homotopy
classes.
This,
for
example,
are g e o m e t r i c a l
manifests itself
in the shape of
and not merely t o p o l o g i c a l spheres.
repre-
the hη- f i b r e s ,
which
It does not take much
e f f o r t to determine them e x p l i c i t l y :
Theorem 12.5.1: Assume J1 is one of the normed algebras 1R, C,IH,(D and η = dim(ü) = 1 , 2 , 4 , 8 . We decompose the t y p i c a l vector X e R ® Jt = R 1+n according to the scheme
X
where the m-part χ = (χ normal basis e
1
=
(x;x) ο
=
(x;x,...,x) ο ι η
χ ) is split up with respect to a f i x e d η
1
ortho-
e : η χ
=
χ
e 1 1
+ . . . + X
η
e η
( a ) We have h _ 1 ( 0 ) = { 0 } , w h i l e the inverse image of a nonzero X under the π canonical Hopf mapping h consists of all (a, b) e 51 χ Jt = R n subject η to the r e s t r i c t i o n s
= 'I h + *oJ
12.5. Geometric
Peculiarities
449
ibi =
[ι χ ι - x 0 J
and
a b
They f o r m a g e o m e t r i c ( b ) For
ι - χ 2
=
(n-l)-sphere.
a l l X 6 H 1 + n as S n , the f i b r e s h _ 1 ( X ) and h ~ \ - X ) η η
pode - X a r e mutually o r t h o g o n a l .
of
X and i t s
anti-
Each of them contains j u s t t h o s e unit
v e c t o r s which a r e perpendicular to the o t h e r :
ΐΤ'ί-Χ)
=
I u € IR2"
|u| = ι
h_1(X)
Proof: T h e o r e m 12.4.1 i m m e d i a t e l y leads to h 1 C0) = {0>. π X * 0 and i n t r o d u c e the a b b r e v i a t i o n s
Both square r o o t s a r e r e a l clidean norm i m p l i e s
* =
[Ι χ 1
" =
[ι χ ι - x oJ
assume
+ x oJ
(and p o s i t i v e ) since the d e f i n i t e n e s s of
the Eu-
|XQ| £ |X|. The condition h n (a, b) = X y i e l d s a f t e r
obvious d e c o m p o s i t i o n the
restrictions
and
2 a b
In the second,
We may t h e r e f o r e
=
χ
we pass over to the norm,
2 I a. I |b|
getting
=
|x|
an
450
12. Hopf Mappings
From the thus obtained system of equations f o r the norms we find |a| = α and |b| = β as the unique solution. The f i b r e of X is therefore characterized by 2ab = χ and |a| = a, the norm of b then automatically has the c o r rect value β. We still have to show that this is an (n-l)-sphere.
On account of the theo-
rems 11.1.4 and 11.2.1, J1 is alternative, while the subalgebra generated by a and b is even associative. This enables us to solve the equation 2ab = χ f o r b:
With the available information on the norm of a, this can also be written b
1
- a
-2
-
ax
The vectors a and b are therefore connected through a bijective linear r e lation.
The set of all solutions of the last equation alone is consequently
an n-dimensional subspace. If we add the constraints |a| = α and |b| = β, it reduces to a sphere, and we have proven part (a). To v e r i f y (b), in
h-1(-X). η
we proceed in the same vein with a pair ( a ' , b ' ) which lies
We find
β
a
a
as well as ι
a' b
2
X
- a b
In complete analogy with the earlier formula b we now get
1
-a
-2
-
ax
12.5. Geometric P e c u l i a r i t i e s
a
451
,
1 "2 r-r - - α χ b
=
2
The scalar product of (a, b) and ( a ' , b ' ) is calculated as
=
=
| a"2
j^- < a , χ b 7 »
+ < ä χ , b ' >J
The stated orthogonality amounts to the same thing as the vanishing of the square brackets,
which in turn is a consequence of the octave identity
Because of the t r i l i n e a r i t y ,
=
we need only check the latter f o r the standard
basis elements u, v, w e {e , e , e , e , e , e ,e ,e }. 0 0 0 1 2 3 4 5 6
Let u = e , x = e , v = e , L Μ Ν
say.
Both sides vanish individually if
J
J
e ,e ,e L Μ Ν
are not contained in a quaternion subalgebra, since then each is perpendicular to the product of the remaining two. This allows us to suppose in addition that u, v, and χ associate with each other. But then 2
=
Re ^u (x
v)j
=
vj
Re ^(ü χ)
= 2
The ideal way to evaluate the Hopf invariant (in the present situation,
we
have η * 1, i . e . η 6 { 2 , 4 , 8 } ! ) is via the definition of i ( h ) as linking π number V(F , F ) of two f i b r e s F and F which can be chosen deliberately. 1 2 1 2 We go f o r the most convenient possibility: F
I
F 2
= h"1 (+1; 0) n
=
Π χ {0} η
=
=
If we apply to Η the identities [AB, C]
=
[A,C] B [B,C]
[A,BC]
=
[A,C] [A,B] C
which hold in every group, we get H'
=
< [h^.h a']*
Since A lies in the centre, Ϊ'
=
i, j e {1
n> ; a , a ' e A ; χ € Η )
[l^a, h a'J = [h^, h ], and thus < [ h ^ h ]K
1. j € { 1
The subgroup Κ of Η generated by the h representatives
of
all
coset
of
A.
by its very construction contains
Hence,
x ' 6 Κ and some a e A such that χ = χ ' a . Χ y =
n> >
ΧA y =
for
each
χ e H,
there
is
an
The rule X y
which is also deduced f r o m A s Z(H), simplifies the description of the commutator group given above to H' In particular,
=
< [hilhj]
i, j e i l
n> >
s
Κ
A s K, and Κ = Η f o l l o w s .
The explicit calculation of the Schur multiplier jK(G) of a ( f i n i t e ) group G in general is an exceedingly d i f f i c u l t problem. Nevertheless,
meanwhile the
multipliers of all simple groups are known (Griess [1980]). We shall be s a t i s f i e d with a special case which is relevant to spinor theory and is already discussed to considerable detail cited at the beginning of this section:
in the paper
of
Schur
456
13.
Spinors
Theorem 1 3 . 1 . 2 : The Schur m u l t i p l i e r
of the s y m m e t r i c group S
η
is t r i v i a l
if
η s 3 and the
( c y c l i c ) g r o u p of o r d e r 2 f o r η a 4.
Proof: We put G = S tion,
Π
and s e l e c t a r e p r e s e n t a t i o n group Η = AaG o f
and H/A = G. To c a l c u l a t e A = MiG),
w e have A s Z ( H ) n H '
presentation
of
rators a . . . . . a 1
S n. An e s p e c i a l l y u s e f u l one is due to Moore. n-l
as w e l l
|i-j|
2.
s
the t r a n s p o s i t i o n s
w e need a It has
gene-
ι
a2 ι
=
1
a
)3
=
i+1
1
as
a
whenever
By d e f i n i -
and the r e l a t i o n s
(a
f o r a l l i,
G.
All
a
ι
=
J
conditions
a
J
hold,
a
ι
e.g.,
if
we
interpret
the
as
( i , i + l ) which i n t e r c h a n g e t w o successive e l e m e n t s of
the
set {1, . . . , n>. That
these
equations
later
as a b y - p r o d u c t
really of
s u f f i c e to
Coxeter
define
theory
r e f r a i n f r o m an e x p l i c i t v e r i f i c a t i o n
S
(see
η
abstractly
section
16.2);
will we
be
found
therefore
here.
T o each a t w e s e l e c t once and f o r a l l a p r e i m a g e A i under the canonical
ho-
momorphism Η —> H/A = G. Then, by t h e o r e m 13. 1. 1, the A^ g e n e r a t e the e n t i r e g r o u p H. They s a t i s f y r e l a t i o n s of the f o r m A2 1 (A
and,
as long as
li-j|
£
2,
certain
elements
A
α
)3 1+1
1
=
β "i
also
A
with
1
=
ι
A
a 1 >ß 1 >3 r 1 j·
j
=
which
r
ιJ all
A
J
A
i
together
lie
in A and hence
in
13.1. Schur Extensions o f the Symmetric Groups
the c e n t r e o f
457
H.
F o r η s 2 w e conclude t h a t Η is c y c l i c and in p a r t i c u l a r m e d i a t e l y leads t o the p r o p o s i t i o n A s H'
abelian.
This
im-
=1.
The n e x t case, η = 3, is somewhat more c o m p l i c a t e d . We now distinguish r e p r e s e n t a t i v e s in a l l 6 c o s e t s of A, e . g . 1, A , A , A A , A A , A A A . With the r 1 2 1 2 Z 1 1 2 1 trivial rules
[ 1, x ]
=
[x,l]
=
[y,x] =
which a r e v a l i d in e v e r y group,
[x,x]
=
1
[x,y]
the 6
= 36 c o m m u t a t o r s of
the A j
reduced to the f o l l o w i n g 10, if we set k = [A^, A^ 1 f o r the sake of
[ Α
ι Λ
A A 1 2
]
[A ,A A ] 1 1 2
=
A A 1 2
=
A A 1 2 A A A 1 2 1
[A ,A A A ] 1 12 1
A A A 1 2 1
=
k
[A ,A A ] 2 12
A A A 2 1 2
=
A"1 a"1 A A A 1 2 2 1 2
[A ,A A ] 2 2 1
A A 2 1
[A ,A A A ] 2 12 1
=
A * A * A" 1 A 1 A A A A 2 1 2 1 2 1 2 1
=
A " A * A A A"1 A 1 A A 2 1 2 1 2 1 2 1
[A A ,A A ]
=
. · · » - ! A a A 2
[A A ,A A A ] 12 12 1
=
A * a * A 1 A 1 A 2 1 2 1
[A A ,A A A ] 2 1 12 1
=
A"
=
A * A * A 1 A 1 2 1 2
2
1
A
1
1
a . A a A 1 2 1
=
a
=
a
k
A A A A 1 2 1 2 1
a 1 2
=
k2
-1-1 a a a 1 2 1 2
A" 1 A" 1 A a A A 2 1 2 1 2 1 1
=
k
=
1 2
brevity:
k
[A ,A A ] 1 2 1
k
=
can be
, 1
458
13. Spinors
These are all powers of k. Since a commutator is unaffected when we multiply i t s arguments by a r b i t r a r y central elements, we conclude H' = . From # ( G ' ) = 3 we moreover deduce that A is the (only) subgroup of index 3 in the cyclic group H': A = The generator k k
=
3
can be simplified f u r t h e r . F i r s t , we have
A"1 A"1 A A 1 2 1 2
=
a" 1 a" 1 (A A f 1 2 12
=
of 1 α"1 β (Α Α Γ 1 1 2 1 1 2
and a f t e r raising to the third power .3 k
=
-3 -3 „3 . . . . -3 α α S (A A ) 1 2 1 1 2
- 3 - 3 -2 α α β 1 2 1
=
On the other hand, (A A ) 3 2 1
=
A"1 (A A ) 3 A = 1 1 2 1 β2 1
=
Α1 β A 1 * 1 1
(A A ) 3 (A A ) 3 12 2 1
=
=
β "l
=
(A A ) 3 12
a3 a3 1 2
3 and this yields k = 1, i. e. A = 1, as required. The remaining "generic" cases (n 2 4) are a bit easier since we have more r e l a t i o n s at our disposal. We begin by showing that the order of the Schur multiplier of G = S
is not larger than 2.
All elements denoted by Greek l e t t e r s lie in A and particularly in the cent r e Z(H). For the r e s t of the proof, we shall always assume |i-j| £ 2. The 2 equat last found equation f o r 0 s t i l l holds, as well as the similar, but more general relation -2 ß ι
From the definition of y
=
3 I
3 a
1+1
, we get A1 A A j ι j
Squaring yields
a
=
y
ιJ
A
ι
13.1. Schur Extensions of the Symmetric Groups
=
A
α
i
A J
A' 1 A2 A J ι J
=
459
r
.2
lj
A ι
2
=
y α "lj ι
or simply
Now we consider a second pair i ' , j ' four
numbers
pairwise
i, i+1, j , j+1 are,
unequal.
The same
with the same properties as i , j .
according
to
is true f o r
the
assumption
i ' , i'+1, j ' , j'+1.
about
Hence,
The
|i-j|.
there
is
some permutation in G = S which transforms the quadruple (i, i+1, j , j+1) in η the right order into ( i ' , i'+1, j ' , j ' + 1 ) .
One of the associated elements in A
will be called X. We then get X"1 A X ι
=
Χ"1 A X J
=
ξ A , l'
s
η A , J
with suitable ξ, ν e A S Z(H). At last, we derive f r o m a comparison of ξ η A , A , 1 J
= X ~
1
A A X 1 J
=
X"1y
iJ
A A X J i
=
r
ξ ν A , A , 1J ^ )' l'
and A j A · 1 ]' the coincidence For simplicity,
if
of
r(J
and 7 l * y
— —
τ # y A g A J i j j' l'
All y-elements
are
consequently
equal.
we w r i t e
I i - j I a 2.
Until now we have only demanded that A ( is in the preimage of a
under the
natural homomorphism A —» A/H £ G; the exact f i x a t i o n has been l e f t open. This gives us some freedom to normalize our relations. If we replace A t by A| = C 1 A [ with ζ ι 6 A, all formulas retain their general shape, but the c o e f f i c i e n t s change their values according to the simple scheme
460
13. Spinors
.
With the shorthand notation X