Operational Spacetime PDF

Operational Spacetime
Fundamental Theories of Physics
An International Book Series on The Fundamental Theories of Physics:

Their Clarification, Development and Application
Series Editors:
GIANCARLO GHIRARDI, University of Trieste, Italy
VESSELIN PETKOV, Concordia University, Canada
TONY SUDBERY, University of York, UK
ALWYN VAN DER MERWE, University of Denver, CO, USA
Volume 163
For other titles published in this series, go to http://www.springer.com/series/6001
Heinrich Saller
Operational Spacetime
Interactions and Particles
123
Heinrich Saller
MPI für Physik
Werner-Heisenberg-Institut
Föhringer Ring 6
80805 München
Germany
hns@mppmu.mpg.de
ISBN 978-1-4419-0897-1 e-ISBN 978-1-4419-0898-8

DOI 10.1007/978-1-4419-0898-8
Springer New York Dordrecht Heidelberg London
Library of Congress Control Number: 2009939325
c Springer Science+Business Media, LLC 2010

All rights reserved. This work may not be translated or copied in whole or in part without the written
permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street, New York,
NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in
connection with any form of information storage and retrieval, electronic adaptation, computer software,
or by similar or dissimilar methodology now known or hereafter developed is forbidden.
The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are
not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject
to proprietary rights.
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)

Contents
0 Introduction and Orientation 1
1 Einstein’s Gravity 17
1.1 Geometrization of Gravity . . . . . . . . . . . . . . . . . . . . 17
1.2 Schwarzschild–Kruskal Spacetime . . . . . . . . . . . . . . . . 20
1.3 Friedmann and de Sitter Universes . . . . . . . . . . . . . . . 24
2 Riemannian Manifolds 29
2.1 Differentiable Manifolds . . . . . . . . . . . . . . . . . . . . . 29
2.1.1 External Derivative . . . . . . . . . . . . . . . . . . . 32
2.2 Riemannian Operation Groups . . . . . . . . . . . . . . . . . 33
2.2.1 Metric-Induced Isomorphisms . . . . . . . . . . . . . . 33
2.2.2 Tangent Euclidean and Poincaré Groups . . . . . . . . 34
2.2.3 Global and Local Invariance Groups . . . . . . . . . . 36
2.2.4 Riemannian Connection . . . . . . . . . . . . . . . . . 39
2.3 Affine Connections . . . . . . . . . . . . . . . . . . . . . . . . 40
2.3.1 Torsion, Curvature, and Ricci Tensor . . . . . . . . . . 41
2.3.2 Cartan’s Stuctural Equations . . . . . . . . . . . . . . 43
2.4 Lie Groups as Manifolds . . . . . . . . . . . . . . . . . . . . . 44
2.4.1 Lie Group Operations . . . . . . . . . . . . . . . . . . 44
2.4.2 Lie Algebra Operations . . . . . . . . . . . . . . . . . 45
2.4.3 The Poincaré Group of a Lie Group . . . . . . . . . . 46
2.4.4 Lie–Jacobi Isomorphisms for Lie Groups . . . . . . . . 46
2.4.5 Examples . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.4.6 Adjoint and Killing Connection of Lie Groups . . . . . 49
2.5 Riemannian Manifolds . . . . . . . . . . . . . . . . . . . . . . 51
2.5.1 Lorentz Covariant Derivatives . . . . . . . . . . . . . . 51
2.5.2 Laplace–Beltrami Operator . . . . . . . . . . . . . . . 52
2.5.3 Riemannian Curvature . . . . . . . . . . . . . . . . . . 53
2.5.4 Einstein Tensor and Conserved Quantities . . . . . . . 54
2.6 Tangent and Operational Metrics . . . . . . . . . . . . . . . . 55
2.6.1 Invariants . . . . . . . . . . . . . . . . . . . . . . . . . 56
v
vi Contents
2.7 Maximally Symmetric Manifolds . . . . . . . . . . . . . . . . 57

2.7.1 Spheres and Hyperboloids . . . . . . . . . . . . . . . . 58
2.7.2 Constant-Curvature Manifolds . . . . . . . . . . . . . 59
2.8 Rotation-Symmetric Manifolds . . . . . . . . . . . . . . . . . 60
2.9 Basic Riemannian Manifolds . . . . . . . . . . . . . . . . . . . 61
2.9.1 Manifolds with Dimension 1 . . . . . . . . . . . . . . . 62
2.9.4 Rotation-Invariant Four-Dimensional Spacetimes . . . 67
2.9.5 Robertson–Walker Metrics . . . . . . . . . . . . . . . . 70
2.10 Covariantly Constant-Curvature Manifolds . . . . . . . . . . . 73
2.10.1 Orthogonal Symmetric Lie Algebras . . . . . . . . . . 73
2.10.2 Real Simple Lie Algebras . . . . . . . . . . . . . . . . 74
2.10.3 Globally Symmetric Riemannian Manifolds . . . . . . 76
2.10.4 Curvature of Globally Symmetric Riemannian
Manifolds . . . . . . . . . . . . . . . . . . . . . . . . . 77
2.10.5 Examples . . . . . . . . . . . . . . . . . . . . . . . . . 79
3 Mass Points 81
3.1 Nonrelativistic Classical Interactions . . . . . . . . . . . . . . 82
3.2 The Symmetries of the Kepler Dynamics . . . . . . . . . . . . 84
3.3 Electrodynamics for Charged Mass Points . . . . . . . . . . . 86
3.4 Einstein Gravity for Mass Points . . . . . . . . . . . . . . . . 87
3.5 Geodesics of Static Spacetimes . . . . . . . . . . . . . . . . . 88
3.6 Gravity for Charged Mass Points . . . . . . . . . . . . . . . . 91
4 Quantum Mechanics 93
4.1 Nonrelativistic Wave Mechanics . . . . . . . . . . . . . . . . . 94
4.2 Harmonic Oscillator . . . . . . . . . . . . . . . . . . . . . . . 96
4.2.1 Position Representation . . . . . . . . . . . . . . . . . 97
4.2.2 Color SU(3) for 3-Position . . . . . . . . . . . . . . . . 97
4.2.3 Harmonic Fermi Oscillator . . . . . . . . . . . . . . . . 99
4.2.4 Bose and Fermi Oscillators . . . . . . . . . . . . . . . 99
4.3 Kepler Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . 100
4.3.1 Position Representation . . . . . . . . . . . . . . . . . 101
4.3.2 Orthogonal Lenz–Runge Symmetry . . . . . . . . . . . 102
4.4 Particles and Ghosts . . . . . . . . . . . . . . . . . . . . . . . 105
4.4.1 Definite Metric, Fock Space, and Particles . . . . . . . 106
4.4.2 Indefinite Metric and Ghosts . . . . . . . . . . . . . . 108
5 Quantum Fields of Flat Spacetime 111

5.1 Electrodynamics of Fields . . . . . . . . . . . . . . . . . . . . 113
5.2 Gravity of Fields . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.3 Gravity and Electrodynamics . . . . . . . . . . . . . . . . . . 117
5.4 Linearized Einstein Equations . . . . . . . . . . . . . . . . . . 118
Contents vii
5.5 Free Particles for Flat Spacetime . . . . . . . . . . . . . . . . 119

5.6 Massive Particles with Spin 1 and Spin 2 . . . . . . . . . . . . 123
5.7 Massless Polarized Photons and Gravitons . . . . . . . . . . . 126
5.8 Quantum Gauge Fields . . . . . . . . . . . . . . . . . . . . . . 130
5.8.1 Fadeev–Popov Ghosts in Quantum Mechanics . . . . . 131
5.8.2 Fadeev–Popov Ghosts for Quantum Gauge Fields . . . 133
5.8.3 Particle Analysis of Massless Vector Fields . . . . . . 134
5.9 Hilbert Representations of the Poincaré Group . . . . . . . . 136
5.10 Normalizations and Coupling Constants . . . . . . . . . . . . 138
5.11 Renormalization of Gauge Fields . . . . . . . . . . . . . . . . 140
5.11.1 Perturbative Corrections of Normalizations . . . . . . 141
5.11.2 Lie Algebra Renormalization by Vacuum Polarization 144
6 External and Internal Operations 147

6.1 Fiber Bundles . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
6.1.1 Fibers and Base . . . . . . . . . . . . . . . . . . . . . 148
6.1.2 Structural and Gauge Groups . . . . . . . . . . . . . . 149
6.2 Nonrelativistic and Relativistic Bundles . . . . . . . . . . . . 150
6.3 Connections of Vector Space Bundles . . . . . . . . . . . . . . 152
6.4 Pure Gauges, Distinguished Frames, and Composite
Gauge Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
6.5 Chargelike Internal Connections . . . . . . . . . . . . . . . . . 156
6.5.1 Currents as Lie Algebra Densities . . . . . . . . . . . 156
6.5.2 Normalizations of Gauge Fields . . . . . . . . . . . . . 158
6.5.3 Gauge Interactions in the Standard Model . . . . . . . 158
6.6 Ground-State Degeneracy . . . . . . . . . . . . . . . . . . . . 161
6.6.1 Electroweak Symmetry Reduction . . . . . . . . . . . 161
6.6.2 Transmutation from Hyperisospin to Electromagnetic
Symmetry . . . . . . . . . . . . . . . . . . . . . . . . . 165
6.7 Lie Group Coset Bundles . . . . . . . . . . . . . . . . . . . . 167
6.8 Electroweak Spacetime . . . . . . . . . . . . . . . . . . . . . . 168
7 Relativities and Homogeneous Spaces 173

7.1 Parametrization of Relativity Manifolds . . . . . . . . . . . . 176
7.1.1 Goldstone Manifold for Weak Coordinates . . . . . . . 176
7.1.2 Orientation Manifolds of Metrical Tensors . . . . . . . 177
7.1.3 Orientation Manifold of Metrical Hyperboloids . . . . 177
7.1.4 2-Sphere for Perpendicular Relativity . . . . . . . . . 178
7.1.5 Mass-Hyperboloid for Rotation Relativity . . . . . . . 178
7.1.6 Spacetime Future Cone for Unitary Relativity . . . . . 179
7.2 Relativity Transitions by Transmutators . . . . . . . . . . . . 180
7.2.1 From Interactions to Particles . . . . . . . . . . . . . . 180
7.2.2 Pauli Transmutator . . . . . . . . . . . . . . . . . . . 181
7.2.3 Weyl Transmutators . . . . . . . . . . . . . . . . . . . 182
7.2.4 Higgs Transmutators . . . . . . . . . . . . . . . . . . . 183
viii Contents
7.2.5 Real Tetrads (Vierbeins) . . . . . . . . . . . . . . . . . 184

7.2.6 Complex Dyads (Zweibeins) . . . . . . . . . . . . . . . 184
7.3 Linear Representations of Relativities . . . . . . . . . . . . . 185
7.3.1 Rectangular Transmutators . . . . . . . . . . . . . . . 186
7.3.2 Representations of Perpendicular Relativity . . . . . . 187
7.3.3 Representations of Electromagnetic Relativity . . . . . 188
7.3.4 Representations of Rotation Relativity . . . . . . . . . 190
7.3.5 Representations of Unitary Relativity . . . . . . . . . 191
7.3.6 Representations of Lorentz Group Relativity . . . . . 191
7.4 Relativity Representations by Induction . . . . . . . . . . . . 194
7.4.1 Induced Representations . . . . . . . . . . . . . . . . . 194
7.4.2 Transmutators as Induced Representations . . . . . . 195
7.5 Hilbert Spaces of Compact Relativities . . . . . . . . . . . . . 196
7.6 Flat Spaces as Orthogonal Relativities . . . . . . . . . . . . . 198
7.6.1 Particle Analysis of Special Relativity . . . . . . . . . 198
8 Representation Coeffficients 201

8.1 Finite-dimensional Representations . . . . . . . . . . . . . . . 202
8.1.1 Metrics of Representation Spaces . . . . . . . . . . . . 202
8.1.2 Metrics of Lie Algebras . . . . . . . . . . . . . . . . . 204
8.2 Group Algebras and Representation Spaces . . . . . . . . . . 205
8.2.1 Finite Groups . . . . . . . . . . . . . . . . . . . . . . . 205
8.2.2 Algebras and Vector Spaces for Locally
Compact Groups . . . . . . . . . . . . . . . . . . . . . 206
8.3 Schur Product of Group Functions . . . . . . . . . . . . . . . 209
8.3.1 Duality for Group Function Spaces . . . . . . . . . . . 209
8.3.2 Hilbert Metrics of Cyclic Representation Spaces . . . 210
8.3.3 Induced Positive-Type Measures . . . . . . . . . . . . 212
8.4 Harmonic Analysis of Representations . . . . . . . . . . . . . 213
8.5 Schur Orthogonality for Compact Groups . . . . . . . . . . . 215
8.6 Translation Representations . . . . . . . . . . . . . . . . . . . 217
8.6.1 Fourier Transformation . . . . . . . . . . . . . . . . . 217
8.6.2 Cyclic Translation Representations . . . . . . . . . . . 219
8.6.3 Spherical and Hyperbolic Positive-Type Functions . . 220
8.6.4 Breit–Wigner Functions . . . . . . . . . . . . . . . . . 222
8.6.5 Gaussian Functions . . . . . . . . . . . . . . . . . . . . 223
8.7 Harmonic Representations of Orthogonal Groups . . . . . . . 225
8.8 Hilbert Metrics for Flat Manifolds . . . . . . . . . . . . . . . 226
8.8.1 Euclidean Position for Nonrelativistic Scattering . . . 227
8.8.2 Minkowski Spacetime for Relativistic Particles . . . . 229
8.9 Parabolic Subgroups . . . . . . . . . . . . . . . . . . . . . . . 231
8.9.1 Discrete and Continuous Invariants . . . . . . . . . . . 233
8.10 Eigenfunctions of Homogeneous Spaces (Spherical Functions) 233
8.10.1 Simple Examples . . . . . . . . . . . . . . . . . . . . . 235
8.10.2 Eigenfunctions of Compact Groups . . . . . . . . . . . 236
Contents ix
8.10.3 Eigenfunctions of Euclidean Groups . . . . . . . . . . 236

8.10.4 Eigenfunctions of Noncompact Groups . . . . . . . . . 236
8.11 Hilbert Metrics for Hyperboloids and Spheres . . . . . . . . . 238
8.11.1 Hyperbolic Position in the Hydrogen Atom . . . . . . 238
8.11.2 Representations of Hyperboloids and Spheres . . . . . 240
8.12 Spherical, Hyperbolic, Feynman, and Causal Distributions . . 244
8.13 Residual Distributions . . . . . . . . . . . . . . . . . . . . . . 246
8.13.1 Macdonald, Bessel, and Neumann Functions . . . . . . 246
8.13.2 Some Special Cases . . . . . . . . . . . . . . . . . . . . 249
8.14 Hypergeometric Functions . . . . . . . . . . . . . . . . . . . . 252
9 Convolutions and Product Representations 255

9.1 Composite Nambu–Goldstone Bosons . . . . . . . . . . . . . 256
9.1.1 Chirality . . . . . . . . . . . . . . . . . . . . . . . . . 256
9.1.2 Chiral Degeneracy . . . . . . . . . . . . . . . . . . . . 258
9.1.3 Massless Chiral Boson . . . . . . . . . . . . . . . . . . 259
9.2 Convolutions for Abelian Groups . . . . . . . . . . . . . . . . 261
9.2.1 Convolutions with Linear Invariants . . . . . . . . . . 262
9.2.2 Convolutions with Self-Dual Invariants . . . . . . . . . 262
9.3 Convolutions for Position Representations . . . . . . . . . . . 263
9.3.1 Convolutions for Euclidean Spaces . . . . . . . . . . . 263
9.3.2 Convolutions for Odd-Dimensional Hyperboloids . . . 264
9.3.3 Convolutions for Odd-Dimensional Spheres . . . . . . 265
9.4 Residual Normalization . . . . . . . . . . . . . . . . . . . . . 266
9.5 Convolution of Feynman Measures . . . . . . . . . . . . . . . 267
9.6 Convolutions for Even-Dimensional Spacetimes . . . . . . . . 269
9.7 Feynman Propagators . . . . . . . . . . . . . . . . . . . . . . 271
9.7.1 Convolutions of Feynman Propagators . . . . . . . . . 272
9.7.2 Convolutions for Free Particles . . . . . . . . . . . . . 273
9.7.3 Off-Shell Convolution Contributions . . . . . . . . . . 274
10 Interactions and Kernels 277

10.1 Invariant Differential Operators . . . . . . . . . . . . . . . . . 279
10.2 Kernels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
10.2.1 Kernels for Bilinear Forms . . . . . . . . . . . . . . . . 281
10.2.2 Group Kernels . . . . . . . . . . . . . . . . . . . . . . 282
10.3 Green’s Kernels . . . . . . . . . . . . . . . . . . . . . . . . . . 283
10.3.1 Linear Kernels for Spacetime . . . . . . . . . . . . . . 284
10.3.2 Laplace–Beltrami Kernels . . . . . . . . . . . . . . . . 285
10.4 Tangent Kernels and Interactions . . . . . . . . . . . . . . . . 287
10.4.1 Position Interactions . . . . . . . . . . . . . . . . . . . 287
10.5 Duality and Normalization . . . . . . . . . . . . . . . . . . . . 288
10.5.1 Schur Products of Feynman Measures . . . . . . . . . 289
10.5.2 Representation Normalizations for Spheres
and Hyperboloids . . . . . . . . . . . . . . . . . . . . . 290
x Contents
10.6 Kernel Resolvents and Eigenvalues . . . . . . . . . . . . . . . 291

10.6.1 Spacetime Normalization in the Nambu–Jona-Lasinio
Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
11 Electroweak Spacetime 295

11.1 Operational Spacetime . . . . . . . . . . . . . . . . . . . . . . 295
11.2 Representations of Electroweak Spacetime . . . . . . . . . . . 297
11.2.1 Harmonic Analysis of the Causal Cartan Plane . . . . 299
11.2.2 Harmonic Analysis of Even-Dimensional
Causal Spacetimes . . . . . . . . . . . . . . . . . . . . 301
11.3 Time and Position Subrepresentations . . . . . . . . . . . . . 304
11.3.1 Projection to Subgroup Representations . . . . . . . . 305
11.3.2 Projection to and Embedding of Particles
and Interactions . . . . . . . . . . . . . . . . . . . . . 306
11.4 Hilbert Spaces for Causal Spacetimes . . . . . . . . . . . . . . 307
11.4.1 Hardy Spaces for the Future and Past . . . . . . . . . 308
11.4.2 Hardy Spaces for Spacetime Cones . . . . . . . . . . . 308
11.5 Spacetime Interactions (Kernels) . . . . . . . . . . . . . . . . 310
11.6 Normalization of Electroweak Spacetime . . . . . . . . . . . . 311
11.6.1 Central Correlation of Hypercharge and Isospin . . . . 311
11.6.2 The Mass Ratio of Spacetime . . . . . . . . . . . . . . 312
11.6.3 Internal Multiplicities . . . . . . . . . . . . . . . . . . 314
12 Masses and Coupling Constants 317

12.1 Harmonic Coefficients of Spacetime . . . . . . . . . . . . . . . 319
12.2 Translation Invariants as Particle Masses . . . . . . . . . . . . 321
12.3 Correlation of Spin and Masses . . . . . . . . . . . . . . . . . 323
12.4 Massless Interactions . . . . . . . . . . . . . . . . . . . . . . . 324
12.4.1 Massless Vector Modes . . . . . . . . . . . . . . . . . . 325
12.4.2 Massless Tensor Modes . . . . . . . . . . . . . . . . . 326
12.5 Spacetime Masses and Normalizations . . . . . . . . . . . . . 327
Bibliography 333
Index 337
Chapter 0
Introduction and Orientation
Einstein’s theory of general relativity identifies spacetime curvature with

a gravitational interaction. The theory is formulated primarily in a geo-
metrical language, while the group-theoretical concepts remain more in the
background. In an operational spacetime approach, the hierarchy will be
reversed.
A reversal of the priorities also for a mathematical treatment of manifolds
is announced by S. Helgason in the preface to his book Groups and Geometric
Analysis:
The role of group theory in elementary classical analysis is a rather
subdued one, the motion group of R3 enters rather implicitly in stan-
dard vector analysis,. . . In contrast our point of view here is to place a
natural transformation group of a given space in the foreground. We
use this group as a guide for the principal concepts.
I could not agree more with such a program, here for the operational treat-
ment of the spacetime manifold. In the following, the basic structures of
physics will be defined and considered, rather restrictively, via the representa-
tions of group and Lie algebra operations, which come in two forms; external,
acting on spacetimelike degrees of freedom, and internal, acting on chargelike
ones. The degrees of freedom are given by the dimensions of the represen-
tation spaces. The representations are characterized by invariants which will
be collected into the concepts of interactions and objects (particles).
A mathematical parallel is given by the Erlangen Program (1872) of Felix
Klein, with an operational characterization of geometries. A Lie group act-
ing on a manifold, e.g., an affine group acting on a vector space, consti-
tutes a Klein space. The related geometry is characterized, especially, by
its invariants, e.g., by dimensions for any affine geometry, and, in addition,
by volumes for special geometry, and, in addition again, by distances and
angles for a Euclidean geometry. If the vector space is real n-dimensonal, Rn ,
H. Saller, Operational Spacetime: Interactions and Particles, 1

Fundamental Theories of Physics 163, DOI 10.1007/978-1-4419-0898-8_1,
2 Chapter 0 Introduction and Orientation
the characterizing operation groups are, respectively, the general linear, the
special linear, and the special orthogonal groups, denoted by GL(n, R) ⊃
SL(n, R) ⊇ SO(n).
Also, the controversy between analytic and algebraic methods to for-
malize and investigate operator groups and Lie algebras, familiar from the
Schrödinger and Heisenberg–Pauli approach in the foundations of quantum
mechanics in the 1920s, is, apparently, not foreign to the mathematicians.
A. W. Knapp writes in the preface of his book Representation Theory of
Semisimple Lie Groups:
Beginning with Cartan and Weyl and lasting even beyond 1960,
there was a continual argument among experts whether the subject
should be approached through analysis or through algebra. Some to-
day still take one side or the other. It is clear from history, though, that
it is best to use both analysis and algebra; insight comes from each.
Apparently, for compact operations, as for electromagnetic phase transforma-
tions, rotations, isospin, and color, the algebraic methods suffice. However,
for noncompact, especially nonabelian, operations as given in the Lorentz
or Poincaré group with their continuous quantum numbers for boosts and
translations, the purely algebraic procedures are sometimes very cumber-
some and difficult to apply and the analytical tools prove extremely useful.
The difficulty of staying with algebraic methods only is illustrated by the
ingenious, but rather complicated, algebraic solution of the nonrelativistic
quantum hydrogen atom by Pauli compared with the analytic differential
equation approach by Schrödinger.
In the following, differential equations, e.g., equations of motion, are not
the basic starting points. I do not think that there exists something like one or
a set of “basic equations.” However, differential equations remain important;
they will be used to characterize representation properties and to solve eigen-
value problems. Also, Lagrangians and the action principle for the derivation
of differential equations will not be used as basic tools. The constitution of
kinetic terms for free objects (particles) and their separation from interaction
terms are, in general, possible only after understanding the origin of those free
objects, e.g., of the atoms as the bound states in the nonrelativistic Coulomb
potential. Interactions are a primary concept, free objects a secondary one —
in parallel to a curved manifold and its flat tangent spaces.
One of the main conceptual difficulties of general relativity is to get rid of
the evolutionarily engrained subconscious “absolutization” of spacetime and
its coordinates as described in 1949 by Einstein himself: “Why were another
seven years required for the construction of the general theory of relativity?
The main reason lies in the fact that it is not so easy to free oneself from
the idea that coordinates must have an immediate metrical meaning.” More
in 1950: “According to general relativity, the concept of space detached from
any physical content does not exist. The physical reality is represented by a
field.” And in 1954: “Space as opposed to ‘what fills space’ which is dependent
Chapter 0. Introduction and Orientation 3
on the coordinates has no separate existence . . . There is no such thing as an

empty space, i.e., a space without field.” Or Born (1962): “It is not space
that is there and that impresses its forms on things, but the things and their
physical laws that determine space.”
A “separate existence” of the coordinates for a spacetime “background”
may suggest a “quantization of spacetime,” e.g., by nontrivial commutators
[xa , xb ] = 0 of spacetime operators (noncommutative spacetime geometry) as
a mathematical generalization of the Born–Heisenberg position-momentum
commutators [x, p] = 0 in nonrelativistic quantum mechanics, which are dis-
tinguished as the historically first operational formulation. Such a “spacetime
quantization” may also lead to discussions of a discrete or grainy or smooth
spacetime — all concepts without an experimentally verifiable meaning, at
least in my understanding of spacetime and quantum.
What is the physical meaning behind the coordinates; what do they para-
metrize? In the following, operations are basic: The spacetime manifold will
be looked at as a reservoir for the parametrization of operation Lie groups or
subgroup classes of Lie groups. For example, the position coordinate does not
describe the actual position of a particle; it parametrizes, for flat manifolds,
the position translation behavior. Or, the curvature and the Ricci tensor
are classical concepts, visualizable with a “rubber spacetime.” In a quantum
framework, the geometrical curvature has, at most, a heuristic value, which
is interpreted via an operational metric, i.e., in terms of bilinear forms of
Lorentz transformations and of tangent translations.
A spacetime dynamics is a representation of the time and space defin-
ing operations. Spacetime is experienced by its operations, e.g., transla-
tions and rotations, implemented by interactions, constituting and acting on
particles.
Newton’s extremely successful approach to a formalization of physics con-
tained three important assumptions idolizing insights that seemed obvious,
but that were seen later to be prejudices and had to be modified.
The first assumption concerns time and space: They were taken to be
absolute, unchangeable — two God-given boxes (sensorium dei) or stages for
the actors of a physical dynamics. Time and space exist on their own. This
was in contrast to Leibniz’s characterization of space and time as relational
concepts. Newton illustrated absolute space or the absoluteness of an accel-
eration, or, in today’s words, the equivalence class of inertial systems, by the
properties of a rotating water-filled vessel.
The second assumption was for the actors: They were idealized by mass
points; i.e., the actors had no extension. Extended bodies were described by
integrals over mass points. This contrasted with Descartes’ view, where ma-
terial bodies were characterized by their space extension. Newton’s material
points were endowed with only one property — they had mass. Connected
with that, it is no wonder that the mass-related gravity structure played the
most important role in Newton’s mechanics: Gravity was the basic physical
law of nature, especially since electromagnetic experiences were restricted to
the strange behavior of amber and magnetic stones. The main task in the
solution of a dynamics was to give the time development of mass points in
space, the script for the actors’ movement on the stage.
The third tacit and “obvious” assumption concerns our language, i.e., how
we can talk about the physical nature and experiments. The relevant episte-
mology can be formulated clearly only in hindsight and with the knowledge
of its change in a quantum theory: In classical physics, the modality structure
in the formulation of the experimental results uses an observer-independent
absolute ontology and the classical logic, in the simplest case with yes–no
values and in an extended form with probabilities, e.g., in thermostatistics.
With the Lorentz group as the main structure in Einstein’s special relativ-
ity, space and time came closer to each other. However, they remain cleanly
separated, no longer as linear spaces, but with the metrical concepts timelike
(bicone) and spacelike. Absolute space and the ether were put to rest by the
Michelson–Morley experiment. Their absolute nature, now together as linear
Minkowski spacetime, was not questioned. The inhomogeneous Galilei group
as characterizing the structure of the time and space boxes was replaced by
the Poincaré group for the spacetime box.
With special relativity, the mass point idealization with time-parametrized
orbits in position gave way to Minkowski spacetime-parametrized fields as
proposed by Faraday and encoded in Maxwell’s electrodynamics. To use
mass points with an eigentime-parametrized motion was still possible, but
somewhat artificially restrictive — they became strangers in a field theory.
With the spacetime-parametrized fields, valued in spaces not necessarily
related to time and position, e.g., with units like Coulomb, came properties
in addition to mass, starting with the electric charge.
Einstein’s general relativity got rid of the absoluteness of flat Minkowski
spacetime.
Quantum theory changed the third epistemological assumption of classical
physics: Physics became a theory of operators. The modality structure of our
statements about experiments is characterized by an observer-relative ontol-
ogy, dependent on the experimental setup (“quantum relativity”), and prob-
ability amplitudes, formalized by scalar products in complex Hilbert spaces.
The essential quantum stucture is the Hilbert space representation of op-
erations, infinite-dimensional for nonabelian noncompact Lie groups. It is
remarkable that the set-oriented structure of measures and probabilities can
be erected on a Hilbert space, i.e., on a linear space with a definite scalar
product. This allows the quantum characteristic concepts of probability am-
plitudes and linear superpositions.
It may well be possible that there remain other unconscious absolutiza-
tions and idolizations in the formulation of our theories.
In the development of physics, the actual experimental precision was
parallel with and allowed useful step-by-step approximations of better theo-
ries, sometimes surprisingly beautiful on each stage. This can be illustrated by
the dynamics of the Kepler potential, where the apparently cyclic planetary
orbits (Copernicus) were improved by Kepler’s ellipses with small eccentric-

ities. They are approximations to the general relativistic rosettes (Einstein)
and were completely reinterpreted by quantum theory (Heisenberg, Pauli,
Schrödinger) after the ad hoc discretized semiclassical Bohr–Sommerfeld ap-
proximation. There is a related remark for the supplementary companionship
of experiment and theory by Maxwell (1864):
For the sake of persons of different types of mind, scientific truth
should be presented in different forms and should be regarded as
equally scientific whether it appears in the robust form and vivid
colouring of a physical illustration or in the tenuity and paleness of a
symbolic expression.
Both in quantum theory with, e.g., Hilbert spaces and C ∗ -algebras for
the formulation of probability amplitudes and probabilities, and in general
relativity with the geometrization of the gravitational interactions, e.g., by
identifying the Einstein curvature tensor with the energy-momentum tensor,
every physicist experiences speechlessly the wonder of an unbelievably deep
and simple, not trivial, mathematical formalization of physics. He or she
ponders the millennia-old question: In which sense are those structures “really
there” or “only” imposed by ourselves and, therefore, reflect our methods for
the understanding of nature, which, however, are, ultimately, also a part of
nature? That groups and Lie algebras in complex representations are such
a strong tool to describe the basic structure of physics is, at least for me, a
deep wonder.
An operational formulation of quantum gravity is still missing. Such a
formulation is proposed to start from operation Lie groups. Given a basic
operation group or Lie algebra, all physical structures can be interpreted
in terms of corresponding realizations or representations. All basic physical
properties are related eigenvalues or invariants. Operation group structures
have to be studied; the familiar action and the Lagrangian-based differen-
tial equations of classical theories are only one, sometimes, but not always,
appropriate formulation for the action of the corresponding Lie algebras.
In an operational spacetime approach, interactions and matter are repre-
sentations of the operations that constitute spacetime. There is no interac-
tion and matter without spacetime — that is easy to comprehend; there is no
spacetime without interaction and matter — that seems to be more difficult
to grasp.
In classical general relativity, the spacetime representations, used, e.g., in
the metrical coefficients or the curvature, are, in general, not unitary; there
is no Hilbert space structure. Quantum gravity will be proposed to rely on
Hilbert representations of spacetime operations. Hilbert representations of a
Lie group are decomposable into cyclic ones that are determined by char-
acteristic cyclic states. The possible basic cyclic ground states in quantum
theory, as exemplified by the weakly degenerate ground state of the standard
model of particles and interactions, may have their classical analogue in the
cosmological models of general relativity as exemplified by the Friedmann

universes. In classical gravity, those models, especially for noncompact uni-
verses with nonabelian operation groups, are described by nonunitary metri-
cal tensors. In a quantum description, their Hilbert representations have to
be considered.
In both theories — general relativity and quantum theory — metrical
structures play an important role: Einstein’s gravity is the dynamics of the
spacetime metric with causal signature (1, 3), whereas quantum theory works
with Hilbert spaces acted on by spacetime representations and a Hilbert
metric (Hilbert space scalar product). In the interaction-free case, i.e., for
flat spacetime, the invariant metric of a Lorentz group representation space is
1
also used for the scalar product of the Hilbert space. For example,the metric

13 ∼ ab 1
= δ for the three definite “spacelike” degrees in the metric 0 −13 of
0
the Minkowski representation is used for the three spin degrees of freedom of a
massive vector field, e.g., for the weak bosons. Matter, representing spacetime
operations, comes with a metric, possibly involving a nontrivial spacetime
function of positive type. This may lead to an alternative interpretation for
the connection of matter and metric as used in Einstein’s gravity.
There are basic numbers in basic physics, e.g., the two polarizations of
the photon, the three spin directions of the weak bosons, or the charges
±1 and 0 of the pions in relation to the electron charge. In addition to
these integers or rationals, ultimately related to winding numbers in the
representations of compact operations (Cartan tori), there seem2 to be ba-
sic numbers from a continuous spectrum like the mass ratios of elementary
particles or the strengths, i.e., coupling constants (normalizations) of ba-
sic interactions. These numbers characterize noncompact group operations
(Cartan planes) with their, in general, complicated Hilbert representations,
infinite-dimensional, if faithful.
A framework with both gravity and electromagnetic interactions has to
face the huge difference in their strengths, illustrated for mass points with
masses m and charges Q = ze by the ratio
EM
GR = − 4πQ 1 Q2
0 Gm1 m2
= −αS m2P mz11 zm22 ,
with the square of the electron charge e in Sommerfeld’s fine structure con-
2
stant αS = 4πe0 c ∼ 137
1
, with integer charge numbers z ∈ Z, and with
Newton’s constant G in the Planck mass m2P = c G . Its huge ratio with usual
m2
elementary particle masses, e.g., for the proton mP2 ∼ (3.6×1019)2 , seems dif-
p
ficult to obtain in a “natural way,” e.g., in polynomial equations P (x) = 0 for
m2 m2
x = mP2 . Even the logarithm of the ratio, e.g., log mP2 ∼ 88, has to face two
p p
decimal orders of magnitude. The most important spacetime operations come
11
n denotes the (n × n)-unit matrix.
2 With experimental errors and the rationals dense in the reals, Q = R, one can never
be sure.
from the orthochronous, i.e., causality-compatible Lorentz group SO0 (1, 3)

and its twofold cover group SL(2, C), i.e., its complex realization. With a
maximal abelian subgroup SO(2) × SO0 (1, 1) for axial rotations and Lorentz
dilations, it has rank 2. Therefore, two invariants characterize its represen-
tations. The historically first real four-dimensional Minkowski representation
D[ 12 | 12 ] of the Lorentz Lie algebra with angular momenta L and boosts B act-
ing on the spacetime translations and its products constitute a very restricted
class. These representations have only one nontrivial independent invariant:
0 ψ1 ψ2 ψ3

ψ1 0 ϕ3 −ϕ2 D[ 12 | 12 ] (L
2 − B 2 ) = −314 ,
D[ 12 | 12 ] (
ϕL + ψ B) = ψ2 −ϕ3
⇒
ψ3 ϕ2
0
−ϕ1
ϕ1
0
D[ 12 | 12 ] (L
B)
= 0.
The later-used complex representations for half-integer spin, faithful for

SL(2, C), e.g., a complex two-dimensional Weyl representation D[ 12 |0] with
Pauli matrices σ , have, at least, a nontrivial invariant also for the noncom-
pact classes of the rotation group, parametrizable by a future hyperboloid,3
SO0 (1, 3)/SO(3) ∼ = Y 3:

D[ 12 |0] (L
2 − B 2 ) = − 3 12 ,
D[ 12 |0] (
ϕL + ψ B) = ϕ
i 2 + ψ 2 ⇒
σ

σ 2
D[ 12 |0] (L
B)
= i 3 12 .
4
The nontrivial Hilbert representations, i.e., with definite unitary boosts,

are, in contrast to Minkowski and Weyl representations, necessarily infinite-
dimensional. The spin and SO(2)-related compact invariant L 2 − B
2 remains
rational as for the rotations. The hyperboloid and SO0 (1, 1)-related noncom-
pact invariant L B
are taken from a continuous complex spectrum.
With an understanding of the quantum structure for spacetime opera-
tions, there has to come a deeper understanding of the phenomenon of masses.
Einstein’s equations identify, up to a constant, the spacetime curvature R
and the energy-momentum tensor T, which, for flat spacetime, gives the
space densities of the spacetime translations P, represented with the mass
m2 as invariant. The one-dimensional eigentime translation group τ ∈ R is,
as Lie group, isomorphic to the dilation group eψ ∈ D(1), which, in the self-
dual form SO0 (1, 1) ∼ = D(1) with contractions and extensions (“Procrustes
transformations”) is a maximal abelian
ψ noncompact
subgroup ofthe Lorentz
group (Lorentz dilations) e ψσ3
= e 0
−ψ
∼
=
cosh ψ sinh ψ
= eψσ1 ∈
0 e sinh ψ cosh ψ
SO0 (1, 1) ⊂ SO0 (1, 3). In a Lorentz group-compatible theory of spacetime
representations, the two-dimensional group eψ0 12 +ψσ3 ∈ D(1) × SO0 (1, 1)
(Cartan spacetime), represented with two continuous invariants, plays an
important role.
It is useful to have a multiplicative and an additive notation for the
abstract real one-dimensional simply connected Lie group D(1) = exp R ∼ =
3 The symbols Ωs , with Ω looking faintly like a circle, and Y s , with Y looking some-
what like a hyperbola, are used for s-dimensional unit spheres and unit hyperboloids,
respectively, elsewhere often denoted by the symbols S s = Ωs and H s = Y s .
R = log D(1). This noncompact group will be called the dilation or causal or
translation group. Its classes with respect to the integers can be parametrized
by the points of a circle R/Z ∼ = Ω1 ∼= SO(2) ∼ = U(1) = exp iR. This com-
pact real one-dimensional group will be called the axial rotation or phase or
electromagnetic group.
An operation group determines its action spaces: A group G action de-
composes a space S into disjoint orbits, G • x for x ∈ S. Each orbit is
isomorphic to subgroup classes G/H, where the isotropy group H ⊆ G
is isomorphic to all fixgroups of y ∈ G • x. Therefore, the study of sub-
group classes G/H (also called homogeneous or symmetric or coset spaces)
and, for complex linear quantum theory, of their associated complex vector
spaces like the closure C(G/H) of its finite linear combinations (cyclic Hilbert
spaces) is of paramount importance. For example, flat Euclidean 3-position
and flat Minkowskian 4-spacetime are operationally described, as symmetric
spaces, by the orthogonal subgroup classes4 SO(3) × R3 /SO(3) ∼= R3 and
4 ∼
SO0 (1, 3) × R /SO0 (1, 3) = R , respectively, not by the manifold isomor-
4 5
phic abelian groups R3 and R4 .

With the action of a group on its irreducible spaces G/H, the choice of a
representative gr ∈ gH in a coset gH ∈ G/H leaves the freedom of the local
fixgroup H. Coset structures are closely related to gauge structures, where
the operations in G/H come in subgroup H-representations.
Representations of Lie operations involve group functions and distribu-
tions as elaborated, especially, by Vilenkin. One has to realize that not
only the basic exponentials like eiEt and e−|mx|, but all relevant, sometimes
very complicated, special functions like Bessel functions
or Neumann and
Macdonald functions for the on-shell contribution d4 q δ(q 2 − m2 )eiqx of a
Feynman propagator, used for basic physical structures, i.e., for quantum the-
ory and general relativity, are representation coefficients (matrix elements) of
Lie groups. With the knowledge of the representational relevance of special
functions, which are used quite often in this book, it is easier to get along
with their complicated expressions.
The connection between Hilbert representations of spacetime and parti-
cles is manifest for interaction-free structures. There, the time and position
translations are formalized by the abelian Lie operations in the groups R and
R3 with energies and momenta, respectively, as eigenvalues of the Hilbert
space representations R ⊕ R3 (x0 , x) −→ eiq0 x0 −iqx ∈ U(1).
“Free particle” structures, in general defined by Hilbert space representa-
tions of translations, start with the quantum harmonic oscillator, which im-
plements the unitary representations of the time translations R t −→ eiEt ∈
U(1), where the equidistanced energy spectrum is given by the invariants of
the product representations R t −→ (eiEt )k ∈ U(1), k = 0, 1, . . . , with the
eigenvectors |kE.
4 This R3 is meant in Helgason’s remark on the first page of this chapter.
5 Such a structural distinction for one underlying set may be illustrated by the plane
R2 , considered either as real vector space, where linearity is constitutive, or as topological
space, where open sets take this role.
More complicated are the representations of the semidirect product

Euclidean group SO(3) × R3 for flat position SO(3) × R3 /SO(3) ∼ = R3 ,
which are used for free scattering states. Nonrelativistic free scattering states
have “eigenvectors” 6 |P 2 , |h|;
ω , h, where the momentum q = P ω contains
the invariant P 2 = q 2 for the position translations R3 and, as eigenvalues,
∈ Ω2 ∼
the directions ω = SO(3)/SO(2) on the unit 2-sphere. For the axial ro-
tation fixgroup SO(2) around the momentum, the helicity |h| is the invariant
and h = ±|h| the eigenvalues for left- or right-handed polarization.
For special relativity, the action of the Galilei group SO(3) × R3 with
rotations and velocity transformations and its inhomogeneous extension by
time and position translations [SO(3) × R3 ] × [R ⊕ R3 ] are expanded to
the semidirect Poincaré group SO0 (1, 3) × R with the homogeneous simple
4
orthochronous Lorentz group SO0 (1, 3) and R4 as translations for relativistic

Minkowski spacetime SO0 (1, 3) × R4 /SO0 (1, 3) ∼
= R4 . According to Wigner,
7
a free elementary particle is described by a complex infinite-dimensional vec-
tor space acted on by an irreducible unitary representation of the Poincaré
cover group SL(2, C) × R4 . The “eigenvectors” |m2 , J; q, J3 contain the eigen-
values momenta q and the invariant mass m for spacetime translations R4 ,
and spin or helicity J with projection J3 , which characterize rotations SU(2)
in a rest system for massive particles and axial rotations SO(2) around the
momentum direction for massless ones.
“Flat” and “curved” are in correspondence with “free” and “interacting”
and, also, with time and space, as given in the following table, which should
not be taken too precisely, but only to indicate a parallelism of concepts:
Dynamics Manifold Group Diff. eq.

Time free flat abelian linear
Space interacting curved nonabelian nonlinear
Flat manifolds come with operation groups, especially translations, whose

representations characterize interaction-free particles. Flat manifolds cannot
“explain” the existence and properties of interactions and bound-state struc-
tures, which represent operations for curved manifolds, especially with non-
abelian action groups. The concept of a bound state with constituents is not
very useful if the bound-state energy is of the order of magnitude of the mass
of the “constituents.” Each group determines its Hilbert spaces. The Hilbert
space for free particles, characterized by the Fock ground state, is inappropri-
ate for nonabelian groups. If the particles reveal the spectrum of nonabelian
spacetime operations, it is very doubtful that they can be understood via
bound states of free particles. Free particles are irreducible Hilbert represen-
tations of the Poincaré group — the actually arising invariants, masses, and
spins (polarizations) cannot be explained by the Poincaré group.
6 In the eigenvector notation |I; w, the representation-characterizing invariants I stand
before the semicolon, the eigenvalues w after it, e.g., the spin SU(2)-eigenvectors |J; J3 .
7 A necessary extension of Wigner’s definition to unstable particles with a width is not
considered in the following.

A related quotation from the last public talk of Heisenberg (1975, my

translation):
It is unavoidable that we use a language originating from classical
philosophy. We ask: What does the proton consist of? Is the quantum
of light elementary or composite?, etc. However, all these questions
are wrongly asked since the words “divide” and “consist of” have lost
almost all their meaning. Therefore, it would be our task to adjust our
language, our thinking, i.e., our scientific philosophy, to this new sit-
uation that has been created by the experiments. Unfortunately, that
is very difficult. Therefore, there creep into particle physics, always
again, wrong questions and wrong conceptions, . . .
We have to come to terms with the fact that the experimental ex-
periences for very small and for very large distances no longer provide
us with an anschauliches Bild and we have to learn to live there with-
out Anschauung (something like “without familiar everyday pictures”).
In this case we realize that the antinomy of the infinitely small for the
elementary particles is resolved in a very subtle way — in a way that
neither Immanuel Kant nor the Greek philosophers could think of —
the words “to divide” lose their sense.
If one wants to compare the insights of today’s particle physics
with any earlier philosophy, it could only be the philosophy of Plato,
since the particles of today’s physics are representations of symmetry
groups — that’s what quantum theory teaches us — and, therefore,
the particles resemble the symmetric Platonic solids.
It is the main prejudice of this book that there is a level for our under-
standing of physics where it no longer makes sense to assume parts of par-
ticles, i.e., particles inside particles, where, however, it makes sense to talk
about interactions or, better, about operations and symmetries. Particles im-
plement only a subclass of physically relevant operations, their operations are
not complete for curved spaces with nonabelian operations for interactions.
Time as a real one-dimensional, necessarily abelian Lie group is flat;
abelian operations cannot implement interactions.
Riemannian manifolds with nonabelian Lie groups as used, e.g., for real
three-dimensional position can be curved, e.g., in the maximally symmetric
form of a compact 3-sphere Ω3 ∼ = SO(4)/SO(3), as used in Einstein’s static
universe, or of a noncompact 3-hyberboloid Y 3 ∼ = SO0 (1, 3)/SO(3) with non-
abelian action (motion) groups SO(4) and SO0 (1, 3), respectively. Both of
these curved positions parametrize classes of SO(3)-subgroups. The hyper-
bolic position Y 3 is represented in the nonrelativistic hydrogen atom with
the Coulomb potential in its Hamiltonian. The bound-state wave functions
−n
r
2J
ψLm (x) ∼ rL Ym L
( xr ) LN 2r
1+2L ( n ) e , with the principal quantum number an
SU(2)-multiplet multiplicity n = 1 + 2J = 1 + L + N , are representation
coefficients of position where the harmonic SO(3)-polynomials x −→ (x)L m =
L
rL Ym ( xr ) with the spherical harmonics Y representing the maximal compact
rotations SO(3) ⊂ SO0 (1, 3) and the remaining exponential e− n with the
r
Laguerre polynomials L the hyperbolic operations SO0 (1, 1) ⊂ SO0 (1, 3) in

a maximal abelian noncompact subgroup. In contrast to the positive momen-
tum invariants q 2 = P 2 (real momenta) for scattering, used in the spherical
d3 q
Bessel functions j0 (P r) = sinP Pr r = 2π|P q 2 − P 2 )eiqx as representation
| δ(
coefficients of the flat position SO(3) × R3 /SO(3), the bound structures for
the hyperbolic position SO0 (1, 3)/SO(3) have negative invariants q 2 = −Q2
(imaginary momenta) as seen at the momentum function dipole singularity in
3
the exponential wave functions Y 3 ∼ |Q|
= R3 x −→ e−|Q|r = dπ2q (q 2 +Q2 )2 eiqx .
These coefficients represent hyperbolic position with eigenvalues i q on a com-
2 2π 2 1
pact 3-sphere Ω3 with measure d3 q (q 2 +Q 2 )2 = |Q| for radius |Q| . The
⎪
⎪
invariant energies −2E = Q2 ∈ { (1+2J) 1 ⎪
2 ⎪J = 0, 2 , . . . } determine the quan-
1
tized curvature of the position hyperboloid Y . 3
The belief in the universality and completeness of flat Minkowski space-

time with its Poincaré group, of the particle fields, and of the Fock state is
reflected by the remark “Each interaction is mediated by a particle.” Such a
strong statement, illustrated by Yukawa’s formulation of nuclear forces with
m c
− π r
the Compton length of “exchanged” pions as range e r is not true and too
narrow. Also, a weakened formulation with the concept of “off-shell” or “vir-
tual particles” for interactions is slippery. Parallel to the distinction between
time(like) and space(like), the distinction between particles and interactions
does not vanish in a relativistic quantum field theory. Spacetime fields can
have non-particle degrees of freedom. For example, the four-component elec-
tromagnetic field contains both the Coulomb and the two photon degrees of
freedom, which, in a particle analysis with the axial rotations SO(2) around
the momenta, come as a rotation scalar and a dublet with polarization ±1
(left- and right-handed), i.e., the Coulomb interaction is not mediated by a
particle. A similar situation occurs for the 10-component gravitational field in
the flat spacetime approach with a ±2-polarization dublet for gravitons (par-
ticles) and eight nonparticle degrees of freedom. including an SO(2)-scalar
for the Newton interaction.
Masses cannot arise only as translation invariants for flat spacetime to
characterize free particles. For example, the masses of quarks, if confined,8
cannot be used as invariants for spacetime translations. What is the opera-
tional meaning of the different kinds of quark masses? What is the meaning
of a Feynman propagator for quarks? Does it make sense to ask if quarks are
stable, e.g., the up quark, or unstable, e.g., the top quark, and if they are
unstable, what is their width? Perhaps, the quarks for the parametrization
of the strong interactions implement only homogeneous (“local”) operations,
but no translations. They are introduced with color, hypercharge, isospin, and
Lorentz group properties, but, if confined, without translation properties; i.e.,
8 So far, there are arguments for color confinement; its rigorous mathematical proof as
a consequence of an unbroken SU(3)-gauge interaction is still missing.

they are not unitary Poincaré group representations, no particles according

to Wigner’s particle definition.
Also, the curvature of manifolds, which can be measured in squared mass
units, must not correspond to a particle. In general, masses from a continuous
spectrum can occur as invariants of any noncompact operation group. For
nonflat space, they may characterize-curvature related interactions and may
be measurable in coupling constants.
In an operational framework, particles are described by representation
coefficients, i.e., functions on a group, whereas interactions are, in general,
distributions, describing the tangent Lie group structure; i.e., they are related
to its Lie algebra and Lie algebra forms. This is familiar from the Lie algebra
structure of the gauge interactions. From a representation point of view, it
is understandable that interactions differ from bound-state wave functions
in the order of the singularity, e.g., the Yukawa interaction for hyperbolic
−|Q|r d3 q 1
position Y 3 with a simple momentum pole e r = 2π q 2 +Q2 e
2
i
q
x
from
the ground-state wave function of the nonrelativistic hydrogen atom, with a
3
momentum dipole e−|Q|r = dπ2q (q 2 |Q| +Q2 )2 e
i
qx

.
The “divergences” of interacting spacetime quantum field theories with the
undefined local products of spacetime distributions are caused by the expan-
sion of interactions, related to nonabelian operations of curved space, with
representation coefficients of abelian operations, especially spacetime trans-
lations for free particles. Renormalizable canonical quantum field theories
with kinetic terms for the free particles and gauge interactions, expanded
with the Fock ground state for free particles, like quantum electrodynam-
ics and the standard model of electroweak interactions, together with the
regularization-by-renormalization procedure, seem to be a viable method to
describe the scattering of free particles. Such a flat spacetime framework,
however, is inappropriate for understanding its ingredients — the parti-
cle spectrum and the interactions themselves. For this purpose, one has to
start from fundamental operations and symmetries, not from fundamental
particles or constituents. Heisenberg (1973): “The fundamental symmetries
define the underlying law which determines the spectrum of elementary par-
ticles. An analogue: The scattering wave functions for flat Euclidean space
SO(3) × R3 /SO(3) ∼ = R3 , e.g., the spherical Bessel functions sinP Pr r for the ro-
tation trivial case, are inappropriate for understanding the bound-state wave
functions of the nonrelativistic hydrogen atom, e.g., for the ground-state wave
function e−|Q|r , which are representation coefficients of the curved hyperbolic
position SO0 (1, 3)/SO(3) ∼ = Y 3.
For general relativistic spacetime, one may expect a generalization from
special relativity with the Lorentz group to the invariance under an even
larger motion group G ⊃ SO0 (1, 3), e.g., under all automorphisms GL(4, R)
of the tangent spaces, i.e., under the structural group of the local frames.
However, such a generalization seems to be inappropriate. As exemplified
by Friedmann universes, there are many different spacetimes with a tangent
Poincaré group. Tangent space groups with translations are rather “passive”;
they are for free objects. They do not describe the operations responsible
for the constitution of particles and their interactions. General relativistic
“nonlinear” spacetime should not be considered as an embedding manifold for
its tangent spaces, e.g., as an expansion of the Poincaré group to the (anti) de
Sitter group SO0 (2, 3) or SO0 (1, 4). The operations that constitute nonlinear
spacetime, not its flat spacetime tangent operations, are “responsible” for and
determine particles and interactions.
The periodic table of the elements is nonrelativistically explained by the
eigenstates of a Hamiltonian with the Coulomb potential. Its operational
background is the Lorentz group SO0 (1, 3), represented by scattering states,
and its compact partner SO(4), represented by bound states, as the corre-
lated product of two SU(2)-groups with the rotations SO(3) and the ad-
ditional Lenz–Runge classes on a 3-sphere Ω3 ∼ = SO(4)/SO(3), classically
visible in the perihelion conservation. The quantum structure of the Kepler–
Coulomb potential 1r is physically relevant, exactly solvable, and aesthetically
appealing. A nonrelativistic nonflat position is implemented by Hilbert space
representations of the 3-hyperboloid Y 3 . The periodic table reflects repre-
sentation invariants of the curved position Y 3 . Analously, the particle table
will be proposed to display invariants of representations of the operational
groups for relativistic four-dimensional spacetime D4 , embedding hyperbolic
position. The invariants of the D4 -representations are used as masses, spins,
charges, and coupling constants for their normalizations.
There are two types of long-range (massless) interactions that come with
either “unbroken” or “broken” symmetries: gauge interactions, like electrody-
namics for unbroken U(1)-symmetry, and Nambu–Goldstone fields for de-
generacy operations of a ground state, like for broken chiral U(1)-symmetry.
Usually, degeneracy operations G/H for a ground state with a symmetry
group H ⊆ G are characterized, quantitively, by a mass; i.e., the massless
Nambu-Goldstone fields in H-multiplets effect also a rearrangement of the
broken dilation D(1) properties. These structures will be proposed for an un-
derstanding of the long-range electromagnetic and gravitational interactions.
The chiral model of Nambu and Jona-Lasionio is an interacting model for
relativistic eigenstates. In addition to the nonrelativistic Kepler interaction,
its structures will play an important role as an illustration for understanding
the particle table, especially the chiral degeneracy of its ground state as an
example of long-range interactions.
Physical operations act both on spacetimelike degrees of freedom, then
called external operations, e.g., Lorentz transformations and translations,
and on chargelike ones, then called internal operations, e.g., hypercharge or
isospin. An understanding of space, time, interaction, and matter has to come
with an “integrative symbiosis” of external and internal operations. A unifica-
tion was first tried by Weyl; he implemented the electromagnetic operations
in the form of a locally acting dilation group D(1). Although experimen-
tally wrong concerning the specific group, the gauge principle proved ex-
tremely fruitful. Together with London, Weyl later replaced the noncompact
dilations by the compact U(1)-phase transformations. As of today, all internal

operations, hyperisospin U(2), and color SU(3), as exemplified by the stan-
dard model of electroweak and strong interactions, are from compact groups,
implemented as gauge operations that accompany each spacetime translation.
The main new and original proposal of this book is an operational
spacetime, called electroweak spacetime D4 , parametrizing the classes of the
internal hypercharge-isospin group U(2) in the general complex linear group
GL(2, C), i.e., the Lorentz cover group SL(2, C), extended by the causal
(dilation) and phase group D(1) × U(1). Its representations and invariants
for the bi-regular action group GL(2, C) × U(2) will be investigated with the
aim of connecting them, qualitatively and numerically, with the properties
of interactions and particles as arising in the representations of its tangent
Minkowski spaces. This is tentatively realized in the last chapter of this book.
Electroweak and gravitational interactions are distinguished representations
of spacetime D4 that rearrange dilation degrees of freedom. The position
curvature-related representations will be connected with the quarks, which,
in the standard model, have been introduced to parametrize the strong
interactions and, if confined, are entities that, with Wigner’s definition, are
not particles, i.e., not translation eigenvectors. The relation between posi-
tion curvature and strong interaction may not be so unnatural if Einstein’s
interpretation of gravity by spacetime curvature is remembered. Even if the
proposal of interactions and particles describing the harmonic analysis of
electroweak spacetime D4 is premature, too simple, or even plainly wrong, I
hope it can illustrate how a unified operational approach for the concepts of
spacetime, interaction, and matter may be concretized.
This book does not discuss astrophysical and cosmological problems; it is
intended as an operational analysis of spacetime with its consequences for in-
teractions and particles, not as an additional book on the classical differential
geometric treatment of relativity and gravitation. For this method, there are
many excellent books, where everybody has his or her favorites. In addition
to the classic of H. Weyl, Raum, Zeit, Materie (1923), with the conception-
ally ground-breaking introduction of the gauge principle, using the physically
wrong dilation group, I favor, as an introduction, the book of R. Sexl and
H. Urbantke, Gravitation und Kosmologie (1981), with a good and competent
mixture of physical and mathematical concepts.
This book addresses graduate students and scientists with an interest in
the structure of those basic physical theories that have some experimental
justification. Apparently, the richness of mathematically appealing forms is
inexhaustible; their beauty may be a trap for physicists who should be pri-
marily interested in their physical relevance. The scientific level of the book
is not undergraduate; it is not written as an introduction. It assumes knowl-
edge of and familiarity with conventional relativity, quantum mechanics and
quantum field theory. In a sense, the book is very conservative; there are no
flashy new and titillating revolutionary ideas, only concepts that have been
used already in the theories mentioned above. However, I try to delve into
them more deeply and more radically. For the mathematical tools, used
in the following, the two books of S. Helgason, Differential Geometry, Lie
Groups and Symmetric Spaces (1978) and Groups and Geometric Analysis
(1984), are recommended. For representation theory, I learned much from the
books of G.B. Folland, A Course in Abstract Harmonic Analysis (1995), and
A. Knapp, Representation Theory of Semisimple Groups (1986); for distribu-
tion theory, from the book of F. Treves, Topological Vector Spaces, Distribu-
tions and Kernels (1967). Since the mathematics is also not undergraduate
and by no means trivial, at least for me, I hope that I didn’t make too many
mistakes — in addition to the usual reckless treatment of mathematics by
physicists.
The following text contains parts considered in a first run in my books
Operational Quantum Theory I — Nonrelativistic Structures (2006) and Op-
erational Quantum Theory II — Relativistic Structures (2006), where some
of the spacetime structures worked with in this book have were introduced.
It also uses the standard mathematical concepts and notation given in those
books in more detail.
The first six chapters give, in an operational language, a short journey
through the conventional theories, while the last six chapters describe how
one should proceed to a more basic understanding of the mutual conditioning
of spacetime and interactions and matter.
Chapter 1
Einstein’s Gravity
In this chapter, Einstein’s gravity with the identification of the curvature of

a Riemannian manifold as a geometrical concept and interaction as a phys-
ical one is shortly exemplified by Schwarzschild–Kruskal spacetime and by
some properties of Friedmann spacetimes with Robertson–Walker metrics.
As a warm-up, the group-oriented operational language will be used in those
important and basic examples for classical macroscopic gravity.
1.1 Geometrization of Gravity

Einstein’s gravity is a dynamical theory of the spacetime metric and curvature
as determined by the energy and momentum of light and matter.
The geometrical curvature concept has its anschaulichen origin in the local
change of (derivative with respect to) an area; therefore, it has the dimension
of an inverse area: An orange skin breaks up if pressed flat on a table, in
contrast to a paper cyclinder, which has no area change. The paper cyclinder
is flat.1 A nontrivial curvature does not exist in one dimension. For example,
a circle Ω1 or a hyperbola Y 1 is flat, equally time, parametrized as a real
one-dimensional manifold. Proper curvature needs a two-dimensional area,
e.g., a submanifold of a real three-dimensional space. The curvature is called
positive or spherical, k = 1, for the compact 2-sphere Ω2 (orange skin) and
negative or hyperbolic, k = −1, for the noncompact 2-hyperboloid Y 2 as given
by a mass hyperboloid in the future cone of (1 + 2)-dimensional Minkowski
spacetime, which crumples (any part
of it) if pressed flat on a table. For n = 4
spacetime dimensions, there are n2 = 6 linear independent areas for three
time-spacelike curvatures and three space-spacelike ones, related to the six
real dimensions of the Lorentz group with three boosts and three rotations.
If an area is measured by a metric, the curvature, i.e., the area change,
contains its 2nd-order derivatives. The curvature R (for Riemann and, also,
1 Its “one-dimensional curvature” is sometimes called external in contrast to the “two-
dimensional internal” curvature of the orange skin described above.

18 Chapter 1 Einstein’s Gravity
∂
Ricci) involves 1st-order derivatives ∂ j = ∂x j
of the Riemannian connection
Γ, which is defined (see Chapter 2) with 1st-order derivatives of the signature
(1, 3)-metric g = ds2 = gli (x)dxl ⊗ dxi ;
1 i jp
2 gkp (∂ g + ∂ j gip − ∂ p gij ) = Γij
k,
i jl jp ip lij
∂ Γk − ∂ Γk − Γil
j il
p Γk + Γjl
p kΓ = R k .
The Einstein tensor Ř• = R• − g2 R•• is the combination of the Ricci tensor
R• with the metric multiplied curvature scalar R•• = gli Rli lij
• = gli Rj . In
Einstein’s gravity, the 2nd-order derivatives of the metric with the dimension
c2
of an area density [R] = m12 , multiplied with κ1 = 8πG ∼ 5.3 × 1027 kgm,
involving Newton’s constant G and the maximal action speed c ∼ 3 × 108 ms ,
are given by the energy-momentum tensor T with trace T• = gli Tli and the
kg
dimension of a mass density [T] = m 3,
Ř• + Λg = −κT, with Ř• = R• − g2 R•• ,

R•• = κT• + 4Λ.
or R• − Λg = −κŤ, with Ť = T − g2 T• ,
The spacetime properties, as given in R, determine the matter properties,

as given in T — or vice versa. Also, a term with the cosmological constant
[Λ] = m12 may be included for a ground state energy-momentum tensor, where
Λ is interpretable as the cosmological “background” curvature (below).
The Einstein and the metrical tensor are covariantly constant, ∇Ř• = 0,
∇g = 0. Only for flat spacetime is the energy-momentum tensor the position
density of the translations P a = c d3 x T0a (x) with the energy-momenta
as eigenvalues and the mass as invariant. There, particles can be defined as
acted on by representations of the Poincaré group SO0 (1, 3) × R4 .
In hindsight, Einstein’s epochal and ingenious identification of the geomet-
rical spacetime curvature with the gravitational interaction via the energy-
momentum tensor, can also be motivated by the insight that already in
classical point mechanics the Lagrangian involves, via the kinetic momen-
x2
tum term p2 = d dt2 , i.e., the free mass point Lagrangian, the flat position
2
metric dx (see Chapter 3) which has to be embedded into a spacetime met-
ric with the tangent property dx20 − dx2 for local flatness (special relativity).
A stationary action involves the extremalization of the length of the mass
point orbits, which is generalized to the geodesics on Riemannian spacetimes
with their extremal length property.
Gravity is characterizable by an action W with the normalization (gravity
−34
coupling constant) given by the Planck area 2 = κ c ∼ (0.8 × 10 m)2 :
grav •
W grav = Wgrav + Wmatter , Wgrav = |g| d4 x 2
2 R• ,
1
|g| = − det g.
2
Here ∼ 1.05 × 10−34 kg−ms is used as dimensional action unit only. The
action can be written with 2nd-order derivatives (Einstein–Hilbert) and 1st-
order ones (Palatini).
1.1 Geometrization of Gravity 19
The real four-dimensional spacetime manifold M(1,3) is assumed with

a causal structure, compatible with the orthochronous Lorentz group
SO0 (1, 3). It is visible by diagonalization2 of the metrical hyperboloid
to a local inertial system with an orthonormal standard form in the tan-
gent spaces Tx (M(1,3) ) ∼ = R4 . The existence of a special relativistic (“flat”)
local Lorentz structure formalizes the principle of equivalence, i.e., the ex-
istence of a local spacetime coordinate system without gravity interactions
(Einstein’s freely falling elevator). The local diagonalization of the metrical
tensor uses a tetrad (4-bein) e with representatives of the 10-dimensional
classes3 e(x) ∈∈GL(4, R)/O(1, 3) of the six-dimensional Lorentz group4 in
the 16-dimensional general linear group;

gli (x) = ela (x)η ab eib (x), η = 10 −1
0
3
, e(x) = (ela (x))l,a=0,1,2,3 ,
GL(4, R) e(x) ∼ = e(x)Λ(x), with Λ(x) ∈ O(1, 3).
Gravity is operationally implemented by the tetrad field. The tetrad op-
erations involve, in classes of the local Lorentz transformations, the maxi-
mal compact group SO(4) ⊂ GL(4, R) with two rotation groups SO(4) ∼
SO(3) × SO(3). The six angles of SO(4) describe the four-dimensional ori-
entation of the metrical hyperboloid. GL(4, R) has real rank 4: In diagonal
form, the 10 local operations of a tetrad representative are reduced to opera-
tions from a maximal noncompact abelian group D(1)4 ⊂ GL(4, R), i.e., to
the four gravity characteristic dilations:
GL(4, R) e(x) = eλ0 (x) O4 (x) ◦ D4 (x), O4 (x) ∈ SO(4), eλ0 (x) ∈ D(1),
−2λ2
e 0 0
e−3λ 0
D4 (x) = λ (x) ∈ SO0 (1, 1)3 , D3 (x) = 0 eλ2 −λ3 0 (x).
0 e D3
0 0 eλ2 +λ3
The four real continuous parameters determine the four lengths of the met-
rical spacetime hyperboloid, i.e., the four units for the time and position
translations in each tangent space Tx (M). They can be arranged in one over-
all dilation,
eλ0 (x) ∈ D(1), with e8λ0 (x) = det g(x) ◦ η = − det g(x) = ( det e(x))2 ,
−3λ
one relative time-position normalization e 0 λ
0
e 13
(x) ∈ SO0 (1, 13 ), and
two relative normalizations of the three position axes D3 (x) ∈ SO0 (1, 12 ) ×
SO0 (1, 1) in the metrical position ellipsoid.
For a spacetimewith SO(3)-space rotation invariance,
there remain three
e2(λ+λ3 ) 0 0
dilations, e.g., g ∼
= 0 −e2(λ−λ3 ) 0 . For a Friedmann universe
0 0 −e 2λ2
12
2 A real matrix is orthogonally diagonalizable, O ◦ M ◦ O T = diag M , O ∈ SO(n), if
and only if it is symmetric M = M T .

3 The doubled symbol g ∈ ∈G/H denotes a representative of a coset (class of a subgroup
H ⊆ G) g ∈ gH ∈ G/H.
4 The indices a, b, c, . . . from the beginning of the alphabet are “Lorentz group-active”;
the indices i, j, k, . . . from the middle of the alphabet are “ GL(4, R)-active.”
witha maximally symmetric space (ahead), there remain two dilations, e.g.,
g∼=
e2λ
0
0
2λ
−e
, and only one dilation, λ2 = λ, for maximally symmetric
21
3
spacetimes (de Sitter universes).
1.2 Schwarzschild–Kruskal Spacetime

There exist four-dimensional spacetimes, which have a geometry with a non-
trivial curvature and a trivial Ricci tensor. The homogeneous Einstein equa-
tions for a rotation-invariant metric of a rotation-invariant mass distribution
like a mass point,
η ab • δ ab
Tab (x) = δ0a δ0b mδ(x) ⇒ Tab (x) − 2 T (x) = 2 mδ(
x),
and their solutions embed the nonrelativistic position equation and, for the
mass point, the Newton potential −∂ 2 1r = 4πδ(x). It is given outside the
Schwarzschild radius 2m = mκ mG
4π = 2 c2 of the mass inside by the Schwarz-
schild metric — in three different parametrizations, called geodesic polar
), Cartesian (t, r, ω
(t, ρ, ω ), and Eulerian cooordinates (t, ψ, ω ):
⎧
⎪
⎪ dρ2
g = (1 − ρ )c dt − 1− 2m + ρ dω
2
m 2 2 2 2
⎪
⎪
⎪
⎪ m 2
ρ
⎪
⎪
⎨ 1− 2r
= 1+ m
m 4
c dt − 1 + 2r dx2
2 2
R• = 0 : 2r
⎪
⎪ e2ψ
= e−2ψ c2 dt2 − sinh 2 2 2
4 ψ m (dψ + sinh ψdω ),
2
⎪
⎪
⎪
⎪ 2 2 2 2 2 2 2 2
⎪
⎪ dω = dθ + sin
2θdϕ , dxψ = dr + r dω ,
⎩
m e
for ρ = r 1 + 2r = m sinh ψ > 2m .
It is remarkable that the time-independent nonrelativistic Newton potential

survives, in geodesic polar coordinates, in the general relativistic framework
in the form of a matrix element of a self-dual dilation:
3
λ3
ρ > 2m : eλ3 (ρ)σ = e 0 e−λ 0
3 (ρ) ∈ SO0 (1, 1),
e2λ3 (ρ) = gtt (ρ) = 1 − 2

m
ρ = V (ρ).
4
In general, the curvature tensor R = η de Rabc = RAB = RBA with
dabc
e
2 = 6 antisymmetric double indices A = da, B = bc, i.e., Rdabc = −Radbc ,
6+1
can be written as a symmetric (6×6)-matrix with 2 −1 = 20 independent

entries where one Biancchi identity condition has to be taken into account.
It is an orthogonally diagonalizable bilinear form for the six linear indepen-
dent spacetime “areas” or the six generators of the tangent Lorentz group
SO0 (1, 3), with the three “timelike” and three “spacelike curvatures” of two-
dimensional submanifolds related to bilinear forms for SO0 (1, 1) and SO(2),
respectively. In contrast to the invariant Killing metric5 of the tangent space
5∧ denotes total antisymmetrization, ∨ total symmetrization.
1.2 Schwarzschild–Kruskal Spacetime 21

Lorentz Lie algebra η ∧ η = − 103 −1 0
3
, the curvature tensor is called an
operational metric since it is related to the Killing metric of the Lie algebra
of the spacetime-characterizing motion group (see Chapter 3).
In the example of the Schwarzschild metric, the bilinear curvature form6
of the Lorentz Lie algebra is invariant under axial rotations SO(2):
⎛ 0101 ⎞
R
⎜ R0202 ⎟
⎜ R0303 ⎟
Rdabc (ρ) ∼
=⎜ ⎟(ρ)
⎝ R2323 ⎠
R1212
⎛ ⎞ R1313
2 0
−12
0 0

m ⎝ 0 0 0
⎠= 1 0 2 0
= ρ3 0 0 −2 0

m
ρ3 0 −1 ⊗ 0 −12 .
0 0 0 12
Schwarzschild spacetime embeds as position manifold a rotation paraboloid

P3 = P1 ×Ω2 , where Ω2 ∼= SO(3)/SO(2) is the 2-sphere. The global symmetry
group (isometry or motion group) is R×SO(3) with the axial rotations SO(2)
as the local invariance group. All matrix elements in metric and curvature
are representation coefficients of the global symmetry group, invariant under
the local group.
In the curvature matrix, the six-dimensional adjoint Lorentz group rep-
resentation is decomposed into two three-dimensional rotation group repre-
sentations, each of which is decomposed into a one- and two-dimensional
SO(2)-representation;

D(SO0 (1, 3)) = Dι (SO(3)) = Dκ (SO(2)),
ι κ
6 = 3 ⊕ 3 = 1 ⊕ 2 ⊕ 1 ⊕ 2.
The metrical components gjk can be connected with the Newton potential
as the nonrelativistic approximation g00 (x) = 1 − 2
ρm and the connection
coefficients Γijk with forces (see Chapter 3).
Manifolds have charts and coordinates which, in general, are useful only
), where ω ∈ Ω2 are the coordi-
locally. The Schwarzschild coordinates (t, ρ, ω
nates of the unit 2-sphere, have a coordinate singularity7 at the event horizon
ρ = 2m , connected with the dilation g00 (x) = e2λ3 (ρ) = 1 − 2
ρm for ρ ≥ 2m .
For a maximal extension of the coordinates for Schwarzschild spacetime, the
geodesics of photons are used (null-like coordinates with g = 0):
2
(1 − 2
ρm )c2 dt2 − 1−dρ2m = 0, dρ
dt = ±c(1 − ρ )
2
m
ρ
ct = +ρ + 2m log | 2
ρm − 1| + const. (outgoing),
⇒
ct = −ρ − 2m log | 2
ρm − 1| + const. (ingoing).
6 The explicit derivation of the curvatures from the metrical tensors is given in
Chapter 2.
7 A simple example for a coordinate singularity is the radial parametrization of the
dρ2
2-sphere dω 2 = dθ 2 + sin2 θdϕ2 = 1−ρ 2 2 2
2 + ρ dϕ with the equatorial singularity at ρ =
sin2 θ = 1.
Via the constants one transforms to either advanced or retarded gton–

Finkelstein coordinates, related to each other by time reversal t+ ↔ −t− ,
⎧
⎨ ct± = ct ± 2m log | 2
m − 1|,
ρ
advanced (t+ , ρ, ω
)
g = (1 − 2
ρm )c2 dt2± ∓ 4
ρm cdt± dρ − (1 + 2
ρm )dρ2
) ⎩
retarded (t− , ρ, ω
− ρ2 dω 2 .
Taking both, Schwarzschild coordinates can be replaced by lightlike ones. In
the metric, the “radial coordinate” ρ is defined implicitly:
(ct, ρ) −→ (ξ+ , ξ− ) : ⎫
ξ± = ct ± (ρ + 2m log | 2
ρm − 1|),⎪
⎬
ξ+ +ξ−
= ct, ⇒ g = (1 − 2
m
ρ )dξ+ dξ− − ρ2 dω 2 .
2
ξ+ −ξ− ⎪
⎭
2 = ρ + 2 m log | ρ
2
m − 1|
A maximal extension involves two charts. The coordinate singularity at the
event horizon ρ = 2m is removed by exponentiation of the lightlike coordi-
nates,
⎧
ξ± ⎨ ρ
− 1, ρ ≥ 2m ,
(ξ+ , ξ− ) −→ e± 4m = v ± u = e 4m 2
m
ρ±ct
⎩ (±1) 1 − ρ , ρ ≤ 2m ,
2
m
leading to the Kruskal coordinates (u, v, ω

) for Schwarzschild–Kruskal space-
time — now extended inside the horizon ρ ≤ 2m ;
⎧
⎨
2
m − 1(sinh 4
m , cosh 4
m ), ρ ≥ 2m ,
ρ ct ct

ρ
(u, v) = e 4 m
⎩ 1 − ρ (cosh ct , sinh ct ), ρ ≤ 2m ,
2
m 4
m 4
m
ρ
u2 − v 2 = (1 − ρ
2
m
)e 2m , 2uv ct
u2 +v 2 = tanh 2
m ,
ρ
−
dρ2
g = e 2m
ρ (du −
2
dv 2 ) − ρ2 dω 2 = (1 − 2
ρm )c2 dt2 − 1− 2ρm
− ρ2 dω 2 .
2m
The radial coordinate ρ is a function of u2 − v 2 . The metric involves a hyper-

−X
bolic Macdonald function k0 (X) = e X with X = 2
ρm and the Schwarzschild
length m = m cG2 , proportional to the mass m, as familiar from the Yukawa
1
potential, there with the Compton length, X = Lrm , Lm = m c , inversely
proportional to the mass m.
With respect to a singularity in the dilation parametrization, there is
an analogy of Kruskal and Schwarzschild coordinates (u, v) ↔ (ct, ρ), on
the one side, with, respectively, translation R × R+ and hyperbolic-dilation
SO0 (1, 1) × D(1)-orbit coordinates (ct0 , r0 ) ↔ (λ3 , ξ0 ) for flat spacetime, on
the other side:
√
−2ξ0 (sinh λ3 , cosh λ3 ), ξ0 ≤ 0 (spacelike),
√
(ct0 , r0 ) =
2ξ0 (cosh λ3 , sinh λ3 ), ξ0 ≥ 0 (timelike),
c2 t20 − r02 = 2ξ0 , 2ct0 r0
c2 t20 +r02
= tanh 2λ3 ,
dξ02
η = c2 dt20 − dr02 − r02 dω 2 = 2ξ0 dλ23 − 2ξ0 − r02 dω 2 .
1.2 Schwarzschild–Kruskal Spacetime 23
r0 is a function of λ3 and ξ0 . The Schwarzschild analogue coordinates (λ3 , ξ0 )

cannot be used on the lightcone ξ0 = 0. An exponential dilation coordinate
is possible either for timelike translations e2λ0 = 2ξ0 or for spacelike ones
e2λ0 = −2ξ0 .
The Kruskal coordinates (u, v, ω ) allow a causal interpretation similar
to flat Minkowski spacetime (ct0 , r0 , ω ): The region with u2 − v 2 < 0 is
outside the horizon (like Minkowskian spacelike), the region with u2 = v 2
(like lightlike) is the horizon, and the region with u2 −v 2 > 0 (like timelike) is
inside the horizon. With this spacelike-timelike analogy, it is understandable
that inside the horizon ρ ≤ 2m , the Schwarzschild–Kruskal metric is not
static; i.e., it has a time dependence, in contrast to the static character outside
ρ ≥ 2m .
The Schwarzschild metric is a special case of the Kerr metric which are
solutions of the homogeneous Einstein equations with only global SO(2)-
axial symmetry. In addition to the Schwarzschild length unit m = m cG2 , they
L
involve a second length unit a = mc , related to the angular momentum L. In
this case, the tetrad contains compact operations from SO(4) in addition to
noncompact dilations.
Other extensions of the Schwarzschild metric lead to the Reissner metric
with a length 2z = z 2 αS G 2 2
c3 = 8πz αS proportional to an electromagnetic
charge number z ∈ Z in a Coulomb potential αρS with Sommerfeld’s fine
structure constant αS ∼ 137 1
(see Chapter 3). All those metrics are special
cases of Kerr–Newman metrics with a trivial curvature scalar that reflects
the tracelessness T• (F) = 0 of the electromagnetic energy-momentum tensor
T (F) = gjk F F − 4 g F Fkj :
c il ij kl 1 il jk
g = (1 − 2
m L
R2 )(cdt − a sin2 θdϕ)2 − 1
1− 2Rm2L
dρ2
,
sin2 θ
− R2 dθ2 − R2 [(ρ
2
+ a2 )dϕ − acdt] 2

R2 = ρ2 + a2 cos2 θ,
with
2m L = 2m ρ + a2 sin2 θ − 2z ,
Kerr
a = 0;
z = 0
R• = 0
Schwarzschild Kerr–Newman
m = 0 : a,
z = 0 a,
z = 0
R• = 0 Reissner
R•• = 0
0; a = 0

z =
R•
• = 0
Globally SO(3)-invariant Reissner spacetime with a trivial curvature

2
2
scalar involves the dilation e2λ3 (ρ) = 1 − 2
ρm + ρz2 for 2
ρm − ρz2 ≤ 1
with Newton’s potential and the Coulomb potential related electromagnetic
contribution. It has a nontrivial SO(2)-invariant Ricci tensor with the
SO(2)-decomposition 4 = 1 ⊕ 1 ⊕ 2:
2

z dρ2
g = 1− 2
m
ρ + ρ2 c2 dt2 − 2
− ρ2 dω 2 ,
1− 2ρm + ρz2
⎛ 2m 32
⎞
ρ − ρ2
z 0 0 0
⎜ 2 ⎟
1 ⎜ −( m
ρ − ⎟
z
(ρ) ∼
0 )12 0 0
Rdabc
= ρ2 ⎜
ρ2
⎟,
⎝ 0 0 − 2ρm +
2
z
ρ2
0 ⎠
2
0 0 0 ( m − z )12
−1 0 0
ρ ρ2
Rab ∼
2z
• (ρ) = ρ4
0 1 0 .
0 0 −12
In general, the Ricci tensor (or the Einstein tensor) is a symmetric bilinear
form of the tangent translations.
In contrast to the “absolute” flat spacetime
1 0
Lorentz metric η = 0 −13 , Rab
ab
• = Rc
abc
is called an operational metric of
spacetime, possibly degenerate. It is trivial for chargeless mass points, z = 0,
i.e., for Schwarzschild and Kerr spacetime. The field equations identify, up
to a normalization, the operational metric with the energy-momentum ten-
sor. The curvature scalar R•• = tr R ◦ (η ∧ η)−1 is the normalization of the
operational Lie algebra metric.
1.3 Friedmann and de Sitter Universes

Robertson–Walker metrical tensors are used for cosmological models. The re-
lated Friedmann universes embed the three maximally symmetric operational
positions, given by spherical, flat Euclidean and hyperbolic manifolds, which
parametrize rotation group classes, with the time dependence given by the
position expansion factor (“radius”) t −→ R(t),
g = dt2 − R2 (t)dσk2 , dθ2 sin2 θ

dρ2 x2
d
dσk2
= 1−kρ2 + ρ dω = 2 2
x2 2
= dr 2
+ r2 dω 2 ,
(1+k 4 )
⎧ 3 dψ 2 sinh2 ψ
⎨ Ω ∼
= SO(4)/SO(3), k = 1,
with R3 ∼
= SO(3) × R3 /SO(3), k = 0,
⎩ 3 ∼
Y = SO0 (1, 3)/SO(3), k = −1.
With the introduction of half-integer spin structures after the Stern–

Gerlach experiments, the real rotation and Lorentz groups can be replaced
by their twofold cover groups, also real Lie groups, with their defining rep-
resentations on complex two-dimensional vector spaces with Pauli and Weyl
spinors, respectively. The maximally symmetric positions can be formulated
as classes of the spin group SU(2), where the isotropy group for Ω3 is the
diagonal SU(2):
Ω3 ∼
= [SU(2) × SU(2)]/SU(2), SU(2)/{±12} ∼ = SO(3),
R3 ∼
= SU(2) × R /SU(2),
3
Y3 ∼
= SL(2, C)/SU(2), SL(2, C)/{±12} ∼
= SO0 (1, 3).
1.3 Friedmann and de Sitter Universes 25
The Lorentz group SO0 (1, 3) is familiar as a special relativistic spacetime

group, i.e., for the spacetime tangent translations R1+3 . As the motion group
of hyperbolic position Y 3 , it does not involve time.
The sphere and the hyperboloid constitute a compact–noncompact pair
(see Chapter 2) as seen in the imaginary–real transition, which relates their
global Lie algebras,
= (iR)3 ⊕ (iR)3 and log SO0 (1, 3) ∼

log SO(4) ∼ = (iR)3 ⊕ R3 .
The Euclidean position arises by “flattening,” i.e., an Inönü-Wigner contrac-

tion, from the curved ones, where, for the hyperbolic position, the boosts are
contracted to the position translations. This is mathematically isomorphic to
the original Inönü–Wigner expansion of the Galilei velocity transformation
0←− 1
into the Lorentz group boosts SO(3) × R3 ←−c SO0 (1, 3) with the speed of
light as the contraction–expansion parameter.
Friedmann universes have two characteristic lengths t,s , or masses,
as intrinsic units for dimensionless coordinates, involving one character-
istic time t = Ht0 , e.g., the Hubble time H10 = R1 dR dt |t=t0 = dt |t=t0 ,
d log R
R(t) ∼ R(t0 )e H0 (t−t0 )

, with H0 ∼ 4.2 × 10 s ∼ 1.4 × 10 yr and
17 10
t = H0 c ∼ 1.2 × 1026 m, and a characteristic length for position, x =

xs ,
and position curvature
k2 . They are invariants for the action of the motion
s
groups of the spacetime manifold, which contains the direct group product
of R (time operations) and, respectively, rotations SO(4), Euclidean opera-
tions SO(3) × R3 , and Lorentz group operations SO0 (1, 3). The two lengths

2
determine one universe-characterizing number
t2 (se Chapter 11).
s
As noncompact manifolds, not as symmetric spaces, hyperbolic and flat
positions are isomorphic, Y 3 ∼ = R3 . The negative curvature −k = 2 is the
2 2
invariant ∂ a = a for representation coefficients a of abelian position sub-
groups, of compact SO(2) = Ω1 with (k, a) = (1, sin θ), = ±i, of non-
compact R with (k, a) = (0, r), and of noncompact SO0 (1, 1) = Y 1 with
(k, a) = (−1, sinh ψ), = ±1.
Cosmological models in general relativity can be seen as the analogue to
ground states with cyclic vectors in quantum theory (see Chapter 8). They
are characterized by a global motion group with a local invariance subgroup
H ⊆ G; e.g., SO(3) ⊂ SO0 (1, 3) for a hyperbolic Friedmann universe.
The local SO(3)-invariance of Friedmann universes shows up, via 13 , in
2
R̈
the curvature with time dependence R for timelike areas and ṘR+k 2 for space-
like ones, and in the Ricci tensor as symmetric bilinear forms (operational
metrics) of, respectively, tangent Lorentz Lie algebra and translations, invari-
ant under the local group, a Lorentz group subgroup,

Rdabc ∼
R̈R13 0
= − R12 2
−(Ṙ + k)13
,
0

Rab ∼
= 1 3R̈R 0 2
, 12 R•• = 3 R̈R+RṘ2 +k .
• R 2
0
2
−(R̈R + 2Ṙ + 2k)13
With the energy-momentum tensor of an ideal cosmological fluid, charac-

terized by a density and pressure, both only time-dependent t −→ (ρ(t), p(t)),
and a velocity field ui ul gil = 1 defining two projectors for time and position,

(ρ + p)ul ui − pgli = ρP0li − pP1li ,
Tli = P0li = ul ui = gli − P1li ,
with
Tab ∼ T• = ρ − 3p,
ρ 0
= 0 p13 ,
the time dependence of the position “radius” t −→ R(t) of a Robertson–

Walker metric is determined by the Einstein–Friedmann equations
η ab •
−Rab
• + 2 R • = κTab + Λη ab
2
3 ṘR+k 0 κρ + Λ 0
⇒ 2
2 +k = 0 (κp − Λ)13
− 2R̈R+ Ṙ
0
R2
13
R̈
= −κ ρ+3p + Λ
3,
⇒ 2
R
Ṙ +k ρ
6
Λ
R2 = κ3 + 3 .
−46
Solar system experiments give the bounds |Λ| < 10m2 for the constant cos-
mological “background” curvature, s = √1 > 1023 m for the length, and
|Λ|
|Λ|
κ < 5.3 × 10−19 m 3 (proton mass mp ∼ 1.67 × 10
kg −27
kg) for the mass
density.
For a static solution, i.e., for a trivial representation of the time transla-
tions and trivial pressure,
g = dt2 − R2 dσk2 , R2 = const., p = 0 ⇒ k

R2 = κρ
2 = Λ,
the ground-state density coincides, up to a factor, with the cosmological con-

stant ρ = 2Λκ . A static Friedmann universe is either Minkowskian, i.e., it has
trivial density and cosmological constant,
R4 : k = 0 ⇒ ρ = 0, Λ = 0,
2
2π
or its position is a 3-sphere with volume √ Λ3
. This is the Einstein universe
with global R × SO(4) and local SO(3)-invariance, i.e., the four-dimensional
cylinder, R × Ωs for s = 3,
k=1 ⇒ κρ
2 =Λ= 1
R2 > 0, Tab = 2Λ a b
κ δ0 δ0 ,
1 dρ2
R × Ω3 : g = dt2 − Λ ( 1−ρ2 + ρ2 dω22 ) = dt2 − 1 2
Λ dω3 ,

Rdabc ∼ ∼ 1 •
03 0 0 0
= −Λ 0 −13 , Rab
• = 2Λ 0 −13 , 2 R• = 3Λ.
2
The mass of Einstein’s universe is κ4π√ > 2 × 1052 kg. Its curvature embeds
Λ
the Killing form of the local invariance Lie algebra log SO(3); the Ricci form
embeds the Euclidean metric of the position translations. As a manifold, the
1.3 Friedmann and de Sitter Universes 27
3-sphere is isomorphic to the spin group Ω3 ∼

= SU(2), not as symmetric space,
where Ω3 has the action group SO(4). The cosmological term, which in its
scientific history oscillated between life and death in its presumed physical
relevance, was first used by Einstein in his Eselei (folly), which proposed,
for the first time, with the 3-sphere Ω3 a finite compact position without
boundary.
Einstein’s universe is unstable. For a nonflat stable Friedmann universe,
the representation of time translations has to be nontrivial.
A curved space can be embedded into a curvature-free spacetime: This is
possible for a Friedmann universe with hyperbolic space where the “radius”
has to have a linear time dependence R(t) = R0 ± t:
R = 0 ⇐⇒ Ṙ2 = −k = 2 ⇒ R(t) = R0 ± t,

k = −1 : ρ = −p = − Λ κ.
A curvature-free Friedmann universe with spherical space is not possible since

it would need a complex formulation R(t) = R0 ± it.
The Killing form η ∧ η of the Lorentz Lie algebra as curvature and the
Lorentz metric η of the tangent translations as Ricci tensor, both up to a
factor,

Rdabc ∼
= −RR̈ 13
0
0
−13 , Rab ∼ R̈
• = 3 R
1
0 −13
0
, 12 R•• = 6 R
R̈
,
arise for the hyperbolic, trigonometric, and linear time dependence of the
scale factor with integration constants mΛ and c0 :
Ṙ2 + k ⇒
R̈R = ⎧
⎧ 2
⎪
⎨
sinh(mΛ t+c0 ) sin(mΛ t+c0 )
, , ±t + c0 , k = −1, ⎨ mΛ , −m2Λ , 0,
mΛ mΛ
R̈
R(t) = cosh(mΛ t+c0 ) , k = 1, = m2 ,
⎪
⎩ c emm Λ
R ⎩ Λ
0
Λt
, k = 0, m2Λ .
R̈
The constant R is the invariant for the time representation.
These operational metrics characterize the three maximally symmetric
universes (de Sitter, Minkowski, anti-de Sitter) with a 10-dimensional global
symmetry and a local six-dimensional Lorentz group invariance. They are
the solutions of the Einstein–Friedmann equations for a constant energy-
momentum tensor and negative pressure,
Ṙ2 +k
p = −ρ ≤ 0 ⇒ R̈
R = R2 = κρ+Λ
3 = Im2Λ = const., I ∈ {1, −1},
e.g., without matter (ρ, p) = 0 and a possibly nontrivial ground-state energy-

momentum with m2Λ = |Λ| 3 .
R̈
The metric of a de Sitter universe with invariant R > 0 for a hyperbolic
time representation R −→ SO0 (1, 1) by the position “radius” arises in three
parametrizations for k = −1, 1, 0, reflecting the possible position submani-
folds Y 3 , Ω3 , and R3 , related to the SO0 (1, 4)-subgroups SO0 (1, 3), SO(4),
and SO(3) × R3 , respectively, that of an anti-de Sitter universe with invari-

ant R < 0 for a spherical time representation R −→ SO(2) by the position
R̈
“radius” in one parametrization for k = −1 with position Y 3 , and that of

R̈
a flat Minkowski spacetime with trivial invariant R = 0 for a linear time
representation R −→ R by the position “radius” in one parametrization for
k = −1. The corresponding global groups are the de Sitter group SO0 (1, 4),
the anti-de Sitter group SO0 (2, 3), and the Poincaré group SO0 (1, 3) × R4 .
All these universes have a Lorentz group as the local invariance group, i.e.,
they are constituted by Lorentz group classes. The “radius” grows or shrinks
exponentially, stays constant, or oscillates, respectively, with time:
de Sitter, Im2Λ > 0 : R(t) ∼ e±mΛ t , SO0 (1, 4)/SO0 (1, 3) ∼

= Y (1,3) ,
Minkowski, m2Λ = 0 : R(t) ∼ 1, SO0 (1, 3) ×
R4 /SO0 (1, 3) ∼
= R4 ,
anti-de Sitter, ImΛ < 0 : R(t) ∼ e
2 ±imΛ t
, SO0 (2, 3)/SO0 (1, 3) ∼
= Y (3,1) .
For s ≥ 1, both the de Sitter and anti-de Sitter universes are a one-
shell hyperboloid, spacelike SO0 (1, 1 + s)/SO0 (1, s) ∼
= Y (1,s) , and timelike
SO0 (2, s)/SO0 (1, s) ∼
= Y (s,1)
. As homogeneous spaces, they are isomorphic
for s = 1 and different for s ≥ 2. For s = 0, the de Sitter and anti-de Sitter
universes are hyperbolic time SO0 (1, 1) = Y 1 and cyclic time SO(2) = Ω1 ,
respectively.
The three maximally symmetric universes are related to each other by
Inönü–Wigner contraction with the time representation invariant m2Λ as the
contraction parameter:
sinh mΛ t
→ t ← sin mΛ t
mΛ ,
m2Λ → 0 : mΛ
SO0 (1, 4) → SO0 (1, 3) × R ← SO0 (2, 3).
4
In the (anti-)de Sitter spacetime, the translations of Minkowski spacetime

are expanded into a part of a simple group.
m2Λ = |Λ+κρ|
3 as intrinsic unit gives the normalization of the Ricci and
the curvature tensor, which, for maximal symmetry, coincide, up to a factor,
with the metrical tensor and its antisymmetric product, i.e., in orthogonal
coordinates with the Lorentz metric of Minkowski spacetime and the Killing
form of the Lorentz Lie algebra log SO0 (1, 3). The maximally symmetric de
Sitter and anti-de Sitter spacetimes are conformal (dilation-equivalent) to flat
Minkowski spacetime:

R = −Im2Λ g ∧ g ∼ 1 0
= −Im2Λ 03 −13 , I ∈ {1, −1},

R• = 3Im2Λ g ∼ = 3Im2Λ 0 −13 , 12 R•• = 6Im2Λ ,
1 0
dx2
g= (m2Λ x2 −I)2
, with x2 = x20 − x2 .
Chapter 2
Riemannian Manifolds
In this chapter, general structures of differential manifolds are given,

especially the operational aspects of Riemannian manifolds. Manifolds as
such are rather amorphous structures. Physically important and manifold-
characterizing are the operation groups, they are parametrizing. A Riemannian
manifold has a global and a local invariance group in addition to its tangent
Poincaré or Euclidean group; e.g., for the sphere Ω2 and the hyperboloid
Y 2 given by the rotations SO(3) and Lorentz transformations SO0 (1, 2),
respectively, as the global groups (motion groups), and, for both, the axial
rotations SO(2) as the local group and the Euclidean group SO(2) × R2 as
the tangent group. The Einstein cosmos R × Ω has R × SO(4) as the global
3
group, SO(3) as the local group, and SO(1, 3) × R4 as the tangent Poincaré
group.
After a discussion of Riemannian manifolds with maximal symmetries
and constant curvature, i.e., spheres, flat spaces and timelike hyperboloids,
the relationship between coset spaces of real simple Lie groups and manifolds
with a covariantly constant curvature as classified by Cartan is presented.
2.1 Differentiable Manifolds

General relativistic spacetime is assumed to be a differentiable manifold.
Real manifolds M ∈ dif R , assumed in the following as finite-dimensional,
connected, and “as smooth as necessary,” “look locally” like open subsets of
Rn — as topological spaces. They describe structures used for Lie groups
lgrpR ⊂ dif R and their tangent Lie algebras lgrpR −→ lagR (ahead).
The most important manifold structure is the unital commutative ring
with the manifold functions {f : M −→ R},

C(M) = (x, Cx (M)), Cx (M) ∼
= R,
x∈M

30 Chapter 2 Riemannian Manifolds
involving the manifold parametrizations by charts. There are many vector

spaces with manifold functions and distributions, which will be used in later
chapters. The functions C(M) will be assumed “as smooth as necesssary.”
With respect to manifold structures, the important difference between
R-linear with the real field R and C(M)-linear with the ring C(M) has to be
taken into account. R-linear structures that are not C(M)-linear are called
gauge dependent (more ahead).
Differentiable manifolds have “infinitesimal structures”, i.e., tangent struc-
tures. The R-linear derivations of the manifold functions from the unital
R-algebra C(M),
C(M) ∈ aagR : v : C(M) −→ C(M), v(f g) = v(f )g + f v(g),
are the vector fields of the manifold, denoted in short by T = T(M) for
tangent bundle (see Chapter 6). They constitute an R-Lie algebra via the
commutator. T is a module over the ring of functions C(M), patched together
by the local tangent spaces, which are R-Lie algebras lagR Tx (M) ∼
= Rn :

T = T(M) = der C(M) = (x, Tx (M)) ∈ modC(M) ,
x∈M
T ∈ lagR : T ∧ T −→ T, [ei , ej ] = ij k
ke .
In general bases, e.g., left-invariant vector fields for nonabelian Lie groups,
the Lie bracket must not be trivial, as in a locally always possible “translation”
basis [∂ j , ∂ k ] = 0. The Lie algebra structure constants ijk in anholonomic
bases are called anholonomy coefficients.
The manifold morphisms dif R (M1 , M2 ) = {ϕ : M1 −→ M2 } are consti-
tuted by the differentiable mappings. In general, the diffeomorphism group
o
of a manifold dif R (M, M), involving the reparametrizations, is too big to
form a Lie group in any reasonable topology. Manifold morphisms induce
ring morphisms and linear mappings of the tangent spaces:
M
⏐1 C(M1 ) T(M
⏐ 1) Tx (M
⏐ 1)
⏐ϕ ⏐ ⏐ ϕ , with ⏐ ∂ j |x −→ ϕj∗a (x)∂ a |ϕ(x) ,
−→ ⏐◦ϕ −→ ∗ ,
ϕj∗a (x) = ∂ϕ∂x
a (x)
.
M2 C(M2 ) T(M2 ) Tϕ(x) (M2 ) j
ϕ∗
Diffeomorphisms give automorphisms Tx (M) ↔ Tϕ(x) (M) of the tangent
spaces, with ϕ∗ (x) ∈ GL(n, R).
The tangent spaces of a direct product manifold are isomorphic to the
direct sum of the individual tangent spaces T(x1 ,x2 ) (M1 × M2 ) ∼ = Tx1
(M1 ) ⊕ Tx2 (M2 ).
Each point has a neighborhood U x where the vector fields T(U ) con-
stitute a free C(U )-module of dimension n = dimR M = dimC(U) T(U ). There
exist local dual bases, (n-beins, moving frames, repères mobiles), for vector
fields v = vj ej ∈ T and its C(M)-linear forms ω = ω j ěj ∈ TT with a local
decomposition of the identity
2.1 Differentiable Manifolds 31
idT(U) = ej ⊗ ěj , ěi , ej = δij .
Two frames with dual bases idT(U) = ej ⊗ ěj = ea ⊗ ěa with j, a = 1, . . . , n,

are related to each other by automorphisms,
ea (x) = eaj (x)ej (x), eaj (x) = ěj (x), ea (x),

j
ěa (x) = ea (x)ěj (x), eja (x) = ěa (x), ej (x),
a i i
ej (x)ea (x) = δj , ej (x)ejb (x) = δba ,
a
e11 (x) . . . en1 (x)
e(x) = 1 ...
n
∈ GL(n, R).
en (x) ... en (x)
An example is the different holonomic bases for different parametrizations:

⎧ j
⎨ dxi , ∂ = δi ,
⎪ j
b
idT(U) = ∂xj ⊗ dxj = ∂xa ⊗ dxa ,
∂ ∂
dxa , ∂ = δab ,
⎪
⎩ dx , ∂ = ∂xa (x) = ej (x).
a ∂xj ∂xj a
A parallelizable manifold even has a global frame, and the tangent bun-
dle is a free module, T(M) ∼ = C(M)n , not only locally. Vector spaces are
parallelizable manifolds. Parallelizable spheres are exactly1 Ω1 , Ω3 , and Ω7 .
Co- and contravariant tensor fields constitute the C(M)-linear tensor al-
! "
n
∼ C(M) contains
gebra (T ⊕ TT ). The highest Grassmann power TT =
the volume elements, invariant for the action of SL(n, R) × Rn and related

to each other by a dilation factor, e.g., in holonomic bases,
"
n
TT ě1 ∧ · · · ∧ ěn = |e| dn x, with ěa = eja dxj , |e| = det eja .
More generally, a volume element leads to isomorphisms between tangent and

cotangent products of equal dimension; for k = 0, . . . , n,
"
k "
n−k
: TT −→ T, ěl1 ∧ · · · ∧ ělk −→ l1 ...lk lk+1 ...ln elk+1 ∧ · · · ∧ eln .
The action of the vector fields on the ring of scalars and on themselves is
uniquely extended by Leibniz’s rule as R-linear Lie derivations of all co- and
contravariant tensor fields, denoted by Lv a = ad v(a) = [v, a]:
! !
v ∈ T : ⎧ (T ⊕ TT ) −→ (T ⊕ TT ),
⎪
⎪ [v, f ] = v(f ), f ∈ C(M),
⎨
[v, ω, w] = [v, ω], w + ω, [v, w], ω ∈ TT ,
with
⎪
⎪ [v, a1 ⊗ a2 ] = [v, a1 ] ⊗ a2 + a1 ⊗ [v, a2 ],
⎩
e.g., [v, f w] = v(f )w + f [v, w].
1 These three spheres SO(1 + s)/SO(s) ∼ Ωs ⊂ R1+s are associated with the three
=
division R-algebras for 1+s = 2n , n = 1, 2, 3, which exist in addition to R with the “sphere”
Ω0 = {±1}: the complex numbers C ∼ = R2 ⊃ Ω1 , Hamilton’s nonabelian quaternions
H∼= R4 ⊃ Ω3 , and Cayley’s nonassociative and nonabelian octonions O ∼ = R8 ⊃ Ω7 .
! !
Lie derivation is compatible with tensor grading, deg k T l TT = k − l.
One has the explicit expressions in a local chart:
vj ∂ j ∈ T : [vj ∂ j , ∂ k ] = −(∂ k vj )∂ j ,
[vj ∂ j , dxl , ∂ k ] = 0 = [vj ∂ j , dxl ], ∂ k + dxl , −(∂ k vj )∂ j
⇒ [vj ∂ j , dxl ], ∂ k = ∂ k vl , [vj ∂ j , dxl ] = (∂ j vl )dxj ,
e.g., [∂ j , ∂ k ] = 0, [∂ j , dxl ] = 0.
A tensor field a is invariant under Lie derivation for [v, a] = 0. This leads
to the invariance Lie algebra of a tensor field, a Lie subalgebra of T:
⎪
⎪
La = {v ∈ T⎪ ⎪[v, a] = 0} ∈ lagR .
By the definitions above, a local dual frame is invariant under all Lie deriva-
tions [v, ej ⊗ ěj ] = 0.
For a Lie subalgebra L ⊆ T, the unital algebra with its invariants is a
subalgebra of the full tensor algebra:
! ! ⎪
⎪
INVL (T ⊕ TT ) = {a ∈ (T ⊕ TT )⎪ ⎪[L, a] = {0}} ∈ aagR .
2.1.1 External Derivative

The external derivative from functions to 1-forms,
d : C(M) −→ TT (M), f −→ df with df, v = v(f ), v ∈ T(M),

holonomic bases: df = (∂ j f )dxj ,
acts via nilquadratic extension by Leibniz’s rule on the Grassmann algebra

of the contravariant tensor fields:
⎧
⎪
⎪ df, v = v(f ), f ∈ C(M),
⎪
⎪
# T # T ⎨ "k "
1+k
d : T −→ T ; ω k
∈ T T
⇒ dω k
∈ TT ,
⎪
⎪
⎪
⎪ d(ω k ∧ ω l ) = dω k ∧ ω l + (−1)k ω k ∧ dω l ,
⎩
d2 = 0.
The external derivative of a 1-form ω has as action on two vector fields (v, w):
ω ∈ TT : 2dω(v, w) = v(ω(w)) − w(ω(v)) − ω([v, w])

= 2vi wj (∂ i ω j − ∂ j ω i ),
for ω = ω k dxk , dω = ∂ l ω k dxl ∧ dxk , v = vi ∂ i , w = wj ∂ j .
This can be expressed in local dual bases with the anholonomy coefficients
by the Maurer–Cartan formula:
2děk (ei , ej ) = −ěk ([ei , ej ]) = −ij

k,
děk = − 12 ij
k ě i ∧ ě j .
2.2 Riemannian Operation Groups 33
"
p
The general Maurer–Cartan expression for p-forms ω ∈ TT reads
$
1+p
(1 + p)dω(v 1 , . . . , v 1+p ) = (−1)1+a v a ω(v 1 , . . . , v̂ a , . . . , v 1+p )
$
a=1
+ (−1)a+b ω([v a , v b ], v 1 , . . . , v̂ a , . . . , v̂ b , . . . , v 1+p ).
a<b
2.2 Riemannian Operation Groups

In addition to real parametrizability and differentiability, manifolds can come
with more specific operations, e.g., related to metrical structures.
A manifold is endowed with a Riemannian structure (M, g) ∈ rdif R by a
tensor field g, which yields symmetric nondegenerate bilinear forms as metrics
of the tangent spaces:
g(x) : Tx ∨ Tx −→ R, g(el (x), ei (x)) = gli (x),
g = gli ěl ⊗ ěi ∈ TT ∨ TT .
The metric of a direct product of two Riemannian manifolds (Mi , gi ) is a

linear combination of the factor metrics with any nontrivial scalar factors
R αi = 0:
M1 × M2 = M : g(x1 , x2 ) = α2 g1 (x1 ) + α1 g2 (x2 ).
If a Riemannian manifold is embedded in a bigger one, there exists a projec-

tion M −→ M1 also for the metric g(x) −→ g1 (x1 ).
By multiplication with the appropriate power of the metric determinant
for the dilations D(1) ∼= GL(n, R)/SL(n, R), the SL(n, R) × Rn -invariant
measures d x become the up-to-a-constant unique GL(n, R) ×
n Rn -invariant
measure | det g(x)|d x.
n
2.2.1 Metric-Induced Isomorphisms

A nondegenerate metric d of a finite-dimensional vector space involves an iso-
d
morphy with the dual vector space V ∼
= V T . Therefore, tangent and cotangent
spaces of a Riemannian manifold are g-isomorphic:
T ej −→ gjk ěl ∈ TT .
The cotangent (inverse) metric is g−1 = gli el ⊗ ei ∈ T ∨ T. In composi-

tion with the dual isomorphy, the volume element-induced isomorphies

n give

isomorphisms of tangent space powers with equal dimensions nk = n−k :
"
k "
n−k "
n−1
T ∼
k = 0, . . . , n : g ◦ : T −→
T T
T , e.g., T = TT .
The external powers of (dxi )ni=1 can be used for the integration of tensor
fields over the manifold and submanifolds.
The metrical reflection is defined for symmetric tensor fields with even
grade, starting with the scalars C(M) f −→ −f , and the grade-2 sym-
metric tensor fields Sij = Sji , which use the metric multiplied trace with a
dimension-dependent coefficient n2 :
⎧ • •
⎨ Š = S − n gS , S = Š − n gŠ ,
2 2
⎪
TT ∧ TT S ↔ Š ∈ TT ∧ TT , S• = gij Sij = −Š• = −gij Šij ,
⎪
⎩ ˇ
Š = S.
The reflection-symmetric tensors are traceless, the reflection-antisymmetric

tensors are the metric with the normalized trace, while the reflected metric
is its negative:
S = Š ⇐⇒ S• = 0, S = −Š ⇐⇒ S − ng S• = 0,
ǧ = −g.
Only in four spacetime dimensions, where n2 = 12 , does the reflected Ricci

tensor Ř• coincide with the Einstein tensor R• − g2 R•• , which is defined with
the factor 12 in any dimension. In causal spacetime, M(1,3) , a reflected “time-
like” tensor in a Minkowski basis is proportional to the definite Euclidean
form δ ab , not to the indefinite Lorentz form η ab :
η ab • δ ab
M(1,3) : Sab = δ0a δ0b M ↔ Šab = Sab − 2 S = 2 M, S• = M = −Š• .
2.2.2 Tangent Euclidean and Poincaré Groups

The metrical structure of a Riemannian manifold defines three operation
groups: its characterizing global group and local invariance group and the
not-so-specific tangent Poincaré or Euclidean group.
Two metrics of a Riemannian manifold define isomorphisms of the tangent
spaces:
g ◦ g−1 : T −→ T, ej −→ gjk gkl
l
e,
−1
g (x) ◦ g (x) ∈ GL(n, R).
With a fixed subgroup H ⊆ GL(n, R), they are H-equivalent if g (x) ◦
g−1 (x) ∈ H for all x ∈ M. This equivalence relation collects the metrics
into disjoint classes GL(n, R)/H, the n-bein manifold modulo H. The local
transformation from one metric to the other gives representatives of the equiv-
alence classes, defined up to local H-transformations,
g = eT ◦ g ◦ e, e(x) ∈∈GL(n, R)/H.

A special case is given by the center GL(R)1n ⊆ GL(n, R) as direct

factor for the volume invariance group GL(n, R) ∼ = GL(1, R)×SL(n, R). Two
metrics are conformal to each other (dilation-equivalent) if the isomorphisms
above are the identity up to a scalar factor, a dilation from the unit connection
subgroup D(1) ⊂ GL(1, R):
D(1)
g ∼ g ⇐⇒ g ◦ g−1 = e2γ idT(U) , gjk (x)g kl (x) = e2γ(x) δjl .
The local orthonormalization of a Riemannian metric displays a character-

istic signature2 (t, s) and allows the definition of a local n-bein field (boldface
e) with respect to a fixed constant metric η:

1t 0
g = ds2 = η ab ěa ⊗ ěb , gjk (x) = eja (x)η ab ekb (x), η = 0 −1s .
This defines an orthogonal Lorentz group O(t, s) with the unit connection
subgroup SO0 (t, s). It is a rotation group for definite signature ts = 0 and
a proper Lorentz group for causal signature (t, s) = (1, s). In contrast to the
Lorentz group, the overall dilations D(1n ) cannot be characterized as the
invariance group of a metric.
The symmetric tensor g has 1+n 2 real parameters, which are taken over
completely by n-beins from the equally dimensional manifold GL(n, R)/SO
0
(t, s) as parameters for the local orthonormalization by n dilations and n2
rotations, i.e., by a representative, defined up to local Lorentz transforma-
tions:
g = eT ◦ η ◦ e, e(x)
∈ D(1)×SO
n
0 (1, 1) n
× SO(n) ∈ GL(n, R)/SO0 (t, s),
n−1
1+n
2 =n − 2 =n+ 2 .
2
The Lie algebra log SO0 (t, s) is simple for n ≥ 3 with the exception
of the two semisimple Lie algebras log SO(4) = log SO(3) ⊕ log SO(3)
and log SO0 (2, 2) = log SO0 (1, 2) ⊕ log SO0 (1, 2). There are characteristi-
cally different Riemannian manifolds M(t,s) with isomorphic tangent groups,
e.g., spheres Ωs and hyperboloids Y s , both with Euclidean tangent groups
SO(s) × Rs .

Together with the tangent translations Rn , one obtains the 1+n2 para-
metric tangent Poincaré group with homogeneous Lorentz group, sometimes
called pseudo-Euclidean and pseudo-Riemannian for ts = 0,
SO0 (t, s) ×
Rn , t + s = n ≥ 1.
It is a proper Euclidean group with metrical ellipsoids for ts = 0, e.g., for

the spheres Ωs and the hyperboloids Y t , and a proper Poincaré group with
metrical “timelike” hyperboloids for (t, s) = (1, s).
Each Lie group with a semisimple Lie algebra has “its” Lorentz and
Poincaré group (ahead).
2 Obviously, the natural numbers (t, s) should remind us of time and space. In general,
it will be difficult to confuse the dimension t ∈ N with a coordinate t ∈ R.

The Lie algebra of the tangent Poincaré group,

Rn ∼ Tx ∼ Rn ∼
n 1+n
log SO0 (t, s) ⊕ = Tx ∧ Tx ⊕ = R( 2 ) ⊕ = R( 2 ) ,
⎪
⎪
has the brackets in an orthonormal basis {Lab , P a ⎪
⎪a, b = 1, . . . , n}:
[Lab , Lcd ] = η ac Lbd − η bc Lad − η ad Lbc + η bd Lac ,
[P a , P b ] = 0,
[Lab , P c ] = η ac P b − η bc P a .
The defining n-dimensional Minkowski representation of the Lorentz (orthog-
onal) group with A = 1, . . . , t and α = t + 1, . . . , t + s = n,

λAB = −λBA
Dn (L) ∼
λAα = λαA
log SO0 (t, s) L = λab Lab −→ = λβα = −λαβ
t λβB
R ⊗ Rt R ⊗R
t s
∈ Rs ⊗ Rt Rs ⊗ Rs
= AL(Rn ),

e.g., for log SO0 (1, 3): D4 (L) ∼
0 λ0α = λα0
= λβ0 λβα = −λαβ
, α, β = 1, 2, 3,
is definite unitary only for the compact rotation groups SO(n). The related
(1+n)-dimensional representation of the Poincaré group with the noncompact
translations

Rn −→ Dn0(L) P0 ∈ AL(R1+n )
log SO0 (t, s) ⊕
is never definite unitary.

The faithful definite-unitary, i.e., Hilbert representations of the Euclidean
and Poincaré groups, especially of SO(3) × R3 for nonrelativistic scattering
in 3-position and of SO0 (1, 3) × R for particles in Minkowski spacetime,
4
massive and massless, are infinite-dimensional. The tangent Poincaré group

contains the operations for free particles.
2.2.3 Global and Local Invariance Groups

The morphisms for Riemannian manifolds ϕ : M1 −→ M2 are compatible
with the metrics g2 (ϕ.x, ϕ.y) = g1 (x, y). The metric-associated global invari-
ance group Gg of (M, g) ∈ rdif R consists of the metric-compatible diffeomor-
phisms with g ◦ (ϕ × ϕ) = g (isometries, Riemannian automorphisms).
The infinitesimal formulation: A metrical tensor field defines its Killing
vector fields by the Lie derivations from its invariance Lie algebra. This leads
to a covariant functor from Riemannian manifolds to real Lie algebras:
⎪
⎪
rdif R (M, g) −→ Lg = {v ∈ T⎪ ⎪[v, g] = 0} ∈ lagR .
1+n

With the 2 metric components, one obtains a system of 1+n 2 differential

equations for the n functions {vp }np=1 :
[vp ∂ p , gij dxi ⊗ dxj ] = [vp (∂ p gij ) + gpj (∂ i vp ) + gpi (∂ j vp )]dxi ⊗ dxj ,
vp ∂ p ∈ Lg ⇐⇒ vp (∂ p gij ) + gpj (∂ i vp ) + gpi (∂ j vp ) = 0
for all i, j ∈ {1, . . . , n}.
Each one-dimensional manifold has one Killing field and R as invariance

Lie algebra:
g = b2 (τ )dτ 2 : v∂τ b2 + 2b2 ∂τ v = 2b∂τ (vb) = 0 ⇒ v∂τ = 1b ∂τ .
For two dimensions, there are three equations for the Killing fields v = v0 ∂τ +
v1 ∂ρ :
⎧
⎨ (v0 ∂τ + v1 ∂ρ )b2 + 2b2 ∂τ v0 = 0,
g = b dτ ± a dρ :
2 2 2 2 (v0 ∂τ + v1 ∂ρ )a2 + 2a2 ∂ρ v1 = 0,
⎩
b2 ∂ρ v0 ± a2 ∂τ v1 = 0.

For the metrical tensor η = ηab dxa ⊗ dxb ∼
1t 0
= 0 −1s of the vector
space R as Riemannian manifold, the Killing fields constitute a Poincaré Lie
n
algebra:
(Rn , η) −→ Lη = log SO0 (t, s) ⊕ Rn ,
va ∂ ∈ Lη ⇐⇒ ∂ v + ∂ b v a = 0 ⇒ v a (x) = (λab − λba )xb + ξ a ,
a a b
va ∂ a = λab (η ac xc ∂ b − η bc xc ∂ a ) + ξa ∂ a ∼
= λab Lab + ξa P a .
The Poincaré Lie algebra is represented by Lie derivations,
Lab −→ η ac xc ∂ b − η bc xc ∂ a , P a −→ ∂ a .
For a metric without Killing vectors, the manifold has no symmetry, Lg =
{0}. If a metric does not depend, in a parametrization, on a coordinate,
∂xk g (x) = 0 for all i, j ∈ {1, . . . , n}, the corresponding derivative v = ∂
∂ ij k
characterizes a one-dimensional invariance Lie algebra. More generally, for

each Killing field v ∈ Lg , there exists a parametrization with v = ∂ k . Such a
one-dimensional Lie algebra R leads by integration locally to the noncompact
covering Lie group exk ∈ D(1) = eR ∼ = SO0 (1, 1) eσ3 xk or to its compact
quotient group e iσ2 xk
∈ SO(2) ∼
= R/Z.
Linear independent Killing vectors {v K }dK=1 of (M, g) define a d-dimen-
sional invariance Lie algebra Lg of the metric with cover group:
rdif R (M, g) −→ Lg −→ exp Lg ∈ lgrpR .
Under appropriate smoothness and connectivity conditions, at least lo-
cally in a chart, a locally isomorphic d-dimensional quotient Lie group Gg
(classes of exp Lg ) is the global symmetry or motion or isometry group of the
Riemannian manifold:
o ⎪
⎪
exp Lg ∼ Gg = {ϕ ∈dif R (M, M)⎪ ⎪g(ϕ.x, ϕ.y) = g(x, y)}.
1+n
Its maximal dimension is 2 : The maximal global symmetry group is either

an orthogonal group, e.g., for spheres and hyperboloids (ahead), or a Poincaré
group for a flat vector space,

dimR Gg ≤ 1+n 2 = dimR Gmax (t, s),

G max
(t, s) ∈ {SO0 (t, 1 + s), SO0 (1 + t, s), SO0 (t, s) ×
Rn }.
max

Gg must not be a subgroup of G
The actual motion group (t, s). It is
1+n
a subgroup of an 2 -dimensional product of maximal motion groups —
possibly also from smaller-dimensional manifolds.
Gg acts on the manifold M, which, therefore, is the disjoint union of Gg -
orbits with fixgroups (isotropy subgroup) Hgι ⊆ Gg for orbit representatives,
% %
M∼ = Gg • xι ∼
= Gg /Hgι .
representatives xι ι
The fixgroups of the points of one orbit, g1,2 • x ∈ Gg • x, are conjugate to

each other, i.e., related by inner Gg -automorphisms with g1 g2−1 . There may
be disjoint orbits with isomorphic fixgroup.
The local invariance group of a Riemannian manifold is defined by the
fixgroup of all manifold points, i.e., by the intersection of appropriate repre-
sentative fixgroups for all orbit types (examples ahaed):
&
Hg = Hgκ ⊆ Gg .
κ
For a maximal global symmetry group, the local invariance group is maximal
and isomorphic to the tangent Lorentz group. In general, the local invariance
group is a Lorentz subgroup:
Hg ⊆ H max (t, s) = SO0 (t, s).
The metrical coefficients are representation matrix elements of the global
symmetry group Gg , invariant under the action of the local group Hg :
%
Gg /Hgι ∼= M x −→ gjk (x), Gg /Hgι ⊆ Gg /Hg .
ι
Some examples: A manifold with trivial global symmetry group {1} is the
%
union of its points M = {x}. A one-dimensional manifold is isomorphic to
x∈M
R∼= SO0 (1, 1) ∼

= Y 1 or to SO(2) ∼= Ω1 . A torus is a product of axial rotation
groups T = SO(2) × · · · × SO(2).
n
Spheres and hyperboloids are one orbit. The metrical coefficients are
spherical harmonics Y : SO(1 + s)/SO(s) ∼ = Ωs −→ R and their hyper-
∼
bolic counterparts SO0 (1, s)/SO(s) = Y −→ R. For example,
s
the Ω2 -
√
metric involves the spherical harmonics 4πY00 (ϕ, θ) = 1 and 4π 2
5 Y0 (ϕ, θ) =
1 − 32 sin2 θ. Its three Killing fields are given by the angular momenta — in
polar coordinates,
∂ ∂
cos ϕ cot θ ∂ϕ + sin ϕ ∂θ
Ω : 2
dω22 2 2 ∼
= dθ + sin θdϕ , L2
= ∂
sin ϕ cot θ ∂ϕ − cos ϕ ∂θ
∂
, [La , Lb ] = − abc Lc .
∂
∂ϕ
The local invariance group SO(2) ⊂ SO(3) is parametrizable by a third

coordinate in (α, ϕ, θ). This coordinate does not show up — neither in the
jk
metrical coefficients, ∂g 2
∂α = 0, nor in the basis as dα .
In contrast to the metrical coefficients for the compact spheres, which are
definite-unitary matrix elements, i.e., of Hilbert representations, those of the
noncompact hyperboloids are not Hilbert representation coefficients; e.g.,
Y 2 : dy22 = dψ 2 + sinh2 ψdϕ2 .
Minkowski spacetime R1+s , s ≥ 2, has metrical coefficients dx2 = dx20 −

dx2 with a trivial representation of the Poincaré group. The action of the
Lorentz group SO0 (1, s), s ≥ 2, on the translations has four fixgroup types
for the “many” orbits: SO0 (1, s) for the trivial translation, SO(s) for timelike
translations, SO0 (1, s−1) for spacelike translations, and SO(s−1) × Rs−1 for
lightlike translations. The common fixgroup for all translations is SO(s − 1),
i.e., the axial rotations SO(2) for four-dimensional spacetime. Vector spaces
Rn are one orbit with respect to their own action as translations. Vector
spaces Rt+s as Riemannian manifolds have the Poincaré group as both the
tangent and motion groups; the local invariance group is the Lorentz group
SO0 (t, s) × Rn /SO0 (t, s) ∼
= Rn .
Summarizing: The three Riemannian operation groups for a Riemannian
manifold (M, g) are
Global symmetry (motion) group Gg
Tangent Poincaré group
Gmax (t, s) ∈ {SO0 (t, 1 + s), SO0 (1 + t, s),
SO0 (t, s) ×
Rn
SO0 (t, s) ×
Rn }
M(t,s) :
Local invariance group
Hg ⊆ Gg ∩ SO0 (t, s)
In a physical context, the global group Gg and its generators give the sym-
metries and conserved quantities of a dynamics, like energy for time trans-
lations and static spacetimes or angular momenta for rotation-symmetric
spacetimes (see Chapter 3), whereas the local subgroup Hg will be related to
gauge transformations, e.g., to electromagnetic and isospin transformations
(see Chapter 6).
A structural group SO0 (t, s) for the tangent frames and signature (t, s)
is more specific than all tangent space automorphisms GL(n, R) for dimen-
sion n. However, the tangent group does not differentiate between manifolds
of equal signature (M(t,s) , g), but between different global or local invariance
groups.
2.2.4 Riemannian Connection

The conditions for the Killing fields of a Riemannian manifold,
(M, g) : vp ∂ p ∈ Lg ⇐⇒ gpj (∂ i vp ) + gpi (∂ j vp ) + vp (∂ p gij ) = 0,
can be written in the simple derivative form of “flat” vector spaces,
(Rn , η) : va ∂ a ∈ Lη ⇐⇒ ∂ a v b + ∂ b v a = 0, v a = η ab vb ,
by introducing the Riemannian connection,

(M, g) : vp ∂ p ∈ Lg ⇐⇒ ∇i v j + ∇j v i = 0, v i = gij vj ,
with ∇i v j = ∂ i v j − Γij k
k v .
It takes into account a possible “nonflatness” by including local changes via

derivatives of the metrical tensor via the connection coefficients,
Γij 1 i pj
k = 2 gkp (∂ g + ∂ j gpi − ∂ p gij ).
The Killing field condition for the metric motivates the introduction of a
second derivative on a Riemannian manifold in addition to and constructed
with the Lie derivative.
2.3 Affine Connections

A Riemann derivative is a special case of an affine connection of a dif-
ferentiable manifold, in general: “On top” of the derivations of the scalar
functions C(M) and in addition to the Lie derivations, a manifold may be
equipped with an affine connection ∇ which defines derivations of vector
fields T = der C(M), i.e., derivations of derivations,
∇ : T −→ T ⊗ TT , ⎧v − → ∇v ,
⎪
⎪ w −→ ∇v w,
⎨ ∇ (f w) = v(f )w + f ∇v w,
v
∇v : T −→ T,
⎪
⎪ ∇ i ej = ∇i ej = Γij k
ke ,
⎩ i ek ij
∇ (vk e ) = [e (vk ) + Γk vj ]ek ,
i
∇ = Γij
k iě ⊗ e k
⊗ ěj ∈ TT ⊗ T ⊗ TT ∈ vecR .
The R-linear tangent bundle endomorphism ∇v is called the covariant deriva-
tion in the direction of v. In contrast to the Lie derivation ad, where, in
general, ad f v = f ad v, an affine connection ∇ is C(M)-linear:
f, g ∈ C(M) : ∇f v+gw = f ∇v + g∇w .
In general, the Lie derivation ad v is not a covariant derivation.
d
A one-dimensional manifold has the connections f (x) dx with functions
f ∈ C(M).
A covariant derivative is grading compatibly extendable as R-linear
derivation to the full tensor algebra by Leibniz’s rule,
! !
∇v : ⎧ (T ⊕ TT ) −→ (T ⊕ TT ),
⎪
⎪ ∇v f = v(f ),
⎪
⎪ ∇
⎨ v ⊗ b) = ∇v a ⊗ b + a ⊗ ∇v b,
(a
with ∇v ω, w = ∇v ω, w + ω, ∇v w,
⎪
⎪
⎪
⎪ ∇ei ěk = ∇i ěk = −Γij
k ěj ,
⎩ ij k
∇ (ω ěk ) = [e (ω ) − Γk ω ]ěj .
i k i j
2.3 Affine Connections 41
A tensor a whose covariant derivatives are trivial in all directions is called

covariantly constant:
∇a = 0, ∇v a = 0 for all v ∈ T.
A local dual frame is covariantly constant ∇ej ⊗ ěj = 0.
2.3.1 Torsion, Curvature, and Ricci Tensor

With an affine connection, the vector fields have two real Lie algebra struc-
tures via the commutators:
v ∈ T ∈ lagR , with [v, w],
∇v ∈ T ⊗ TT ∈ lagR , with [∇v , ∇w ].
The mapping v −→ ∇v for the tangent fields is analogous to the adjoint
representation of a Lie algebra L l −→ ad l ∈ L⊗LT in its endomorphisms.
∇v and v are morphisms for T as an R-vector space, not for T as a
C(M)-module; in general, ∇v f w = f ∇v w and [v, f w] = f [v, w]. That is the
origin of their local transformation behavior. The two R-Lie algebra struc-
tures are combined in the construction of torsion and curvature, which are
even compatible with the C(M)-module property.
The torsion (tensor) describes the C(M)-linear difference of covariant and
Lie derivations:
⎧
⎨ T (v ∧ w) = ∇v w − ∇w v − [v, w]
T : T ∧ T −→ T, = −T (w ∧ v),
⎩
T (ei ∧ ej ) = Tkij ek ,
T = 12 Tkij ek ⊗ ěi ∧ ěj ∈ T ⊗ TT ∧ TT ∈ modC(M) ,
Tkij = −Tkji = Γij ji
k − Γk − k .
ij
In general, ∇ : T −→ T ⊗ TT is not a Lie algebra morphism; i.e., the dia-

gram with the commutator of vector fields and the commutator of covariant
derivatives in the direction of vector fields
∇⊗∇
T∧T −→ (T ⊗ TT ) ∧ (T ⊗ TT )
[ , ] ↓ ↓ [,]
T −→ T ⊗ TT
∇
is not commutative. The curvature (tensor) denotes the difference of both
commutators
R : T ∧ T −→ T ⊗ TT , R(v ∧ w) = [∇v , ∇w ] − ∇[v,w] = −R(w ∧ v),
R = 12 Rlij
k e ⊗ ěl ⊗ ěi ∧ ěj ∈ T ⊗ T ⊗ T ∧ T ∈ modC(M) ,
k T T T
Rlij
k
lji jl
= −Rk = ei (Γk ) − ej (Γil ij pl il jp jl ip
k ) − p Γk − Γp Γk + Γp Γk
= ∂ i Γjl il jp jl ip
k − ∂ Γk − Γp Γk +Γp Γk (holonomic bases).
j il
It defines endomorphisms of the tangent spaces:

R(ei ∧ ej ) = Rlij k e ⊗ ěl ,
k
R(v ∧ w) : T −→ T, lij k
R(e ∧ e ).e = Rk e .
i j l
There are maximally n2 linear independent coefficients {R(ei ∧ ej )x }i,j .

In general, connection, torsion, and curvature are defined without a met-
rical structure of the manifold. The Ricci tensor, a contraction (partial trace)
of the curvature tensor, is a connection-induced bilinear form of the tangent
spaces
lij
R• : T ⊗ T −→ C(M), R• (el , ei ) = Rli
• = Rj ,
R• = R• ěl ⊗ ěi ∈ T ⊗ T ∈ modC(M) ,
li T T
jl ij pl il jp jl ip
Rli
• = e (Γj ) − e (Γj ) − p Γj − Γp Γj + Γp Γj .
i j il
The Ricci form is not necessarily symmetric. It may be degenerate.

A one-dimensional manifold with connection ∇f = f (x) dx d
has trivial
torsion, curvature, and Ricci tensor.
For a torsionfree connection, the Lie bracket [v, w] coincides with the an-
tisymmetric combination of covariant derivatives. The structure constants of
the Lie derivatives (anholonomy coefficients) coincide with the antisymmetric
part of the affine connection coefficients:
⎧
⎨ ∇v w − ∇w v = [v, w],
T =0: Tkij = 0,
⎩
Γk − Γji
ij ij
k = k .
The two Lie brackets, for vector fields [v, w] and covariant derivatives
[∇v , ∇w ], locally fulfill a Jacobi identity. For a torsionfree connection, the
curvature fulfills the first Bianchi identity, which is equivalent to the Jacobi
identity for the Lie bracket of the vector fields:
⎧
⎨ R(v ∧ w).u + R(w ∧ u).v + R(u ∧ v).w =
T =0⇒ [[v, w], u] + [[w, u], v] + [[u, v], w] = 0,
⎩
Rlij
k + R ijl
k + Rjlik = 0.
The second Bianchi identity for a torsionfree connection is equivalent to the

Jacobi identity for the Lie bracket of the covariant derivatives:
⎧
⎨ ∇z [R(v ∧ w)u] + ∇v [R(w ∧ z)u] + ∇w [R(z ∧ v)u] =
T =0⇒ [∇z , [∇v , ∇w ]]u + [∇v , [∇w , ∇z ]]u + [∇w , [∇z , ∇v ]]u = 0,
⎩
∇p Rlijk + ∇i Rljp
k + ∇j Rlpi k = 0.
The Ricci tensor arises in the contracted second Bianchi identity:
Rkij i lp j lpi
k = 0, ∇ R• − ∇ R• + ∇ Rj = 0.
p li
2.3 Affine Connections 43
2.3.2 Cartan’s Stuctural Equations

In general, an affine connection is C(M)-linear “in one index” only, ∇f v =
f ∇v . With the n-elements of a cotangent basis, an affine connection is deter-
mined by n2 connection 1-forms:
Γjk = Γij
k ěi ∈ T ∈ modC(M) .
T
In general, the n2 connection 1-forms Γjk are not coefficients for C(M)-tensors:
Their parametrization dependence leads to an inhomogeneous contribution in
addition to the “normal” homogeneous C(M)-transformation in the transition
to another dual basis via ea ⊗ ěa = ei ⊗ ěi . This inhomogeneous contribution
is the origin of gauge transformations, here for the full linear group as the
structural group for the frames (see Chapter 6):

ij a k
∼ e (x) : ∇ei ea = Γia b i a
b e = [e (ek ) + Γk ej ]e ,
GL(n, R) e(x) = a
k a ij k
Γia i a k
b = ej Γk eb + e (ek )eb .
In physical language, the inhomogenous derivative transformation behavior

to the n2 connection forms Γjk with respect of the n2 -dimensional structural
group GL(n, R) is called a gauge transformation,
e ∈ GL(n, R) : Γi −→ e ◦ Γi ◦ e−1 + (∂ i e) ◦ e−1 (holonomic bases).
The maximal automorphism group GL(n, R) of the tangent spaces is not very
characteristic of a manifold and, probably, physically irrelevant. It contains
the tangent structures of all reparametrizations.
Cartan’s structural equations relate the external derivative of a cotangent
basis and its connection 1-forms to the 2-forms for torsion and curvature with
the characteristic sum of two contributions:
děk + Γik ∧ ěi = 12 Tk = 12 Tkij ěi ∧ ěj ∈ TT ∧ TT ,
dΓlk + Γpk ∧ Γlp = 12 Rlk = 12 Rlij
k ěi ∧ ěj ∈ T ∧ T .
T T
The torsion equations refine the Maurer-Cartan equations děk = − 12 ij

k ěi ∧ ěj .
In contrast to the n2 fields Γjk with inhomogeneous transformation behavior,
the n 2-forms Tk and the n2 2-forms Rlk are C(M)-tensors. They have a
homogeneous GL(n, R)-transformation behavior without n-bein derivatives.
The covariant derivative on the covariant tensor fields can be written with
the external derivative in the form of an absolute derivative,
⎧
⎪
⎪ Da(v) = ∇v a, v ∈ T,
! ! ⎨
Df = df,
D: T −→ TT ⊗ T,
⎪
⎪ D(a ⊗ b) = ∇a ⊗ b + a ⊗ ∇b,
⎩
D(f w) = df ⊗ w + f Dw,
e.g., the absolute derivative of a vector field
D : T −→ TT ⊗ T, Dv = D(vj ej ) = dvj ⊗ ej + vj Γijk ěi ⊗ e
k
ij
= [e (vk ) + Γk vj ]ěi ⊗ e .
i k
A parameter-dependent vector field R τ −→ v(τ ) ∈ T has the absolute

derivative
Dv = dτ ⊗ dτ (vj ∂ j ) = dxi ⊗ ∇i (vj ∂ j )
= (∂ i vk + Γij ij
k vj )dxi ⊗ ∂ = (dτ vk + Γk vj dτ xi )dτ ⊗ ∂ .
k k
A one-parameter-dependent geodesic R τ −→ xi (τ ) fulfills
dτ (vj ∂ j ) = 0, with vj = dτ xj ⇒ dτ vk + Γij

k vj vi = 0.
2.4 Lie Groups as Manifolds

Lie groups G are special manifolds. A Lie group action on itself restricts the
relevant manifold structures to group action-compatible ones, representable
at the group unit 1 ∈ G.
2.4.1 Lie Group Operations

A group acts on itself by left and right translations (multiplications)
(g1 , g2 ) ∈ G × G : Lg1 ◦ Rg2 : G −→ G, k −→ g1 kg2−1 ,
which are diffeomorphisms for Lie groups. The local invariance group in the
global symmetry (motion) group G × G is the diagonal group that arises as
the fixgroup of the neutral element,
⎪
⎪
k ∈ G : (G × G)k = {(g1 , g2 ) ∈
⎪ G × G⎪
⎪g1 kg2−1 = k}
⎪
⎪
= {(kg, gk)
⎪ ⎪g ∈ G}
∼ ⎪
= (G × G)1 = {(g, g)⎪⎪g ∈ G} = diag [G × G] ∼= G.
Therefore, the group is isomorphic to the orbit of the doubled group as global
symmetry group with the diagonal group as the local invariance group:
G = [G × G]/ diag [G × G].
diag [G × G] is not a normal subgroup; i.e., the classes of the diagonal group
(denoted with “doubled letters” like G) are isomorphic to the group G as
manifold, in general not as Lie group,
G × G (g1 , g2 ) −→ g1 g2−1 ∈ G ∼
= G = [G × G]/ diag [G × G].
An example is the manifold isomorphy of the spin group SU(2) and the
3-sphere:
[SU(2) × SU(2)]/SU(2) ∼
= SO(4)/SO(3) = SU(2) ∼
= Ω3 ∼
= SU(2).
2.4 Lie Groups as Manifolds 45
Ω3 characterizes the directions of the Lenz–Runge perihelion vector in the

nonrelativistic Kepler potential 1r (see Chapters 3 and 4).
The Lie group functions C(G) are acted on by the left and right regular
group representations, C(G) f −→ g1 fg2 with g1 fg2 (k) = f (g1−1 kg2 ). The
functions are both-sided regular representation matrix elements (representa-
tion coefficients) of G × G and can be harmonically analyzed with respect to
the irreducible representations.
2.4.2 Lie Algebra Operations

The left-invariant vector fields of a Lie group define its Lie algebra:
lgrpR G −→ log G = L ∈ lagR .
The Lie algebra can be realized by derivations of group functions at the group
unit l = ∂|g=1 ∈ T1 (G) = L.
A Lie algebra acts on itself in the adjoint representation
L × L −→ L, ad l(m) = [l, m],
written in dual Lie algebra bases with the structure constants ab a b
c = ( )c as
matrices,
(la , ľa )na=1 , ľa ,lb = δab ,
dual bases of L, LT :
n
holonomic bases at 1 ∈ G: ∂
∂αa = ∂ , dαa
a
,
a=1
c l ⊗ ľb .
ad la = ab c
adjoint action:
The adjoint action can be extended, as usual, to the tensor algebra by dual
product invariance and Leibniz’s rule
! !
[l, ] : ⎧(L ⊕ LT ) −→ (L ⊕ LT ),
⎪
⎪ [l, α] = 0, α ∈ R,
⎨
[l, ω] = −( ad l)T (ω), ω ∈ LT ,
with
⎪
⎪ [la , ľc ] = −ab
c ľb ,
⎩
[l, a1 ⊗ a2 ] = [l, a1 ] ⊗ a2 + a1 ⊗ [l, a2 ].
The identification of the Lie bracket with the tensor product commutator
in the tensor algebra of the Lie algebra defines its enveloping algebra:
'
E(L) = L modulo {[l, m] − (l ⊗ m − m ⊗ l) for all l, m ∈ L},
in E(L) : [l, m] = l ⊗ m − m ⊗ l.
Its center contains the Lie algebra invariants,
⎪
⎪
⎪ I] = {0}}.
centr E(L) = {I ∈ E(L)⎪[L,
A Lie group acts adjointly on its Lie algebra, illustrated in a representation
with log G ⊆ AL(Rn ) and G ⊆ GL(n, R):
G × log G −→ log G, (g, l) −→ Ad g(l) = g ◦ l ◦ g −1 .
2.4.3 The Poincaré Group of a Lie Group

The Lie algebra intrinsic bilinear form, invariant under Lie algebra action
(inner derivations), is the symmetric Killing form,
κ : L ∨ L −→ R, κ(l, m) = tr ad l ◦ ad m = κ(m, l),
κ([k, l], m) + κ(l, [k, m]) = 0,
in dual bases κ = κab ľa ⊗ ľb = κab dαa ⊗ dαb ∈ LT ∨ LT ,
κab = κba , ab c κ
cd
+ bd
c κ
ac
= 0.
The Killing form is trivial for abelian Lie algebras and nondegenerate pre-
cisely for semisimple Lie algebras; there the adjoint representation is injective
L∼ = ad L.
Linear forms LT of a semisimple Lie algebra inherit, with the Killing
isomorphism, the inverse Killing form as bilinear form,
L −→ LT , l −→ lκ = κ(l, ),
−1
κ : L ∨ LT −→ R, κ−1 (lκ , mκ ) = κ(l, m).
T
The diagonalized Killing form displays the dimensions of noncompact,

null-

1t 0 0
like (including abelian), and compact degrees of freedom κ =∼ 0 0 l 0 .
0 0 −1s
The adjoint representation of a semisimple Lie algebra is a subalgebra of
the Lie algebra of an orthogonal group,
semisimple Rn ∼ = ad L ⊆ log SO0 (t, s) ∼
=L∼
n
= R( 2 ) .
With the exception of the smallest simple Lie algebras log SO(3), log SO(1, n
2)
∼ R 2 ≥ n
3
= , the invariance Lie algebra of the L-Killing form is strictly larger
than the Lie algebra L, e.g., log SU(n) ⊂ log SO(n2 − 1) for n > 2, and
log SO0 (1, 3) ⊂ log SO0 (3, 3) for the Lorentz group. The invariance group of
the Killing form (“signature group”) for a semisimple Lie algebra defines the
Lorentz and Poincaré group of the Lie group G:
log G ∼
= Rn , sign κ = (t, s) : SO0 (t, s) ×
Rn , log SO0 (t, s) ⊇ log G.
2.4.4 Lie–Jacobi Isomorphisms for Lie Groups

The isomorphisms g∗ ∈ GL(n, R), induced by the left multiplications (also
left translations), transport all structures of the tangent space at the unit to
any group element g ∈ G, T1 ∼ = Tg ,
G
⏐ T(G)
⏐
⏐g −→ ⏐ g ,
∗
G T(G)
G1 − → g · 1 = g (left multiplication),
= log G = L = T1 (G) la −→ l(g)j = g∗ ja la ∈ Tg (G) ∼
Rn ∼ = L,
e.g., the Lie algebra structures at the group unit with a basis {la }na and
brackets [la , lb ] = ad la (lb ) = ab c
c l to isomorphic Lie algebras with trans-
formed structure constants,
[l(g)i , l(g)j ] = (g)ij k ij i j ab c
k l(g) , (g)k = g∗ a g∗ b c g∗ k , g∗ a |g=1 = δj .
j a
The Lie–Jacobi isomorphisms g∗ = e(g) ∈∈GL(n, R)/ Ad G represent

classes (not all classes) of the adjoint group Ad G ∼ = G/ centr G in the
general group. They are n-beins for the Lie group with a Lie algebra para-
metrization at least in the neighborhood of the group unit,
$ ( ad l)k
Ad G Ad g = e ad l = k! ,
$ ( ad l)k
k≥0
1
g∗ = e∗ad l = (1+k)! = 1n + 2 ad l + . . .
k≥0
∼
=“ ∂ Ad g
”=“ e ad l −1n
” (symbolic notation).
∂ ad l ad l
The adjoint group is trivial for abelian G. The adoint Lie algebra ad L ∼ =
L/ ad L is trivial for abelian L, and isomorphic to the Lie algebra in the
semisimple case L ∼ = ad L.
The Haar measure of the adjoint group can be written, at least in the
neighborhood of the group unit, with an overall dilation as the volume factor:
for Ad G : dg = | det g∗ |dn l.
The Lie–Jacobi transportation of the Killing form defines, for semisimple
Lie algebras, a Riemannian structure of the Lie group, (G, κ) ∈ rdif R :
Tg × Tg −→ R, κ(g)jk = tr ad lj (g) ◦ ad lk (g) = g∗ ja κab g∗ kb .
2.4.5 Examples
The compact space rotation group SO(3) has the angular momentum Lie
algebra, block-diagonalizable in the real and diagonalizable in the complex to
the rotation eigenvalues, i.e., third spin components {1, 0, −1}, in a Euclidean
parametrization:
R3 ∼ ⎧ SO(3) O = ad O
= log
−α2 0 −1
⎪
⎪O = −α3
0 α3 0
⎨ 0 α1 = αR ◦ 0 0 0 ◦ RT = iαu ◦ Δ ◦ u ,
⇒ 1α2 0 −α01 0 1 0 0
⎪
⎪
⎩Δ = 0 0
k 0
k
, R ∈ SO(3), u ∈ SU(3), α2 = α
2 = − 12 tr O2 .
0 0 (−1)
It leads to the group parametrization:

O
O2= e = 13 + α O +2 α2 2O ∈T SO(3),2
sin α 1−cos α 2
O = Ad
O = O ◦ O = −α R ◦ Δ ◦ R = −α u ◦ Δ2 ◦ u ,
with
O = O ◦ O ◦ O = −α2 O.
3
The determinant of the Lie–Jacobi isomorphisms O∗ of SO(3) as

representative of GL(3, R)/SO(3),
O∗ = eO
∗ =
∂O
e“iα∂O ” = 13 − sin
α3 O +
α−α 2 1−cos α
O = 13 + 12 O + . . .
−1
0 0

α2
sin α 1−cos α
iα α 0 α
=u◦ 0 1 0 ◦ u = R ◦ 0 1 0 ◦ RT ,
e−iα −1 − 1−cos α
0 sin α
0 0
α
2
−iα
α
2 α α
sin sin
det O∗ = α
2
, d3 O = α
2
dα1 dα2 dα3 ,
2 2
involves the spherical Bessel function, which arises in the representation of

three-dimensional quantum scattering structures (see Chapter 8):
sin α 3
j0 (r) = α 2 = d2πq δ(q 2 − 1)eiqx , r = |x| = α2 .
2
The Killing form for the angular momenta,
κ(O, O) = tr O2 = −2αa δ ab αb , κ(Oa , Ob ) = κab = −2δ ab ,
gives the Killing metric for the rotation group. It is the product of Lie–Jacobi
isomorphisms with their transposed ones and contains the spherical Bessel
function as metrical coefficient:
2 −α
4 sin2 α 2
α) = 12 O∗ ◦ κ ◦ O∗T = −13 +
g( O ◦ OT
α4 α 2
sin 2
= −( ααdα )2 − α α2
[d − ( α αdα )2 ].
2
The volume element and metrical tensor in Euler angles ( α2 , θ, ϕ) show the
manifold isomorphy SU(2) ∼ = Ω3 :
⎧ 1 3 2 α 2 α
⎪⎪
⎨ 8 d O = sin 2 d ω d 2 ,
2
α1 cos ϕ sin θ with d ω = sin θ dϕ dθ,
α2 = α sin ϕ sin θ ,
α3 cos θ ⎪
⎪ α) = −dα2 − 4 sin2 α2 dω 2 ,
g(
⎩
with dω 2 = dθ2 + sin2 θdϕ2 .
The noncompact partner for the 3-rotations is the Lorentz group SO0 (1, 2)
in one and two dimensions with real and imaginary angles for the two
Cartan subgroup types SO0 (1, 1) and SO(2), respectively, diagonalizable
for det Q = 0 in the complex to the eigenvalues, either boost components or
polarization components, in a Euclidean parametrization:
R3 ∼ ⎧ SO0 (1,2)0 Qβ = ad

= log Q
⎪
⎪ Q = β1
1 β2
= γuγ ◦ Δγ ◦ uγ , uγ ∈ SU(3),
⎪
⎪
0 α3
⎪
⎪ β −α 0
⎪
2 3
⎪
⎨
⇒ γ 2 = β12 + β22 − α23 = 12 tr Q2 ,
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ diag (1, 0, −1) (boosts), 2 2
γ , γ > 0,
⎩ Δγ = for γ =
diag (0, 1, −1) (rotations) i −γ , γ 2 < 0.
2
The group and the Lie–Jacobi isomorphisms are parametrized by the

exponent:
Q = Ad Q = eQ = 13 + sinh γ Q−
γ
Q ∈ SO0 (1, 2),
1−cosh γ 2
γ2
Q
Q∗ = e∗ = “ ∂Q ” = 13 + γ 3 Q − γ 2 Q = 13 + 12 Q + . . . ,
∂Q sinh γ−γ 2 1−cosh γ
2 2
sinh γ2 sinh γ2
det Q∗ = γ , d3 Q = γ dβ1 dβ2 dα3 .
2 2
The determinant contains for γ ∈ R the hyperbolic Macdonald function with

the Yukawa potential and for γ ∈ iR with sinh iα = i sin α the spherical
Bessel function:
e−ξ
k0 (ξ) = , k0 (−ξ)−k 0 (ξ)
= sinh ξ
ξ ,
ξ d3 q 1 i 2
k0 (r) = q2 +1 e
2π 2
qx
, r = |x|.
The Lie algebra Killing form gives the Killing metric for the Lorentz group
SO0 (1, 2):

1 0 0
ab ∼
κ(Q , Q ) = κ = 2 0 1 0 , γδ = γ1 δ1 + γ2 δ2 − γ3 δ3 ,
a b
0 0 −1

(γdγ)2 sinh γ2 2 2
g(γ) = 2 Q∗ ◦ κ ◦ Q∗ = γ 2 +
1 T
γ [dγ 2 − (γdγ)
γ2 ]
2
= dγ 2 + 4 sinh2 γ2 dω 2 .
2.4.6 Adjoint and Killing Connection of Lie Groups

For Lie groups G ∈ lgrpR , the geometrical concepts of connection, torsion,
and curvature can be given by properties at the group unit T1 (G) = L ∈ lagR .
A Lie group has many connections: There is a bijection between bilin-
ear Lie algebra mappings γ : L × L −→ L and left-invariant connections
∇l m = γ(l, m). The Lie bracket gives inner Lie algebra derivations. For a
semisimple Lie algebra, all derivations are inner. Distinguished Lie bracket-
related connections are defined, up to a real factor, by an inner derivation:
⎧
⎨ T (l ∧ m) = (2α − 1)[l, m],
at 1 ∈ G : ∇l = α ad l, α ∈ R ⇒ R(l ∧ m) = α2 [ ad l, ad m] − α ad [l, m]
⎩
= α(α − 1) ad [l, m].
Precisely for abelian groups with L = Rn (“translations”), Lie bracket-induced

connections are trivial. There are two “extreme” cases for Lie bracket-related
connections, either with trivial curvature or with trivial torsion,
α = 1 ⇒ R = 0,
α = 12 ⇒ T = 0.
A direct identification of a covariant derivative with the adjoint action

α = 1 defines the adjoint connection. It identifies the covariant and Lie
derivation. The Lie bracket is the torsion. The mapping l −→ ∇l is a Lie

algebra morphism; i.e., the curvature is trivial,

T (l ∧ m) = [l, m],
at 1 ∈ G : ∇l = ad l ⇒
R(l ∧ m) = 0.
Some structures of the adjoint connection arise in teleparallel theories.

The roles of torsion and curvature are “reversed” for the Killing connection
of a Lie group. A connected Lie group has a unique connection, invariant
under left and right translations and group inversion. It is given by inner
derivations with trivial torsion:
at 1 ∈ G : ∇l = 1
2 ad l ⇒ T (l ∧ m) = 0.
A nontrivial connection with one half of the adjoint representation is not

a Lie algebra morphism:
at 1 ∈ G : ∇ : L −→ L ⊗ L , l −→ ∇l = 2 ad l,
T 1
∇l m = 2 ad l(m) = 12 [l, m],

1
∇l : L −→ L,
c l , Γc = −Γc = 2 c ,
∇la lb = 12 ab c ab ba 1 ab
∇ = 12 ad = 12 ab ˇ
c l ⊗ ľb ∧ la ∈ L ⊗ L ∧ L .
c T T
For the Killing connection, the manifold is constituted by the classes of the
left–right operation group with the diagonal group G = G × G/ diag (G × G)
∼
= G. The tangent structures at the unit are characterized by the vector space
with the corresponding classes in the doubled Lie algebra:
L ⊕ L (l1 , l2 ) −→ l1 − l2 ∈ L ∼
= L = L ⊕ L/ diag (L ⊕ L).
In general, the classes are isomorphic L ∼

= L as vector space, not as Lie algebra.
The Killing curvature at the group unit is, up to a factor − 14 , the com-
mutator in the adjoint representation:
⎧
⎪ [∇l , ∇m ] = 12 ∇[l,m] = 1
ad [l, m],
at 1 ∈ G : ⎪
⎪
4
⎪
⎪
⎨ R(l ∧ m)
⎪ = − 41 ad [l, m] = − 41 [ ad l, ad m],
R : L ∧ L −→ L ⊗ L , T R(l ∧ m).n = − 41 [[l, m], n],
⎪
⎪
⎪
⎪ R(la ∧ lb ) − 41 ab ˇ
c ad l = − 4 c e l ⊗ ld ,
c 1 ab cd e
⎪
⎪
=
R(l ∧ m) : L −→ L, ⎩
Rdab
e = − 4 c e = − 4 (e c − e c ).
1 ab cd 1 ac bd bc ad
The name of the connection is motivated by the fact that the Ricci tensor
(bilinear form) at the group unit coincides up to a constant 14 with the Killing
form:

at 1 ∈ G : R• (l, m) = 14 tr ad l ◦ ad m = 14 κ(l, m),
R• : L ∨ L −→ R, R• (ld , la ) = Rda
• = Rb
dab
= 14 ab dc 1 da
c b = 4 κ .
2.5 Riemannian Manifolds 51
In contrast to the adjoint connection ∇ = ad (at 1 ∈ G), the Killing

connection ∇ = 12 ad (at 1 ∈ G) yields, for nondenerate Killing form, i.e., for
a semisimple Lie algebra, a Riemannian connection for G ∼ = G:
at 1 ∈ G,
R• = 14 κ, R•• = n
4.
semisimple L ∼
= Rn :
An example is the rotation group:

⎧
⎨ ab c = − , a, b, c ∈ {1, 2, 3}, 123 = 1,
abc
at 13 ∈ SO(3),
Rdab
= − 4 = − 14 (δ ad δ be − δ bd δ ae ),
1 abc cde
log SO(3) ∼
= R3 : ⎩ ead
R• = 14 κ(Oa , Od ) = 14 ab
c b = − 2 δ .
dc 1 ad
2.5 Riemannian Manifolds

2.5.1 Lorentz Covariant Derivatives
The Riemannian connection of a Riemannian manifold (M(t,s) , g) ∈ rdif R ,
motivated above by the Killing fields of its metric, is uniquely characterized
by its torsion freedom and a covariantly constant (derivation trivial) metric.
It is given in a holonomic basis ej = ∂ j by
⎧
⎪
⎪ Γij
k = Γk
ji
⎨
T = 0, (jli) = ∂ j gli − Γjl
p g − Γp g = 0,
pi ji lp
∇g = 0, ⎪ ⎪ (jli) −(lij) − (ijl) = 0,

⎩
Γij 1
k = 2 gkp (∂ g
i pj
+ ∂ j gpi − ∂ p gij ).
Both the identity and the metric (dual isomorphism) are covariantly constant
∇ei ⊗ ěi = 0, ∇gil ěl ⊗ ěi = 0, ∇gil ei ⊗ el = 0. With the invariant dual
isomorphisms, one has gli ∇ai = ∇al .
A (pseudo-)Riemannian manifold with torsion cannot be embedded into
a (pseudo-)Euclidean space.
The transformation of the n2 connection 1-forms from a general, e.g.,
a holonomic basis ∂ j ⊗ dxj = ea ⊗ ěa , to orthonormal bases with invari-
ance Lorentz group involves derivatives of the n-bein (parametrization de-
pendence),
e(x) ∈∈GL(n, R)/SO0 (t, s) :

TT Γjk = Γij cj
k dxi = Γk ěc −→ Γ b = Γ b dxi = Γ b ěc ∈ T ,
a ia ca T
a ij k a ij
Γ ia
b = ej Γk eb + (∂ ek )eb ⇒ ∂ ek + ej Γk − Γ b ek = 0,
i a k i a ia b
Γij
k
i j b i j ij k
b ek + (∂ eb )ek ⇒ ∂ eb − Γk eb + eaΓ b = 0.
= ejaΓ ia b j ia
Since the n-bein are equivalence classes with the representatives defined up
to local SO0 (t, s)-transformations, the Lorentz degrees of freedom show up
in an inhomogeneous SO0 (t, s)-transformation behavior,
⎧
⎨ eck − → Λca eak , ejb −→ ejd Λ−1db ,
Λ(x) ∈ SO0 (t, s), −1d −1d
Γ ia − → ΛacΓ ic
d Λ
i a
b + (∂ Λc )Λ b,
⎩ bi −1 −1
Γ − → ΛΓ i i
Γ Λ + (∂ Λ)Λ .
Einstein’s relativity, as formulated by Weyl for spinor fields, defined in

representation spaces for the orthogonal covering group SO0 (1, 3), uses the
Lorentz group as the structural group. The Fock–Iwanenkov coefficients Γ ab
for e ∈∈GL(n, R)/SO0 (t, s) are given by the tangent c
n
space forms Γ b with
Lorentz group gauge transformation behavior. The 2 linearly independent
elements, written with ec = ∂ c = ecj ∂ j ,
b ηca = Γ ab = −Γ
Γ ic i Γiba
= ee (ηac ηbd − ηbc ηad )(η df eej ∂ c ejf + η ef edj ∂ c ejf − η df ecj ∂ e ejf ),
i
can be used for any Lie algebra representation of the tangent Lorentz group
SO0 (t, s) on a vector space V ,
log SO0 (t, s) Lab −→ DV (Lab ) ⊆ AL(V ),

ab Dn (L ) ∈ Tx ∧ Tx ⊆ AL(R ).
Γ i ab n
e.g., Minkowski representation:
!
The Lorentz covariant derivative of the tensor algebra (T ⊕ TT ) with
the tangent fields and its forms, induced by the affine connection, can be
defined, via the Fock–Iwanenkov coeficients, also for the spinor representa-
tions, one of SO(t, s) for odd t + s, e.g., for the Pauli spinors for SO(3) ∼=
SU(2), and two for even t + s, e.g., for the Weyl spinor representations of
SO0 (1, 3) ∼
= SL(2, C), for their sums and product representations, i.e., for
all fields acted on by a finite-dimensional representation of the Lorentz cover
group SO0 (t, s) and its Lie algebra.
The tetrad is more appropriate than the metrical tensor for the formu-
lation of the gravitational interaction (general relativity). It implements the
10 degrees of freedom in the form of operations and allows the inclusion of
spinor structures. The Lorentz group as subgroup of the maximal structural
group GL(4, R) is physically relevant for the spinor structures.
2.5.2 Laplace–Beltrami Operator

g
With the isomorphy of tangent and cotangent fields TT ∼ = T, defined by
the metric, the external derivative of a function defines a vector field as its
gradient:
d g−1
grad = g−1 ◦ d : C(M) −→ TT (M) −→ T(M),
f −→ grad f = gkj (∂ j f )∂ k (in holonomic bases),
from df, v = v(f ) = g(grad f, v) for v ∈ T(M).
The trace of the covariant derivative,

jk
∇k v = ∇k vi ∂ i = (∂ k vj + Γki j k ki k
j vi )∂ , ∂ vk + Γk vi = (∂ + Γj )vk ,
2.5 Riemannian Manifolds 53
leads to the divergence from vector fields to scalar fields,
div : T(M) −→ C(M), v = vj ∂ j −→ div v = (∂ k + Γjk

j )vk ,
f ∈ C(M) : div f v = v(f ) + f div v.
It can be written as a metric-normalized derivative,

|g| = − det gij = |e|2 , |e| = det eia = |g|,
gij dgij = −gij dgij = 2eai deia = −2eia deai = 2Γjk
j dxk
1
= |g| d|g| = d log |g| = 2d log |e|,

⇒ div v = (∂ k + Γjkj )vk =
√1 ∂ k |g|vk .
|g|
The manifold-integrated divergence of any vector field is trivial:

M
|g|dn x div v = 0.
A conformal vector field has the metric as eigentensor. The eigenvalue is,
up to a dimension-related factor, the divergence of the vector field:
v ∈ T : [v, g] = ρ(v)g ⇒ ρ(v) = 2

n div v.
Therefore, a Killing vector field for a global invariance has trivial divergence.
The Laplace–Beltrami operator for functions on a Riemannian manifold
is the divergence of the gradient. It contains the Lorentz covariant derivative
of a vector field and is invariant under the global group Gg :
grad div
∂g2 : C(M) −→ T(M) −→C(M),
∂g2 = div grad = √1 ∂ k |g| gkj ∂ j
|g|
= gkj (∂ k ∂ j − Γkj
i ∂ ) = (∂a − Γ ab )∂ .
i b a
It is integration-symmetric:

M |g|dn x f1 (x)(∂g2 f2 )(x) = M |g|dn x (∂g2 f1 )(x)f2 (x).
The Laplace–Beltrami operator for a Lie group with semisimple Lie algebra
is closely related to its Killing form (see Chapter 10).
2.5.3 Riemannian Curvature

The Riemannian connection is given by the 1-forms in the Cartan equation
for trivial torsion, which allows the computation of the curvature 2-forms —
in orthogonal bases,
g = η ab ěa ⊗ ěb ,
děc + Γ bc ∧ ěb= 0, η acΓ bc = Γ ab = −Γ
Γba ,
Γdc + Γ ac ∧ Γ da
dΓ = 2 Rc ěa ∧ ěb ,
1 dab
Γ − ηabΓ ca ∧ Γ db
dΓ cd
= 12 Rcdab ěa ∧ ěb = Rcd = −Rdc.
The tangent space endomorphisms, defined by the Riemannian curvature,

leave the metric invariant. This leads to an additional antisymmetry of the
curvature tensor:

g(R(v ∧ w).u, z) + g(u, R(v ∧ w).z) = 0,
R(v ∧ w) : T −→ T :
gkp Rlij
p =R
klij
= −Rlkij .
Thus, the curvature defines a symmetric bilinear form of the antisymmetric

tensors Tx ∧ Tx ∼
n
= R( 2 ) :
R : (T ∧ T) ∨ (T ∧ T) −→ R, R(ek ∧ el , ei ∧ ej ) = Rklij = Rijkl .
For dimR M = n = 1, 2, 3,
4, . . . , the maximal number of independent compo-
nent takes into account n4 first Bianchi identities, caused by the Lie structure
of the derivations and relevant for dimension n ≥ 4:
(n)+1
n
n2 (n2 −1)
2
2
− = 4 12 = 0, 1, 6, 20, . . . .
The Riemannian Ricci tensor is a symmetric bilinear form of the tangent

translations:
R• : T ∨ T −→ R, R• (ei , el ) = Rlij
j = R
klij
gkj = Rli
• = R• .
il
The trace of the composition of inverse metric and Ricci tensor defines the
curvature scalar:
R• ◦ g−1 : T −→ T, R•• : M −→ R, R•• = gli Rli

• ∈ C(M).
There exist stationary coordinates at each point of a Riemannian manifold

M P0 ∼= x0 , where the Taylor expansion of the metrical tensor gives the or-
thonormal standard metric, trivial connection coefficients, and the curvature
as second-order coefficients (area change),
gil (x) = η il + 13 Riklp (x0 )(xk − x0k )(xp − x0p ) + . . . ,

gil (x0 ) = η il , Γil 0
k (x ) = 0.
2.5.4 Einstein Tensor and Conserved Quantities

The second Bianchi identity, as a twice-contracted Jacobi identity, leads, for
any dimension, to a covariantly divergence-free Einstein tensor,

gli •
gpl ∇p Rli • − 2 R • = 0.
With the exception of the metrical tensor ∇p gli = 0, the Einstein tensor is,
in four dimensions, the only covariantly conserved second-order tensor, that
can be built from g and its first- and second-order derivatives.
2.6 Tangent and Operational Metrics 55
For nonflat Riemannian manifolds in general, there are no conserved

currents as familiar from flat spacetime theories. The Killing fields (v K )dK=1
of the global invariance Lie algebra of a Riemannian manifold allow the con-
struction of conserved quantities (V K )dK=1 in the Killing field directions of
the Einstein tensor if it vanishes sufficiently asymptotically, e.g., for “asymp-
totically flat” spacetimes M(1,s) . The integration over an (n − 1)-dimensional
submanifold, e.g., an s-dimensional position submanifold, uses the volume
elements, given with the isomorphies g ◦ ,

gli •
v K = viK ∂ i ∈ Lg : V K = Mn−1
(dn−1
x) v
l i
K
Rli
• − 2 R• ,
"
n−1
(dn−1 x)l ∈ TT , e.g. for n = 4 : dxl ↔ (d3 x)l = glr rijk dxi ∧ dxj ∧ dxk .
For Minkowski spacetime, one obtains the 10 conserved quantities of the

Poincaré group: four momenta, three angular momenta, and three boosts.
In Einstein’s gravity, the integration goes over the energy-momentum tensor
gli •

• − 2 R• = −κT − Λg , e.g., R3 d x T (x).
Rli li li 3 0i
2.6 Tangent and Operational Metrics

The tangent metrics of a Riemannian manifold are the metric of the trans-
lations and the Killing form of the orthogonal tangent group for n ≥ 3.
They are symmetric, nondegenerate, and invariant under the full Lie algebra

log SO0 (t, s) ∼

n
= R( 2 ) with dimension n2 = ts+ 2t + 2s , the number of linear
independent areas,

η(P a, P b ) = ηab ,
R × R −→ R,
n n
∼ 1t 0
η= 0 −1s ,
⎧
⎨ κ(Lda , Lbc ) = −(η
η − η dc
db ac
η ab ),
log SO0 (t, s) × log SO0 (t, s) −→ R,
⎩ κ = −η ∧ η ∼
1ts 0
= 0 −1t s . +
2 2
The curvature and Ricci tensors give additional, possibly degenerate, sym-
metric bilinear forms of the tangent Lorentz Lie algebra and the translations,
respectively, called the operational metrics of the manifold:
log SO0 (t, s) × log SO0 (t, s) −→ R, R(Lda , Lbc ) = Rdabc

= edk eal Rklij ebi ecj ,
Rn × Rn −→ R, R• (P a , P b ) = Rab
•
= ebi eal Rli
• = ηdc R
dabc
.
In contrast to the signatures of the tangent metrics (Killing form and

Lorentz metric) on an analytic and connected manifold, the signatures of
the operational metrics (curvature and Ricci tensor) may change from point
to point, e.g., spherical, flat, and hyperbolic (saddlelike) areas on a deflated
rubber ball.
In an orthonormal

tangent basis, half the curvature scalar is given by the
sum of the n2 diagonal elements of the curvature, with signs η(A) = ±1:
⎧
⎪
⎪ $
n
⎪
⎪ Rab
• = η(d)Rdabd ,
⎪
⎪
⎪
⎨ d=1
η ab = η(a)δ ab , η(a) = ±1 ⇒ $ (n2 )
$
⎪
⎪ 1 •
⎪
⎪ 2 R• = η(d)η(a)Rdaad = η(A)RAA ,
⎪
⎪
⎪
⎩ d<a A=1
with A = da, η(A) = −η(d)η(a).
At any manifold point, the global symmetry group with the local invariance
group is represented by linear transformations Gg −→ GL(n, R), Tx ∼ = Rn .
An n-bein with a left action by the global symmetry group and a right action
by the tangent Lorentz group transmutes from GL(n, R) to SO0 (t, s). The
components of the metric, curvature, and Ricci tensor are representation
coefficients of the global symmetry group, invariant under the local invariance
group Hg ⊆ H max (t, s) ∼
= SO0 (t, s):
%
Gg /Hgι ∼
= M x −→ g(x), R(x), R• (x), R•• (x),
ι ⎧
⎪
⎪ R([H, L1 ], L2 ) + R(L1 , [H, L2 ]) = 0,
⎨
L1,2 ∈ log SO0 (t, s),
H ∈ log Hg :
⎪
⎪ R • ([H, P 1 ], P 2 ) + R• (P 1 , [H, P 2 ]) = 0,
⎩
P1,2 ∈ Rn .
2.6.1 Invariants
In general, a bilinear vector space form is invariant under a linear group H for
γ : V × V −→ R, γ(h.v, h.w) = γ(v, w), h ∈ H ⊆ GL(V ),

h ◦ γ ◦ hT = γ.
With another H-invariant bilinear nondenerate form δ : V × V −→ R, the

coefficients Ik of the characteristic polynomial, especially the determinant I0
and the trace I1 , for the (γ, δ)-composed V -endomorphisms are H-invariants:
γ ◦ δ −1 : V −→ V, h ◦ (γ ◦ δ −1 ) ◦ h−1 = γ ◦ δ −1 ,
$
dim RV
−1
det (γ ◦ δ − λ1V ) = Ik λk .
k=0
2.7 Maximally Symmetric Manifolds 57
Therefore, invariants of the local invariance group Hg of a Riemannian

manifold can be obtained from the characteristic polynomials with the cur-
vature tensor, which defines endomorphisms of the Lorentz Lie algebra, and
the Ricci tensor, which defines endomorphisms of the tangent translations:

log SO0 (t, s) −→ log SO0 (t, s), det [R ◦ (g ∧ g)−1 − λ1(n) ],
2
eb ∧ ec −→ Rbcda e ∧ e ,
d a
det [Ref bc (η ∧ η)ef da − λ(1n ∧ 1n )bc
da ],

Rn −→ Rn , det (R• ◦ g−1 − λ1n ),
eb −→ R• ba ea , det (Rf• b ηf a − λδab ).

The composed mappings in the characteristic polynomials of degree n2 and n

can be taken in an orthonormal basis with det e ◦ f ◦ e−1 = det f and
tr e ◦ f ◦ e−1 = tr f . Functions, especially R-polynomials of invariants, are
invariants too. The curvature scalar R•• = tr R• ◦ g−1 is an invariant. The
maximal SO0 (t, s) ⊇ Hg has R = 1, 2, . . . generating invariants for t + s =
n = 2R and n = 2R + 1.
2.7 Maximally Symmetric Manifolds

A Riemannian manifold (M(t,s) , g) where the Ricci tensor coincides, up to
normalization, with the metric is called an Einstein manifold; i.e.,
1 •
λ ∈ R : R• = λ

2 g, 2 R• = n λ
2
2 .
The normalization factor with a length unit can be absorbed in the definition
of g.
A Riemannian manifold has maximal global symmetry or, equivalently,
constant curvature, if the curvature is, up to a constant, the antisym-
metric square of the metric g ∧ g. Then the manifold is Einsteinian
with
γ 1t 0
λ = −k(n − 1), even conformally flat g = e η = e 0 −1s with a scalar
γ
function eγ (more ahead). For n ≥ 3, the curvature is, up to a scalar func-

tion, the Killing form3 of the local invariance Lie algebra log SO0 (t, s). The
invariant scalar R•• involves a constant k ∈ {±1, 0}:
⎧
⎨ R• = −(n − 1)
k2 g,
R =
2 g ∧ g,
k
⇒ Rli = −(n
n
−k 1)
2 g ,
k li
Rklij =
k2 (gki glj − gkj gli ) ⎩ 1 ••
2 R• = − 2
2 .
For an Einstein manifold, the metrically reflected Ricci tensor is the neg-
ative Ricci tensor. The Einstein tensor is trivial exactly for n = 1, 2, i.e., for
3 With Schur’s lemma, the Killing form of a simple complex Lie algebra is, as an in-
variant bilinear form, unique up to a scalar factor.

the trivial SO(1) = {1} and the abelian orthogonal tangent Lie algebras,
compact SO(2), and noncompact SO0 (1, 1):

R• − n2 gR•• = Ř• = −R• ,
R• =
2 g ⇒
λ
R• − g2 R•• = − n−2 λ
2
2 g.
2.7.1 Spheres and Hyperboloids

A simply connected complete Riemannian manifold with definite metric and
constant curvature is isometric either to a sphere, if compact, or, if noncom-
pact, to a Euclidean space or to a one-shell “timelike” hyperboloid,
⎧ s
⎨ Ω ∼= SO(1 + s)/SO(s), k = 1,
s = 1, 2, . . . : Rs ∼
= SO(s) ×
R s
/SO(s), k = 0,
⎩ s ∼
Y = SO0 (1, s)/SO(s), k = −1.
The structures for s = 1 with SO(1) = {1} are abelian. As manifolds, not
as symmetric spaces, hyperboloids and flat spaces are isomorphic, Y s ∼
= Rs .
One can define for s = 0 the discrete groups Ω = {±1}, R = {0}, and
0 0
Y 0 = {1}.
The invariant metric and measure for the Euclidean space Rs are dxs2
and ds x. Unit spheres and hyperboloids can be parametrized by polar coor-
dinates from the Eulerian parametrization of their global symmetry groups
with nonabelian degrees of freedom for s ≥ 2:

s ≥ 1 : Ωs ωs = cos θ
∈ R1+s , ω0 = 1,
sin θ ωs−1

Y s ys = cosh ψ
sinh ψ ωs−1
∈ R1+s .
Therefrom, one obtains metric and measure,
⎧ 2 ⎧ 2
⎨ dθ , s = 1, ⎨ dψ , s = 1,
dωs2 = dθ2 + sin2 θ dϕ2 , s = 2, dy2s = dψ 2 + sinh2 ψ dϕ2 , s = 2,
⎩ 2 ⎩ 2
dθ + sin2 θ dωs−1
2
, dψ + sinh2 ψ dωs−1
2
,
ds ω = sins−1 θ dθ ds−1 ω, ds y = sinhs−1 ψ dψ ds−1 ω.

The measures involve sin θ and sinh ψ as “radii” of spheres.Geodesic polar
coordinates are obtained by a transformation of the “leading angle” to a length
coordinate:
⎧
⎨ dω 2 = dρ22 + ρ2 dω 2 ,
s 1−ρ s−1
θ ∈ [− 2 , 2 ],
π π
sin θ = ρ ∈ [−1, 1] :
⎩ ds ω = ρs−1 √ 2 ds−1 ω,
dρ
1−ρ
⎧
⎨ dy2 = dρ22 + ρ2 dω 2 ,
s 1+ρ s−1
ψ ∈] − ∞, ∞[, sinh ψ = ρ ∈] − ∞, ∞[:
⎩ ds y = ρs−1 √ 2 ds−1 ω.
dρ
1+ρ
The compact spheres have a coordinate singularity at ρ2 = 1.

2.7 Maximally Symmetric Manifolds 59
Spheres and hyperboloids are conformally Euclidean, as seen explicitly in

a Cartesian parametrization,
⎧ dρ2 dxs 2
⎪
⎪
2 2
2 + ρ dωs−1 = ,
⎪
⎪
1−kρ (1+k x2 2

4 )
⎪
⎪ 2 2 2 2
⎪
⎪ where dxs = dr + r dωs−1 ,
⎪
⎨ with ρ = r
, r 2
= x2
,
∼
2
k = (1, 0, −1) = (Ω , R , Y ) :
s s s
1+k r4
⎪
⎪ and r = 2 tan θ
, r, 2 tanh ψ
⎪
⎪ 2 2 ,
⎪
⎪ s
⎪
⎪ (d ω, d x, d y) = (1+k x2 )s ,
s s s d x
⎪
⎩ 4
ds x = rs−1 dr ds−1 ω.
The disk r < 2 for the 2-hyperboloid Y 2 (Poincaré’s model) has the non-
2
Euclidean geometry dxr22 2 where the conformal factor for the length ex-
(1− 4 )
pansion goes to infinity at the boundary r → 2. For half of the 2-sphere Ω2 ,
2
projected on the disk r ≤ 2, the non-Euclidean geometry dxr22 2 arises with
(1+ 4 )
the conformal factor for the length contraction. The finite distances on the
nonflat geodesics are given by the “leading angle”:

r dR 4 arctan r2 = 2θ, k = +1,
d(0, r) = 0 =
1+k R4
2
2 log 2−r = 4 artanh r2 = 2ψ, k = −1.
2+r
The one-shell spacelike hyperboloids have as pseudometric and invariant

measure:

s ≥ 1 : SO0 (1, 1 + s)/SO0 (1, s) ∼ = sinh
= Y (1,s) y(1,s)
ψ
cosh ψ ωs ∈ R
2+s
,
dy2(1,s) = dψ 2 − cosh2 ψ dωs2 ,
d(1,s) y = coshs ψ dψ ds ω.
2.7.2 Constant-Curvature Manifolds

More generally, precisely for a constant curvature, the Riemannian manifold
is a general hyperboloid (sphere for positive signature), or a flat vector space,
Y (t,s) ∼
= SO0 (t, 1 + s)/SO0 (t, s), k = 1,
Rn ∼
= SO0 (t, s) ×
Rn /SO0 (t, s), k = 0,
Y (s,t) ∼
= SO0 (1 + t, s)/SO0 (t, s), k = −1.
The numbers (t, s) and (s, t) denote the (noncompact, compact) dimensions
in the hyperboloids, e.g., noncompact t(1 + s) − ts = t for Y (t,s) . Examples
are the (anti-)de Sitter universes Y (1,3) and Y (3,1) . The flat spaces R(t,s) =
Rn /SO0 (t, s) ∼
SO0 (t, s) × = Rn have to be distinguished, as symmetric
spaces, from the abelian groups Rn for n ≥ 2.
The tangent Poincaré groups, isomorphic for these three manifolds, arise
as Inönü–Wigner contractions of the global symmetry groups,
⎫ SO0 (1 + t, s) → SO0 (t, s) ×

Rn ← SO0 (t, 1 + s),
(t, s) = (1, 0),⎬
cos θ c sin θ 1 x cosh ψ c sinh ψ
(θ, ψ) = xc , − → ← 1 sinh ψ cosh ψ .
⎭
1
c sin θ cos θ 0 1 c
c →∞
In the complex, the two hyperboloids are related to each other by the
compact–noncompact transition iθ ↔ ψ. All three maximal global symmetry
groups are subgroups of SO0 (1 + t, 1 + s), characterizable as fixgroups of
nontrivial vectors x ∈ R2+t+s with negative, trivial, and positive square x2 as
familiar from the SO0 (1, 3)-subgroups SO0 (1, 2), SO(2) × R2 , and SO(3) as
fixgroups for spacelike, nontrivial lightlike, and timelike x ∈ R4 , respectively.
The metrical tensor is invariant under the maximal local group SO0 (t, s):
i.e., it is conformal to a flat bilinear form. The overall Lorentz invariant
dilation factor for the metrical tensor depends on the invariant x2 :
gli (x) = e(x)2 η li , e(x) = eγ(x) 1n = 1n

2 ∈ D(1) with x2 = xr ηrm xm ,
1+k x4
2 2
γ(x) = − log(1 + k x4 ) = k x4 + . . . ,
R = kg ∧ g = ke2γ η∧η ∼
= Rklij (x) = ke2γ(x)(η ki η lj − η ki η lj ).
Is is possible to use points with x = 0.

From the cases with definite metric Y (0,s) = Ωs and Y (s,0) = Y s , one
generalizes to hyperboloids Y (t,s) and Y (s,t) with ω0 = 1, dω02 = 0, and
d0 ω = 0 (sloppy ω-notation):

t ≥ 1, SO0 (t, 1 + s)/SO0 (t, s) ∼ = sinh
= Y (t,s) y(t,s)
ψ ωt−1

cosh ψ ωs ∈ R1+t+s ,
dy2(t,s) = dψ 2 + sinh2 ψdωt−1 2
− cosh2 ψ dωs2 ,
d(t,s) y = sinht−1 ψ dt−1 ω coshs ψ ds ω dψ,
s ≥ 1, SO0 (1 + t, s)/SO0 (t, s) ∼
= Y (s,t)
y
(s,t) = cosh ψ ωt
sinh ψ ωs−1
∈ R1+t+s ,
dy2(s,t) = dψ 2 − cosh2 ψdωt2 + sinh2 ψ dωs−1
2
,
d(s,t)
y = cosh ψ d ω sinh
t t s−1 s−1
ψ dψd ω.
2.8 Rotation-Symmetric Manifolds

The Schwarzschild metric characterizes a spacetime manifold that embeds a
parabolic position manifold P3 . Starting from the one-dimensional parabola,

P1 : ζ 2 = 40 (ρ − 2), 0 = 0, p = ρζ = ± 4
0ρ(ρ − 2
) ∈ R2 ,

0 dρ2
dp2 = dρ2 + dζ 2 = 1 + ρ−2
dρ2 = 1− 2 for 0 = 2,
ρ
2.9 Basic Riemannian Manifolds 61
paraboloids arise in Euclidean R1+s by axial rotations. They have as geodesic

polar and Cartesian parametrization:

Ps : ps = ± 4
0 (ρ − 2
)
∈ R1+s , ωs−1 ∈ Ωs−1 , 4
ρ2
2 ≥ 1,
ρ ω s−1

2
0 2 2 2
dps = 1 + ρ−2
dρ + ρ dωs−1
4 2
dρ2 2 2

= 1− 2 + ρ dωs−1 = 1 + 2r dx2s for 0 = 2 and ρ = r 1 + 2r .

ρ
The rotation axis of the not simply connected paraboloid is inside, but not
part of, the paraboloid (visualize with s = 2). The parameter space of Ps is
twice the Euclidean⎪ space Rs (projection of Ps on Rs ) up to a ball around
s⎪
⎪
the origin {x ∈ R ⎪x < 42 }. The rotation paraboloid Ps has SO(s) as the
2
global symmetry (motion) group with SO(s) as the local invariance group,
%
Ps ∼
=2 {ρ} × Ωs−1 .
ρ≥2
As manifold, Ps is isomorphic to R × Ωs−1 — however, not as homogeneous

space, since its global symmetry group is not R × SO(s).
In general, a line M1 , rotated in Euclidean R1+s with O(s) for s = 1, 2, . . . ,

M1 : ζ = f (ρ), ls = ρ fω(ρ) s−1

∈ R1+s , ωs−1 ∈ Ωs−1 ,
is an s-dimensional manifold with the metric:
Ms ⊃ Ωs−1 : dm2s = a2 (ρ)dρ2 + ρ2 dωs−1

2
with a2 (ρ) = 1 + f (ρ)2 ,
e.g., cylinders R × Ωs−1 with f = 1 = a or a 2-torus with the two radii

c0 ≤ c1 ,
T2 ∼
c20
= Ω1 × Ω1 : dt22 = c2 −(ρ−c 1)
2 2 2
2 dρ + ρ dϕ .
0
In general, the transformations for SO(s − 1)-invariant metrics between

geodesic polar and Cartesian parametrizations read with integration constant
c0 ∈ R,

dξ ξ
2 2 2 2 ξ 2 (r) 2 2 2
a (ρ)dρ + ρ dωs−1 = r2 (dr + r dωs−1 ), a(ξ)
dξ
dr = r ,
ξ a(ξ) = log c0 r.
2.9 Basic Riemannian Manifolds

A Riemannian metric, symmetric and nondegenerate, determines “its” con-
nection. The related curvature, Ricci tensor, and curvature scalar involve up
to second-order derivatives of the metric
∂2
g −→ R, R• , R•• .
This transition will be considered in all details for the manifolds with
dimensions 1 and 2 and for rotation-invariant manifolds of dimensions 3 and
4 in an orthonormal basis (n-bein), g = ds2 = η ab ěa ⊗ ěb , ěa (x) = eja (x)dxj .
The curvature and Ricci tensors with representation coefficients of the global
symmetry group Gg yield symmetric bilinear forms of the orthogonal Lie
algebra Tx ∧ Tx ∼ = log SO0 (t, s) and of the translations Tx ∼ = Rn , re-
spectively, invariant under the action of the local invariance group of the
manifold Hg ⊆ Gg ∩ SO0 (t, s). The curvature scalar, if nontrivial, defines an
Hg -invariant dilation.
The solution of Einstein’s equations requires the opposite transition (in-
tegration) from curvature and Ricci tensor (energy-momentum tensor) to
Riemannian metric.
2.9.1 Manifolds with Dimension 1

A “sufficiently smooth” dilation factor for a one-dimensional Riemannian
manifold with abelian tangent group R can be absorbed by a reparamet-
rization:
τ
M1 : g = b2 (τ )dτ 2 = dt2 with t = c0 ds b(s), ∂t = 1b ∂τ .
The manifold is isomorphic either to the circle Ω1 ∼

= SO(2) ∼
= R/Z or to the
one-branch hyperbola and real one-dimensional line Y 1 ∼= SO0 (1, 1) ∼
= R.
The curvature is trivial.

Two-dimensional Riemannian manifolds, visualizable in ordinary 3-position
space, are the origin of the curvature concept with the change of an area and
the nontrivial real groups SO(2) for axial rotations and SO0 (1, 1) for Lorentz
dilations. The Riemannian operation groups for nonflat manifolds with (0, 2)-
or (1, 1)-signature are basic for the curvature concept in n ≥ 2 dimensions:
Tangent Poincaré group Global symmetry group Gg

SO(2) × R2 Gmax (0, 2) ∈ {SO(3), SO0 (1, 2)}
Manifold SO0 (1, 1) ×
R2 Gmax (1, 1) ∼
= SO0 (1, 2)
M(0,2)
M(1,1) Local invariance group
Hg ⊆ Gg ∩ SO(2)
Hg ⊆ Gg ∩ SO0 (1, 1)
The transition from a Riemannian metric to its Ricci tensor reproduces the
tangent space metric up to a factor, given by the one curvature component.
The derivatives are denoted by ȧ = ∂τ a, a = ∂ρ a, etc.:

ě1
g = b2 (τ, ρ)dτ 2 ± a2 (τ, ρ)dρ2 , = ab dτ
dρ ,
ě2
dρ ∧ dτ b
dě1
dě2 = bȧ dτ ∧ dρ , Γ
12
= ȧb dρ ∓ a dτ = 1
ba (ȧě2 ∓ b ě1 ),

R12 = dΓΓ12 = (∂τ ȧb ± ∂ρ ba )dτ ∧ dρ =
κ± ě1 ∧ ě2 ,

ab ∼ 1 0 κ+ 1 2 , 1 •
R1212
= κ± , R• = ±κ± 0 ±1 = 2 R• = ±κ± .
−κ− η2 ,
There is only one linear independent area for two dimensions. Therefore,
curvature, Ricci tensor, and curvature scalar contain the same relevant
coefficient:

1
κ± (τ, ρ) = ba (∂τ ȧb ± ∂ρ ba ).
The Einstein tensor of two-dimensional gravity with abelian Lorentz and local
invariance group is trivial:
R• − g2 R•• = 0.
The following three examples are characteristic for the curvature concept:
the two-dimensional sphere (Riemannian) and the two one-shell hyper-
boloids, timelike (Riemannian and simply connected) and spacelike (pseudo-
Riemannian and doubly connected). They have the smallest real simple
Lie groups as maximal operation groups and the abelian Cartan subgroups
SO(2) and SO0 (1, 1) as local invariance groups:
Ω2 =∼ SO(3)/SO(2) : dω22 = dθ2 + sin2 θdϕ2 ,
2 ∼
Y = SO0 (1, 2)/SO(2) : dy22
= dψ 2 + sinh2 ψdϕ2 ,
Y (1,1) ∼
= SO0 (1, 2)/SO0 (1, 1) : dy2(1,1) = dψ 2 − cosh2 ψdϕ2 .
The curvature κ± = aä = −k is the representation-characterizing invariant

−k = λ2 , λ = (±i, ±1), of the compact and noncompact local abelian group
SO(2) and SO0 (1, 1) with the second-order derivative,

(sin θ, 1),
(∂ 2 + k)a = 0 : (a, k) =
(sinh ψ, −1), (cosh ψ, −1).
This gives the metric proportional curvature and Ricci tensors:
Ω2 : R1212 = −1, Rab
• = −12 ,
Y :
2
R1212 = +1, Rab
• = +12 ,
Y (1,1) : R1212 = +1, Rab
• = −η2 .
With ∂τ ∈ Lg as the global symmetry group, e.g., for a static manifold,

one obtains

g = b2 (ρ)dτ 2 ± a2 (ρ)dρ2 , κ± (ρ) = ± ba
1
∂ρ ba .
Reciprocal metrical coefficients in a self-dual dilation group SO0 (1, 1)
eλ3 σ3 with eλ3 = a,
dρ2
g = b2 (ρ)dτ 2 ± b2 (ρ) , ab = 1 ⇒ κ± (ρ) = ± 21 ∂ρ2 b2 ,
are used in the Schwarzschild geometry, which, for (1, 1)-spacetime, has a
nontrivial Ricci tensor:

dρ2
g = (1 − 2ρ )dτ 2 − 1− 2 ,
M(1,1) ∼
= R × P 1
: ρ
R1212
= κ− = ρ3 , R• = − ρ23 η2 .
2 ab
Two-dimensional rotation manifolds have SO(2) as the global symmetry

group, e.g., paraboloid and torus,

g = a2 (ρ)dρ2 + ρ2 dϕ2 ,
M ⊃Ω :
2 1
a
R1212
= κ+ = ∂ρ∂ 2 a12 = 2ρ
1
∂ρ a12 = − ρa 3, Rab
• = κ+ 1 2 ,
P2 : 1
a2 = 1 − 2
ρ , κ + = 1
ρ3 ,
1 c0 −ρ
T2 = Ω1 × Ω1 : a
1
2 = 1 − ( ρ−c0 2
c1 ) , κ + = c2 ρ
.
1
A constant curvature arises for
b2 (ρ) = 1
a2 (ρ) = c2 ρ2 + c1 ρ + c0 ⇒ R1212 = κ± = ±c2 .

Three-dimensional nonflat manifolds have nonabelian Lorentz groups and the
following Riemannian operation groups for (0, 3)- or (1, 2)-signature:
Tangent Poincaré group Global symmetry group Gg

SO(3) × R3 Gmax (0, 3) ∈ {SO(4), SO0 (1, 3)}
R3
SO0 (1, 2) × Gmax (1, 2) ∈ {SO0 (1, 3), SO0 (2, 2)}
Manifold
M(0,3) Killing form
M(1,2) Local invariance group
log SO(3) : κ ∼
= −1
3 Hg ⊆ Gg ∩ SO(3)
∼ 12 0 1 0
log SO0 (1, 2) : κ = = Hg ⊆ Gg ∩ SO0 (1, 2)
0 −1 0 η2
The local invariance group can be nonabelian and the Einstein tensor
nontrivial.
A (0, 3)-manifold with a 2-sphere factor dω22 = dθ2 + sin2 θdϕ2 has the
rotations in the global symmetry group and the axial rotations in the local
invariance group:
M(0,3) ⊃ Ω2 : Gg ⊇ SO(3),H
g ⊇ SO(2),

ě1 a dρ
g = a2 (ρ)dρ2 + ρ2 dω22 , ě2 = ρ dθ ,

ě3
ρ sin12θ dϕ 1 1
dě1 0 Γ = a dθ = ρa ě2 ,
dě2 = dρ ∧ dθ, ⇒ Γ 13 = a 1
sin θ dϕ = 1
ρa ě3 ,
dě3 [sin θ dρ + ρ cos θ dθ] ∧ dϕ Γ 23 = cos θ dϕ = cot θ
ě3
ρ

It arises by rotations (above) of ζ = f (ρ) with f (ρ) = a2 (ρ) − 1. With the
2-forms by exterior derivative,

Γ12 = − aa2 dρ ∧ dθ
dΓ = − ρa
a
3 ě1 ∧ ě2 ,

a
Γ13 = [− aa2 sin θ dρ +
dΓ 1
a
cos θ dθ] ∧ dϕ = [− ρa3 ě1 + ρ21a cot θ ě2 ] ∧ ě3 ,
Γ23 = − sin θ dθ ∧ dϕ
dΓ = − ρ12 ě2 ∧ ě3 ,
one obtains the curvature 2-forms that display the local SO(2)-invariance
2 ↔ 3:

Γ12 + Γ 13 ∧ Γ 23 = dΓ
R12 = dΓ Γ 12 = − ρa
a
3 ě1 ∧ ě2 =
∂ 1
ě1 ∧ ě2 ,
∂ρ2 a2
a
R 13
Γ
= dΓ 13
+Γ 12
∧Γ 32
Γ
= dΓ 13
−Γ 12
∧Γ 23
= − ρa 3 ě1 ∧ ě3 = ∂ 1
∂ρ2 a2
ě1 ∧ ě3 ,
Γ23 + Γ 21 ∧ Γ 31
R23 = dΓ = 1
( a12 − 1) ě2 ∧ ě3 .
ρ2
The (3 × 3) curvature matrix as a bilinear form of the tangent Lie algebra

log SO(3) decomposes as 3 = 1 ⊕ 2 into SO(2)-invariant spaces:

R1212 1 1
2ρ ∂ρ a2 12 0
R dabc ∼
= R
1313
= 1 1 .
0 ( a2 − 1)
R2323 ρ2
The two sums of the two curvature elements determine the Ricci tensor for
the SO(2)-invariant bilinear form of the translations R3 :

R11
• 1 1
ρ ∂ρ a2 0
Rab ∼
= 22
R• = ,
• 0
1
[ 2ρ ∂ρ a2 +
1 1 1
( a2 − 1)]12
R33
• ρ2

− ρ12 ( a12 − 1) 0
1 • 1 ab • ∼
2 R• − 1), Rab − 2 δ R•
1 1 1 1
= ρ ∂ρ a2 + ρ2 ( a2 • = 1 1 .
0 − 2ρ ∂ρ a2 12
A trivial curvature scalar characterizes the rotation 3-paraboloid:
ρ ∂ρ a2 + ρ2 ( a2 − 1) = 0 ⇐⇒ a2 = 1 −
1 1 1 1 1 2c1
⎧ ρ ,
2
⎪
⎪ dp3 = 1− 2ρ + ρ dω2 , a2 = 1 − ρ
2 dρ 2 2 1 2
⎨
P3 : ⇒ −2 aa3 = ρ22 = − ρ1 ( a12 − 1),
⎪
⎪
⎩ Rdabc ∼ = 13
−2 0
, Rab ∼
= 13
2 0
, R•• = 0.
ρ 0 12 • ρ 0 −12
The 3-sphere, Euclidean 3-space, and 3-hyperboloid have a maximal

global symmetry group and all the local invariance group SO(3):
⎫
Ω3 ∼
= SO(4)/SO(3), ⎬
1
= 1 − kρ2 ⇒ ∂ρ∂ 2 a12 = −k = ρ12 ( a12 − 1),
a2
R3 ∼
= SO(3) × R /SO(3), Rdabc ∼
= −k13 , Rab ∼
• = −2k13 ,
3
∼ ⎭
Y3 = SO 0 (1, 3)/SO(3), 1 •
2 R• = −3k, R• − 12 δ ab R•• ∼
ab
= k13 .
A spherical parametrization of a manifold, conformal to a 3-sphere, gives

the curvature and the Ricci tensor. Now R = ∂α R,
g = R2 (α)(dα

2
+ sin2 αdω22 ),
R
∼ 2 − 2(cot α + ∂α ) 0
Rab
• = − R12 R
R R
.
0 (2 − 3 cot α R − R )12
A hyperbolic parametrization of an SO(2)-invariant (0, 3)-manifold,

conformal to the 3-hyperboloid, is obtained by noncompact–compact transi-
tion with ψ ↔ iα and R ↔ −iR :
g = R2 (ψ)(dψ

2
+ sinh2 ψdω22 ),
R
R dabc ∼
= 1 1 + (2 coth ψ + R )R
R 0
,
R2 0 [1 + (coth ψ + ∂ψ ) R
R ]12

Rab ∼
= 1 2 + 2(coth ψ + ∂ψ ) R
R 0
.
• R2 R
0 (2 + 3 coth ψ RR + R )12
For three dimensions, a diagonal metric can lead to a nondiagonal Ricci

tensor, as exemplified with a causal (1, 2)-manifold M(1,2) ⊃ Ω1 with global
symmetry group SO(2) ⊆ Gg [details ahead as special case of the (1,3)-case]:

g = b2 (τ, ρ)dτ 2 − [a2 (τ, ρ)dρ2 + ρ2 dϕ2 ], κ− = 1
(∂ ȧ
ba τ b
− ∂ρ ba ),
⎛ 0101 ⎞ ⎛ 2 ⎞
R ρa κ− 0 0
b
Rdabc ∼
=⎝ R0202 R0212 ⎠ = ρa2 ⎝
1
0 −b − ȧb
⎠,
R0212 R1212 a
0 −b ⎛ − a
ȧ
⎞
⎛ ⎞
R00
• R01
• 1 0 0 2 bb 2 ȧ 0
⎜ ⎟
b

Rab ∼
= ⎝ R• R11 ⎠ = −κ−
10 1
• • 0 −1 0 + ρa2 ⎝ 2 b ȧ
2 aa 0 ⎠.
22 0 0 0 b a
R• 0 0 −b + a
Static spacetime with the motion group SO0 (1, 1) has reciprocal metrical
time-independent coefficients,
2
g = b2 (ρ)dτ 2 − [ bdρ 2 2
2 (ρ) + ρ dϕ ], ab = 1, ḃ = 0, κ− = − 12 ∂ρ2 b2 ,

R0101 − 12 ∂ρ2 b2 0
R dabc ∼
= R 0202 = 1 ,
R1212 0 − ρ ∂ρ b2 η2

R00
• ( 12 ∂ρ2 b2 + 1 2
ρ ∂ρ b )η2 0
Rab ∼ R11 = ,
• = •
2
22 0 − ρ ∂ρ b2
R•
1 •
2 R• = 12 ∂ρ2 b2 + 2ρ ∂ρ b2 .
(1, 2)-Schwarzschild spacetime has a nontrivial Ricci tensor:
⎧ dρ2
⎪
⎪ g = (1 − ρ2 )dτ 2 − [ 1− 2 2
2 + ρ dϕ ],
⎪
⎨ ρ
M(1,2) ∼ Rdabc ∼
η 0
= R × P2 : = ρ23 02 1 ,
⎪
⎪
⎪
⎩ ∼ 4 02 0 , 1 R• = 23 .
Rab
• = ρ3 0 −1 2 • ρ
2.9.4 Rotation-Invariant Four-Dimensional Spacetimes

A rotation-invariant nonflat causal (1, 3)-spacetime has as Riemannian oper-
ation groups:
Tangent Poincaré group Global symmetry group Gg ⊇ SO(3)
SO0 (1, 3) ×
R4 Gmax (1, 3) ∈ {SO0 (1, 4), SO0 (2, 3)}
M(1,3) ⊃ Ω2 :
Killing
form of log
SO0 (1, 3) Local invariance group
κ=∼ 13 0
= η2 ⊗ 13 SO(2) ⊆ Hg ⊆ Gg ∩ SO0 (1, 3)
0 −13
A metric can be given with an orthonormal basis of the cotangent spaces,

ě b dτ
0
ě1 a dρ
g = b2 (τ, ρ)dτ 2 − [a2 (τ, ρ)dρ2 + ρ2 dω22 ] ⇒ ě2 = ρ dθ .
ě3 ρ sin θ dϕ
The nontrivial connection 1-forms are with ḃ = ∂τ b, b = ∂ρ b, etc.:

⎧
⎪ b b
⎪
⎪ Γ 01 = ȧ
dρ + dτ = ȧ
ě1 + ě0 ,
⎛ ⎞ ⎪ b a ba ba
dě0 ⎛ ⎞ ⎪
⎪
b dρ ∧ dτ ⎪
⎨ Γ 12
⎜dě1 ⎟ ȧ dτ ∧ dρ = 1
dθ = 1
ě2 ,
⎜ ⎟ ⎝ ⎠⇒ a ρa
⎝dě2 ⎠ = dρ ∧ dθ, ⎪ Γ 13 = 1
sin θ dϕ = 1
ě3 ,
dρ + ρ cos θ dθ] ∧ dϕ ⎪
⎪ a ρa
dě3
[sin θ ⎪
⎪
⎪
⎪
⎩ Γ 23 = cos θ dϕ = cot θ
ě3 ;
ρ
with the 2-forms by exterior derivative:

Γ01 = (∂τ ȧb − ∂ρ ba ) dτ ∧ dρ
dΓ = κ− ě0 ∧ ě1 ,
Γ12 = [∂τ a1 dτ + ∂ρ a1 dρ] ∧ dθ
dΓ = 1ρ [ 1b ∂τ a1 ě0 + a1 ∂ρ a1 ě1 ] ∧ ě2 ,
Γ13 = [∂τ a1 sin θ dτ + ∂ρ a1 sin θ dρ +
dΓ 1
a
cos θ dθ] ∧ dϕ = 1ρ [ 1b ∂τ a1 ě0 + a1 ∂ρ a1 ě1
+ ρa 1
cot θ ě2 ] ∧ ě3 ,
Γ23 = − sin θ dθ ∧ dϕ
dΓ = − ρ12 ě2 ∧ ě3 ;
with the (1, 1)-curvature

κ− = 1 ȧ
ba (∂τ b − ∂ρ ba ).
Thus, one obtains the curvature 2-forms [there is the local SO(2)-
invariance 2 ↔ 3]:
R01= dΓΓ01 + Γ 02 ∧ Γ 12 + Γ 03 ∧ Γ 13 = dΓ
Γ01 = κ− ě0 ∧ ě1 ,
b
R = dΓ
02
Γ + Γ ∧ Γ + Γ ∧ Γ = −Γ
02 01 21 03 23
Γ01 ∧ Γ 12 = −[ ba ě0 + ba
ȧ
ě1 ] ∧ ρa
1
ě2 ,
R = dΓ
12
Γ − Γ ∧ Γ + Γ ∧ Γ = dΓ
12 10 20 13 23
Γ 12
= [ ρb ∂τ a ě0 + ρa ∂ρ a ě1 ] ∧ˇe2 ,
1 1 1 1
Γ23 − Γ 20 ∧ Γ 30 + Γ 21 ∧ Γ 31 = dΓ
R23 = dΓ Γ23 + Γ 12 ∧ Γ 13= ( a12 − 1) ρ12 ě2 ∧ ě3 ,
with the (6×6)-curvature matrix as a bilinear form of the Lorentz Lie algebra
log SO0 (1, 3). It decomposes into SO(2)-invariant subspaces of dimensions 1
and 2:
⎛ ⎞
R0101
⎜ R0202 R0212 ⎟
⎜ R0303 R0313 ⎟
R dabc ∼
=⎜ ⎟
⎜ R2323 ⎟
⎝ ⎠
R0212 R1212
R0313 R1313
⎛ ρa2 κ− 0 0 0
⎞

− bb 12 − ȧ
1 ⎜ ⎟
0 0 b 12
= ρa2 ⎝ 0 0 1−a2
0
⎠.
ρ

0 − ȧ
b 12 0 − aa 12
The bilinear Ricci tensor gives a (4 × 4)-matrix with subtraces of the cur-
vature as an SO(2)-invariant bilinear form of the spacetime translations R4 :

R00
• = −R0101 − 2R0202 = −κ− + ρa 2b
2b ,

R11
• =R 0101
− 2R 1212 2a
= κ− + ρa3 ,
R01
• = −2R0212 = ρa2ȧ2 b ,

b 1−a2
R22
• = R33
• = R
0202
− R1212 − R2323 = ρa12 ( aa − b) − ρ2 a2 ,
⎛ R00 R01 ⎞
• •
∼ ⎜ R10 R•
11
⎟
Rab
• =⎝ •
22
⎠
R•
R33
• ⎛ ⎞
1 0 0
2 bb

2 ȧ
b 0
1 ⎝ 2 ȧ
⎠.
= −κ− 0 −1 0 + ρa2 b 2 aa 0
0 0 02 a 1−a2
0 0 (− bb + a − ρ )12
A trivial Ricci tensor characterizes Schwarzschild (1, 3)-spacetime:

R0212 = 0,
R• = 0 :
2R0202 = −R0101 = −2R1212 = R2323

ȧ = 0, ∂ρ ba = − ρ2 ba , ba = constant,
⇒ b a ⇒
− = = 1−a2 1
a2 = b2 = 1 − 2cρ1 .
b a 2ρ
It has position paraboloids P . The global symmetry group is R × SO(3),

3
with SO(2) as the local invariance group:

⎧ 2
⎪
⎪ g = (1 − 2ρ )dτ 2 − 1−
dρ 2 2
2 + ρ dω2 ,
⎪
⎨ ⎛ ρ ⎞
M(1,3) ∼
2 0 0 0
= R × P3 : 0 −12
⎪
⎪ Rdabc ∼
= ρ13 ⎝ 0
0 0
⎠ = 13 η2 ⊗ 2 0
.
⎪
⎩ 0 −2 0 ρ 0 −1 2
0 0 0 12
For only two position dimensions with abelian rotations SO(2) and (1, 2)-
signature, a trivial Ricci tensor requires a flat spacetime — there is no ana-
logue for Schwarzschild (1, 3)-spacetime,

ȧ = 0, bb = aa = 0,
(1, 2)-spacetime: R• = 0 ⇒
g = b2 (τ )dτ 2 − (dρ2 + ρ2 dϕ2 ).
The curvature and Ricci tensor are diagonal for time independent position
dilation ȧ = 0, they lead to static spacetimes, and the time dependence in b
is ineffective,
g = b2 (τ, ρ)dτ 2 − [a2 (ρ)dρ2 + ρ2 dω22 ],
⎛ −ρ a ∂ b 0 0 0
⎞
b ρ a
⎜ b
−b ⎟
Rdabc ∼ 1 ⎜ ⎟,
0 12 0 0
= ρa2 ⎝ ⎠
1−a2
0 0 ρ 0

⎛ 0 0 0 − aa 12 ⎞
b b
ρ a
b ∂ρ a + 2 b 0 0
Rab ∼ 1 ⎝ −ρ a b
+ 2 aa ⎠,
• = ρa2
0 b ∂ρ a 0
a 1−a2
0 0 −( bb − a + ρ )12
1 • b 2 b a 1−a2
2 R• −
1
= ab ∂ρ a + ρa2 ( b a) + ρ2 a2 ;
e.g., Einstein’s static universe with b = 1 and 1

a2 = 1 − ρ2 ,
M(1,3) ∼
= R × Ω3 : g = dτ 2 − dω32 .
A position dependence of time only leads to
⎧ ⎛ b ⎞
⎪
⎪ − b 0 0 0
⎪
⎪ ⎜
⎟
⎪
⎪ R dabc ∼
=⎝ 0 − 1 b
ρ b 12 0 0
⎠,
⎪
⎪
⎪
⎪ 0 0 0 0
⎨ ⎛ 0 0 0 0 ⎞
g = b2 (ρ)dτ 2 − (dρ2 + ρ2 dω22 ), b 2 b
b + ρ b 0 0
⎪
⎪ ab ∼ ⎝ ⎠,
⎪
⎪ R = 0 − b
0
⎪
⎪
• b
b
⎪
⎪ 0 0 −ρ
1
12
⎪
⎪
b
⎩ 1 R• = b + 2 b .
2 • b ρ b
Static manifolds with time-independent dilations SO0 (1, 1) like Schwarz-

schild spacetime have an (η2 ⊕ 12 )-Ricci form. They are given in the form
of endomorphisms of the Lorentz Lie algebra and the tangent translations:
2
g = b2 (ρ)dτ 2 − [ bdρ 2 2
2 (ρ) + ρ dω2 ],
⎧ ⎛ − 1 ∂ 2 b2 ⎞
⎪
⎪ 2 ρ 0 0 0
⎪
⎪ ⎜ ⎟
⎪
⎪ dabc ∼ ⎜ 0 −
1 2
2ρ ρ b 12
∂ 0 0
⎟,
⎪
⎪ R =⎝ ⎠
⎪
⎨ 0 0 b2 −1
0
ρ2
1 2
⎪ 0 0 0 2ρ ∂ρ b 12
⎪
⎪ ( 12 ∂ρ2 b2 + 1 2
ρ ∂ρ b )η2 0
⎪
⎪ Rab ∼ ,
⎪ • = b2 −1
⎪
⎪ 0 −( ρ
1
∂ρ b2 + )12
⎪
⎩ 1 • b2 −1
ρ2
2 R•
1 2 2 2 2
= 2 ∂ρ b + ρ ∂ρ b + ρ2 .
A trivial curvature scalar,

b2 −1
1 2 2
2 ∂ρ b + ρ2 ∂ρ b2 + ρ2 = 0 ⇐⇒ b2 (ρ) = 1 − 2c1
ρ + c2
ρ2 ,
characterizes Reissner spacetimes with Schwarzschild length c1 = m and

charge area c2 = 2z :
2

g = 1− 2c1
ρ + c2
ρ2 dτ 2 − 1− 2cdρ1 + c2 + ρ2 dω22 ,
⎛ 2c1 −
3c2
0
ρ ρ2
0 0
⎞
ρ
1 ⎜ ⎟
c2
Rdabc ∼
= ρ3 ⎝
0 −(c1 − ρ )12 0
c2
0
⎠,
0 0 −2c1 + ρ 0
c2
(c1 −
0
0 0 ρ )12
1 0 0
Rab ∼ c2
−1 , R•• = 0.
• = ρ4 0 0
0 0 12
A curvature tensor with a factor η2 ⊗ requires the metric for a Schwarzschild

spacetime with a cosmological constant
k2 = Λ3 ,
b −1 2
1 2 2
2 ∂ρ b = ⇐⇒ b2 (ρ) = 1 + kρ2 − 2cρ1 ,
ρ2 2

g = 1 + kρ2 − 2cρ1 dτ 2 − 1+kρdρ2 − 2c1 + ρ2 dω22 ,
2c
ρ
dabc ∼ k − ρ31 0
R = η2 ⊗ 0
c
(k + 1 )1
,
ρ3 2
Rab ∼ 1 •
• = 3kη4 , 2 R• = 6k.
Local SO(3)-invariance implies maximal local SO0 (1, 3)-invariance for max-
imal symmetric (anti-)de Sitter and flat spacetimes:
⎧ 2
⎪
⎨ g = b2 (ρ)dτ 2 − [ bdρ 2 2
2 (ρ) + ρ dω2 ],
(1,3) ∼
M = (Y (1,3)
,R ,Y
4 (3,1)
): b −1
2
1
= 2ρ ∂ρ b2 ⇒ b2 (ρ) = 1 + kρ2 ,
⎪
⎩ dabc
ρ2
R ∼
= k1(3,3) for k = (1, 0, −1).
Nondiagonal elements arise for only time-dependent dilations. The time

dilation factor can be absorbed in a reparametrization, b = 1:
g = dτ 2 − [a2 (τ )dρ2 + ρ2 dω22 ],

1 0 0
0 2ȧ 0
ab ∼
R• = − a 0 −1 0
ä
+ ρa12 2ȧ 0 0
2
.
0 0 02 0 0 − 1−a
ρ 12
2.9.5 Robertson–Walker Metrics

Examples for SO(2)-invariant (1, 3)-spacetimes with hyperbolic or spherical
coordinates,
− RT (τ,
2
g = T2 (τ,
ψ)dτ
2 2
ψ)(dψ 2 2
+ sinh ψdω2 )
ě 0 dτ
ě1 R dψ
⇒ ě2 = R sinh ψ dθ ,
ě3 R sinh ψ sin θ dϕ
are the SO(3)-invariant Friedmann universes with position manifolds Ω3 ,

R3 , and Y 3 , the global symmetry groups SO(4), SO(3) ×
R3 , and SO0 (1, 3),
respectively, and the local invariance group SO(3). They have a nontrivial
time representation coefficient, t −→ R(t):
g = dt2 − R2 (t)dσk2 = T 2 (τ )(dτ 2 − dσk2 ),

dt
with dτ = R(t) = T (τ ),
dρ2
dσk = (dω32 , dy23 , dx3 2 ) = 1−kρ
2 2 2
2 + ρ dω2 , k = (1, 0, −1).
The nontrivial connection 1-forms of (1, 3)-spacetimes with local SO(2)-

invariance are with Ṫ = ∂τ T , T = ∂ψ T , etc.:
⎛ ⎞
⎛ ⎞ T dψ ∧ dτ
dě0 ⎜ ⎟
⎜dě1 ⎟ ⎜ Ṙ dτ ∧ dψ ⎟
⎜ ⎟ = ⎜ [Ṙ sinh ψ dτ + (R sinh ψ + R cosh ψ) dψ] ∧ dθ, ⎟
⎝dě2 ⎠ ⎜ ⎟
⎝[Ṙ sinh ψ sin θ dτ + (R sinh ψ + R cosh ψ) sin θ dψ ⎠
dě3
+R sinh ψ cos θ dθ] ∧ dϕ
⎧ Ṙ T Ṙ T
⎪
⎪ Γ01 = dψ + dτ = ě1 + ě0 ,
⎪
⎪
T R TR TR
⎪
⎪
⎪
⎪
⎪
⎪ Γ 02 = Ṙ
sinh ψ dθ = Ṙ
ě2 ,
⎪
⎪ T TR
⎪
⎨ Γ 03 = Ṙ
sinh ψ sin θ dϕ = Ṙ
ě3 ,
T TR
⇒
⎪
⎪
1 R
⎪
⎪ Γ 12 = ( RR sinh ψ + cosh ψ) dθ =
⎪ R ( R + coth ψ) ě2 ,
⎪
⎪ R
⎪
⎪ Γ 13
= ( sinh ψ + cosh ψ) sin θ dϕ = 1 R
R( R + coth ψ) ě3 ,
⎪
⎪
R
⎪
⎪
⎩ 23 cot θ
Γ = cos θ dϕ = R sinh ψ ě3 ,
with the 2-forms by exterior derivative:

Γ01 = (∂τ Ṙ
dΓ T
− ∂ψ TR ) dτ ∧ dψ = 1
TR
(∂τ Ṙ
T
− ∂ψ TR ) ě0 ∧ ě1 ,
Γ02 = [sinh ψ∂τ Ṙ

dΓ T
dτ + ∂ψ ( Ṙ
T
sinh ψ) dψ] ∧ dθ
= R [ T ∂τ T ě0 + R sinh
1 1 Ṙ 1
∂
ψ ψ T
( Ṙ sinh ψ) ě1 ] ∧ ě2 ,
Γ03 = [sinh ψ∂τ Ṙ
dΓ T
sin θ dτ + ∂ψ ( Ṙ
T
sinh ψ) sin θ dψ + Ṙ T
sinh ψ cos θ dθ] ∧ dϕ
= R [ T ∂τ T ě0 + R sinh ψ ∂ψ ( T sinh ψ) ě1 + T R sinh ψ cot θ ě2 ] ∧ ě3 ,
1 1 Ṙ 1 Ṙ Ṙ

Γ12 = [sinh ψ∂τ RR dτ + ∂ψ ( RR sinh ψ + cosh ψ) dψ] ∧ dθ
dΓ

= R [ T ∂τ RR ě0 + R sinh
1 1 1
∂ ( R sinh ψ + cosh ψ) ě1 ] ∧ ě2 ,
ψ ψ R

Γ13 = [sinh ψ∂τ RR sin θ dτ + ∂ψ ( RR sinh ψ + cosh ψ) sin θ dψ
dΓ

+( RR sinh ψ + cosh ψ) cos θ dθ] ∧ dϕ

= R [ T ∂τ RR ě0 + R sinh
1 1 1
∂ ( R sinh ψ + cosh ψ) ě1
ψ ψ R
R
+ R sinh ψ ( R + coth ψ) cot θ ě2 ] ∧ ě3 ,
1
Γ23 = − sin θ dθ ∧ dϕ = − R2 sinh

dΓ 1
2 ψ ě2 ∧ ě3 .
Thus, one obtains the nontrivial curvature 2-forms [there is the local
SO(2)-invariance 2 ↔ 3]:
R01 = dΓ
Γ01 + Γ 02 ∧ Γ 12 + Γ 03 ∧ Γ 13 = dΓ
Γ01
T
= T R (∂τ T − ∂ψ R ) ě0 ∧ ě1 ,
1 Ṙ
Γ02 + Γ 01 ∧ Γ 21 + Γ 03 ∧ Γ 23 = dΓ
R02 = dΓ Γ02 − Γ 01 ∧ Γ 12
T 1 R
= −[ T R ě1 + T R ě0 ] ∧ R ( R + coth ψ) ě2 +[ T1R ∂τ Ṙ
Ṙ
T
ě0 + R2 sinh
1
∂
ψ ψ
×( Ṙ
T
sinh ψ) ě1 ] ∧ˇe2 ,
Γ12 − Γ 10 ∧ Γ 20 + Γ 13 ∧ Γ 23 = dΓ
R12 = dΓ Γ12 − Γ 01 ∧ Γ 02

= −[ TṘR ě1 + TTR ě0 ] ∧ TṘR ě2 + [ T1R ∂τ RR ě0 + R2 sinh
1
∂
ψ ψ

×( R sinh ψ + cosh ψ) ě1 ] ∧ˇe2 ,
R
Γ23 − Γ 20 ∧ Γ 30 + Γ 21 ∧ Γ 31 = dΓ
R23 = dΓ Γ23 − Γ 02 ∧ Γ 03 + Γ 12 ∧ Γ 13
1 R 1 R
= − T R ě2 ∧ T R ě3 + R ( R + coth ψ) ě2 ∧ R
Ṙ Ṙ
( R + coth ψ) ě3
− R2 sinh
1
2 ψ ě2 ∧ ě3 ,
with the following nontrivial curvature components:

g = T 2 (τ, ψ)dτ 2 − R2 (τ, ψ)(dψ 2 + sinh2 ψdω22 ),
T
R T − ∂ψ R ],
= T1R [∂τ Ṙ
0101
R T
R0202 = R0303 = T1R [∂τ Ṙ
T − (coth ψ + R ) R ],

R0212 = R0313 = T1R [∂τ RR − TT RṘ ],
2
R
R1212 = R1313 = R12 [1 − Ṙ
T 2 + (coth ψ + ∂ψ ) R ],
2
R R

R2323 = R12 [1 − Ṙ
T 2 + (2 coth ψ + R ) R ].
The Ricci tensor has the following nontrivial components:

R00
• = − T1R [3∂τ Ṙ
T − ∂ψ R ] + R2 (coth ψ + R ) T ,
T 2 R T
T Ṙ2
R11
• = + T R [∂τ T − ∂ψ R ] − R2 [2 − 2 T 2 + 2(coth ψ + ∂ψ ) RR ],
1 Ṙ 1

R01
• = − T1R [2∂τ RR − 2 TT RṘ ],
Ṙ2 T R R T R
R22
• = R33
• T − R2 [2 − 2 T 2 + coth ψ( T + 3 R ) + R + T R ].
= + T1R ∂τ Ṙ 1
For the three Robertson–Walker metrics, one obtains — in addition to

hyperbolic position k = −1 — the metric for spherical k = 1 and flat position
k = 0:
⎧
⎪ dabc ∼ 1 R̈R13 0
⎪
⎨ R = ,
0 −(Ṙ + k)13
R 2 2
g = dt − R (t)dσk ,
2 2 2
R ab ∼
= 1 −3 R̈R 0
,
⎪
⎪ • R2 0 2
(R̈R + 2Ṙ + 2k)13
⎩ 1 • R̈R+Ṙ2 +k
2 R• = −3 R2 ,
⎧
⎪
⎪ Rdabc ∼ 1 T − ( T ) ]13
[ T̈ Ṫ 2
0
⎪
⎨
= T2 −[( T
Ṫ 2 ,
0 ) + k]13

g = T 2 (τ )(dτ 2 − dσk2 ), Rab ∼
= 1 −3[ T̈
T − (T ) ]
Ṫ 2
0
,
⎪
⎪ • T2 [ T̈ Ṫ 2
⎪
⎩ 1 •
0 T + (T ) + 2k]13
2 R• = − T32 ( T̈T + 2k).

2.10 Covariantly Constant-Curvature Manifolds 73
The de Sitter, flat, and anti-de Sitters manifold with maximal symmetry
arise for
⎧
⎪
⎪ g = dt2 − R2 (t)dy23
⎪
⎨ = T 2 (τ )(dτ 2 − dy23 ),
∼ (Y (1,3) , R4 , Y (3,1) ) :
M(1,3) =
⎪ R(t) = (sinh t, t, sin t)
⎪
⎪
⎩ = T (τ ) = √ eτ 2τ , eτ , √ eτ 2τ .
1−e 1+e
2.10 Covariantly Constant-Curvature

Manifolds
A torsionless manifold with covariantly constant curvature is called affine
locally symmetric,
T = 0, ∇R = 0.
Special cases of manifolds with covariantly constant curvature are globally

symmetric Riemannian manifolds with a definite metric, completely classified
by Cartan. They are isomorphic to symmetric spaces G/K with classes of a
compact local invariance subgroup K ⊆ G in a simply connected real Lie
group as the global symmetry (motion) group and inherit adjoint structures
of G. Their algebraic origin lies in orthogonal symmetric Lie algebras.
2.10.1 Orthogonal Symmetric Lie Algebras

An orthogonal symmetric Lie algebra (L, R) is a real Lie algebra with a⎪non-
⎪
trivial involutive automorphism where the fixed elements C = {c ∈ L⎪ ⎪R(c)
= c} constitute a compact Lie algebra. It is called effective if the distinguished
compact subalgebra does not contain central elements centr L ∩ C = {0}.
An orthogonal symmetric Lie algebra has a Killing form orthogonal direct
decomposition into involution eigenspaces, the compact Lie subalgebra C, and
the R-antisymmetric vector subspace V , compatible with the Lie bracket,

[C, C] ⊆ C, [C, V ] ⊆ V, [V, V ] ⊆ C,
L= LR+ ⊥LR− = C⊥V,
κ(C, V ) = {0}.
There are three types of effective orthogonal symmetric Lie algebras, de-
noted with subindex (c, nc, 0),
( ⇒ (Lc , Rc ) of compact type,

Lc semisimple and compact
Lnc semisimple and noncompact,
⇒ (Lnc , Rnc ) of noncompact type,
Lnc = C⊥V is a Cartan decomposition
[V, V ] = {0}, V abelian ideal ⇒ (L0 , R0 ) of Euclidean type.
Therefore, all semisimple real Lie algebras are effective orthogonal symmetric.
Every effective orthogonal symmetric Lie algebra has a Killing form orthogo-
nal decomposition into ideals of compact, noncompact, and Euclidean types:
(L, R) = (Lc , Rc )⊥(Lnc , Rnc )⊥(L0 , R0 ).
In the spherical, hyperbolic, and flat example the antisymmetric sub-

spaces contain the classes of Lie algebras for orthogonal groups O = −OT ∈
log SO(s). They are the Rs -isomorphic tangent spaces of sphere, hyperboloid,
and Euclidean space:

0 θ
Rs −θT O ∈ log SO(1 + s) = log SO(s)⊥ log Ωs ,

0 β
Rs T
β O
∈ log SO0 (1, s) = log SO(s)⊥ log Y s ,

0
x
Rs 0 O
Rs ] = log SO(s) ⊕
∈ log[SO(s) × Rs .
The polar decompositions of the complex linear groups give rise to orthogonal
symmetric, not effective, Lie algebras:
log[GL(n, C)] = log U(n)⊥R(n), R(n) ∼

2
= Rn .
For an orthogonal symmetric Lie algebra (L, R), the complexification

(L∗ , R∗ ) with “imaginary” iV is, as a real Lie algebra, orthogonal symmetric:

L = C⊥V, R ↔ L∗ = C⊥iV, R∗ (C + iV) = C − iV .
There is a compact–noncompact duality: If (L, R) is of the compact type, its

dual (L∗ , R∗ ) is of the noncompact type, and vice versa.
An orthogonal symmetric Lie algebra (L, C) is irreducible if L is semisim-
ple and if C does not contain a nontrivial L-ideal. Then [V, V ] = C, and
ad L C acts irreducibly on V .
2.10.2 Real Simple Lie Algebras

The transition from a complex to real simple Lie algebra is characterized by
a Lie algebra involution (conjugation).
According to Cartan, there are two types of real simple Lie algebras. The
first type arises from a complex simple Lie algebra Lr with rank r by using
the three possible kinds of involutions R (anticonjugation, symplectic, and
orthogonal) of its compact form:
lagR (compact) Lcr −→ LRr ∈ lagR (noncompact, simple).

L Complex Lr Compact Lc
r Anti LR
r Symplectic LH
r Orthogonal Lp,q
r
dimC L = d in log SL(n, R) in log SL(n, H)
p+q =1+r
for 1 + r even
r≥1 log SL(1 + r, R) log SU(p, q)
log SU(1 + r) log SU (1 + r)
A log SL(1 + r, C) ⊃ logSO(1
+ r) ⊃ log SU(p) × U(1)
r(2 + r) ⊃ log SpU(1 + r)
r(2 + r)
1+r ×SU(q)
2 2+r
2 p2 + q 2 − 1
p+q = r
r≥3 log Sp(2r) log SpU(2p, 2q)
2r logSpU(2r)

C log Sp(C ) ⊃ log U(r) = log Sp(2r) ⊃ log SpU(2p)
1+2r
1+2r ×SpU(2q)
2
2 r2
2p+1 2q+1
2
+ 2
p + q = 1 + 2r
r≥2
log SO(1 + 2r) log SO(p, q)
B log SO(1 +2r, C) = log SO(1 + 2r) −
1+2r ⊃ log SO(p)
1+2r 2 ×SO(q)
p
2
2
+ q
2
p + q = 2r
r≥4 log SO (2r)
log SO(2r) log SO(p, q)
log SO(2r,
D C) 2r = log SO(2r) ⊃ log U(r)
⊃ log SO(p) × SO(q)
2r 2 r2
p
2
2
+ q
2
Simple complex and real Lie algebras

R H
of real form type I: Lc p,q
r , Lr , Lr , Lr
(⊃ maximal compact Lie subalgebra with dimension)
(without the exceptional Lie algebras)
For small rank r ≤ 4, there are Lie algebra isomorphies, for the complex
algebras given by
A1 ∼
= B1 ∼
= C1 , B2 ∼
= C2 , D2 ∼
= A1 ⊕ A1 , D3 ∼
= A3 .
One has the inclusions
Ar ⊂ D1+r ⊂ B1+r ∩ C1+r .
Via their adjoint representations and the orthogonal invariance group of the
Killing form, all semisimple Lie algebras can be considered orthogonal sub-
algebras:
Cd ∼
=L∼ = ad L ⊆ log SO(d, C) ∼
d
= C( 2 ) .
In addition to the real forms of simple complex Lie algebras, there are
the simple real Lie algebras, which are the canonically complexified (with the
doubled reals CR = R ⊕ iR) compact Lie algebras Lcr . The basic compact
Lie algebras are maximal in the doubling L(r,r) = Lcr ⊕ iLcr with doubled
dimension 2dc and doubled rank 2r.
Lie algebra Lie group Dimension d = 2dc
A(r,r) ∼
= log SL(1 + r, CR ) 2r(2 + r)

C(r,r) ∼
= log Sp(2r, CR ) 2 1+2r
2
1+2r
B(r,r) ∼
= log SO(1 + 2r, CR ) 2 2

D(r,r) ∼
= log SO(2r, CR ) 2 2r
2
Simple real Lie algebras

of canonical complexification type II: L(r,r) = Lcr ⊕ iLcr
(without the exceptional Lie algebras)
2.10.3 Globally Symmetric Riemannian Manifolds

Corresponding to the classification of the real simple Lie algebras, there are
two types, I and II, of irreducible orthogonal symmetric Lie algebras (L, R).
Both types come in pairs (c, nc) connected with each other by the compact–
noncompact duality.
The compact Lc are
type Ic : Lc simple, R any involutive Lc -automorphism,

R
type IIc : Lc = L1 ⊕ L2 with simple ideals L1 ↔ L2 .
The two isomorphic compact ideals L1 ∼

= L2 give a compact subalgebra with
the reflection-symmetric elements C = L+ = [L± , L± ] and a vector subspace
with reflection-antisymmetric ones, V = L− = [L+ , L− ]. There is the vector
space isomorphy V ∼ = C.
The duality-related noncompact Lnc are
type Inc : Lnc simple with simple complexification,

R keeps fixed a maximal compactly embedded subalgebra,
type IInc : Lnc = L1 ⊕ iL2 with C • [L1 ⊕ iL2 ] = L ∈ lagC simple,
R is an L-conjugation
with respect to a maximal compactly embedded subalgebra.
A pair (G, K) of a connected Lie group with a compact connected subgroup

is associated with an orthogonal symmetric Lie algebra L = C⊥V for (L, C) =
(log G, log K). For each G-invariant positive definite metric g (which ex-
ists), (G/K, g) is a locally symmetric Riemannian manifold, i.e., torsionfree
with covariantly constant curvature. The tangent space at the unit coset
1K ∈ G/K is the antisymmetric vector subspace T1K (G/K) = V . The uni-
versal covering group defines the universal covering manifold exp L/K for
all (L, C)-associated locally symmetric Riemannian manifolds; it is a globally
symmetric Riemannian manifold. The Cartan classification above of the real
simple Lie algebras and the irreducible orthogonal symmetric Lie algebras
gives all globally symmetric Riemannian manifolds.
Subtype Noncompact Inc Compact Ic Dimension n Real rank r

(d+2)(d−1)
AI SL(d, R)/SO(d) SU(d)/SO(d) 2 d−1
A II SU∗ (2d)/SpU(2d) SU(2d)/SpU(2d) (2d + 1)(d − 1) d−1
A III SU(p, q)/SU(p) SU(p + q)/SU(p) 2pq min(p, q)
×U(1) × SU(q) ×U(1) × SU(q)

C I Sp(2d)/U(d) SpU(2d)/U(d) 2 d+1

2 d
C II Sp(2p, 2q)/SpU(2p) SpU(2p + 2q)/SpU(2p) 4pq min(p, q)
×SpU(2q) ×SpU(2q)
BD I SO0 (p, q)/SO(p) SO(p + q)/SO(p) pq min(p, q)
×SO(q) ×SO(q)

D III SO∗ (2d)/U(d) SO(2d)/U(d) 2 d

2 [d
2]
Riemannian globally symmetric manifolds of type I

(without the exceptional manifolds)
Examples for BD I are the timelike one-shell hyperboloids Y s and their

compact partners the spheres Ωs with the same compact subgroup SO(s) for
the classes.
Starting from a Cartan factorization of a noncompact group with simple
Lie algebra into maximal compact and maximal abelian noncompact sub-
groups whose dimension is the real rank
G = KAK, dimR (G, K, A) = (dG , dK , r),
the Riemannian manifold is given by the maximal compact classes
type Inc : M ∼
= G/K, dimR M = n = dG − dK .
For the compact manifold, the abelian group A has to be replaced by its
compact partner; e.g., SO0 (1, 1) eβσ3 by eiασ3 ∈ SO(2).
The type II involves a doubling: The compact manifolds contain the
diagonal group classes of the doubling of the simple, compact Lie groups
K = exp L, e.g., SU(n) = SU(n) × SU(n)/SU(n). They are isomorphic to
the group K ∼
= K.
Noncompact IInc Compact IIc Dimension Real rank
K × K ∗ /K [K × K]/ diag [K × K] dK rK
Riemannian globally symmetric manifolds of type II
The IInc noncompact partner manifolds are G/K with a connected Lie
group G and maximal compact subgroup K, where the complexification
log G = log K + i log K is simple; e.g., SL(n, C)/SU(n).
2.10.4 Curvature of Globally Symmetric Riemannian

Manifolds
The adjoint Lie algebra structures can be taken over by the involution anti-
symmetric subspace V of an orthogonal symmetric Lie algebra,
n=dimR L
(la )a=1 ,
(log G, log K) = (L, C), L = C⊥V, bases:
(C α )s=dim
α=1
RC
, (V A )dim RV
A=1 .
The doubled adjoint V -action defines endomorphisms for V and C:
[[V, V ], V ] ⊆ [C, V ] ⊆ V,
[[V, V ], C] ⊆ [C, C] ⊆ C.
The antisymmetric subspace V is an example for a triple Lie subspace, defined

in general by a double adjoint stability [[V, V ], V ] ⊆ V .
The exponential mapping of V parametrizes the manifold G/K ∼ = exp V .
The n-bein for the group action 1K −→ gK relate to each other the tangent
spaces. They involve the even contribution of e∗ad l ∼

= “ e ad−1
ad l
l ” with the
n
adjoint square:
$
gK ∼
= eV = Vk
k! ,
k≥0
V = T1K (G/K) V A −→ V j (gK) = (gK∗ )jA V A ∈ TgK (G/K) ∼ = V,
$ (( ad V)2 | )k
gK∗ ∼
= e∗ad V = (1+2k)!
V ∼
= “ sinh ad V
ad V |V ” (symbolic notation)
k≥0
= 1n−s + 16 ( ad V)2 |V + · · · ∈ GL(n − s, R).
The Ad G-invariant measure on the coset space Ad G/K can be written in

orthogonal coordinates:
for Ad G/K : dgK = | det e∗ad V |dn−s ρ.
Since, in general, [V, V ] ⊂ V , the Killing connection cannot be taken

over for G/K. Via the triple Lie property, the V -restricted adjoint square
defines the Killing curvature at the neutral class 1K ∈ G/K of the globally
symmetric Riemannian manifold:

R(V ∧ V ) = − ad |V [V, V ],
R : V ∧ V −→ V ⊗V , T
R(V A ∧ V B ) = −AB
α ad |V l = −α E V ⊗ V̌D ,
α AB αD E
⎧
R(V ∧ V )(V ) = −[[V, V ], V ],
⎨

R(V ∧ V ) : V −→ V, R(V A ∧ V B ).V D = −ABα E V ,
αD E
⎩
REDAB
= −α E .
AB αD
Killing Ricci tensor: RDA

• = R DAB
B = AB Dα
α B .
For V as a representation space of the compact Lie algebra C, there exist

strictly positive definite symmetric C-invariant bilinear forms:

γ(V, V) > 0, γ(V A , V D ) = γ AD ,
γ : V ∨ V −→ R,
γ([C, V], V ) + γ(V, [C, V ]) = 0 for C ∈ C.
Any such form can be used for a G-invariant Riemannian metric
G/K gK ∼
= eV −→ g(eV )jk = (e∗ad V )A γ AD (e∗ad V )kD ,
j
and all G-invariant metrics on G/K arise from a C-invariant positive definite
bilinear form on V . All G-invariant metrics g on G/K lead to the same
Riemannian connection and to the Killing curvature tensor on the associated
symmetric space G/K.
For a semisimple L = C⊥V the L-invariant symmetric Killing form, re-
stricted to V , adds up two C-invariant contributions, related to the two
subspaces and reflected to each other. They are not necessarily symmet-
ric. The Killing form of L, restricted to V , is the symmetrized Killing Ricci
tensor
⎧
⎪ V ∨ V −→ R, κ(V A , V D ),
⎪
⎪
⎪
⎨ κAD = Aa Db AB Dα Aα DB
b a = α B + B α
semisimple G/K, = R• + R• ,
DA AD
L = C⊥V, ⎪
⎪ R •
• = κAD R• = n−s
AD
⎪
⎪ 2
⎩ RDA +R•AD
1 AD •
•
2 − 2 κ R • = − n−s−1 AD
2 κ .
With the appropriate renormalization γ = ∓κ for compact and noncom-

pact orthogonal Lie algebras, one obtains a positive definite Riemannian
structure.
2.10.5 Examples
The simplest type Ic example is the 2-sphere Ω2 ∼
= SO(3)/SO(2):
⎧
⎪
⎪
⎪ bases: (Oa ∼ = −abc )3a=1 , O3 ,
⎨
(log SO(3), log SO(2)) : (T A = OA )A=1,2 ,
⎪
⎪ [O , O ] = 0, [O3 , T 1,2 ] = ∓T 2,1 ,
3 3
⎪
⎩
[T 1 , T 2 ] = O3 ,
0 0 −θ2
T = ad T = 0 0 θ1 , R2 ∼ = log Ω2 θ22 ,
−θ
θ2 −θ1 0
sin θ −θ2
eT = 13 + sinθ θ T + 1−cos
2 T
θ 2
, θ 2
= θ 2
1 + θ 2
2 , Ω 2
θ2 = sin θ cos ϕ
,
−θ2 θ θ θ
0

θ
sin ϕ
1 2 1 0 0
2 T | R2
2
0
T2 = θ1 θ2 −θ12 0 = 0 −θ 2
= −θ2 R ◦ 0 0 0 ◦ RT .
0 0 −θ 2 0 0 1
The 2-bein for Ω2 is orthogonally diagonalizable with its determinant for the
dilation:
sin θ
eT∗ = 12 − sinθθ−θ 3 T 2 |R2 = R ◦ 0θ 1 ◦R ,
0 T
T
e∗ ∈ GL(2, R), R ∈ SO(2),
det eT∗ = sinθ θ , d2 ω = sinθ θ dθ1 dθ2 = d cos θ dϕ.
The 2-sphere has the Ricci tensor
AB3 D3B +A3B DB3

Ω2 : RAD
• = 12 κ(T A , T D ) = 2 = −δ AD .
As example for types Inc and IInc , the 3-hyperboloid Y 3 ∼

= SO0 (1, 3)/SO
(3) ∼
= SL(2, C)/SU(2) with the action of the orthochronous Lorentz group
and the classes of the rotation groups is associated with the Lie algebra in a
Cartesian parametrization. The rotations are unitarily diagonalizable in the
complex displaying the spin eigenvalues, whereas the boosts are orthogonally
diagonalizable in the real and display the boost eigenvalues:
[Oa , Ob ] = −abc Oc

, [Oa , B b ] =
− B , [B , B ]= O ,
abc c a b abc c
−
◦ u∗ ,
α 0 Δ 0
O = ad O = − = iα u ◦
0 α
0
Δ
1 0 0
α
∼
= αa abc , α2 = α2a , Δ = 0 0 0 , u ∈ SU(6),
0 0 −1

◦ r∗ ,
0 β Δ 0
B = ad B = β T 0 = β r ◦ 0 −Δ

β ∼
= βa abc , β 2 = βa2 , r ∈ SO(6),
Killing form: κ(B a , B b ) = 4δ ab = −κ(Oa , Ob ).
The boost exponents give a Y 3 -parametrization:

0 β3 −β2
B = β T 0 , R3 ∼
0
β
= log Y 3 β = −β3 0 ,β1
β2 −β1 0
cos ϕ sin θ
eB = 16 + sinh β
B + 2 B
1−cosh β 2
, Y 3

β
sinh β ∼
= sinh β sin ϕ sin θ .
β β β
cos θ
The Lie–Jacobi isomorphism leads to the 3-bein eB ∗ for Y :

3
sinh β
β 0 0
eB
∗ = 13 + β3 B |R3
sinh β−β 2
=R◦ 0 1 0 ◦ RT ,
sinh β
0 0 β
B 2 |R3 = β 2 R ◦ Δ2 ◦ RT , eB∗ ∈ GL(2, R), R ∈ SO(3),

B sinh2 β sinh2 β
det e∗ = β 2 , d y = β 2 dβ1 dβ2 dβ3 = sinh2 βdβd2 ω.
3
The structures for the 3-sphere Ω3 ∼

= SO(4)/SO(3) = SU(2) (type IIc
example) are obtained with the noncompact–compact duality:
B ↔ i
↔ iT , β α, sinh β ↔ i sin α.
The starting point is a direct product Lie algebra log SO(3) ⊕ log SO(3)
with the diagonal Lie algebra O ∈ log SO(3) and the isomorphic vector sub-
space T ∈ V ∼= R3 :
[Oa , Ob ] = −abc Oc , [Oa , T b ] = −abc T c , [T a , T b ] = −abc Oc ,

L1,2 = O±T
2 , [L 1,2 , L1,2 ] =
a b
− L1,2, [L1 , L2 ] = 0,
abc c
0 i
α γ1,2
±iγ1,2

T = ad T = iαT 0 , L1,2 = ±iγ1,2
T γ1,2

,
Killing form: κ(T a , T b ) = −4δ ab = κ(Oa , Ob ).
Chapter 3
Mass Points
A dynamics is a representation of spacetime operations. A nonrelativistic

mass point dynamics is a representation of the time translations. The
equations of motion, of second order in time for position as basic observable
and of first order for the position-momentum pair, express the Lie algebra
d
action dt of the time operations. The classical mass point orbits as solutions
are realizations of the (eigen)time translation group t ∈ R ∼ = D(1) et in
∼
position, faithful noncompact with image R = SO0 (1, 1) = Y 1 , e.g., for free
mass points and hyperbolic orbits of never-returning comets, and unfaithful
compact with image R/Z ∼ = SO(2) = Ω1 for periodic orbits, e.g., for elliptic
orbits of planets. Newton’s idealization of mechanics working with mass
points was successful even after the introduction of the electromagnetic fields
by Faraday and Maxwell, and of the metrical tensor field in Einstein’s gravity.
The time orbits of mass points in position are derivable by an extremaliza-
tion of an action, leading to the Euler–Lagrange equations of motion. For
a general relativistic mass point dynamics, this extremalization merges into
the property of geodesics to have an extremal length; the Lagrangian is
essentially the spacetime metric.
The gravitational interactions in the general relativistic formulation were
first tested by the geodesics of mass points and light rays in a Schwarzschild
geometry, in the classical tests of the perihelion shift of the planets and
the light ray bending at the sun’s boundary. The metrical tensor arises
as the GL(R4 )-orbit of the flat Minkowski metric for free mass points
(particles) by linear operations gli (x) = Vab li
(x)η ab with the tetrad prod-
uct Vab (x) = ea (x)eb (x), where e(x) ∈∈GL(4, R)/O(1, 3) contains compact
li l i
SO(4)-rotations with six parameters and the four noncompact dilations

D(1)4 (see Chapter 1). Therefore, the geodesics as free mass point eigentime
orbits in curved spacetime can be interpreted as orbits in flat position under
the influence of a tetrad implemented interaction V : Curvature becomes
the interaction for flat space; the metrical coefficients of a static spacetime

82 Chapter 3 Mass Points
describe interactions that can be formulated in terms of a nonrelativistic

potential. A nonrelativistic precursor for such a reinterpretation is the cen-
trifugal potential.
The motion group with the global invariances of spacetime is generated by
the conserved quantities of a dynamics, e.g., the motion group R×SO(3) with
the time translations generated by the Hamiltonian for the conserved energy
and the rotations in the position motion group generated by the conserved
angular momenta.
In contrast to quantum physics, measurements in classical physics are
complete and absolute. All observables, like positions, momenta, energy, an-
gular momenta, etc., have their values in the real or complex numbers K =
R, C, which are directly the possible results of their measurement. In a quan-
tum language, the classical observables are simultaneously diagonalizable;
they commute with each other. Vectorial observables, V ∼ = Kn , are measur-
able with all their coefficients; the values in different bases can be computed
from each other by basis transformations.
3.1 Nonrelativistic Classical Interactions

Newton space–time is operationally characterized by classes of the inhomoge-
neous Galilei group, isomorphic as flat manifold, not as homogeneous space,
to the time and position translations,
)
[SO(3) × R3 ] ×
[R ⊕ R3 ] SO(3) × R3 ∼
= R ⊕ R3 .
The (doubled) semidirect group multiplication law looks rather compli-

cated — more suggestive in the four- and five-dimensional faithful represen-
tations:

Galilei group: SO(3) × R3 (O, v ) −→ O0 v1 ∈ GL(4, R),
(O1 , v1 ) ◦ (O2 , v2 ) = (O1 ◦ O2 , v1 + O1 .v2 ),
inhomogeneous: [SO(3) × R3 ] ×
[R ⊕ R3 ] −→ GL(5, R),

O
v
x
(O, v ; t, x) −→ 0 1 t ,
0 0 1
(O1 , v1 ; t1 , x1 ) ◦ (O2 , v2 ; t2 , x2 )
= (O1 ◦ O2 , v1 + O1 .v2 ; t1 + t2 , x1 + v1 t2 + O1 .x2 ).
A time translation R and rotation SO(3)-invariant dynamics of a nonrel-

ativistic mass point in a potential V are characterized by an action W with
time derivatives ξ˙ = dξ
dt = dt ξ, etc.,
⎧ ˙
p

⎨ x = m ,
px˙ − H) ∼
˙2
W = dt ( = dt[m x2 − V (r)],
p˙ = −∂V
= − x ∂V ,
2
p
H = 2m + V, ⎩ ¨ r ∂r
x = − xr ∂mr
∂V
.
3.1 Nonrelativistic Classical Interactions 83
Parallel with the metrical tensor dx2 = dr2 + r2 (dθ2 + sin2 θdϕ2 ) of flat
x2
p 2 = d
position, the kinetic energy dt2 with the momenta p can be decomposed
with polar coordinates R3 = R+ × Ω2 into the contributions with radial
momentum pr and with angular momenta L:
−θ̇ sin ϕ − ϕ̇ sin θ cos θ cos ϕ
sin θ cos ϕ
x = r sin θ sin ϕ , L = x × p = mr2 θ̇ cos ϕ − ϕ̇ sin θ cos θ sin ϕ ,
cos θ ϕ̇ sin2 θ
+ Lr2 ,
2
2
p =
p2r p2r = m ṙ , L
2 2 2 = m2 r4 (θ̇2 + ϕ̇2 sin2 θ).
The equations of motion are derived by a stationary action, i.e., as Euler–
∂ξ − dt ∂ ξ̇ = 0:
Lagrange equations ∂L d ∂L
2 2
2 (θ̇ + ϕ̇ sin θ) − V ]
2
W = dt [ m2ṙ + mr 2 2
⎧ d 2 2
⎨ dt r ϕ̇ sin θ = 0,
⇒ d 2 2 2
dt r θ̇ = r ϕ̇ sin2θ cos θ,
⎩
2 (θ̇ + ϕ̇ sin θ) − V ].
∂ 2
r̈ = ∂mr [ mr 2 2
Rotation invariance is equivalent to angular momentum conservation. The

Poisson–Lie brackets, [F, G]P = ∂F x − ∂
∂G ∂G ∂F
∂
p ∂ x , normalized as [
p ∂ p, x]P = 13 ,
vanish for angular momenta and Hamiltonian:
∈ log SO(3) : dt L
L P = 0, [La , Lb ]P = −abc Lc .
= [H, L]
Therefore, in classical physics, all orbits with conserved angular momentum
dt L
= 0 are planar since L
x = 0, Lp = 0 (first Kepler law). For the mass
point motion, one can choose an “equatorial” plane:
dt L
= 0, L 2 = L2 = constant,
(θ, θ̇) = ( 2 , 0) ⇒ L = mr2 ϕ̇.
π
The time translations R, generated by the Hamiltonian H, have the energy

as their invariant:
2
L2
H = m2ṙ + 2mr 2 + V = E.
The remaining one-dimensional radial position translations are characterized

by a radial action with an additional repulsive centrifugal potential VL (r) =
L2
2mr 2 ,
L2 2
L2
mr̈ = − ∂r
∂
(V + 2mr 2 ), Wr = dt( m2ṙ − 2mr 2 − V ).
The centrifugal potential originates from the rotation degrees of freedom. It

arises also in a free theory with V = 0.
The planar orbits have the time parametrization t −→ r(t),
r(t)
− mL2 r2 ⇒ t − t0 = r(t0 ) √
2(E−V ) 2
( dr 2
dt ) = m
mrdr
,
2 2 2m(E−V )r −L
and the polar representation ϕ −→ r(ϕ),

r4 L2
ṙ = dr
dϕ ϕ̇ = L dr
mr 2 dϕ ⇒ ( dϕ
dr 2
) = L2 [2m(E − V ) − r 2 ],
r(ϕ)
ϕ − ϕ0 = √ Ldr
r(ϕ0 ) r 2m(E−V )r 2 −L2
.
3.2 The Symmetries of the Kepler Dynamics

Mass point mechanics, by itself, has no principle to distinguish “basic” poten-
tials. In hindsight, the basically most important nonrelativistic interactions
for mass points are described by the rotation-invariant Kepler potential. It is
2
used as a Newton potential with coupling constant G = κc 8π for a gravistatic
interaction, only attractive, and as a Coulomb potential for an electrostatic
interaction, attractive or repulsive,

−m1 m2 G with masses m1,2 ,
V (x1 − x2 ) = |x1 −x2 | , γ0 =
γ0
1
Q1 Q2 4π 0
with charges Q1,2 .
The dimensionless interaction constants γc0 , with as dimensional unit only,

allow, for the Coulomb potential, a factorization in the normalization of the
electromagnetic interaction with Sommerfeld’s fine structure constant αS and
integer charge numbers with respect to the electron charge e:

γ0
−m1 m2 m12 with Planck mass m2P = c G = cκ ∼ (6.1 × 10
8π −8
kg)2 ,
c = 2
P
2
g
z1 z2 4π e
with g 2 = c 0
= 4παS ∼ 10.9
1
, z= Q e ∈ Z.
In contrast to the integer charge numbers z ∈ Z, related to the compact-

ness of the electromagnetic operation group U(1) (see Chapter 5), there
arise continuous numbers mmP for the noncompact operations of gravity with
the masses m2 ∈ R+ as invariants of the spacetime translation group R4 .
The Planck mass with m2P 2 = 8π c2 defines the limit for the relative magni-
2
tude of Schwarzschild and Compton lengths:

m ≥ Lm ⇐⇒ m2 ≥ m2P ,
m = κm , L = , with
8π m cm m ≤ Lm ⇐⇒ m2 ≤ m2P .
The Kepler potential determines a dynamics of two classical mass points

with position-momentum pairs (xi , pi )i=1,2 in Euclidean position, compatible
with the inhomogeneous Galilei group [SO(3) × R3 ] ×
R4 for nonrelativistic
theories. A Lagrangian encodes the time development by the Hamiltonian H:
p21 p22
L(1, 2) = p1 dt x1 + p2 dt x2 − H(1, 2), H(1, 2) = 2m1 + 2m2 + V (x1 − x2 ).
It is the sum of the free center of mass motion (xc , pc ) for the translations
of flat position and the reduced dynamics (x, p) for the position representing
Kepler interaction:
⎧
pc 2 ⎨ L(x, p ) = pdt x − H(x, p ),
L(1, 2) = pc dt xc − 2M + L(x, p ), 2
p
H(x, p ) = 2m + V (x ),
M = m1 + m2 , ⎩
M m = m1 m2 .
3.2 The Symmetries of the Kepler Dynamics 85
Only for gravity is the center of mass transformation compatible with the
transition to mass sum and reduced mass:
2 2
V (r) = −m1 m2 8πr
c κ
= −mM 8πr
c κ
= −mc2
M
r .
There is no analogue factorization into “sum charge” and “reduced charge”

for the integer charge numbers in the product z1 z2 .
The Kepler potential is distinguished by being the inverse of the invariant
Laplace operator (Laplace kernel) for flat position R3 ∼ = SO(3) ×
R3 /SO(3)
with respect to the point-supported Dirac position distribution (see
Chapter 10):
−∂ 2 1 = 4πδ(x) = 4πδ(
r ω) 12 δ(r). r
A classical dynamics with the Kepler potential md2t x = γ0 rx3 has one
3
intrinsic unit γm0 = (−c2 M , z1mz2 αS c) with the dimension ms2 . For bound
orbits, it gives the relation of the orbit “radius” to the periodic orbit time as
illustrated by Kepler’s third law, where the reduced mass m drops out.
The equations of motion for the reduced system of the Kepler dynamics
give as action of the time translations

2
+ |γr0 |δ ,
p p

H = 2m dt x = [H, x]P = m ,
δ = ±1 (repulsion, attraction),
dt p = [H, p]P = −∂V = γ0 rx3 ,
with the first-order position derivatives of the potential (Kepler force γ0 rx3 ).
The Kepler Hamiltonian is distinguished by real six-dimensional global
position groups G ⊃ SO(3) with the rotations as subgroup. It has an invariant
perihelion vector P,
the Lenz–Runge vector in the orbit plane,
P
= 1
m|γ0 |
p ×L
+ δ x , L
r
P = 0, [P,
H]P = 0.
The energy E as time translation invariant for the Hamiltonian is a function of

the angular momentum and perihelion invariant, which both describe position
properties,
mγ02 P 2 −1
P
2 = 1 + 2
mγ02
HL
2 ⇒E= 2 L2 , with P
2 = P 2, L
2 = L2 .
The time orbits in position space are conic sections, given by polar equations
with one focus as origin (second Kepler law):
P
x = P r cos ϕ = 1
p × L)
m|γ0 | (
x+ δr = 1
m|γ0 | L
2
+ δr
2
⇒ r(ϕ) = 1 L
m|γ0 | P cos ϕ−δ .
x directs to the peri- and aphelion for ϕ = 0 and ϕ = π, respectively.

As will become more clear in quantum theory, the invariance group of the
Kepler dynamics is a consequence of the operational structure of position. Its
classical representation by the mass point orbits depends on the energy sign.
There arise the three maximal global invariance groups of three-dimensional

Riemannian manifolds, i.e., SO(4) (rotations), SO0 (1, 3) (Lorentz transfor-
mations), and SO(3) × R3 (Galilei group) as the motion groups of, respec-
tively, sphere Ω , hyperboloid Y 3 , and flat R3 , used as position manifolds in
3
Friedmann universes, all with SO(3) as the local invariance group. The Lie
algebra for a time-translation eigenvalue E ∈ spec H,
[La , Lb ]P = −abc Lc , [La , P b ]P = −abc P c ,
2H abc c ∼ 2E abc c
[P a , P b ]P = mγ 2 L = mγ 2 L ,
0 0
is semisimple for negative energy (E) = |E| E

= −1 and cyclic (elliptic)
“bound” orbits, characterized by the subgroup SO(2) ⊂ SO(4) and illustrated
by planets, simple for positive energy (E) = 1 and hyperbolic “scattering”
orbits, characterized by the subgroup SO0 (1, 1) ⊂ SO0 (1, 3) and illustrated
by comets, and semidirect for trivial energy E = 0 and parabolic “scattering”
orbits, characterized by the subgroup R ⊂ SO(3) × R3 :
2
E = 0 : B = mγ0 P, [La , B b ]P = −abc B c , [B a , B b ]P = (E)abc Lc ,
2|E|
E=0: [La , P b ]P = −abc P c , [P a , P b ]P = 0.
The contractions of the noncompact–compact pair for trivial energy lead to
the Galilei group,
E→0
R3 0←E
SO0 (1, 3) −→ SO(3) × ←− SO(4),
which is isomorphic to the Euclidean group with the contracted boosts P
as
flat position momenta.
3.3 Electrodynamics for Charged Mass Points

In special relativistic electrodynamics, compatible with the Poincaré group,

SO0 (1, 3) × R4 (Λ, x) −→ Λ0 x1 ∈ GL(5, R),
(Λ1 , x1 ) ◦ (Λ2 , x2 ) = (Λ1 ◦ Λ2 , x1 + Λ1 .x2 ),
and its Lorentz group classes SO0 (1, 3) × R4 /SO0 (1, 3) ∼ = R4 as flat
Minkowski spacetime, the time-independent Coulomb potential is embedded
into the spacetime-dependent vector potential A. The orbit of a charged mass
point τ −→ Xa (τ ) = (ct, X(t))
is parametrized by the eigentime [τ ] √ = s;
i.e., the orbit X as a spacetime “field” depends only on τ = (x0 )ϑ(x2 ) x2 .
Electrodynamics is characterizable by an action as the sum of the free actions
and the interaction:
elmag
Wmatter = Welmag + Wmatter + Wint ,
ab
Welmag = d4 x g12 (Fab ∂ A −∂
a b b a
2
A
+ Fab4F ),

]∼
a
Wmatter = dτ [P a dτ Xa − P2m
aP
= dτ m 2 dτ Xa dτ X
a

Wint = −z dτ Aa (X)dτ Xa .
3.4 Einstein Gravity for Mass Points 87
The vector potential and field strengths, acted on by a four-dimensional

Minkowski and a six-dimensional adjoint representation of the Lorentz group,
1
have the dimensions of a length and an area density, [A] = m , [F] = m12 . The
Lorentz invariant interaction is the line integral along the spacetime coordi-
nates of the mass point:

z dτ (dτ Xa )Aa (X) = z dXa Aa (X) = d4 x Ja (x)Aa (x).
1
The current for a mass point with the dimension of a volume density [J] = m3 ,

Ja (x) = z dτ dτ Xa δ(x − X) = z dXa δ(x − X) = z PPa0 δ(x − X(t)),

is proportional to its energy-momentum Pa = mdτ Xa = * mc (1, Vc ) and
2
1− Vc2
the Lorentz invariant inverse energy-multiplied Dirac position distribution
1
x − X).
P0 δ(

The equations of motion display the action of spacetime translations with
×B(X)

the Lorentz force m F (X)Pb = E(X)+
1 ba *
V

effecting the mass point orbit,
2
1− Vc2
∂ a Ab − ∂ b Aa = Fba , − g12 ∂ b Fba = Ja ,

Pa
dτ Xa = m , dτ P a = mz
Fba (X)Pb ,
⇒ [δa dτ − m F (X)]dτ Xb = 0.
b z ba
The system involves the mass point equations with the Coulomb force
as an electrostatic approximation with Aa (x) = (A0 (x), 0) for a current
J0 (x) = z1 δ(x):
g2
− g12 ∂2 A0 (x) = z1 δ(x), A0 (x) = z1 4πr ,
cg2 X
dt P = −z2 c∂A (X) = z1 z2 4π R3 .
0
3.4 Einstein Gravity for Mass Points

In general relativity with a Riemannian manifold (M(1,3) , g), gravity for a
mass point on the orbit τ −→ Xk (τ ) is characterizable by the action

W = |g|d4 x 2κ
c
R•• + Wmatter
grav
,

W grav m li
= dτ g (X)dτ Xl dτ Xi = ∼ dτ (P l dτ Xl − 1 gli P l P i ).
matter 2 2m
Its variation leads to the geodesic orbits (shortest paths):

1 ij
m gki P , dτ gki P = − m Γk (X)gir gjl P P ,
1 i i r l
dτ Xk =
⇒ [δki dτ + Γij
k (X)dτ Xj ]dτ Xi = 0.
The energy-momentum tensor of the mass point is proportional to the energy-

momentum square,
P lP i P lP i
|g|Tli (x) = m cdτ m 2 c2 δ(x − X) = cP 0 δ(x − X),

1 li •
−κTli = Rli• − 2 g R• .
In contrast to the charge dependence of the electromagnetic interaction via

z
minert , the gravitative interaction is independent of the mass of the mass point
mgrav
via minert = 1. However, there remains a difference between mass points with
2
m > 0 and light rays with m = 0.
Flat spacetime gives the free Newtonian mass point in a special relativistic
framework:
1
g = η : dτ Xk = m Pk , dτ Pk = 0.
For the nonrelativistic Newton interaction
00of mass points,
there is the
gravistatic approximation with gki (x) ∼=
g (
0
x) 0
−13
for an energy-
momentum tensor T00 (x) = m1 δ(x):
− κ2 ∂ a Γ00 x) = κ1 ∂2 g00 (x) = m1 δ(x), g00 (x) = 1 − m1 4πr

a (
κ
=1− 2
1
r ,
2
c 00 2
c κ X
dt P = m2 2 ∂g (X) = m1 m2 8π R3 .
3.5 Geodesics of Static Spacetimes

Spacetime M(1,3) with the time translations and the rotations in the motion
group R × SO(3) ⊆ Gg , e.g., Schwarzschild or Reissner spacetime outside the
Schwarzschild radius, has the metric
g = c2 dτ 2 = e2λ3 (ρ) c2 dt2 − e2λ(ρ) dρ2 − ρ2 (dθ2 + sin2 θ dϕ2 ).
The geodesics are parametrizable by eigentime τ for ξ˙ = dξ

cdτ = 1c dτ ξ etc. in
the action

grav
Wmatter = dτ 2
g (X) dX
m li l dXi
dτ dτ
= c2 dτ m 2λ3 2
2
[e ṫ − e2λ ρ̇2 − ρ2 (θ̇2 + sin2 θ ϕ̇2 )].
The metrical coefficients are radial-dependent normalizations, i.e., dilations

ρ −→ (eλ3 (ρ) , eλ(ρ) , ρ, ρ sin θ) ∈ D(1)4 of the “energy-momenta” (ṫ, ρ̇, θ̇, ϕ̇).
The geodesics as shortest paths arise from a stationary action, i.e., from
the Euler–Lagrange equations for the time coordinate t and for the position
coordinates (ρ, θ, ϕ) with the metrical tensor as Lagrangian L(τ ) = cg(τ )
2 dτ 2 :
⎧
⎨
d
cdτ ρ2 ϕ̇ sin2 θ = 0,
d 2λ3 d 2 2 2
cdτ e ṫ = 0 and cdτ ρ θ̇ = ρ ϕ̇ sin θ cos θ,
⎩ d 2λ 2λ3 2
∂
cdτ e ρ̇ = ∂ρ [−e ṫ + e2λ ρ̇2 + ρ2 (θ̇2 + sin2 θ ϕ̇2 )].
3.5 Geodesics of Static Spacetimes 89
The Killing fields for the global invariance group Gg of spacetime give rise
to a planar motion with invariant eigenvalues E for time translations R and
L for rotations SO(3):
E = e2λ3 ṫ, (θ, θ̇) = ( π2 , 0), L = ρ2 ϕ̇.
The eigentime parametrization of the metrical coefficients gives the
R × SO(3)-invariant normalization of the Lagrangian (GR for “general
relativistic”):
GR: g
c2 dτ 2 = e2λ3 ṫ2 − e2λ ρ̇2 − ρ2 (θ̇2 + sin2 θ ϕ̇2 )
1 for mass points m2 > 0,
= E 2 e−2λ3 − e2λ ρ̇2 − Lρ2 = ϑ(m2 ) =
2
0 for light m2 = 0.
It is the radially renormalized extension of the mass as the translation invari-
ant of Minkowski spacetime translations and generalizes the nonrelativistic
expression for the time translations and the position translations with the
Euclidean group (NR for “nonrelativistic,” SR for “special relativistic”)
R4 : p20 − p2 = E2 − p2r − L2
2 2
SR: SO0 (1, 3) × c r = m2 ,
2 2
p L 2
NR: R × [SO(3) × R3 ] : E −
2m = E − 2m − 2mr 2 = V (r).
p
r
The general relativistic invariant for the eigentime translations can be com-
pared with the corresponding relation in nonrelativistic mass point mechanics
with time translations:
L2
GR with dτ : E 2 e−2λ3 − e2λ ρ̇2 − ρ2 = ϑ(m2 ),
2
L2
NR with dt: E − m ṙ2 − 2mr 2 = V.
The second-order radial equations of motions arise by time and eigentime

derivation, respectively,
2
NR: r̈ = − 21 ∂r
∂
( mL2 r2 + 2V
m ),
∂ L2
GR: ρ̈ = − 21 ∂ρ [ ρ2 + ϑ(m ) − E 2 e−2λ3 ]e−2λ .
2
They can be derived as Euler–Lagrange equations of a radial action with an

effective potential:
2
L2
NR: Wt = dt m [ ṙ2 − Veffm(r) ], 2Veff = mr 2 + 2(V − E),
ρ̇2 L2
GR: Wτ
c2 = dτ m [ 2 − c2 ], 2 c2 = [ ρ2 + ϑ(m2 ) − E 2 e−2λ3 ]e−2λ .
veff (ρ) veff
The effective potential involves the additional centrifugal potential for a

rotation-invariant nonrelativistic dynamics, connected with SO(3) and the
conserved angular momentum, and, in the general relativistic case, a “gen-
uine” potential for static spacetimes, connected with the invariance under
time translations R:
∂ Veff (r) ṙ 2
NR: r̈ = − ∂r m , with 2 + Veffm(r) = 0,
∂ veff (ρ) ρ̇2
GR: ρ̈ = − ∂ρ c2 , with 2 + veffc2(ρ) = 0.
The effective potential is given by tetrad action on the flat Minkowski

spacetime tensor (free theory), here by dilations eλ(ρ) , eλ3 (ρ) :
L2
NR: V = 0 ⇒ 2Veff = mr 2 − 2E,
L2
GR: g = η ⇒ 2 vceff
2 = ρ2 − ϑ(m )
2
− E 2.
The remaining integration gives the planar orbits in position t, ϕ −→
r(t), r(ϕ) and τ, ϕ −→ ρ(τ ), ρ(ϕ):
NR: dt = √ mdr
, dϕ = √
Ldr
,
−2mVeff (r) r2 −2mVeff (r)
GR: dτ = √ dρ
, dϕ = √cLdρ
.
−2veff (ρ) ρ2 −2veff (ρ)
In the opposite direction, a given potential can be translated into a met-

rical spacetime tensor via Veff ∼ veff .
Thus, a static spacetime metric describes a gravity interaction whose
geodesics can be expressed as caused by a potential of a nonrelativistic dy-
namics, e.g., for reciprocal metrical coefficients of time and position, i.e., with
a self-dual dilation ρ −→ eσ3 λ3 (ρ) ∈ SO0 (1, 1):
g = e2λ3 (ρ) c2 dt2 − e−2λ3 (ρ) dρ2 − ρ2 dω 2
= [ϑ(m2 ) + Lρ2 ]e2λ3 − E 2 .
2
⇒ 2 veffc2(ρ)
The classical example is Schwarzschild spacetime, i.e., outside the event

horizon ρ ≥ 2 , with the angular momentum-dependent corrections Lρ
3 of
2
L2
the Newton potential 1
mc2 [Veff (r) − 2mr 2 ] = −
r :
2
2 2 dρ2
g = (1 − ρ )c dt − 2 − ρ2 dω 2 ,
1− ρ
E 2 −ρ̇2 L2
ϑ(m2 ) = 2 − ρ2
1− ρ
L2 2

⇒ 2 veffc2(ρ) = [ϑ(m2 ) + ρ2 ](1 − ρ ) − E 2.
For Schwarzschild spacetime without the Lenz–Runge invariance, the elliptic
orbits of mass points for a Kepler dynamics with SO(4)-invariance are re-
placed by the geodesic rosette orbits with an only rotation SO(3)-invariant
dynamics and position as a rotation paraboloid:
L2 2

m2 > 0 : 2 veffc2(ρ) = (1 + ρ2 )(1 − ρ ) − E 2.

The perihelion shift Δϕ• ∼ R •
is of the order of magnitude of the Schwarz-
schild length of the central mass, e.g., the sun with ∼ 1.5 × 103 m (general
relativistic correction), divided by the radius of the planet’s orbit (Newton
gravity), e.g., Δϕ• ∼ 2.6 × 10−8 for mercury with R• ∼ 5.8 × 1010 m.
Also, light with trivial eigentime invariant ϑ(m2 ) = 0 has a nontrivial
effective potential for nontrivial angular momenta,
L2
m = 0 : 2 veffc2(ρ) = ρ2 (1 − 2

ρ ) − E 2.
3.6 Gravity for Charged Mass Points 91
There is no nonrelativistic contribution from Newton’s gravity. The bending

angle Δϕ ∼ r

is of the order of magnitude of the Schwarzschild length of
the central mass divided by the distance of the light ray to its center, e.g.,
Δϕ ∼ 2.1 × 10−6 for the Schwarzschild length of the sun divided by the sun
radius r ∼ 7 × 108 m in the case of a ray at the sun’s boundary.
Other examples are static spacetimes with spherical, flat, or hyperbolic
position:
dρ2
g = e2λ3 (ρ) c2 dt2 − 1−kρ2 − ρ dω ,
2 2
2
L2
ϑ(m2 ) = E 2 e−2λ3 − 1−kρ
ρ̇
2 − ρ2
L 2
⇒ 2 veffc2(ρ) = ρ2 + [ϑ(m ) − E e
2 2 2λ3
](1 − kρ2 ) − kL2 .
The static universe with flat position uses the inverse metrical coefficient
− g001(r) , i.e., the eigentime derivation of time as effective potential,
g = e2λ3 (ρ) c2 dt2 − dρ2 − ρ2 dω 2 ,

= E 2 e−2λ3 − ρ̇2 − Lρ2
2
2
ϑ(m )
= Lρ2 + ϑ(m2 ) − 2E 2 e−2λ3 , E 2 e−2λ3 = ṫ =
2
⇒ 2 veffc2(ρ) dt
cdτ ,
¨ 2
ξ = E ξ ∂ e−2λ3 for ξ2 = ρ2 .
2 ρ ∂ρ
The nonflat static Friedmann universes, especially for a spherical position

k = 1 (Einstein’s universe),
2
R × (Ω3 , Y 3 ) : g = c2 dt2 − 1−kρ
dρ
2 − ρ dω
2 2
⇒ 2 veffc2(ρ) = Lρ2 + [ϑ(m2 ) − E 2 ](1 − kρ2 ) − kL2 ,

2
lead to a harmonic oscillator potential for k[E 2 − ϑ(m2 )] > 0. In this case, the
geodesics are oscillator orbits with one condition for the integration constants:
k[E 2 − ϑ(m2 )] = ω 2 ⇒ 2 veffc2(ρ) = L
2
ρ2 + ω ρ − k(ω + L ),
2 2 2 2
¨ ˙
ξ 2 = ρ2 : ξ + ω 2 ξ = 0 with ξ 2 + ω 2 ξ 2 = k(ω 2 + L2 ),
˙
ξ = c+ eiωt + c− e−iωt ⇒ ξ 2 + ω 2 ξ 2 = 4ω 2 c+ c− , c+ c− = k ω 4ω
2
+L2
2 .
3.6 Gravity for Charged Mass Points

For the Reissner metric,
g = e2λ3 (ρ) c2 dt2 − e−2λ3 (ρ) dρ2 − ρ2 dω 2 ,
the Lie algebra of the dilations involves two contributions:

2z
e2λ3 (ρ) = 1 − 2
m
ρ + ρ2 .
The gravitative Newton potential − 2

ρm with the Schwarzschild length
κ
2
m = m 8π = m cG2 = 8πLm , with Lm = mc Compton length
is a solution of the homogeneous Einstein equations R• ba = 0, e.g., outside a

mass point, with the position derivation
−∂ 2
ρm = 4πm δ(x).

2
The electromagnetic contribution ρz2 arises from the Coulomb potential
of a charged mass point with a radial field strength:
A0 = z αρS , Fα0 ∼
= −∂ zαρS = zαS ρx3 ,
ab
− κ1 Rab ab
• = T (F) =
1
g2 (ηcd F F − η Fcd F )
ac bd
1 0 40 cd
∼
= αS
z 2 8πρ 4 0 −1 0 .
0 0 12
It is normalized by an area, where the Planck area is multiplied by the elec-

g2
tromagnetic normalization αS = 4π ∼ 137
1
and the squared charge number
Q
z = e of a mass point:
S κ
2z = z 2 α8πc = z 2 αScG
3 = z 2 α8πS 2 .
The position dependence ρ12 originates from the second-order integrated

traceless energy-momentum tensor of the electromagnetic field with
2

2z
∂ 2 2ρz2 = ρ4 = (∂
ρz )2 .
The validity of the parametrization is dependent on the ratio of electro-

magnetic and gravitative area involved:

2
2
0 = 1 − 2
ρm + ρz2 ⇒
ρm = 1 ± 1 −
2z ,
m
zLm √ 5.1×1033

z
with
m =
8παS ∼ zLm m .
For electromagnetism “stronger” than gravity as in
mz ≥ 1, e.g., for charged

elementary particles, there is a “naked” singularity, i.e., without event horizon,
and a static universe,

proton: Lp = mp c ∼ 6 × 10−16 m, zp2 = 1 ⇒ (
mz )p ∼ 3 × 1018 .
The effective potential for a charged mass point is a fourth-order polyno-

mial in ρ1 :

L2
2z
m2 > 0 : 2 veffc2(ρ) = 1 + ρ2 1− 2
m
ρ + ρ2 − E 2.
Chapter 4
Quantum Mechanics
It is not really a good strategy to follow the historical development of classical

theory to quantum theory by “quantizing” a classical theory. Quantum the-
ory is not only the change of a government; it is even the change of the
constitution.
For example, a classical Lagrangian with the separation in a free kinetic
and an interaction term is not a good starting point to solve a quantal bound-
state problem. The ontology and interpretation of a quantum description are
completely different from those of a classical one. A classical–quantum rela-
tionship and distinction may be found in the different representation struc-
tures of time and position operations. In classical mechanics, the time orbits
are valued in position, t −→ x(t), mass points have a position, and the con-
cept of a “point particle” makes sense. In quantum mechanics, the time orbits
are valued in a Hilbert space with probability amplitudes. Now, in quantum
mechanics, there are also orbits of position (“of,” not “in”): The Schrödinger
wave functions (“information catalogues”) for bound-state vectors are matrix
elements of infinite-dimensional Hilbert representations of noncompact posi-
tion operations x −→ ψ(x). The concept of a mass point is very restricted.
For example, it does not make sense to call an electron, e.g., “in” a hydrogen
atom, a point particle, as there are no orbits in position.
Nonrelativistic quantum mechanics is a theory of representations of time
and position operations acting on a Hilbert space. The position-momentum
operators (x, p) come with the duality-induced nontrivial commutator
[x, p] = i13 , which defines the Heisenberg Lie algebra. The solution of
a dynamics involves the determination of the energy eigenvalues of a Hamil-
tonian, generating time translations in irreducible Hilbert representations,
2

p
H = 2m + V (x),
E ∈ spec H ∩ R, R t −→ eiEt ∈ U(1).
In nonrelativistic mechanics, time translations are implemented as

d
dt = i ad H; i.e., by the adjoint action of the Hamiltonian, built by position
94 Chapter 4 Quantum Mechanics
operations, with the corresponding relations of eigenvalues and representation

invariants for time and position operations.
The vectors, acted on by time translations, especially the bound-state
eigenvectors |E, lie in a Hilbert space that is related to the quantum algebra
generated by position and momentum (x, p) with appropriate topological
features.
The familiar Hilbert space with the square-integrable function classes
L2 (Rs ) of position translations x ∈ Rs , s = 1, 2, . . . , as eigenvalues of x and
the momentum representation p −→ −i∂ is defined by the faithful Hilbert
representations of the noncompact real (1+2s)-dimensional Heisenberg group
H(s) ∼ = eR ×
s
R1+s , a semidirect product affine subgroup. Its nilpotent Lie al-
gebra log H(s) ∼ = R1+2s is characterized by s nontrivial brackets [x, p] = 1s I,
for illustration in a nonunitary faithful real (2+s)-dimensional representation:
H(s) −→ GL(R2+s ),
0 α γ
1 α γ+α
β
p + γI) −→ exp 0 0
αx + β
exp( β = 0 1
β ,
0 0 0 0 0 1
x+(β1 +β2 )
p+(γ1 +γ2 −β1 α2 )I
eα1 x+β1 p+γ1 I ◦ eα2 x+β2 p+γ2 I = e
(α1 +α2 )
,
H(s) ∼ = eR ×
s p+γI
R1+s eα x+β = eα x ◦ eβp+γI .
Here, the position-momentum
commutator is represented by a nilquadratic
0 0 1
matrix I −→ 0 0 0 . In the faithful complex infinite dimensional Hilbert
0 0 0
representations, the invariant basic element I −→ i1 takes one nontrivial
spectral value i ∈ spec I = iR as action unit (Planck’s constant). Then the
time translation eigenvectors are given by position-dependent Schrödinger
functions:
= ψE ∈ L2 (Rs ) : ψE (t) = eiEt ψE , H|E ∼

|E ∼ 2 2
= (− 2m ∂ + V )ψE = EψE .
The Schrödinger functions lead to position measures ds x |ψ(x)|2 for proba-

bilities.
A dynamics for nonrelativistic quantum mechanics comes with three in-
trinsic units — in addition to the universal action unit and the “individual”
mass m, there is a “normalization unit” of the specific potential. With those
units, only dimensionless variables will be used in the following.
4.1 Nonrelativistic Wave Mechanics

For a rotation-invariant dynamics with time translation generator (Hamilto-
2
= 0, the position translations are decomposed
nian) H = p2 + V (r), [H, L]
R = R+ × Ω into radial translations and rotations. In a derivative repre-
3 2
sentation the radial and angular momentum squares act on differentiable

functions as follows (see Chapter 10):
4.1 Nonrelativistic Wave Mechanics 95
⎧
⎨ ipr ∼
= 1r dr r = dr + 1r ,

pr ∼
= −d2r − 2r dr ,
2 2
= 0,
p2 = p2r + Lr2 , [p2 , L]
⎩ 2 ∼
L = − sin12 θ [(sin θ ∂θ
∂ 2 ∂ 2
) + ( ∂ϕ ) ].
The rotation representations for Ω2 are harmonically analyzable by the

spherical harmonics as Hilbert basis of the 2-sphere functions
∞
⎪ dense
L⎪⎪ ∼
{Ym ⎪L = 0, 1, . . . ; |m| ≤ L} basis of L 2
(Ω 2
) = C1+2L ,
⎧ 2 L (ϕ, θ)YL (ϕ, θ) = δ LL δ
L=0
⎪
⎪ d ω Y mm ,
⎪
⎪ ∞
m m
⎪
⎪ $ $L
⎨ L (ϕ, θ)YL (ϕ , θ ) = δ(
Ym ω − ω )
m
with
⎪
⎪
L=0 m=−L = δ(ϕ − ϕ ) sin1 θ δ(θ − θ ),
⎪
⎪ Y (ϕ, θ) = L(1 + L)YL (ϕ, θ),
⎪
⎪ L 2 L
⎩ m m
−i ∂ϕ
∂ L
Ym (ϕ, θ) = mYm L
(ϕ, θ).
In contrast to the spherical harmonics, appropriate for asymptotically

free scattering waves, with the r → 0 ambiguity in xr , the harmonic SO(3)-
polynomials as product with the corresponding radial power are defined also
for x → 0. They are eigenfunctions for a trivial translation invariant and,
therefore, appropriate for bound waves,
(
2 YL (ϕ, θ) = L(1 + L)YL (ϕ, θ),
L
L L L
(x)m = r Ym (ϕ, θ) : m m ⇒ ∂ 2 (x)L
m = 0.
p2r rL = −L(1 + L)rL−2
The x-homogeneous harmonic polynomials span the irreducible SO(3)-repre-

sentation spaces C1+2L . Position polynomials as a sum of homogeneous poly-
nomials can be decomposed into harmonic polynomials and r2 -parametrized
invariant coefficients (see Chapter 8).
Representations of the radial translations R+ in bound waves come after
the separation of the harmonic polynomials, which are acted on by the irre-
ducible SO(3)-representations. This leads to the Schrödinger equations for
the representation coefficients of radial translations R+ :
∞
$ $
L
2(1+L)
m dL (r) ⇒ [dr +
(x)L − 2V (r) + 2E]dL (r) = 0.
2
ψ(x) = r dr
L=0 m=−L
The representations can involve an R+ -reparametrization with a mono-

tonic function ξ:
R+ r −→ ξ(r) ∈ R+ .
The R+ -representation coefficient is a product with an exponential,1
−→ dL (r) = FL (ξ) e− 2 ,
ξ
R+ ξ
2
1 The ξ
normalization 2
is chosen with respect to the Laguerre polynomials below.
with the Schrödinger equation for the remaining function:

+
(dr ξ)2 d2ξ + d2r ξ − (dr ξ)2 + 2(1+L) dr ξ dξ
r ,
+ 14 (dr ξ)2 − 12 d2r ξ + 2(E − V ) − 1+L
r dr ξ FL (ξ) = 0.
Quantum numbers of bound states are determined with the condition of

square integrability L2 (R3 ) of the position wave functions.
For both the harmonic oscillator and the Kepler potential (below), there
arises a Laplace differential equation,
2
2 + (1 + λ − ξ) dξ + N ]LN (ξ) = 0,
d d λ
[ξ dξ N = 0, 1, . . . ,
which is solved by the Laguerre polynomials of degree N (radial quantum

number, knot number) and real order λ = −1, −2, . . . . The λ-dependence is
used for the angular momentum L-dependence. The Rodrigues formula for
the Laguerre polynomials contains derivatives of the function ξ −→ ξ λ e−ξ :
N N
LλN (ξ) = (ξ −λ eξ dξ
d λ −ξ N ξ
ξ e ) N ! = N1 ! ξ −λ eξ dξ
d
N ξ
λ+N −ξ
e
$N
Γ(1+λ+N ) (−ξ)k
= Γ(1+N −k)Γ(1+λ+k) k! .
k=0
The bound waves involve the representation coefficients (see Chapter 8)
−→ LλN (ξ) e− 2 ,
ξ
R+ ξ
2
which give a basis for a Hilbert space for each λ ∈ −N,

ξ⎪⎪
L2 (R+ , R) has basis {ξ −→ ξ 2 LλN (ξ)e− 2 ⎪
λ
⎪N = 0, 1, . . . }
⎧ ∞
⎪ λ −ξ λ λ Γ(1+λ+N )
⎨ 0 ξ e dξ LN (ξ) LN (ξ) = N! δN N ,
∞
$
with −λ ξ
⎪
⎩
N! λ λ
Γ(1+λ+N ) LN (ξ) LN (ξ ) = ξ e δ(ξ − ξ ).
N =0
4.2 Harmonic Oscillator

The three- dimensional isotropic harmonic oscillator Hamiltonian,
p2 +
x2
H= 2 ,
has only bound waves, no scattering solutions. The normalization of the po-
2 2
tential V (r) = k02r = r2 yields an intrinsic frequency unit [ km0 ] = s12 .
4.2 Harmonic Oscillator 97
4.2.1 Position Representation

An irreducible exponential with squared radial dependence as the radial rep-
resentation coefficient ξ(r) = r2 , dL (r) = FL (r2 )e−r determines the har-
2
monic oscillator potential, normalized with a momentum unit |Q|:

2
2(1+L)
[d2r + r dr − 2V (r) + 2E]FL (r2 ) = 0
2 2
− Q 2r r2 4
2
for F0 (r ) = e ⇒ V (r) = 2 Q , E = 32 Q2 .
The momentum unit can be chosen to be L-independent; e.g., Q = 1. The

general square-integrable solutions are products of a Laguerre polynomial of
degree N with a quantum number independent exponential:
⎧ 1+2L r2
⎪
⎪ FL (r2 ) = LN 2 (r2 )e− 2 ,
⎪
⎨
ELN = 32 + k = 32 + L + 2N,
[d2r + 2(1+L) dr − r2 + 2E]FL (r2 ) = 0 ⇒
r ⎪
⎪ L, N = 0, 1, . . . ,
⎪
⎩ k 1+2L r2
ψLm (x) ∼ ( x)m LN 2 (r2 ) e− 2 .
L
The harmonic oscillator solutions for each angular momentum L =

0, 1, 2, . . . constitute a Hilbert space basis:
1+2L ξ⎪
⎪
L2 (R+ , R)-basis : {ξ −→ ξ 4 LN 2 (ξ)e− 2 ⎪
1+2L
⎪N = 0, 1, . . . },
∞ 1+2L
−r 2+2L
1+2L
1 ∞ 1+2L
1+2L 1+2L
LN (r ) = 2 0 ξ 2 e−ξ dξ LN 2 (ξ) LN 2 (ξ)
2
2 2
0
dr LN (r ) e
2
r 2
Γ(N + 32 +L)
= 2N ! δN N .
4.2.2 Color SU(3) for 3-Position

The harmonic oscillator Hamiltonian with creation and annihilation opera-
tors (ua , ua ), a = 1, 2, 3,
p2 +
x2 {ua ,u
a} −ipa
xa√
H= 2 = 2 , ua = 2
, [ub , ua ] = δba ,
generates time orbits of the Hilbert vectors with k quanta. The creation
operator polynomials of degree k, acting on the Fock ground-state vector |0,
give the Schrödinger wave functions as position representation with degree-k
polynomials:
1 n1 2 n2 3 n3
|k; n1 , n2 , n3 = (u ) (u√)k! (u ) |0
∼ r2
= {x −→ Hk (x)e− 2 } ∈ L2 (R3 ), k = n1 + n2 + n3 ,
ua |0 = 0,
ua (t) = eit ua , |k; n1 , n2 , n3 (t) = eikt |k; n1 , n2 , n3 .
2 Usually, 3 2
the additive term V0 = 2
Q is introduced as ground-state energy.
The complex embedding of the 3 positions and momenta R3 ⊕ R3 −→ C3

leads to a color SU(3)-invariance with Gell–Mann matrices {λA }8A=1 :
χ8
χ3 + √3 χ1 − iχ2 χ4 − iχ5
, C = 2i ua λba ub , [C, H] = 0.
χ
χA λA = χ1 + iχ2 −χ3 + √83 χ6 − iχ7
χ
χ4 + iχ5 χ6 + iχ7 −2 √83
The irreducible representations are characterized by two integers as generat-

ing invariants for the SU(3)-operations with rank 2:
⎪
⎪
⎪ 1,2 ∈ N},
irrep SU(3) = {[2C1 , 2C2 ]⎪2C
dimC [2C1 , 2C2 ] = (1 + 2C1 )(1 + 2C2 )(1 + C1 + C2 ).
The harmonic oscillator representations [k, 0] (singlet, triplet, sextet, etc.)
with one trivial SU(3)-invariant C2 = 0 are the totally symmetric products
of SU(3)-triplets,
-k -k
[k, 0] = [1, 0] ∈ C3 ∼
2+k
= C( 2 ) ,
2+k
dimC [k, 0] = 2 = 1, 3, 6, . . . , k = 0, 1, 2, . . . .
The embedded rotation group, generated⎪ by the transposition antisym-
⎪
metric Lie subalgebra log SO(3) = {ϕa La ⎪
⎪ϕa ∈ R},
⎧
⎪
⎨ 0 χ2 χ5 0 ϕ3 −ϕ2
χA i λ −(λ
A A T
)
= −χ2 0 χ7 = −ϕ3 0 ϕ1 ,
SO(3) → SU(3) with 2
−χ5 −χ7 0 ϕ2 −ϕ1 0
⎪
⎩
La = abc ub u c , (L1 , L2 , L3 ) = (C7 , −C5 , C2 ),
comes with the real five-dimensional orientation manifold, given by the rota-
tion group

orbits A(3) = SU(3)/SO(3) in the color group, which describes
the 42 − 1 relative phases of the three angular momenta directions in com-
plex quantum structures. A(s) = SU(s)/SO(s) are the globally symmetric
compact Riemannian manifolds of subtype A I (see Chapter 2).
The energy (suitably normalized) as principal quantum number E = k =
L + 2N (polynomial degree) is the power of the product representations
-
k
[1, 0] of the time translations R. It is the sum of the angular momen-
tum quantum number L for SO(3) and the radial quantum number (knot
number) N for the rotation group classes in the real five-dimensional space
SU(3)/SO(3). One has with the angular momentum
degeneracy 1 + 2L the
energy degeneracy given by the dimension 2+k 2 of the SU(3)-representa-
tion [k, 0]:
⎧ ⎧ $
⎪ ⎪
⎪ (1 + 2L),
⎪ [L], k = 0, 2, . . . , ⎪
SO(3) ⎨
⎨ ···
[k, 0] ∼=
L=0,2,...,k
with 2+k = $
⎪
⎪ [L], k = 1, 3, . . . , 2 ⎪
⎪
⎩ ⎪
⎩ (1 + 2L),
L=1,3,...,k
···
(k, 0), (k − 2, 1), . . . , (0, k2 ),
ELN − 2 = k = L + 2N ⇒ (L, N ) =
3
(k, 0), (k − 2, 1), . . . , (1, k−12 ).
4.2 Harmonic Oscillator 99
4.2.3 Harmonic Fermi Oscillator

The free fields of Minkowski spacetime (see Chapter 5) are harmonically
analyzed with momentum-dependent creation and annihilation operators
q ), u (
(u( q )) as eigenvector distributions for the translations:
x ∈ R3 : (u( q )) −→ (eiqx u(q ), e−iqx u (q )).
q ), u (
The harmonic Bose oscillators of the former section are used for Bose
particle fields and, in addition, harmonic Fermi oscillators for Fermi particle
fields. They have the quantum mechanical Hamiltonian for the time transla-
tion representation:
[ua ,u
a]
H= 2 , {ub , ua } = δba , {ub , ua } = 0, a, b = 1, 2, 3.
There is no position representation for Fermi oscillators. In three dimensions,
they also have SU(3)-symmetry with the Lie algebra representation,
C = 2i ua λba ub , [C, H] = 0.
The eigenvectors are given by the totally antisymmetrized products,
where, because of the anticommutators in the quantization, there arise only
singlets, a triplet, and an antitriplet (Pauli exclusion principle for fermions):
[1, 0] ∧ [1, 0] = [0, 1], [1, 0] ∧ [1, 0] ∧ [1, 0] = [0, 0],
dimC [0, 0] = 1, dimC [1, 0] = dimC [0, 1] = 3.
For the time translation eigenvectors in the Hilbert space, the creation oper-
ators are applied to the Fock ground state with ua |0 = 0:
⎫
singlet: |0, ⎪⎪
triplet: ua |0 = |1; a, ⎬ with |k; . −R→ eikt |k; . ,
b c
antitriplet: abc2u u |0 = |2; a, ⎪⎪ k = 0, 1, 2, 3.
⎭
singlet: u1 u2 u3 |0 = |3,
The singlet and triplet representation dimensions are also valid for the rota-
tion subgroup SO(3) ⊂ SU(3) with L = 0, 1.
4.2.4 Bose and Fermi Oscillators

In general for isotropic harmonic Bose and Fermi oscillators: The rotation
group embedding SO(s) → SU(s) in s = 2, 3, . . . positions, Rs ⊂ Cs ∼= V 1,
leads to an energy degeneracy, i.e., the s-position comes with a quantum-
implemented SU(s)-symmetry. There arise the totally (anti)symmetric prod-
ucts of the defining representation, for Fermi the two trivial (k = 0, s) and
the (s − 1) fundamental SU(s)-representations,
-
k -
k

Bose : [1, 0, . . . , 0] with dimC [1, 0, . . . , 0] = s+k−1

k , k = 0, 1, . . . ,
"
k "
k
s
Fermi : [1, 0, . . . , 0] with dimC [1, 0, . . . , 0] = k , k = 0, 1, . . . , s.

The invariant of the time translation representation in U(1s ) ⊆ U(s) is

related to a representation invariant of SU(s) since both groups have the
center of SU(s) as nontrivial intersection, i.e., the cyclotomic group I(s),
⎪
⎪ s
U(1s ) ∩ SU(s) ∼= I(s) = {z ∈ C⎪ ⎪z = 1},

in the SU(3)-Bose example: E = k, dimC [k, 0] = 2+E 2 . A similar central

correlation shows up in the hypercharge-color relation of the quarks in the
standard model of elementary particles (see Chapter 6).
The creation operators, acting on the Fock state vector, give orthogonal
bases for the (anti)symmetric powers
(u1 )n1 · · · (us )ns |0 = |k; n1 , . . . , ns ∈ V k ,

$ s
ni = k, for Fermi ni ∈ {0, 1}.
i=1
The direct sum of those finite-dimensional Hilbert spaces is the finite dimen-
sional Hilbert space for the Fermi oscillator,
"
k
s "
V1 ∼ V1 ∼
s s
Fermi: V k = = C( k ) , Vk = = C2 .
k=0
The direct sum for the Bose oscillator gives the complex polynomials in the
position coordinates. Its completion with the Fock ground state |0 ∼
r2
= e− 2
in the scalar product is isomorphic to L2 (Rs ):
-
k ∞
-
V1 ∼ V1 ∼
= C[x1 , . . . , xs ] ∼
s+k−1
Bose: V k = = C( k ) , Vk = = Cℵ0 ,
k=0
C[x1 , . . . , xs ] = L2 (Rs ).
4.3 Kepler Dynamics

The Kepler dynamics with time translation generator,
p2
H=
2 + δr , δ = ±1 repulsive, attractive,
has both bound waves and scattering solutions. The normalization of the po-
tential V (r) = |γr0 |δ = δr gives an intrinsic velocity unit [ γ0 ] = ms with γ0 =
cαS z1 z2 , αS ∼ 137
1
, for a Coulomb potential and charge numbers z1,2 ∈ Z.
The binding energies contain a factor α2S ∼ 6 × 10−5 that reduces the elec-
tron mass-energy me c2 ∼ 0.5 MeV to the Rydberg energy 12 MR c2 ∼ 14 eV,
MR = me α2S . The maximally symmetric noncompact position as represented
by the wave functions of the nonrelativistic hydrogen atom is not the flat
4.3 Kepler Dynamics 101
Euclidean position R3 , but the hyperbolic position Y 3 ∼ = SO0 (1, 3)/SO(3),

i.e., the rotation group classes in the Lorentz group. They are isomorphic as
manifolds, not as homogeneous spaces,

cosh ψ
sinh ψ ω2 ∈ Y ⊂ R , x = rω2 ∈ R3 , r = | sinh ψ|.
3 4
This will be discussed in more detail in Chapters 8 and 10.
4.3.1 Position Representation

In the position representation, the Kepler Hamiltonian is the sum of the
Laplacian and the Laplacian kernel:
d3 q 1 iqx
∼ 1 2 p2 ∼
= −∂ 2 , 1r = 2π q2 e ,
H = −2∂ + r, δ 2

with ∂ r = −4πδ(x).
2 1
For an irreducible noncompact radial representation r −→ e−r as bound

solution of 3-position, an attractive Kepler potential is necessary to compen-
sate the contribution ( 1r dr r)2 − d2r = 2r dr , related to the SO(3)-rotations,
[d2r + 2(1+L)
r dr − 2V (r) + 2E]dL (r) = 0,
for d0 (r) = e−|Q|r ⇒ V (r) = − |Q|
r , 2E = −Q .
2
After separating, in general, for a rotation-invariant interaction V (r),

an exponential with a complex “momentum” iq, there remain the radial
equations:
dL (r) = FL (ξ)e− 2 , ξ2 = iqr, q ∈ C,

ξ
ξV (r)
⇒ [ξd2ξ + (2 + 2L − ξ)dξ − (1 + L − 2q2 ) + 4ξ (1 − 2E
q2 )]FL (ξ) = 0.
δ
For the Kepler potential V (r) = r and a purely “kinetic” energy, given by
half of the squared “momentum,”
q2
E= 2 ⇒ [ξd2ξ + (2 + 2L − ξ)dξ − (1 + L − i qδ )]FL (ξ) = 0,
there arises a Laplacian differential equation, solved, for r = 0 regularity,

by confluent hypergeometric functions (see Chapter 8), which, for negative
integer α, are Laguerre polynomials:
[ξd2ξ + (γ − ξ)dξ − α] f (ξ) = 0; ξ, α, γ ∈ C,
∞
$ Γ(α+k) Γ(γ) ξ k
f (ξ) ∼ 1 F1 (α; γ; ξ) = Γ(α) Γ(γ+k) k! = 1 F1 (γ − α; γ, −ξ)e ,
ξ
k=0
Γ(1+λ)
1 F1 (−N ; 1 + λ; ξ) = N ! Γ(1+λ+N ) LN (ξ) for N ∈ N.
λ
One obtains for the Kepler potential

V (r) = δ
r ⇒ dL (r) = 1 F1 (1 + L − i δq ; 2 + 2L; 2iqr)e−iqr .
For the attractive interaction V (r) = − 1r and states with negative energy
(binding energy) and square-integrable wave functions, the radial imaginary
“momentum” and, therefore, the energy is “quantized” (quantum numbers in
the original sense) as seen in the Laguerre polynomials,
⎧
⎪
⎪ dL (r) = L1+2L (2|Q|r)e−|Q|r ,
δ = −1, q = −i|Q|, ⎨ 1
N
Qn = n = 1 + 2J = 1 + L + N,
2
E = − Q2 < 0, ⇒
⎪
⎪ L, N = 0, 1, . . . ,
2 = |Q|r,
ξ ⎩ 2J
( n ) e− n .
r
ψLm (x) ∼ ( nx )L
m LN
1+2L 2r
In contrast to the bound-state exponential e−|Q|r , the scattering solutions

in the Kepler potential V (r) = δr , with positive energy and real momentum
from a continuous spectrum, are obtained after the separation of a unitary
exponential eiP r . The radial equations have the real (r = 0)-regular solutions:
⎧
⎪
⎪ dL (r) = 1 F1 (1 + L − i Pδ ; 2 + 2L; +2iP r)e−iP r
q = P, ⎨
2 = 1 F1 (1 + L + i Pδ ; 2 + 2L; −2iP r)e+iP r ,
E = P2 > 0, ⇒
⎪
⎪ P ∈ R, L = 0, 1, . . . ,
ξ
2 = iP r
⎩ P
ψL,m (x) ∼ (P x)L
m 1 F1 (1 + L + i P ; 2 + 2L; −2iP r)e
δ iP r
.
The large distance behavior [44] does not lead to spherical Bessel functions
for free particles, representing Euclidean position R3 (see Chapter 8). For the
hyperbolic position Y 3 -representations, the asymptotic Bessel functions are
modified by a radial-dependent logarithm and a constant phase αL :
r → ∞ : (P x)L
m 1 F1 (1 + L + i P ; 2 + 2L; −2iP r)e
δ iP r
2 − P log 2P r+αL )
sin(P r− Lπ δ
∼ Ym
L
(ϕ, θ) Pr .
4.3.2 Orthogonal Lenz–Runge Symmetry

In an algebraic quantum treatment,3 the Kepler Hamiltonian with angular
momentum and Lenz–Runge vector,
p2
= x × = L×
p×L− p
H= 2 + δr , L p, P 2 + δ xr ,
build the same three Lie algebra structures as in the classical case (see
Chapter 3):
⎧ a
⎨ [iL , iL ]
b
= −abc iLc ,
= 0, [H, P]
[H, L] = 0, [iL , iPb ]
a
= −abc iPc ,
⎩
[iPa , iPb ] = 2Habc iLc ,
3 There is another algebraic treatment of the quantum hydrogen atom by Pauli —
working, in some analogy to the harmonic oscillator, with creation and annihilation
operators.
4.3 Kepler Dynamics 103
with the Lorentz group for scattering, the orthogonal group for bound states,
and the Galilei or Euclidean group as contraction:
SO0 (1, 3) → SO(3) ×

R3 ← SO(4),
E ∈ spec H : E>0 E=0 E < 0.
The additional i-factors in the Lie brackets are related to the different dual
normalization in the Poisson bracket [p, x]P = 1 and the quantum commuta-
tor i[p, x] = 1.
Again, the squares of angular momentum and Lenz–Runge vector as in-
variants for position operations determine the invariant Hamiltonian for the
time translations,
2 + 1) ⇒ − 1 = 1 + L
2 = 1 + 2H(L 2 − 2
P
P 2H 2H ,
with an additional constant 1 compared to the classical case.

For attraction, δ = −1, and negative energies, one has representations of
the doubled compact Lie algebra log SO(3) ∼= Ac1 ∼
= R3 :
⎧
⎪ = √P
⎨ B

−2H

, J± = L±2 B ,
spec H E < 0 : [iJa± , iJb± ] = − abc iJc± , [Ja+ , Jb− ] = 0,
⎪
⎩
invariants: L 2 +B 2 = 2(J+ 2 + J− 2 ), L = J+ 2 − J− 2 ,
B
∼
log[SO(3) × SO(3)] = A1 ⊕ A1 .
c c
The quantum algebra for the space with the defining four-dimensional re-
presentation of Ac1 ⊕ Ac1 is generated by two pairs of Pauli spinors (creation
and annihilation operators) with Bose statistics,
nontrivial: [uA , uB ] = δA
B
, [aA , aB ] = δA
B
, A, B = 1, 2.
The double-“spin” Lie algebra is implemented by the six basic vectors
log[SU(2) × SU(2)] : iJ+ = iu σ2 u , iJ− = ia σ2 a ,

log SO(4) : L = J+ + J− , = J+ − J− .
B
The angular momenta are a basis of the diagonal Lie algebra.

The Hilbert space uses four creation operators:
⎪
⎪
basis {(u1 )n1 (u2 )n2 (a1 )m1 (a2 )m2 |0⎪
⎪n1,2 , m1,2 = 0, 1, . . . },
with uA |0 = 0 = aA |0, 0|uA u |0 = δA
B B
= 0|aA aB |0.
The eigenvalue of a time-conserved operator Q, i.e., [H, Q] = 0, for a

simultaneous eigenvector of the Hamiltonian H and Q is denoted in the
following as Q = E|Q|E.
The weight diagrams of the irreducible SU(2) × SU(2)-representations,
irrep [SU(2) × SU(2)] = irrep SU(2)⎪ × irrep SU(2)

⎪
= {(J1 , J2 )⎪
⎪J1,2 = 0, 12 , 1, . . . },
occupy (1 + 2J1 )(1 + 2J2 ) points of a rectangular grid. The two invariants
determine the occurring representations. The triviality of the invariant
P
L = 0 (classical orthogonality of angular momentum and Lenz–Runge
perihelion vector) “synchronizes” the centers I(2) = {±1} of both SU(2)s
(central correlation) and leads to the relevant group SO(4) with the integer
spin sum in the irreducible representations:
SU(2)×SU(2) ∼
= SO(4), with I(2) = {(1, 1), (−1, −1)} ⊂ SU(2) × SU(2),
I(2) ⎪
⎪
⎪
irrep SO(4) = {(J1 , J2 )⎪J1,2 = 0, 12 , 1, . . . , with J1 + J2 = 0, 1, . . . }.
The orthogonality condition enforces even the equality of both SU(2)-
invariants J+ = J− = J:
B
0=L = J 2 − J 2 ⇒ J 2 = J 2 = J(1 + J), J = 0, 1 , 1, 3 , . . . .
+ − + − 2 2
Therefore, the energy-degenerated representations are of the type (J, J); the
multiplets of both Ac1 -representations have equal dimension 1 + 2J. The
SU(2)-multiplet dimension is the principal quantum number n = 1 + 2J.
The weight diagrams occupy (1 + 2J)2 points of a square grid:
⎪
⎪
irrep Kep SO(4) = {(J, J)⎪
⎪J = 0, 12 , 1, . . . }.
These Kepler or harmonic SO(4)-representations (see Chapter 8) are the
totally symmetrized products of the defining four-dimensional SO(4)-re-
presentation ( 12 , 12 ) with the decompositions into irreducible ones with the
dimensions dimC ( L2 , L2 ) = (1 + L)2 :
⎧ ⎧$
⎪ (L L ⎪
⎪ (1 + L)2 ,
- ⎪
⎨ 2 , 2 ), 2J = 0, 2, . . . , ⎪
⎨ ···
2J
3+2J
( 12 , 12 ) =
L=0,2,...,2J
with =
⎪
⎪ (L L 3 ⎪
⎪ $
⎩ 2 , 2 ), 2J = 1, 3, . . . , ⎪ (1 + L)2 .
L=1,3,...,J
⎩
···
Similar to the one-dimensional harmonic oscillator with the 1-quantum

state vector |1 = u|0 as the defining U(1)-orbit, there is the state vec-
tor with the defining fourdimensional SO(4)-representation ( 12 , 12 ) (s- and
p-shell) for the atomic bound-state vectors. The highest-weight vector in an
irreducible representation space comes with highest “spins” j± = J:
(u1 a1 )2J |0 = |J; J|J; J.
Its extremality involves the triviality for the action of the two raising
operators:
± = (u u2 , a a2 ), (L , P ) = u u2 ± a a2 ,
raising: J+ 1 1 + + 1 1
− − −
lowering: J± = (u u1 , a a1 ), (L , P ) = u u1 ± a2 a1 .
2 2 2
By the two lowering operators (J− ±)

J−j±
|J; J = |J; j± (in the weight dia-
gram: horizontal to the left and vertical downwards), one reaches all eigen-
vectors of a square grid:
⎪
⎪
basis of C(1+2J) : {|J; j+ |J; j− ⎪
2
⎪j± = −J, . . . , J}.
4.4 Particles and Ghosts 105
The energy eigenvalues are given with the inverse of the Casimir invariant
1 = −2H(1 + L2 + B 2 ), i.e., by the normalization 1 = −2En n2 :
− 2H
1
= 1 + 2J+2 + J−2 = 1 + 4J(1 + J), J = 0, 12 , 1, 32 , . . . ,
En = − 2n1 2 , multiplicity: n2 = (1 + 2J)2 = 1, 4, 9, 16, . . . .
As seen in experiments, there is an additional twofold degeneracy in the

atoms. It originates from an additional “internal” spin SU(2)-property of the
electron not contained in the nonrelativistic scheme above. It can be added
by an ad hoc doubling leading to doubled multiplicities 2n2 = 2, 8, 18, . . . .
The SO(4)-representations are decomposable with respect to the posi-
tion rotation SO(3)-properties into irreducible representations of dimension
= J+ + J− :
(1 + 2L) with integer L = 0, 1, . . . for angular momentum L
SO(3)
2J $
2J
(J, J) ∼
= [L], (1 + 2J)2 = (1 + 2L).
L=0 L=0
The Lenz–Runge invariance-related difference 2J −L = N , characterizing the

classes SO(4)/SO(3) ∼ = Ω3 , is the radial quantum number or knot number.
The SO(4) multiplets comprise all wave functions ψ 2J with equal sum
L + N = 2J for the principal quantum number n = 1 + 2J with angular
momentum (1 + 2L)-multiplets for SO(3) and radial quantum numbers N
for Lenz–Runge classes, parametrizable by the 3-sphere Ω3 ∼ = SO(4)/SO(3).
There is an orthogonal basis transformation from eigenvectors of double-
“spin” {J3± } to eigenvectors of angular momentum {L 2 , L3 }:
n = 1 + 2J : |J; j+ |J; j− ∼ |n; L, m, m = j+ + j− ,

2 |n; L, m = L(1 + L)|n; L, m, L = 0, . . . , 2J,
L
L3 |n; L, m = m|n; L, m, m = −L, . . . , +L.
All vectors of this basis are obtained from the highest vector |J; J|J; J =
|n; 2J, 2J with the angular momentum and Lenz–Runge lowering operators
(in the weight diagram: diagonal and skew-diagonal downwards, respectively):

|n; L, L = (P− )2J−L |n; 2J, 2J, L = 0, . . . , n − 1,
n = 1 + 2J : with
|n; L, m = (L− )L−m |n; L, L, m = −L, . . . , +L.
4.4 Particles and Ghosts

Particles are defined with irreducible Hilbert representations of time transla-
tions. Nonparticle degrees of freedom use indefinite unitary, nondecomposable
time representations, e.g., the two nonparticle degrees in the four-component
electromagnetic field (see Chapter 5).
4.4.1 Definite Metric, Fock Space, and Particles

The irreducible Hilbert representation of the time translations with real en-
ergy E ∈ R,
R t −→ eiEt ∈ U(1) ⊂ GL(V ),
as used for the harmonic oscillator act on a complex one-dimensional space V
with the creation operator u as basis. The dual representation R t −→ e−iEt
acts on the dual space V T , spanned by the annihilation operator u :
V = Cu, u(t) = eiEt u, V T = Cu , u (t) = e−iEt u .

⎪
⎪
The product representations with energy eigenvalues {zE ⎪
⎪z ∈ Z} act on
! T ∼ 2
the tensor algebra V of the self-dual space V = V ⊕ V = C , i.e., on all
complex linear combinations of all products of u and u . The Bose and Fermi
quantum algebras Q (V) for = ∓1, respectively, are the quotient algebras
of the tensor algebras with respect to the equality of the dual product of the
basic vectors,
duality of V and V T : u , u = 1, u, u = 0, u , u = 0,
and their corresponding (anti-)commutators, [a, b]! = a ⊗ b + b ⊗ a. This

can be implemented by going over to classes of V with respect to an
appropriate ideal (the elements of the ideal constitute the 0-class):
S = {[u , u] − 1, [u, u]

! , [u , u ] },

I(S ) = minimal ideal in V, generated by S ,

' C4 , = +1 (Fermi quantum algebra),
Q (V) = V/I(S ) ∼ =
Cℵ0 , = −1 (Bose quantum algebra).
In the finite-dimensional Fermi quantum algebra, the Pauli exclusion principle

is formalized by nilquadratic basic vectors:
= +1 : (u)2 = 0 = (u )2 ,
basis of Q+ (C2 ) : {1, u, u , [u, u ]}.
'
0
The quantum algebras are associative and unital 1 ∈ C = V. Their U(1)-
conjugation with (αa) = αa for α ∈ C and (ab) = b a is induced by

the conjugation of the basic vectors u ↔ u . The familiar (anti-)commutators
hold in the quantum algebras, e.g., for the adjoint action of the Hamiltonian,
constructed with the quantum-opposite commutator:
⎧
⎨ [u , u] = 1, [u, u] = 0 = [u , u ] ,
2
in Q (C ) : H = EI1 , I1 = 12 [u, u ]− ,
⎩
[I1 , (u)k (u )l ] = (k − l) (u)k (u )l , k, l = 0, 1, 2, . . .
The Bose quantum algebra can be generated by position and momentum

as basic hermitian combinations:
√−u

u
for Q− (C2 ) : x = u√ +u
2
, ip = 2
⇒ [ip, x] = 1,
p2 +x2
H =E 2 , [iH, x] = Ep, [iH, p] = −Ex.
The quantum algebras inherit the Hilbert metric of the basic representa-
tion space,
V × V −→ C, u|u = 1,
in the form of the Fock state, a linear, conjugation-compatible form:
⎧
⎨ a + b = a + b,
Q (C2 ) −→ C, a −→ a with αa = αa, α ∈ C,
⎩
a = a,
defined by [I1 , a] = 0, (u u)k = 1 for k = 0, 1, 2, . . . ⇒ (u )k (u)l = δkl k!.
It makes the quantum algebras pre-Hilbert spaces with a semidefinite
product:
Q (C2 ) × Q (C2 ) −→ C, a|b = a b,
⇒ a|a = a a ≥ 0,
e.g., u|u = u u = 1,
u |u = uu = 0.
The classes of the quantum algebras with respect to the annihilator left ideal
Q (C2 )u with Q (C2 )u = 0 is a vector space with a definite scalar prod-
uct. Its Cauchy completion is the familiar Fock space. The ground-state vector
|0 is the class of the unit, and the vectors |k with k quanta arise by the
creation operator products:
⎧
⎨ |0 = 1 +k Q (C )u ,
⎪ 2
Fock (C2 ) = Q (C2 )/Q (C2 )u : |k = (u)

√ |0, u |0 = 0,
⎪
⎩ k|l = δ k!.
kl
The two-dimensional Fermi Fock space has the ground-state vector and the
1-quantum vector as basis. The Bose Fock space has countably infinite Hilbert
dimension:
∞

Fock+ (C2 ) = C|0 ⊕ C|1, Fock− (C2 ) = C|k.
k=0
The position representation for the Bose oscillator comes with the Hermitian
polynomials Hk :
⎧
⎪
2
−x
⎪
⎪ u |0 = 0 ⇒ |0 ∼
= e√√ ,
2
⎪
⎨
u ∼
2 2
x
√ x = −e 2 √ dx − 2 x
= x−d 2 2
e , π
2
−x
|k = √k! |0 ∼
k k
= √2k k! e√√ ,
(u) H (x)
∼ x+d ⎪
2 2 2
√ x = e− 2 √
x dx x2
u = e , ⎪
⎪
2 2 ⎪
⎩ k
π
H (x) = ex (−dx )k e−x .
2 2
All free particles, embedded in relativistic fields (see Chapter 5), use this
structure, characterized by the (anti-)commutator and the Fock form:
Q (C2 ) with [u , u] = 1 and u u = 1.
The time-dependent Fock state value with a = 0|a|0 can be formulated
as an integral with a Dirac energy measure or as the residue of a simple pole:
0|[u (0), u(t)] |0 = 0|[u (0),

u(t)]− |0 = 0|u

(0)u(t)|0
. 0 1 iq t
=e iEt
= dq0 δ(q0 − E)e iq0 t
= dq 2iπ q0 −E e
0
,
(dt − iE)e iEt
= 0.
The time-ordered operator products have a pole structure in the upper and
lower complex energy planes:
0
ϑ(±t)0|u (0)u(t)|0 = ϑ(±t)eiEt = ± dq 1
2iπ q0 ∓io−E e
iq0 t
,
(dt − iE)ϑ(±t)e iEt
= dt ϑ(±t) = ±δ(t).
ϑ(−t)0|u (0)u(t)|0 connects time order with creation-annihilation order —

first creation, then annihilation. For free particle fields, those integrals
show up as energy-momentum integrals, e.g., in Feynman propagators (see
Chapter 5).
4.4.2 Indefinite Metric and Ghosts

Nonparticle degrees of freedom come with reducible, but nondecomposable
translation representations, the simplest ones being complex two-dimensional.
The faithful image of noncompact time R is in an indefinite unitary group.
In addition to the energy eigenvalue E ∈ R, there occurs an arbitrary real
nontrivial nilconstant (gauge-fixing constant) ν ∈ R, ν = 0:

R t −→ eiEt 10 iνt
1 ∈ U(1, 1) ⊂ GL(W ).
W and its dual space W with dual representation t −→

T
The vector
space
e−iEt −iνt 1 can be spanned by two basic vectors (b, g) and their dual
1 0
vectors (g× , b× ):
W = Cb ⊕ Cg ∼ = C2 and W T = Cg× ⊕ Cb× ∼ = C2 ,

× ×
duality (nontrivial): g , b = 1, b , g = 1,
which are eigenvectors (g for “good”) and nilvectors (b for “bad,” no eigen-
vector):
g(t) = eiEt g, b(t) = eiEt (b + iνtg),
× −iEt ×
g (t) = e g , b (t) = e−iEt (b× − iνtg× ).
×
The quantum algebras for the complex four-dimensional self-dual basic

space W = W ⊕ W T ∼ = C4 with all complex linear combinations of products
of the four basic vectors are constructed, as in the irreducible case in the
former subsection, with the (anti-)commutators implemented by the dual
products:
nontrivial for Q (C4 ) = Q (C2 ) ⊗ Q (C2 ): [g× , b] = 1, [b× , g] = 1.
Again, the quantum algebras are associative and unital. They inherit the
×
U(1, 1)-conjugation × from the basic vectors (g, b) ↔ (g× , b× ). The Fermi
quantum algebra is finite-dimensional, Q+ (C4 ) ∼= C16 , in contrast to the Bose
4 ∼ ℵ0
quantum algebra, Q− (C ) = C .
The Hamiltonian is the sum of a diagonal part and a nilpotent part:
× ×

HN = EI2 + νN = E [g,b ]− +[b,g ]−
+ ν
[g, g ×
] ∼ E ν
2 2 − = 0 E ,
[HN , b] = Eb + νg, [HN , g] = Eg.
The action of N with the admixture of good eigenvectors is the precursor
of the gauge transformations in quantum field theories. Its adjoint action
ad N = [N, ] is nilquadratic:
[N, HN ] = 0, [N, b] = νg, [N, g] = 0, (ad N)2 (b, g) = 0.
The sesquilinear U(1, 1)-form is indefinite:

b|b b|g 0 1
W × W −→ C, g|b g|g = 1 0 ,

b + g|b + g b + g|b − g 1 0
b − g|b + g b − g|b − g =2 0 −1 .
The eigenvector and nilvector constitute a ghost pair (Witt pair) with a
neutral signature metric, as familiar from a lightlike basis {e± = e0 ± e3 } for
the lightcone in Minkowski spacetime with the indefinite O(1, 1) ⊂ O(1, 3)
Lorentz metric.
As in the former subsection, the sesquilinear U(1, 1)-form can be extended
to the quantum algebras, which, because of the indefiniteness, does not lead
to a nontrivial Hilbert space.
In an integral formulation of the time representation matrix elements, the
indefinite metric shows up in the derivative of a Dirac distribution and the
higher-order pole (dipole):
×
g (0)b(t) b× (0)b(t) 1 iνt iEt δ(q0 − E) −νδ (q0 − E)
× ×
g (0)g(t) b (0)g(t)
= e = dq 0 δ(q0 − E) eiq0 t
0 1
10
. 0 q0 −E (q −E) ν
= dq2iπ 0
0
1
2
eiq0 t .
q0 −E
Particles and ghosts can be combined by using decomposable time repre-

sentations and the corresponding product quantum algebras, e.g.,
iωt
e 0 0
R t −→ 0 eiEt iνt eiEt ∈ U(1) × U(1, 1) ⊂ U(2, 1) ⊂ GL(C3 ),
iEt
0 0 e
'3
nontrivial for Q (C6 ) = Q (C2 ): [u , u] = 1, [g× , b] = 1, [b× , g] = 1.
The linear combinations of eigenvectors (particles) and ghost pairs have time
representation matrix elements that combine simple and higher-order poles,
for example, with α ∈ C:
. ν|α|2
ψ =u+α b− 1
2α g ⇒ ψ ∗ (0)ψ(t) = 2iπ [ q0 −ω − q0 −E + (q0 −E)2 ]e
dq0 1 1 iq0 t
. dq0 (ω−E) 2
iq0 t
= 2iπ (q0 −ω)(q 0 −E)
2e
for |α| = ν > 0.

2 E−ω
Chapter 5
Quantum Fields of Flat

Spacetime
Spacetime theories are field theories as conceptionally initiated by Faraday

and first formulated by Maxwell. Charged or neutral mass points with or-
bits x −→ X(τ ) in spacetime,
√ i.e., depending only on the Lorentz invariant
2
eigentime τ = (x0 )ϑ(x ) x2 , and the related point-supported Dirac dis-
tributions δ(x − X(τ
)) for position in the electromagnetic current and the
energy-momentum tensor are strangers for relativistic electromagnetic and
gravity fields, x −→ A(x), g(x). As underlined by Einstein himself and sug-
gested by quantum theory, such dichotomic theories have to be replaced by
pure field theories, i.e., with fields also for matter. Einstein (1936): “What
appears certain to me, however, is that, in the foundations of any consistent
field theory, the particle concept must not appear in addition to the field
concept.” Classical gravity is a real theory. The interpretation of quantum
theory with probability amplitudes is established by a complex Hilbert space
formulation. Einstein probably thought primarily of classical fields. Neverthe-
less, he also considered complex structures in his attempts to unify gravity
and electrodynamics.
Fields can be valued in “new” spaces. They have, in general, external
and internal degrees of freedom. Quantum fields can also implement, in
addition to global operations like rotations SO(3) in classical and quan-
tum mechanics, local operation groups, e.g., a gauge symmetry like U(1)-
electromagnetism or the electroweak U(2)-interactions in the standard model
of particle interactions.
Quantum theory does not use pointlike particles and time orbits in posi-
tion as basic concepts. In nonrelativistic quantum mechanics, the position is
an operator that may — but does not have to — be diagonalized by an appro-
priate experimental setup as exemplified by the double-slit experiments. The
position dependent wave functions yield probability catalogues for the pos-
sible results of position measurements. In relativistic quantum field theory,
112 Chapter 5 Quantum Fields of Flat Spacetime
neither time nor position comes as basic operator. Quantum fields are opera-
tors, not spacetime-dependent probability amplitudes (wave functions). The
name “second quantization” is misleading, as there is only one “quantiza-
tion” — in two steps: Quantum mechanics is the quantum structure of time-
dependent operators, built by position and momentum, which give rise to
time representation coefficients; quantum field theory is the quantum struc-
ture of spacetime-dependent operators, built by, e.g., electron-positron or
gauge fields, which give rise to spacetime representation coefficients.
The time-dependent position-momentum pairs t −→ (x, p)(t) of quantum
mechanics are replaced, in canonical quantum field theories, by spacetime-
dependent field pairs, e.g., by the pair with electromagnetic potential and field
strength x −→ (Aa , Fba )(x), or by a pair with a scalar field and its derivative
x −→ (ΦΦ , ∂ aΦ )(x), or by a Dirac field with its conjugate x −→ (Ψ Ψ, Ψ ∗ )(x).
Interactions, in mechanics described by position-dependent potentials x −→
V (x), e.g., ω2 x2 (harmonic oscillator) or |x| δ
(Kepler potential), are given,
in the field theories used, by field polynomials, (A, Φ , Ψ ) −→ P (A, Φ , Ψ ),
2
implementing operation invariants, e.g., by m2 Φ 2 (mass term) or g80 (Φ Φ ∗Φ −
2 2 a
M ) (Higgs potential) or by gauge vertices Ψ γaΨ A .
To characterize quantum field theory as a theory of pointlike particles is
wide of the mark. It is also not obvious how to concretize operationally the
concept of a “pointwise” interaction as attributed, e.g., to the local product
of an electron-positron field with a gauge field Ψ (x)γaΨ (x)Aa (x), used as
gauge vertex for an electromagnetic interaction. Also a naive interpretation
of Feynman diagrams with pointlike particles propagating in spacetime may
lead to wrong associations.
The Poincaré group SO0 (1, 3) × R4 for Minkowski spacetime with the
translation subgroup, or, better, its twofold cover SL(2, C) × R4 , is im-
plemented by interaction-free elementary particles. All spacetimes with
Minkowski tangent spacetimes can be endowed with free particles à la
Wigner, i.e., as irreducible unitary Poincaré group representations acting on
infinite-dimensional Hilbert spaces. Massive particles with spin 0, e.g., pions
as an example for integer spin, and with spin 12 , e.g., electron-positrons as an
example for half-integer spin, are relativistically embedded, respectively, in a
scalar field Φ and a Dirac spinor field Ψ with dimension of a length density
1
for integer spin, [Φ Φ] = m , and of a square root volume density for halfinteger
√1
Ψ] =
spin, [Ψ 3
.
m
The embedding of particles in relativistic fields is closely related to the
induction of Hilbert representations of the Poincaré group from Hilbert repre-
sentations of a “little” group and translations, given by the spin rotations and
timelike translations for massive particles in a rest system and by the axial
rotations around the momentum and lightlike translations for massless parti-
cles in a “lightsystem.” The Fourier expansion of a free field, i.e., its particle
analysis, involves creation and annihilation operators for particles (better:
operator distributions) with a definite momentum q ∈ R3 . Those transla-
tion eigenoperators do not describe the position of a pointparticle. For free
5.1 Electrodynamics of Fields 113
particles, the position coordinate characterizes the translation behavior in

the irreducible unitary representations R3 x −→ eiqx ∈ U(1).
The Wigner definition of gravitons as particles makes sense only for flat
spacetime, i.e., for an expansion of Einstein’s gravity in absolute Minkowski
spacetime. Obviously, the related definition of special relativistic fields for
interaction-free gravitons does not solve the problem of the quantization of
gravity. In addition and as well known, such a perturbative flat spacetime
approach of gravitative interactions leads to a nonrenormalizable framework.
Special relativistic quantum field theory has Planck’s action constant
and the speed of light c as two universal intrinsic units. They reduce the
“individual” units of a special dynamics to a mass or a length unit with the
Compton conversion factor c ∼ 3.3 × 10−42 kg m.
5.1 Electrodynamics of Fields

Flat spacetime field theory uses the gauge principle for the construction of
“basic” interactions. However, there is no principle for the determination of
the “basic” gauge groups.
In contrast to electrodynamics with charged mass points, which is usually
formulated in the real, electrodynamics with charged matter fields is formu-
lated more appropriately in the complex with a manifest U(1)-origin of the
electromagnetic gauge interactions. Since real electrodynamics does not sug-
gest immediately the compact phase group U(1), it is understandable that
Weyl, in his first attempt, gauged the locally isomorphic noncompact dila-
tion group D(1) = exp R ∼ = R, the simply connected Z-fold covering group of
U(1) = exp iR ∼ = R/Z.
Maxwell’s electrodynamics
is characterized by an action, given by the
∼
Killing form η ∧ η = 0 −13 0
13 with signature (3, 3), for the field strengths F
in an adjoint representation of the Lorentz Lie algebra. The Killing form is
normalized with the dimensionless Sommerfeld constant, in the second-order
derivative Lagrangian:

Welmag = d4 x L(A), L(A) = 4g12 Fab Fab , g 2 = 4παS ∼ 137 4π
∼ 0.9 × 10−1 .
The field strengths Fab = ∂ b Aa − ∂ a Ab arise from the vector potential A,

which is acted on by the Minkowski representation of the Lorentz group. In
a first-order derivative Lagrangian, the vector potential and field strengths
are used as independent variables:
2 ab
g2 g F = ∂ b Aa − ∂ a Ab ,
L(A, F) = Fab ∂ A −∂
a b b a
A
+ Fab Fab
⇒
2 4 ∂a Fab = 0.
With opposite charge numbers ±z = ± Q e ∈ Z in dual irreducible repre-

sentations of the group U(1) eiα −→ e±izα , charged fields come in complex
Φ, Φ ∗ ) and (Ψ
pairs, e.g., (Φ Ψ, Ψ ), involving particles and antiparticles. The fields
have an electromagnetic interaction via a covariant derivative1 with the vector
potential. Each spacetime translation is accompanied by an electromagnetic
U(1)-operation. This replaces the Lorentz force for the charged mass points:

elmag
Wmatter Φ ) = d4 x [(∂ a + izAa )Φ
(Φ Φ∗ (∂a − izAa )Φ
Φ − m2ΦΦ ∗ ],
(∂a − izAa )(∂ a − izAa )ΦΦ = −m2Φ ,
elmag
Wmatter Ψ) = d4 x i[Ψ
(Ψ Ψγa (∂ a − izAa )Ψ
Ψ − imΨ ΨΨ ],
γa (∂ − izAa )Ψ
a Ψ = imΨ Ψ.
The electromagnetic current for a charged mass point is given by a Dirac

distribution in position:

3 Ja (x) = m dτ Pa δ(x − X(τ )) = z dXa δ(x − X) = z P0 δ(x − X(t)),
z Pa
d x J0 (x) = z.
For matter fields, the point-supported Dirac distribution is spread to a space-

time field by a U(1)-invariant product of dual fields with nontrivial U(1)-
transformations:
Ja (Φ Φ∗Φ − Φ ∗ (∂a − izAa )Φ
Φ) = iz[(∂a + izAa )Φ Φ],
Ψ) = zΨ
Ja (Ψ ΨγaΨ .
The eigentime orbits x −→ τ −→ X(τ ) of charged mass points have trivial

U(1)
electromagnetic U(1)-properties X −→ X. Fields x −→ (ΦΦ(x), Ψ (x)) “open
up local spaces” for local transformation groups with chargelike (internal)
U(1)
degrees of freedom, here (ΦΦ, Ψ ) −→ eizα (Φ
Φ, Ψ ) (see Chapter 6).
The current involves the density for the U(1)-Lie algebra:

i d3 x J0 (x) = l0 ∈ log U(1).
∂L
It canbe considered to be the answer Ja = ∂A a of a classical field Lagrangian
4
W = d x L to the change of the vector potential (gauge field). The gauge
interaction vertex Ja Aa implements the representation of the U(1)-Lie alge-
bra acting on the vector spaces with the matter fields (see Chapter 6).
A real formulation exists for integer spin: Dual U(1)-representations can
be combined in a representation of the U(1)-isomorphic axial rotation group
Φ2
SO(2) with a basis of the real 2-dimensional space Φ = Φ 1√ +iΦ
2
:

ϕ = Φ Φ 2 −→ e
1 zIα
ϕ , I = 01 −1 0 , e
zIα
∈ SO(2) ∼
= U(1),
L(ΦΦ) = 2 [(∂ a − zIAa )ϕ
1
ϕ ]T (∂a − zIAa )ϕ
ϕ − m2ϕ 2 ,
Φ) = −zϕ
Ja (Φ ϕT I(∂a − zIAa )ϕ ϕ.
T denotes transposition; e.g., ϕ T = (Φ

Φ1 , Φ 2 ).
1 To save parentheses, derivatives act only on the immediately following field.
5.2 Gravity of Fields 115
5.2 Gravity of Fields

Einstein’s gravity is characterized by an action with the curvature scalar in
terms of the metrical tensor, normalized with the Planck area or the squared
Planck mass m2P 2 = 8π c2 , in the second-order derivative Lagrangian,
2

Wgrav = d4x L(g),
kj with 2 = 8πG
∼ (0.8 × 10−34 m)2 ∼ 8π
.
L(g) = 1
2
2 |g| g kj R• , c3 m2P
The Planck mass is given in proton mass mp units by mP

mp ∼ 1020 or
m2P
log m2p ∼ 88.
A first-order derivative Lagrangian with a tetrad e(x) ∈∈GL(4, R)/SO0
(1, 3) and the Fock–Iwanenkov connection Γ for a structural group SO0 (1, 3)
as independent variables reads
i] [k i]
L(e, Γ ) = 12 |e|eak ebi (∂ [kΓ ab − 2Γ ac η cdΓ db ), |e| = det eic
⇒ 2Γ iab = (ηac ηbd − ηbc ηad )eie (η df eej ∂ c ejf + η ef edj ∂ c ejf − η df ecj ∂ e ejf ).
The gravitative equations of motion use the covariant derivative for the
tangent translations of the spacetime manifold, e.g., for the scalar field:
grav
Wmatter (ΦΦ) =
|g| d4 x 12 (gkj ∂ kΦ ∂ j Φ − m2Φ 2 )
=− |g| d4 x 12 (Φ
Φ∂g2Φ + m2Φ 2 ),
∂g2Φ = −m2Φ .
In mass point gravitation, the mass drops out in the geodesic equations. The
field equations of motion contain the masses.
The field equation involves the invariant Laplace-Beltrami operator (see
Chapter 2):

∂g2 = √1 ∂ k |g| gkj ∂ j = (∂j − Γki
j gki )∂ = (∂a − Γ ab )∂ .
j b a
|g|
Six of the 16 connection forms act as gauge fields of the tangent Lorentz
group SO0 (1, 3):
Γk Λ−1 + (∂ k Λ)Λ−1 .
Λ ∈ SO0 (1, 3) : Γ k −→ ΛΓ
The Lorentz covariant derivative is defined for the two fundamental Weyl
spinor representations [ 12 |0] and [0| 12 ], for their sums and product representa-
tions, i.e., for all fields acted on by a finite-dimensional representation [L|R]
of Lorentz group and Lie algebra:
log SO0 (1, 3) Lab −→ D[L|R] (Lab ) ∈ AL(V ), V = ∼ C(1+2R)(1+2L) ,

log SO0 (1, 3) Γ ab L −→ Γ ab D[L|R] (L ) = Γ [L|R] ∈ AL(V ),
k ab k ab k
Lorentz group covariant derivative: 1V ∂ k − Γ k[L|R] .

Integer spin particles are in fields with a Lorentz group representation that
can be chosen as real, i.e., V ∼
= R(1+2R)(1+2L) for the integer sum L + R ∈ N.
For the action on the vector field ∂ iΦ with the scalar field derivatives, the
Lorentz group gauge fields are valued in the Lie algebra of the real four-di-
mensional Minkowski representation:

Γk k
Γk[ 1 | 1 ] ∼
0 0α = Γ α0
= Γk Γ αβ = − Γ k
k ∈ AL(R4 ), α, β = 1, 2, 3.
2 2 β0 βα
1
The energy-momentum tensor with dimension [T] = m4 for the gravitative
interaction
gli • gli
• −
−
12 (Rli 2 R• ) = Tli , −
2 R•
1 li
= Tli − 2 T•
is given by the derivative of the Lagrangian with respect to the metric, e.g.,
for a scalar field,
li
√2 Φ)
∂L(Φ
Φ ) = ∂ lΦ ∂ iΦ − √g
= Tli (Φ Φ)
L(Φ
|g| ∂gli |g|
li
= ∂ l Φ ∂ iΦ − g
2 (∂j Φ ∂ j Φ − m2Φ 2 ),
T• (Φ
Φ ) = −∂j Φ ∂ j Φ + 2m2Φ 2 ,
gli gli 2 2
Tli − 2 T• = ∂ l Φ ∂ iΦ − 2 m Φ .
The fields spread the Dirac position distribution in the energy-momentum

tensor of a mass point to spacetime. In flat spacetime, it involves the position
density of the translation generators,
l i
|g|Tli (x) = m
1
dτ P l P i δ(x − X(τ )) = PPP0 δ(x − X(t)),

3 0a a
g =η: d x T (x) = P .
For orthonormal
bases, the metric derivative can be replaced by the tetrad
derivative and |g| = |e|:
grav
Wmatter = |e|d4 x 12 (∂aΦ ∂ aΦ − m2Φ 2 ),
Φ)
1 ∂L(Φ eia
|e| ∂ea
Φ) = ∂aΦ ∂ iΦ −
= Tia (Φ Φ
|e| L(Φ ), eal Tia (Φ
Φ) = Til (Φ
Φ).
i
The energy-momentum tensor for the gravitative interactions contains, com-

pared with the current for the electromagnetic interactions, an additional
derivative.
Dirac fields for half-integer spin particles can be defined in orthonormal
bases with the Dirac representation [ 12 |0] ⊕ [0| 21 ] of the Lorentz cover group
SL(2, C) and an action Ψ −→ e− 4 [γ ,γ ]λab Ψ . They gravitate via the tetrad
1 a b
grav
Wmatter Ψ) = |e| d4 x [iΨ
(Ψ Ψγc eck (∂ k − Γ k )Ψ
Ψ + mΨ
ΨΨ ],
γc eck (∂ k − Γ k )Ψ
Ψ = γc (∂ c − Γ c )Ψ
Ψ = imΨ
Ψ.
5.3 Gravity and Electrodynamics 117
Their covariant derivative involves the gauge fields Γ k valued in the Lorentz
Lie algebra of the complex four-dimensional Dirac representation:
Γ k = −Γ
Γkab 41 [γ a , γ b ] ∈ AL(C4 ),
Lorentz Lie algebra: L −→ i d3 x Jab
ab
0 ( x),
ab [γ a ,γ b ]
Ψ
Jc = iΨ γc 4 Ψ .
The Lagrangian change for a tetrad variation reads

Ψ)
1 ∂L(Ψ Γc eia
|e| ∂ea = iΨ γa ∂ − γc ∂ea Ψ − |e| L(Ψ ).
i ∂Γ
Ψ Ψ
i i
5.3 Gravity and Electrodynamics

In the action of electromagnetic and gravitative fields,

W grav,elmag = |g| d4 x [ 2
12 gli Rli 1 kl ji
• + 4g2 gli gkj F F ],
the Ricci tensor contains second-order derivatives of the metric. The first-
order derivative of the vector potential in the field strength Fil = ∂ l Ai −
∂ i Al as an external derivative F = dA has no metric-dependent connection
contributions. The field strengths have the equation of motion

√1 ∂ k |g| gli gkj Fij = ∂ k Flk + Γkj kj
k Flj − Γl Fjk = 0.
|g|
The energy-momentum tensor of the electromagnetic field is traceless:

gli kj
Tli (F) = 1 kl ji
g2 (gkj F F − 4 F Fkj ), T• (F) = 0.
The electromagnetic and gravitative interaction of a charged scalar matter

field,
grav,elmag
Wmatter = |g| d4 x [ 2
12 gli Rli 1 kl ji
• + 4g2 gli gkj F F
k Φ∗ j Φ − m2ΦΦ ∗ ],
+gkj (∂ + izA )Φ (∂ − izAj )Φ
k
comes with the electromagnetic current and the energy-momentum tensor of

the matter field, both invariant under phase U(1) ∼= SO(2)-gauge transfor-
mations:
− g12 (∂ b Fba + Γbc
b Fca − Γa Fcb ) = Ja (Φ
bc Φ ),
η ab •
−
12 (Rab
• − 2 R• ) = Tab (F) + Tab (Φ
Φ),
Φ) = iz[(∂a + izAa )Φ
Ja (Φ Φ∗Φ − Φ ∗ (∂a − izAa )Φ
Φ ],
b Φ∗ a
T (Φ ) = (∂ + izA )Φ (∂ − izAa )Φ
ab Φ b Φ
Φ∗ (∂c − izAc)Φ
− η ab [(∂ c + izAc )Φ Φ − m2Φ ∗Φ ].
5.4 Linearized Einstein Equations

For a linear approximation of gravity on the background of absolute
Minkowski spacetime with fixed Lorentz metric (M(1,3) , η), the metric is
expanded with an order parameter λ ∈ R by a 10-component symmetric
tensor field R4 x −→ λE(x) = E(x) (order parameter λ is included):
gil = δci δdl (η cd + 2λEcd ) + . . . = δci δdl (η cd + 2Ecd ) + . . . .
It is the first-order approximation of the tetrad,
gil = eic η cd ejd , eic = δai (δca + Eac ) + . . . .
The gravity tensor field will be put side by side with a flat spacetime
vector field x −→ A(x) as used for the electromagnetic gauge field. The
analogue to the electromagnetic field strengths with first-order derivatives is
the linearized connection coefficients:
Fcb = ∂ b Ac − ∂ c Ab ,
Γcdb
Γ = Γ cd
a η
ab
= −∂ b Ecd + ∂ c Ebd + ∂ d Ecb .
The derivative of the field strength corresponds to the linearized curvature

tensor with only spacetime derivatives and without the characteristic non-
pi li pj
p Γk − Γp Γk = O(λ ):
linearity, i.e., without the self-coupling squares Γlj 2

∂a Fcb = ∂a ∂ b Ac − ∂a ∂ c Ab ,
(A, F, ∂F) :
∂a Fca = ∂ 2 Ac − ∂ c ∂a Aa ,
⎧ cdb
⎪
⎪ R a − ∂ Γa ,
= ∂ dΓ cb b cd
⎨ acd
R• = ∂ dΓ ca − ∂ a cd
Γa
(E, Γ , R) : a
⎪
⎪ = ∂ 2 Ecd − ∂ c ∂a Eda − ∂ d ∂a Eca + ∂ c ∂ d E• ,
⎩ 1 •
2 R• = ∂ 2 E• − ∂c ∂a Eca ,
with the trace denoted by E• = Eaa .

The Einstein tensor Ř• of the curvature, which, in four spacetime dimen-
sions, coincides with the reflected Ricci tensor (see Chapter 2),
Rab η ab • ˇ =R ,
• ↔ Ř• = R• − 2 R• , Ř
ab ab
• •
involves derivatives only of the reflected linearized tensor (Einstein combina-

tion Ě):
∂a Fca = ∂ 2 Ac − ∂ c ∂a Aa ,
Rcd
• = ∂ 2 Ecd − (δbd ∂ c + δbc ∂ d )∂a Ěab ,
Řcd
• = ∂ 2 Ěcd − (δbd ∂ c + δbc ∂ d − η cd ∂b )∂a Ěab ,
with Ěab
ab
= Eab − η2 E• , Ě ˇ = E.
5.5 Free Particles for Flat Spacetime 119
The field strength derivative in the Maxwell equations is determined by the

electromagnetic current, and the Einstein tensor by the energy-momentum
tensor:
∂a Fca = −g 2 Jc ,
Řcd
• = − T .
2 cd
In the lineariziation of Ř• = −2 T − Λg no cosmological term arises if the

constant is assumed of more than second-order Λ = O(λ2+k ), k ≥ 0.
The analogue to the Lagrangian of electrodynamics,
Lelmag = L(A) + Lmatter = 1
4g2 Fab F
ab
− Ac Jc + Lmatter,
is the Lagrangian of gravity, linearized with |g| = − det g = 1 − 2E• , and a

gravitational normalization 2 of second-order in the flat space expansion:,
Lgravity = L(E) + Lmatter + O(λ2 ),
• • E = λE = O(λ),
2
2 E R•
1 cd
L(E) = + Ecd T , with
2 = O(λ2 ), Λ = O(λ2+k ).
The Minkoswki spacetime expansion can be considered as an expansion in
Newton’s constant λ2 ∼ 2 = 8πGc3 .
The linearized gravitative interaction is illustrated by a scalar field:
grav
Wmatter Φ) = |g| d4 x 12 (gkj ∂kΦ ∂j Φ − m2Φ 2 )
(Φ
= d4 x (1 − E• ) 12 [(η cd + 2Ecd )∂cΦ ∂dΦ − m2Φ 2 ] + O(λ2 ),
Φ) = (1 − E• ) 12 (∂ cΦ ∂cΦ − m2Φ 2 ) + Ecd ∂ cΦ ∂ dΦ

L(Φ
= 12 (∂ cΦ ∂cΦ − m2Φ 2 ) + Ecd Tcd (Φ
Φ),
η cd
with T (Φ ) = ∂ ∂ − 2 (∂c ∂ − m2Φ2 ).
cd Φ c Φ dΦ Φ cΦ
The field equations are compared with those of electrodynamics for a charged
scalar field:
ab
∂ 2Φ + (2∂aΦ ∂b + 2∂a ∂bΦ )(Eab − η2 E• ) = −(1 − E• )m2Φ ,
∂ 2Φ − 2iz∂aΦ Aa − z 2Φ Aa Aa = −m2Φ .
5.5 Free Particles for Flat Spacetime

Canonical quantum field theory works with free particle fields. They are used
also as asymptotic fields for an expansion of interactions. A free scalar quan-
tum field has the Lorentz transformation behavior
Λ
Λ ∈ SO0 (1, 3) : Φ −→ Φ Λ , with Φ Λ (x) = Φ (Λ−1 .x).
The particle mass is the translation invariant in the flat spacetime equation:

∂a ∂ aΦ = −m2Φ ,
g=η: ab
Tab (ΦΦ) = ∂ aΦ ∂ bΦ − η2 (∂cΦ ∂ cΦ − m2Φ 2 ).
The Feynman propagator with a normalization factor ρ(m2 ) = 0,

d4 q ρ(m2 )
0|Φ
Φ(y)Φ
Φ (x)|0Feynman = i
π (2π)3 q2 +io−m2 e
iq(x−y)
,
involves the on-shell contribution (particles) from the Dirac distribution in

the decomposition
i 1
π q2 +io−m2 = δ(q 2 − m2 ) + i
π qP
1
2 −m2 .
This contribution with the expectation value 0| . . . |0 of the anticommutator
and the Fock ground-state vector |0 for translation eigenvectors is a represen-
tation matrix element of the Poincaré group SO0 (1, 3) × R4 (see Chapter 8),
for a scalar field nontrivial for the translations R only:
4
d4 q
R4 x − y −→ 0|{ΦΦ(y), Φ (x)}|0 = ρ(m2 ) (2π) 3 δ(q − m )e
2 2 iq(x−y)
.
The off-shell contribution (“virtual particles”) with the principal value part,
d4 q ρ(m2 ) iq(x−y)
−(x0 − y0 )0|[Φ
Φ (y), Φ (x)]|0 = πi (2π) 3 q 2 −m2 e
P
= 0 for (x − y)2 < 0,
is not a coefficient of a Poincaré group representation. It is a distribution,
arising by causal ordering from the causally supported, i.e., x ∈ R4+ ∪ R4− ,
on-shell quantization of the free scalar field via the commutator:
d4 q
Φ(y), Φ (x)] = ρ(m2 ) (2π) 3 (q0 )δ(q − m )e
2 2 iq(x−y)
[Φ
4 2
d q ρ(m ) iq(x−y)
= −(x0 − y0 ) πi (2π) 3 q 2 −m2 e ,
/
/
P
4
2 −i x−
[i∂ Φ (y), Φ (x)]/
a
= ρ(m ) (2π)3 (q0 )q δ(q − m )e
2 d q a 2 q ( y)
x0 =y0
= ρ(m2 )δ0a δ(x − y ).
For free fields, the normalization ρ(m2 ) = 0 can be chosen arbitrarily, e.g.,
ρ(m2 ) = 1 for a position translation normalization with d3 x δ(x) = 1.
For a particle field, the time projection, effected by position integration,
gives as real part a harmonic oscillator cosine, i.e., a time representation
coefficient with the mass as invariant. The position projection, effected by
time integration, is nontrivial for the off-shell contribution (imaginary part)
only. It gives as position representation coefficient a Yukawa potential with
the inverse mass as its range:
3 d4 q −i|mx0 |
1
d x iπ(2π)3 −q2 −io+m 2e
iqx
= cos mx0 −(x
|m|
0 )i sin |m|x0
= e |m| ,
d4 q 1 iqx
4
d q 1 iqx
dx0 iπ(2π)3 −q2 −io+m 2e = dx0 iπ(2π) 3 −q 2 +m2 e
d3 q −i
P
−|m|r
1
= 4iπ3 q2 +m 2e
qx
= −i e 2πr .
The position dependence of the Yukawa potential comes from the time pro-
jection of a causally supported x2 ≥ 0 distribution, not from a spacelike
x2 < 0 dependence.
5.5 Free Particles for Flat Spacetime 121
The on-shell function for particles fulfills a homogeneous equation for

the translation invariant of the Poincaré group representation ∂ 2 = −m2 ,
whereas the off-shell distribution for “virtual particles” and interactions is a
Green’s kernel of the corresponding inhomogeneous equation,

d4 q δ(q2 − m2 ) 0
(∂ 2 + m2 ) (2π)3
i 1
π q2 −m2
eiqx = −2iδ(x) .
P
A free-field Feynman propagator has three parts: In addition to the on-shell

contribution with the spherical Bessel function sin r|q |r , there is the causally
supported off-shell contribution for energies under the mass threshold with
−|Q|r
the Yukawa potential e r and for energies over the mass threshold with
the spherical Neumann function cos r|q |r , both singular at r = 0:

d4 q δ(q2 − m2 )
(2π)3
i 1
π q2 −m2
eiqx
P
+ ,
sin |
q |r 0
= dq0
(2π)2 ϑ(q0 − m )
2 2 r
cos |
q |r
+ ϑ(m − q0 )
2 2
−|Q|r eiq0 x0 ,
−i e r
−i r
with |
q | = q02 − m2 and |Q| = m2 − q02 .
The harmonic expansion of a free field, i.e., its particle analysis, uses cre-
ation and annihilation eigenoperators (u(q ), u (q )) of the spacetime transla-
tions in the direct integral decomposition,
d3 q

⊕
Φ (x) = 2q0 (2π)3 q ) + e−iqx u (
[eiqx u( q )], with q0 = q 2 + m2 ,
with the quantization and the Hilbert space metric involving

the “inverse”
of the Lorentz invariant momentum measure 2(2π)3 q 2 + m2 δ(q ) ↔
3
3
√q 2 2 ,
d
2(2π) q +m

−
[u (p ), u(q )] = ρ(m2 )2q0 δ( q2πp
),
q −
0|{u (
q )}|0 = 0|u (
p ), u( p )u(q )|0 = m ⎪
2
; p |m ; q = ρ(m )2q0 δ( 2πp ),
2 2
⎪
distributive Hilbert basis: {|m2 ;
q = u(q )|0⎪
⎪q ∈ R3 }.
The field action on the Fock ground state gives cyclic vectors (see Chapter 8),
3 3
Φ (x)|0 = ⊕ 2q0d(2π)
q
3 e
iqx
|m2 ; Φ(x) = ⊕ 2q0d(2π)
q , 0|Φ q
3 e
−iqx
m2 ; q |.
Corresponding structures occur for a Dirac field, Ψ = r ⊕ l, acted on

by a right-left decomposable SL(2, C)-representation [ 12 |0] ⊕ [0| 21 ] (ahead),
as defined by a commutative diagram,
Ψ
R⏐4 −→ C⏐4 s
⏐ ⏐ Ψ −→ Ψ s ,
s ∈ SL(2, C) : Λ(s) s ,
Ψ(Λ−1 (s).x),
with Ψ s (x) = s.Ψ
R4 −→ C4
Ψs

with s −→ s = s
0
0
ŝ ∈ SL(2, C) ⊕ SL(2, C), ŝ = s−1∗ ,
s −→ Λ(s) ∈ SO0 (1, 3),
Λba (s) = 12 tr σa sσ̌ b s∗ , Weyl matrices: σa = (12 , σ ), σ̌a = (12 , −σ ),
and SU(2) rotation properties for its spin- 12 particles. The flat spacetime
equation with translation invariant mass and the Feynman propagator are
⎧
⎪
⎪ γc ∂ cΨ = imΨΨ,
⎪
⎪
⎪
⎨ Tba (Ψ Ψγa ∂ bΨ − δab i(Ψ
Ψ) = iΨ Ψγc ∂ cΨ − imΨΨΨ),
g=η:
⎪
⎪ 0|Ψ
Ψ(y)ΨΨ(x)|0Feynman = ρ(m2 ) πi (2π)
4 a
d q γa q +m iq(x−y)
3 q 2 +io−m2 e
⎪
⎪
⎪
⎩
= 0|[Ψ
Ψ(y), Ψ (x)]|0 − (x0 − y0 )0|{ΨΨ(y), Ψ (x)}|0.
The harmonic analysis of the Dirac spinor field with left and right irreducible
Weyl spinor contributions involves translation and rotation eigenoperators
for particles (uA ( q )), e.g., negatively charged, z = −1, and antipar-
q ), uA (
A
ticles (aA (
q ), a ( q )), then positively charged, z = +1, with spin direction
(eigenvalue) “up” and “down,” A = 1, 2:
A q iqx C
A √ ⊕ d3 q sC ( m ) [e q ) + e−iqx a C (
u ( q )]
∼ r
Ψ (x) = Ȧ (x) = m 3 ,
l 2q0 (2π)
ŝȦ q
C ( m ) [e
iqx C
q ) − e−iqx aC (
u ( q )]
s−1 C ( q ) [e−iqx u (q ) + eiqx a

√ ⊕ d3 q C (
q )]
Ψ ∗ (x) ∼
rȦ Ȧ m C
= l (x) =
m 2q0 (2π)3 ,
A
ŝ−1 C q
A ( m ) [e
−iqx
q ) − eiqx aC (
uC ( q )]
q 2 + m2 .
with q0 =
q q
s( m ) and ŝ( m ) are the momentum dependent boost representations (Weyl
transmutators; see Chapter 7), which embed the spin group SU(2) into the
Lorentz group SL(2, C).
The particle operators have the quantization and Hilbert metric:
−
{uA (
p ), uB (q )} = ρ(m2 )2q0 δA
B
δ( q2πp
),
−
0|uA ( q )|0 = m2 , 12 ; −1, p, A|m2 , 12 ; −1, q, B = ρ(m2 )2q0 δ AB δ( q2π
p )uB ( p
),
−
{aB (
p ), aA (q )} = ρ(m2 )2q0 δA
B
δ( q2π p
),
q−
0|a B
( q )|0 = m , 2 ; +1, p, B|m2 , 12 ; +1, q, A
p )aA ( 2 1
= ρ(m )2q0 δAB δ( 2πp ),
2
⎪
⎪
distributive Hilbert basis: {|m2 , 12 ; z, q, A⎪
⎪z = ±1, q ∈ R3 , A = 1, 2}.
5.6 Massive Particles with Spin 1 and Spin 2 123
The Hilbert metric involves the SU(2)-scalar products δ AB and δAB for the
two spin degrees of freedom.
In general, the finite-dimensional irreducible representations [L|R] of the
Lorentz cover group SL(2, C), in parallel to its compact partner SU(2) ×
SU(2), have complex dimensions (1 + 2L)(1 + 2R):
⎪
⎪
irrep fin SL(2, C) = {[L|R]⎪
⎪L, R = 0, 12 , 1, . . . }.
They are indefinite unitary for [L|R] = [0|0]. In a rest system for a massive
particle, the Lorentz group representations are decomposable into irreducible
(1 + 2J)-dimensional SU(2)-representations [J]:

SU(2) J=L+R
[L|R] ∼
= [J].
J=|L−R|
The subset with the representations, faithful only for the proper Lorentz
group SO0 (1, 3) ∼
= SL(2, C)/I(2), has integer spins SO(3) ∼ = SU(2)/I(2),
in analogy to the central correlation for the SO(4)-representations (see
Chapter 4):
⎪
⎪
⎪ R = 0, 1 , . . . ; L + R = 0, 1, . . . }.
irrep fin SO0 (1, 3) = {[L|R]⎪L, 2
A four-component vector field is acted on by a Minkowski representation

[ 12 | 12 ]. Its totally symmetric products contain the irreducible Minkowski
representations, which are the harmonic SO0 (1, 3)-representations (see
Chapter 8). They are the noncompact partners of the Kepler (harmonic)
SO(4)-representations,
⎪
⎪
⎪ = 0, 1 , . . . },
irrep Mink SO0 (1, 3) = {[J|J]⎪J 2
⎧
⎪
⎪ [ L2 | L2 ], 2J = 0, 2, . . . ,
-
2J ⎨
[ 12 | 12 ] =
L=0,2,...,2J
⎪
⎪ [ L2 | L2 ], 2J = 1, 3, . . . .
⎩
L=1,3,...,2J
5.6 Massive Particles with Spin 1 and Spin 2

The Lorentz-compatible projectors S(q) and V(q) from a Lorentz vector
q ∈ R4 to its spin J = 0 and spin J = 1 contributions, respectively, give a
time-space decomposition in a rest system of a massive particle q 2 = m2 > 0,
SO(3)
[ 12 | 12 ] ∼= [0] ⊕ [1], 14 = S + V, S ◦ V = 0,
∼ qc qa
S = Sa = q2
c q0 =m
=
1 0
, S.q = q, S ◦ S = S, Scc = 1,
0 03
V ∼
c q0 =m 0 0
= Vac = δac − q qq2a = 0 13 , V.q = 0, V ◦ V = V, Vcc = 3.
The projectors arise as products of the Minkowski boost representations [ 12 | 12 ],

parametrized by energy-momenta:
q
) ∼
q 1 0 qβ
Λ( m =m qα δαβ m +
qα qβ , α, β = 1, 2, 3,
m+q0
qc qa qc qa
⇒ Λcγ ( m
q q
)δ γα Λaα ( m ) = −η ca + m2 = −V ca , Λc0 ( m
q q
)Λa0 ( m )= m2 = S ca .
For higher-order Lorentz group representations, the projectors to rota-

tion group representations can be combined by S and V, e.g., the projector
S cd Scd = 1 from [0|0] to spin 0 and the projector P2 from [1|1] to spin
2, where the spin-0 contribution has to be subtracted from the symmetric
SO0 (1, 3)-combination V ∨ V = [1] ∨ [1] = [0] ⊕ [2]:
SO(3)
[1|1] ∼
= [0] ⊕ [1] ⊕ [2],⎧
ab = 3 V Vab ,
1 cd
⎪
⎪ (P0 )cd
⎪
⎨ Vac Vbd +Vad Vbc
(P2 )cd
ab = − 13 V cd Vab ,
19 = P0 + P1 + P2 , 2
Va Vb −Va Vb
⎪
c d d c
⎪
⎪ (P1 )cd = ,
⎩ ab 2
(PA )ab (PB )ab
cd
ef = δAB (PA )cd ef ,
1 γδ αβ

P0cd,ab 3δ δ
q q δγα δδβ +δγβ δδα q q
P2cd,ab = Λcγ ( m )Λdδ ( m ) 2 − 13 δ γδ δ αβ Λaα ( m )Λbβ ( m ).
P1cd,ab δγα δδβ −δγβ δδα
2
For a massive spin-1 and spin-2 particle the Lorentz group covariant
equations (with q ∼
= i∂ in the projectors) for the embedding fields Z and Y
involve the transversality conditions from V.q = 0:
Vac ∂ 2 Za = −m2Z Zc ⇒ ∂c Zc = 0,
(P2 )cd 2
ab ∂ Y
ab
= −m2Y Ycd ⇒ ∂c Ycd = 0.
The Z-boson mass, neglecting its width, is given in proton mass units by
m2Z
mp ∼ 97, log m2p ∼ 9.1.
mZ
In the Feynman propagators for the massive particles with normalizations

ρ(m2 ) = 0,
d4 q −V ca
0|Zc (y)Za (x)|0Feynman = ρ(m2Z ) πi (2π)3 q2 +io−m2Z e
iq(x−y)
,
d4 q (P2 )cd,ab
0|Ycd (y)Yab (x)|0Feynman = ρ(m2Y ) πi (2π)3 q2 +io−m2Y eiq(x−y) ,
the on-shell contribution from the Dirac distribution is a representation ma-

trix element of the Poincaré group for a massive particle with fixgroup SO(3)
for the energy-momenta with q 2 = m2 > 0. To avoid long range interactions
via massless fields, the on-shell projectors
c
with the mass c
are also used for the
off-shell contribution, e.g., Vac = δac − qmq2a , not δac − q qq2a .
Z
The projectors in the numerators, Vca for spin-1 particles and (P2 )ab
cd for
spin-2 particles, embed, in the respective forms −V ca and (P2 )cd,ab , the
5.6 Massive Particles with Spin 1 and Spin 2 125
Hilbert metric for the particle degrees of freedom. The Lorentz invariant
indefinite metric is used for the metrical tensor of the spin degrees of free-
dom (subindex ±). The vector representation has sign (−η) = (3, 1) and the
tensor representation sign η ∨ η = (7, 3):
⎧
⎨ −η = −1 0 ,
0 13
metric of [ 12 | 12 ] :
⎩ 1 1 SO(3)
[2|2] ∼ = [0]+ ⊕ [1]− ,
⎧ ⎛ 1 0 0 0
⎞
⎪
⎪
⎪
⎨ =⎝ ⎠.
0 1 0 0
η∨η 0 0 −13 0
metric of [ 12 | 12 ] ∨ [ 12 | 12 ] :
⎪
⎪ 0 0 0 15
⎪
⎩ SO(3)
[0|0] ⊕ [1|1] ∼
= 2 × [0]+ ⊕ [1]− ⊕ [2]+ .
They contain the definite metrical subtensors 11+2J , projected to for massive
13 , and for massive spin-2 particles, 15 , enclosed by double
spin-1 particles,
lines ×
.
In general, the metrical structure of the Lorentz group [J|J]-representa-
tion is given by the signature of indefinite orthogonal groups:

[J|J] SO0 (t, s), J = 0, 1, . . . ,
SO0 (1, 3) −−−→
SO0 (s, t), J = 12 , 32 , . . . ,

1+2J
with (t, s) = ( 2+2J

2 , 2 ), t + s = (1 + 2J)2 , t − s = 1 + 2J.

The maximal compact subgroup SO( 2+2J2 ) × SO( 1+2J2 ) contains, for in-

2J
teger J, the rotation SO(3)-representations [L] (positive definite) and
L=0,2,...

2J−1
[L] (negative definite), and, with opposite association, for half-integer J.
L=1,3,...
The spin-1 particle field has a harmonic expansion with translation and
rotation eigenoperators for energy-momenta q and spin 1 with three directions
in the quantization and Hilbert metric:
3
Za (x) = ⊕ 2q0d(2π) q a q
3 Λα ( m )[e u (q ) + e−iqx uα (q )],
iqx α
−
Z
[uα ( q )] = ρ(m2Z )2q0 δαβ δ( q2π
p ), uβ ( p
), with q0 = m2Z + q 2 ,
−
0|uα ( q )|0 = m2Z , 1;
p )uβ ( p, α|m⎪
2
Z , 1; q , β = ρ(m2Z )2q0 δ αβ δ( q2π p
),
⎪
⎪
distributive Hilbert basis: {|mZ , 1;
2
q , α⎪q ∈ R , α = 1, 2, 3}.
3
The spin-2 particle field has a similar harmonic expansion for the particles
with five spin directions:
3
Yab (x) = ⊕ 2q0d(2π)q a q b q
3 Λα ( m )Λβ ( m )
Y Y
γα δβ γβ δα
×( δ δ +δ δ
− 13 δ γδ δ αβ )[e⎪ u (q ) + e−iqx uγδ (q )],
iqx γδ
2
⎪
⎪
distributive Hilbert basis: {|mY , 2;
2
q, A⎪q ∈ R3 , A = ±2, ±1, 0}.
5.7 Massless Polarized Photons and Gravitons

With Wigner, particles — here, polarized massless photons and gravitons for
the electromagnetic and gravitational field, respectively — are defined for flat
Minkowski spacetime only.
Massless particles have no rest system. Projectors like V ∼
c
= δac − qqq2a for
spin 1 and S ∼
c
= q q2a for spin 0 are not defined for q 2 = 0. With the fixgroup
q
SO(2) × R2 ⊂ SO0 (1, 3) of the nontrivial energy-momenta with q 2 = 0, and
the fixgroup in the fixgroup SO(2) ⊂ SO(2) × R2 , massless particles have no
spin SO(3), they have only polarization SO(2) around the flight direction,
e.g., polarization (±1) for the photons and polarization (±2) for gravitons.
The four-component Lorentz vector for the electromagnetic field decomposes
into an SO(2)-dublet (±1) for the two polarized particle degrees of freedom
and two SO(2)-singlets for the Coulomb and gauge degree of freedom. The
10-component symmetric Lorentz tensor for the flat spacetime gravitative
field contains an irreducible Lorentz group nonet [1|1] and singlet [0|0] and
decomposes with respect to polarization SO(2) into two particle degrees of
freedom (±2) and eight nonparticle degrees of freedom as follows:
SO(2)
Aa : [ 12 | 12 ] ∼
= (±1) ⊕ 2 × (0),
SO(2)
Eab : [1|1] ⊕ [0|0] ∼
= (±2) ⊕ 2 × (±1) ⊕ 4 × (0).
The free equation for the electromagnetic vector field contains the ∂ 2 -
multiplied spin-1 projector,
∂a Fca = (δac ∂ 2 − ∂ c ∂a )Aa = Vac ∂ 2 Aa = 0,
whereas the free equation for the Einstein tensor field does not contain pure
projectors, but instead contains the following combination of the derivatives
(spin-2 and spin-0 projectors for massive particles):
Řcd
• = [δa δb ∂ − δb ∂ ∂a − δb ∂ ∂a − η ηab ∂ + η ∂b ∂a + ηab ∂ ∂ ]E
c d 2 d c c d cd 2 cd c d ab
Vac Vbd +Vad Vbc

=( 2 − V cd Vab )∂ 2 Eab
[(P2 )cdab − 2(P0 )ab ]∂ E
cd 2 ab
= = 0.
With the divergenceless antisymmetric field strengths and the divergence-
less Einstein tensor,
∂c ∂a Fca = 0,
∂c Řcd
• = 0,
the vector and tensor fields for massless particles are determined up to the
divergences of a scalar and a vector field, respectively:
Ac −→ Ac + ∂ c α ⇒ Fcb −→ Fcb ,
(
Ecd −
→ Ecd + ∂ c ξ d + ∂ d ξ c ,
• −→ Ř• .
⇒ Řcd cd
Γb −
cd
→ Γ cd c d
b + 2∂ ∂ ξb
5.7 Massless Polarized Photons and Gravitons 127
The electromagnetic current and the flat spacetime energy-momentum tensor

are conserved, ∂c Jc = 0 and ∂c Tcd = 0.
The freedom in the definition of the polarization (±1)-embedding Lorentz
vector field A is connected with the U(1)-gauge transformation of the electro-
magnetic field. The corresponding freedom in the definition of the polarization
(±2)-embedding Lorentz tensor field E in the tetrad expansion e = 1 + E
can be related to the reparametrizations of the spacetime manifold:
xb −→ xb (x) = xb + λξb (x), ecb = ∂ c xb = δbc + λ∂ c ξb .
These reparametrizations up to order λ are the remainder of the parametriza-
tion independence of Einstein’s gravity, i.e., of the homogeneous C(M)-
module transformations gil −→ eij gjk elk with the structural group eij ∈
GL(4, R) for the local frames (see Chapter 2).
The gauge transformation analogy between the electromagnetic potential
A and the metrical tensor g, or the tetrad e, is superficial: With respect
to the Lorentz group SO0 (1, 3) for orthogonal frames, not the tetrad e, but
the connection Γ is the analogue to the U(1)-connection, the electromag-
netic gauge field A, both with inhomogeneous transformation behavior (see
Chapter 6):
e
⏐
⏐
∂
⏐A ↔ Γ
⏐
⏐ ⏐
U(1) ∂ ∂ SO0 (1, 3).
F ↔ R•
The matter equations for linearized gravity are invariant up to order λ only
with a corresponding coordinate reparametrization, e.g., Φ (x) −→ Φ (x). A
full reparametrization invariance requires the full nonlinear theory.
The tetrad field can be considered as a Goldstone field for the rearrange-
ment of its 10 operations e ∈∈GL(4, R)/O(1, 3),
especially for the four met-
rical dilations (see Chapter 1). The 10 = 4 + 42 operations, dilations D(1)4 ,
and rotations SO(4), come in four O(1, 3)-vector fields (ela )a=0,1,2,3 with re-
spect to the ground-state Lorentz invariance for a flat spacetime metric η
(see Chapter 7).
The free-field equations can be reduced to Klein–Gordon equations
∂ 2Φ = 0 for the translation-invariant mass m2 = 0, where the Lorentz vector
properties of the scalar squared spacetime derivative ∂ 2 do not combine
with the Lorentz properties of the fields. To obtain such a decoupling for
electrodynamics, the divergence of the vector field has to vanish. For gravity,
harmonic coordinates are used, where the divergence of the Einstein tensor
vanishes:
for ∂c Ac = 0 ⇒
∂ 2 Ac = −g 2 Jc ,
dc
η cd • ∂ 2 (Ecd − η2 E• ) = −2 Tcd ,
for ∂c (E − 2 E ) = 0 ⇒
cd
dc
∂ 2 Ecd = −2 (Tcd − η2 T• ).
The Schwarzschild metric for energy-momentum tensor Tcd (x) = mδ0c δ0d δ
(x) is considered in this framework: A reflected “timelike” tensor is propor-
tional to the Euclidean δ cd , not to the Lorentz metric η cd :
η cd • δ cd •
Tcd = δ0c δ0d T• ⇒ Tcd − 2 T = 2 T .
The gravity field and the linearized metric for a static “timelike” energy-
momentum tensor ∂ 0 T• = 0 is given as follows:
2
∂ 2 Ecd (x) = −∂ 2 Ecd (x) = −
2 T• (x)δ cd
2 T• (
y)
⇒ Ecd = E00 δ cd with E00 (x) = −
2 d3 y 4π|x− y| ,
g = (1 + 2E )c dt − (1 − 2E )dx .
00 2 2 00 2
For a mass point, one obtains the linearized Schwarzschild metric and, anal-
ogously, the Kepler potential:
2
Tcd (x) = mδ0c δ0d δ(x) ⇒ Ecd (x) = −m 8πr

δ cd ,
g2 with − ∂ 2 4πr
1
= δ(x).
c
J (x) = zδ0 δ(x) ⇒ A (x) = z 4πr δ0c ,
c c
The propagators of massless vector and tensor fields in spacetime are, up

to gauge terms,
d4 q −ηca iq(x−y)
0|Ac (y)Aa (x)|0Feynman = ρA (0) πi (2π) 3 q 2 +io e ,
4 ηca ηdb +ηda ηcb −ηcd ηab
0|Ecd (y)Eab (x)|0Feynman = ρE (0) πi (2π)
d q
3
2
q2 +io eiq(x−y) .
They can be obtained from the projectors for massive spin-0, -1, and -2
fields by omitting the gauge-related contributions q⊗q
m2 in the vector projector
qc qa
Vc = δc − m2 . The projector inverse of the kinetic term in the free equation
a a
for the Einstein tensor field,
ab − 2(P0 )ab ]∂ E
[(P2 )cd cd 2 ab
= 0,
(P2 − 2P0 )(P2 − 12 P0 ) = P2 + P0 ,
gives the relevant combination [20]:
δca δdb +δda δcb −ηcd η ab
(P2 − 12 P0 )ab
cd = 2 + q⊗q
m2 -terms.
The nonrelativistic Kepler potential 1r can be embedded into massless

relativistic fields with any Lorentz group representation [J|J]: The Coulomb
and Newton potentials are the position projections — via time integration —
of the off-shell contributions (“virtual particles”) in the vector and tensor field
propagators,
d4 q 1 iqx =
d4 q 1 iqx =
d3 q 1 −iqx 1
dx0 iπ(2π) 3 −q 2 −io e dx0 iπ(2π) 3 −q 2 e q2 e
4iπ 3 = 2iπr ,
P
which is, up to −2i, the inverse of the invariant spacetime Laplacian (see
Chapter 10),
d4 q
∂ 2 iπ(2π) 1
3 −q 2 −io e
iqx
= −2iδ(x).
5.7 Massless Polarized Photons and Gravitons 129
There are no projectors from a four-component Lorentz vector field Ac

to the two polarized massless photons, which can be expressed with their
energy-momenta, and, similarily, no projectors from a 10-component sym-
metric Lorentz tensor field Ecd to the two polarized massless gravitons. The
q
boost representations Λ( m ) for a transformation to a rest system are not de-
fined for m = 0. The embedding of the two particle degrees of freedom with
polarization into Lorentz group-compatible fields with four and 10 compo-
nents, respectively, involves two and eight additional nonparticle components.
For the four-component electromagnetic field, the propagator involves the
four-dimensional unit
[ 12 | 12 ], 14 = 1[ 12 | 12 ] ∼
= δca .
The indefinite metric −η ca contains a definite Hilbert submetric 12 (lower
right corner in the metrical matrix) for the two polarized particle components
(±1) (subindex ± gives the metric for −η):
⎧ −1 0 0
⎪
⎨ −η = 0 1 0 ,
metric of [ 12 | 12 ] : 0 0 12
⎪
⎩ 1 1 SO(2)
[2|2] ∼
= (0)− ⊕ (0)+ ⊕ (±1)+ .

The other two components constitute a Witt pair −1 0 ∼
=
0 1
0 1 1 0
with neutral signature for the Coulomb and gauge degree of freedom, both
nonparticlelike.
For the 10-component gravity field, the 10-dimensional symmetric tensor
unit can be decomposed into the two harmonic units (see Chapter 8) for the
irreducible Lorentz group representations, 10 = 9 + 1:
[ 12 | 12 ] ∨ [ 12 | 12 ] = [1|1] ⊕ [0|0],
14 ∨ 14 = 1[1|1] + 1[0|0] ∼
δ a δ b +δ a δ b ηcd η ab ηcd η ab
= [ c d2 d c − 4 ] + 4 .
The graviton propagator comes with a projector combination:
δca δdb +δda δcb −ηcd η ab
2 = (1[1|1] − 1[0|0] )ab
cd .
1[1|1] gives the pair of polarized gravitons (±2) with definite metric 12 (lower
right corner in the metrical matrix). There is a triplet of Witt pairs with
neutral signature (13 , −13 ). The remaining degree of freedom with metric 1
is paired with the one degree of freedom in −1[0|0] with metric −1:
⎧ ⎛ −13 0 0 0
⎞
⎪
⎪
⎪
⎨ 0 1 0 0
(η ∨ η)[1|1] = ⎝ 0 0 13 0
⎠,
metric of [1|1] :
⎪
⎪ 0 0 0 12
⎪
⎩ SO(2)
[1|1] ∼
= (0)−⊕ 2 × (0)+⊕ (±1)−⊕ (±1)+⊕ (±2)+ ,

−(η ∨ η)[0|0] = −1,
metric of [0|0] : SO(2)
[0|0] ∼
= (0)− .
In analogy to the
one neutral
pair for electrodynamics,
the four neutral Witt
pairs in gravity −η 0 ∼
=
0 η
for η = 1 0
, with the Newton
0 η η 0 0 −13
and gauge degrees of freedom, are nonparticlelike.
The Lorentz group representations for flat spacetime gravity constitute
an irreducible real 10-dimensional SL0 (R4 )-representation (see Chapter 7),
which is considered with a Lorentz invariant metric of signature (6,4):
SO0 (1,3)
∼
) [2, 0, 0] ⊕
4
irrep SL0 (R = [0|0] [1|1],
−14 0 ∼
= (−1) ⊕ −13 0
.
0 16 0 16
For long-range interactions in flat spacetime, only two Lorentz group

representations are used — the vector representation [ 12 | 12 ], e.g., for the elec-
tromagnetic field, and the tensor with scalar representation [1|1] ⊕ [0|0] for
the gravitational field, coming, respectively, with one pair of massless pho-
tons of polarization (±1)+ and one Witt pair (0)± and, still to be detected,
one pair of massless gravitons (±2)+ and four Witt pairs (±1)± , 2 × (0)± . So
far, no long-range interactions have been found with harmonic Lorentz group
representations [J|J] for J ≥ 32 , coming together with a massless particle pair
with polarizations (±2J) and (1 + 2J)2 − 2 non-particle modes:
SO(3)
2J SO(2)
[J|J] ∼
= [L] ∼
= (2J + 1) × (0) ⊕ 2J × (±1) ⊕ . . . ⊕ (±2J),
L=0
SO(2)
with [L] ∼
= (0) ⊕ (±1) ⊕ . . . ⊕ (±L).
5.8 Quantum Gauge Fields

The extension of global “phase” transformations eiα ∈ U(1) to spacetime-
dependent local “gauge” transformations {x −→ eiα(x) } ∈ U(1)R is not
4
enough for quantum gauge structures. Gauge transformations for quantum

fields arise in the form of Becchi–Rouet–Stora (BRS)transformations, the
gauge parameters are implemented by quantum fields: the Lorentz scalar
Fadeev–Popov fields with Fermi statistics. The origin of BRS transformations
for four-component massless vector fields lies in the two SO(2)-scalar non-
particle degrees of freedom (Witt pair with neutral signature), which arise in
addition to the two polarized particle degrees of freedom in an SO(2)-dublet.
BRS invariance goes with the projection to the particle degrees of freedom.
BRS nontrivial degrees of freedom are without particle content and can have
equations of motion with “usual,” i.e., not covariant, derivatives.
For a quantum formulation of a gauge field theory, demonstrated for the
simplest abelian case, i.e., for quantum electrodynamics with a massless vec-
tor field A, the spacetime dependence of a classical gauge transformation
with parameter α as invariance of the Lagrangian
5.8 Quantum Gauge Fields 131
ab
2 F Fab
L(Aa , Fab ) = 12 Fab cd
ab ∂c Ad + g 4 ,
A −→ A + ∂ α, Fab −→ Fab ,
a a a
is drastically reduced. A quantum gauge theory has a duality-completing

scalar field (“gauge-fixing” field) S. There remains a transformation with a
massless “Lie parameter field” β :
ab 2
L(Aa , Fab , S) = 12 Fab cd a
ab ∂c Ad + S∂a A + g
2 F Fab
4 − g 2 λ S2 ,
A −→ A + ∂ β , Fab −→ Fab , S −→ S, with ∂ 2β = 0.
a a a
5.8.1 Fadeev–Popov Ghosts in Quantum Mechanics

The “gauge-fixing” part of the dynamics with the gauge transformations,
⎧
2
⎨ ∂a Aa = g 2 λS, ∂a S = 0,
L(A, S) = S∂a A − g λ 2 ,
a 2 S A −→ Aa + γ a , S −→ S,
a
⎩
∂ aβ = γ a , ∂aγ a = 0,
is the Lorentz compatible spacetime distribution of the noncompact time

development for a free mass point:
⎧
2
⎨ dt x = p, dt p = 0,
L(x, p) = pdt x − p2 , x −→ x + γ , p −→ p,
⎩
dtβ = γ , dtγ = 0.
The gauge transformation is the relativistic distribution of a position trans-

lation transformation for the free mass point position.
A noncompact time development has translation eigenvectors and nilvec-
tors. The subspace built by the eigenvectors has a trivial eigenvalue (nildimen-
sion) for the action of the nilpotent part of the Hamiltonian. In the self-dual
vector space V ∼ = R2 , spanned by position and momentum, the Hamiltonian
matrix of the free mass point, a linear 2 × 2 transformation, is nilquadratic:

x∼=x= 1 , p∼
0
=p= 0 ,
1
⎧
p2 ∼ 0 1 ⎪
⎨ dt x = h.x = p,
HB = 2 = h = 0 0 ,
⇒ dt p = h.p=0,
h ◦ h = 0, eht = 10 1t ⎪
⎩ x(t) = eht x = x + tp .
p(t) p p
In a Bose quantum algebra, generated by position and momentum, the

Hamiltonian is not nilquadratic with respect to the quantum product:
2
[ip, x] = 1, HB = p2 , ⇒ dt x = [iHB , x] = p,
but HB = 0
2 d t p = [iHB , p] = 0.
By introducing additional Fermi degrees of freedom as partners for the Bose

position-momentum pair, it is possible to construct nontrivial nilquadratic
quantum operators. To formulate the distinction between translation

eigenvectors and nilvectors (particle- and nonparticle-interpretable), a quan-
tum gauge theory has a Bose–Fermi twin structure: The spinless part of
the gauge Bose field and its “gauge-fixing” dual partner are accompanied by
Fadeev–Popov fields as their Fermi counterparts, whose classical limits are
the spacetime-dependent Lie parameters of the gauge group.
The Bose–Fermi twin structure is discussed first in a nonrelativistic quan-
tum-mechanical model: A noncompact time development for the additional
Fermi degrees of freedom needs two dual pairs:
Bose: [ip, x] = 1, Fermi: {β

β , γ̌γ } = 1 = {γγ , β̌
β },
2
Hamiltonian: HB+F = HB + HF = p2 + iγ̌γ γ .
The equations of motion are

⎧
⎪
⎪ dtβ = [iHB+F , β ] = γ,
⎨
dt x = [iHB+F , x] = p, dtγ = [iHB+F , γ ] = 0,
Bose: Fermi:
dt p = [iHB+F , p] = 0, ⎪ dtβ̌
⎪ β β]
= [iHB+F , β̌ = −γ̌γ ,
⎩
dtγ̌γ = [iHB+F , γ̌γ ] = 0.
They can be derived from a classical Lagrangian (first- or second-order time

derivatives):
2
L(x, p, β , γ ) = pdt x − p2 + iγγ dtβ̌
β + iγ̌γ dt β − iγ̌γ γ ,
1 2
L(x, β ) = 2 (dt x) + i(dtβ )(dtβ̌ β ).
The nilquadratic Becchi–Rouet–Stora charge NBF implementing the gauge

transformation x −→ x + δx with δx = γ is given by the time development
invariant
NBF = γ p ⇒ NBF 2
= 0, [HB+F , NBF ] = 0.
Its linear hybrid adjoint action in a hybrid algebra generated by Bose and
Fermi vectors,

[a, b] ⇐⇒ a or b is Bose,
[[a, b]] =
{a, b} ⇐⇒ a and b are Fermi,
defines the BRS transformations:

⎧
⎪
⎪ δββ = {iNBF , β } = 0,
⎨ δγγ = {iNBF , γ } = 0,
δx = [iNBF , x] = γ ,
Bose: Fermi:
δp = [iNBF , p] = 0, ⎪
⎪ β
δβ̌ β}
= {iNBF , β̌ = ip,
⎩
δγ̌γ = {iNBF , γ̌γ } = 0.
With the Fadeev–Popov number operator for the Fermi degrees of freedom,

[iP, β ] = β , [iP, γ̌γ ] = −γ̌γ ,
βγ ) ⇒
P = i(γ̌γ β + β̌
β ] = −β̌
[iP, β̌ β , [iP, γ ] = γ ,
[P, HB+F ] = 0,
the space Q(HB+F ) spanned by the eigenvectors of the Hamiltonian is defined

by trivial eigenvalues both for the BRS charge NBF and the Fadeev–Popov
number P :
⎪
⎪
Q(HB+F ) = {a⎪ ⎪[[NBF , a]] = 0 and [P, a] = 0}.
5.8.2 Fadeev–Popov Ghosts for Quantum Gauge Fields

The Lorentz-compatible distribution of the nonrelativistic model for the elec-
tromagnetic quantum gauge field,
ab
L(A, F, S, β , γ ) = Fab ∂ A −∂
a b b a 2
2
A
+ S∂a Aa + g 2 ( Fab4F − λ S2 )
β + iγ̌γ a ∂aβ − ig 2 λγ̌γ aγ a ,
+ iγγ a ∂aβ̌ ⎧ a
⎧ b a ⎪ ∂ β = g 2 λγγ a ,
⎨ ∂ A − ∂ a Ab = g 2 Fab , ⎪
⎨ ∂ γa
a = 0,
Bose: ∂a Aa = g 2 λS, Fermi:
⎩ ⎪
⎪ ∂ aβ
β̌ = −g 2 λγ̌γ a ,
∂ Fab − ∂a S = 0,
b
⎩
γa
∂aγ̌ = 0,
β , γ a , γ̌γ a ) with Fermi quantization:

β , β̌
uses Lorentz scalar Fadeev–Popov fields (β
[iS, Aa ](x) = {β
β , γ̌γ a }(x) = {γγ a , β̌
β }(x) = δ0a δ(x).
A second-order derivative Lagrangian reads

L(A, β ) = − 4g12 (∂ a Ab − ∂ b Aa )(∂a Ab − ∂b Aa ) + 1
2g 2 λ
(∂a Aa )2 + i
g2 λ
(∂ aβ )(∂aβ̌
β ).
The hybrid adjoint action of the nilquadratic linear BRS charge generates
the linear BRS transformations
⎧
⎨ δAa = [iNBF , Aa ] = g 2 λδ0aγ 0 ,
Bose: δS = [iNBF , S] = 0,
⎩
δF ab = [iNBF , Fab ] = 0,
NBF = g 2 λ d3 x γ 0 (x)S(x), ⎧
⇒
2
NBF =0 ⎪
⎪ δββ = {iNBF , β } = 0,
⎨ δγγ a = {iN , γ a } = 0,
BF
Fermi:
⎪
⎪ δ β
β̌ = {iN β } = ig 2 λS,
BF , β̌
⎩
δγ̌γ = {iNBF , γ̌γ a } = 0.
a
The subspace with the particle-interpretable degrees of freedom, i.e., with-

out nilvectors, is characterized by a trivial BRS charge and a trivial Faddev–
Popov number:

P = d3 x P0 (x), Pa = i(γ̌γ aβ + β̌
β γ a ).
“Gauge-invariant” fields are harmonically analyzable with translation eigen-

vectors only.
The spinless and “gauge-fixing” Bose degrees of freedom and the Fermi
Fadeev–Popov ones display a twin structure. The BRS current Nb (x) of
the Fermi type has its counterpart in the nonderivative part H(x) of the
Lagrangian (Bose type):
2
Na = g 2 λγγ a S, HB+F = g 2 λ( S2 + iγ̌γ aγ a ).
The dynamics HB+F in the mass point model arises by a BRS transfor-
mation from an operator K connecting Bose and Fermi degrees of freedom:
2
NBF = γ p, HB+F = p2 + iγ̌γ γ ,
HB+F = {NBF , K}, K = β̌β2p + γ̌γ x.
2
Since NBF = 0, the BRS invariance of the Hamiltonian is obvious:
[NBF , HB+F ] = [NBF , {NBF , K}] = [NBF

2
, K] = 0.
The corresponding relativistic field operators arise from the position

distributions:
βS
β̌

K= 2
γ a Aa , (HB+F , NBF , K) =
+ γ̌ d3 x (H, N0 , K)(x).
In parallel to the equations of motion, the BRS transformations for non-

abelian groups, e.g., for a massless SU(2)–Yang–Mills triplet vector field, also
contain nonlinear contributions.
5.8.3 Particle Analysis of Massless Vector Fields

The harmonic analysis (see Chapter 4) of a massless vector field contains
four momentum operators, two with a time representation in U(12 ) (first and
second components) and two with a time representation in U(1, 1) (zeroth
and third components):
⎛ iqx ⎞
e [B( q |x0 G(
q ) + iν| q )] + (1 − ν)e−iqx G× (
q)
⎝ −iqx
⎠,
3 iqx 1
Aa (x) = ⊕ 2|qd|(2π)
q a q
3 Ow ( |
q| )
e u (q)+e u1 (
q ) + e−iqx u (
eiqx u2 (
q)
q)2
q ) + e−iqx [B× (
(1 − ν)eiqx G( q |x0 G× (
q ) − iν| q )]
with q0 = |
q | and λ = 1 − 2ν.
The transition from the Lorentz group SO0 (1, 3) to the axial rotation fixgroup
SO(2) uses the transmutator (see Chapter 7):
1 0

Ow ( |qq | ) = 0 O( q ) ◦ w ∈∈SO0 (1, 3)/SO(2),
|
q|
qA qB qA

δ AB − |
q

O( |q | ) = q |(|
qB
− |
q |+q3 ) |q|
q3 ∈∈SO(3)/SO(2),
q|
|q |
0 0 1 −1 0 0 1 0 1
0 12 0 =w◦ 0 1 2 0 ◦ w , with w = 2
T √1 0 1 2 0 .
1 0 0 0 0 1 −1 0 1
The massless vector fields involve an SO(2)-polarized particle pair (left

and right polarized photons) and an SO(2)-trivial pair without particle
interpretation:
for Aa : R4 × SO(2) −→ U(1, 1) × U(2)

U(1, 1) × U(2) ⊂ U(1, 3) ⊃ SO0 (1, 3).
The first and second components of the massless field with nontrivial polariza-
tion around the momentum q carry two irreducible U(1) time representations
with energy q0 = |q |. They constitute a harmonic U(2)-oscillator:
⎧
⎪
⎪ q |x0 −i
1 0
⎪
⎪ R 4
−→ U(1 2 ) e i| qx

,
⎪
⎪ 0 1
⎨
q , x0 ) = ei|q |x0 uA (q ),
uA (
A, B ∈ {1, 2} :
⎪
⎪
⎪
⎪ R4 × SO(2) −→ U(12 ) ◦ SU(2) = U(2),
⎪
⎪
⎩ −
p ), uB (
[uA ( q )] = g 2 2|q |δAB
δ( q2πp
).
The zeroth and third components with trivial polarization are connected in
a reducible, but decomposable faithful time representation in the indefinite
unitary group U(1, 1) with energy q0 = | q | and nilconstant ν|q | involving the
“gauge fixing” constant 2ν = 1 − λ (see Chapter 4):
⎧
⎪
⎪ 1 iν| q |x0
⎪
⎪ R 4
−→ U(1, 1) ei|q |x0 −iqx ,
⎪
⎪ 0 1
⎨
B(q , x0 ) = ei|q |x0 [B(q ) + iν|q |x0 G(q )],
a, b ∈ {0, 3} :
⎪
⎪
⎪
⎪ G( q , x0 ) = ei|q |x0 G(q ),
⎪
⎪
⎩ × −
[B (
p ), G( q )] = [G× ( p ), B(q )] = g 2 2|q |δ( q2π p
).
A U(1)-time development with particle interpretation has a Fock state:
0, 1; q, B = 0|uA (
p, A|0, 1; q )|0 = 0|{uA (
p )uB ( p ), uB (q )}|0
−
q |δ AB δ( q2π
= g 2 2| p
).
A Fock form also for

the U(1,1)-time
representations
leads to an indefinite
B|B B|G 0 1 ∼ −1 0
metric G|B G|G ∼ 1 0 = 0 1 :
0|B× ( q )|0 = 0|{B× (

p )G( p ), G(
q )}|0
× −
= 0|G ( q )|0 = 0|{G× (
p )B( q )}|0 = g 2 2|q |δ( q2π
p ), B( p
).
It gives a Fock value for the anticommutator and the Feynman propagator:
d4 q
0|{Ac , Aa }(x)|0 = g 2 (2π)3 (−η
ca
− 2νq c q a ∂q∂ 2 )δ(q 2 )eiqx ,
d4 q + −ηca qc qa
,
0|{Ac , Aa }(x) − (x0 )[Ac , Aa ](x)|0 = g 2 πi (2π)3 q 2 +io + 2ν (q 2 +io)2 e
iqx
.
5.9 Hilbert Representations

of the Poincaré Group
As initiated by Wigner, the Hilbert representations of the semidirect product
Poincaré group SO0 (1, 3) × R4 , or of its cover SL(2, C) × R4 , can be induced
from Hilbert representations of direct product subgroups where the fixgroups
SL(2, C)q of spacetime translations or, equivalently, of energy-momenta, are
relevant.
There are four fixgroup types: for the trivial translation, and for time-,
space- and lightlike translations, e.g., for q = 12 , σ 3 , 12 + σ 3 , respectively,

q = qa σ a = q0 12 + qσ = qq10 ++ iqq32 qq10 −− iqq32 , det q = q 2 = q02 − q 2 ,
⎪
⎪
SL(2, C)q = {s ⎪ ∗
⎧ ∈ SL(2, C)⎪s ◦ q ◦ s = q}
⎪
⎪ SL(2, C) ∼ SO 0 (1, 3), q = 0,
⎨
∼ SU(2) ∼ SO(3), q 2 > 0,
=
⎪
⎪ SU(1, 1) ∼ SO 0 (1, 2), q 2 < 0,
⎩
SO(2) × R2 , q = 0, q 2 = 0.
The representations, always Hilbert in the following, with nontrivial

translation representations are induced from the representations of direct
products:
⎧ 2
⎨ q >0: SU(2) × Rt ,
q = 0 : SL(2, C), q = 0 : q2 < 0 : SU(1, 1) × Rs ,
⎩ 2
q = 0 : [SO(2) × R2 ] × Rl .
The representations of the one-dimensional time-, space-, and lightlike trans-

lations Rt,s,l ∼
= R −→ U(1) can by written with four-dimensional ones, R4 ,
imt ∼ iqx
e.g., e = e with q 2 = m2 > 0. The representations of the Poincaré
group integrate the translation representations over the corresponding three-
dimensional energy-momentum orbits:
q2 > 0 : SL(2, C)/SU(2) ∼ = Y3 (energylike hyperboloid),

2
q <0: SL(2, C)/SU(1, 1) ∼
= Y (1,2) (momentumlike hyperboloid),
2
q = 0, q = 0 : SL(2, C)/ [SO(2) ×
R ] (pointed future lightcone).
2
The representations of SU(2) × Rt are used for massive particles; the

irreducible ones are characterized by the invariants (m2 , J) with mass m2 > 0
for the translations and spin 2J ∈ N for the rotations.
The representations of SU(1, 1) × Rs have a momentumlike translation
invariant x −→ eiqx , q 2 < 0. The representations of the homogeneous group
SO0 (1, 2) ∼ SU(1, 1) ∼ SL0 (R2 ) (locally isomorphic) were first given by
Bargmann and are superficially described in the following: SL0 (R2 ) has two
types of Cartan subgroups (see Chapter 8), compact SO(2) and noncompact
SO0 (1, 1), whose representations characterize, respectively, the discrete series
5.9 Hilbert Representations of the Poincaré Group 137
representations with integer invariant (winding number) and the principal

series representations with imaginary continuous invariant iQ ∈ iR+ . In ad-
dition, there is a supplementary series with real invariant 0 < I < 1. Its
representations are nonamenable, i.e., with trivial Plancherel measure. So
far, the representations of the Poincaré group, induced from SL0 (R2 ) × Rs ,
have not been used to describe experiments.
The representations of [SO(2) × R2 ] × Rl have lightlike translation invari-
ant x −→ e , q = 0, q = 0. The representation of the semidirect fixgroup,
iqx 2
a subgroup of the Galilei group, isomorphic to the Euclidean group in two

3 1 2
dimensions (see Chapter 8), eiϕ3 σ +ψ1 σ +ψ2 σ ∈ SO(2) × R2 , with one ro-
tation and two boosts can be induced from subgroup representations with
boost fixgroups:
ψ = ψ1 σ 1 + ψ2 σ 2 ,
⎪
⎪ SO(2), ψ = 0,
⎪ ∗
SO(2)ψ = {o ∈ SO(2)⎪o ◦ ψ ◦ o = ψ} =
{1}, ψ = 0.
The representations with nontrivial boosts start from trivial homogeneous
group representations with a continuous invariant for the boosts Rl ⊂ R2 ×
Rl ⊂ SL(2, C) × R4 . So far, they were not necessary in particle physics. The
fixgroup SO(2)-representations for trivial boosts are characterized by integer
invariant pairs (polarization) (±2J). They are used for massless particles.
Also, the representations of the Lorentz group SL(2, C) as fixgroup for
trivial energy-momenta have not been used for free particles. SL(2, C) has
one type of Cartan subgroup SO(2)×SO0 (1, 1), which leads to the character-
ization of the irreducible representations, as given by Bargman and Gel’fand,
by one integer “compact” invariant 2J ∈ N, and by a “noncompact” continu-
ous imaginary one for the principal series and, for the supplementary series
(nonamenable), by a trivial “compact” invariant and a real continuous one
0 < I < 1.
There are altogether eight types of Hilbert representations of the Poincaré
group, two for fixgroup SL(2, C), one for fixgroup SU(2), three for fixgroup
SU(1, 1), and two for fixgroup SO(2) × R2 :
Translation Translation Homogeneous group
invariant fixgroup invariants
SO(2) × SO0 (1, 1) : N × iR+ (principal)
−−− SL(2, C)
{0}×]0, 1[ (supplementary)
m2 > 0 SU(2) SO(2) : N
SO(2) : N (discrete)
2
m <0 SL0 (R2 ) SO0 (1, 1) : iR+ (principal)
]0, 1[ (supplementary)
SO(2) : N
m2 = 0 SO(2) ×
R2
R: iR+
Hilbert representations of the Poincaré group

SL(2,C) ×
R4 ∼ SO0 (1,3) ×
R4
The real three-dimensional fixgroups for nontrivial translations with rank-1

Lie algebras and one generating invariant, related either to compact SO(2)
or to noncompact SO0 (1, 1) ∼ = R, are subgroups of the real six-dimensional

SL(2, C) with rank-2 Lie algebra and two generating invariants (2J, iQ) ∈
N × iR+ , related to SO(2) × SO0 (1, 1). They have corresponding subsets of
invariants for their representations.
The Wigner classification of the representations of the Poincaré group
with the Minkowski translations as the tangent group of the spacetime
manifold gives no hintas to why only two types are used for flat spacetime,
those for free particles, with both causal translation invariants m2 ≥ 0 and
compact fixgroups,
m2 > 0 : SU(2) × Rt ⊂ SL(2, C) × R4 ,

m = 0 : SO(2) × Rl ⊂ [SO(2) ×
2 R ] × Rl ⊂ SL(2, C) ×
2 R4 .
Flat spacetime theory also cannot explain the particle spectrum, i.e., for
these two types, which invariants occur, i.e., which masses and spins or
polarizations.
5.10 Normalizations and Coupling Constants

Flat spacetime fields with particles are acted on by irreducible Hilbert re-
presentations of the Poincaré cover group SL(2, C) × R4 , which are induced
by representations of SU(2) × R for massive particles and SO(2) × R4 for
4
massless particles.
Coefficients of Hilbert representations for massive particles m2 > 0 have a
spectral decomposition into irreducible components for SU(2) with invariant
spin J and, for fixed J, a Lehmann–Källen decomposition for the translations
with invariant mass m2 , normalized by ρ(m2 ):
∞ d4 q
J = 0, 12 , 1, · · · : 0
dm2 ρ(m2 ) (2π)3
q
ζJ ( m )δ(q 2 − m2 )eiqx .
q
In a rest system, q = (m, 0, 0, 0), the polynomials ζJ ( m ) are spin-SU(2)
units, used as sesquilinear forms for the Hilbert metric of the spin degrees of
freedom:
ζJ (1, 0, 0, 0) = 11+2J .
The SU(2)-embedding polynomials involve real linear combinations of har-
monic Lorentz group polynomials (q)L with maximal degree L = 2J,
⎧ $
⎪
⎪ α2J q L
L ( m ) , J = 0, 1, . . . ,
⎨
q
ζJ ( m )= $
L=0,2,...,2J
α2J
2J = 1,
⎪
⎪ α2J q L
( ) , J = 1 3
, , . . . ,
⎩ L m 2 2
L=1,3,...,2J ⎧
⎪
⎪ (q)0 = 1, ⎪
⎨ 1 ⎪
(q) = q = {qa ⎪ ⎪a = 0, 1, 2, 3},
harmonic SO0 (1, 3)-polynomials:
⎪
⎪ (q) 2
= (q ∨ q) = {qa qb − η4ab q 2 },
⎩ 3
(q) = {qa qb qc − ηab qc +ηac4qb +ηbc qa q 2 }, . . . ,
5.10 Normalizations and Coupling Constants 139
1
for example, for spin 2 with Weyl matrices and for spin 1:
q
ζ 12 ( m ) = σ a qma → σ 0 = 12 ∼
= δAB ,
q2 ∼ q 2
q
ζ1 ( m ) = m2 − ηab m2 = ( m ) − 34 ( m
qa qb
) → 13 ∼
q 0
= δαβ .
The harmonic polynomials are totally symmetric products of the left and
right Weyl representations [ 12 |0] and [0| 12 ] of the boosts (Weyl transmutators;
see Chapter 7):
-
L -
L
(m ) = s∗ ( m
q L q
) ⊗ s( m q
),
+ ,
q1 − iq2
q
s( m ) = q0 +m
2m
σq
12 + q0 +m
= √ 1 q0 + q3 + m
q + iq q − q + m ,
+ , 2m(q0 +m) 1 2 0 3
q
ŝ( m ) = q02m+m
12 − q0σ+m
q

, s(1, 0, 0, 0) = 12 , ŝ(1, 0, 0, 0) = 12 .
They are acted on by the SO0 (1, 3)-representations [ L2 | L2 ] (see Chapter 8).
Feynman propagators, as used for the expansion of interactions with par-
ticle fields, also include the imaginary principal value off-shell part:
∞ d4 q q
ζJ ( m )
J = 0, 12 , 1, · · · : 2 2 i
0 dm ρ(m ) π (2π)3 q2 +io−m2 e
iqx
.
Representations of stable particles with different masses are Schur-
orthogonal (see Chapter 8), as illustrated for scalar particles with the
divergent volume of the mass hyperboloid,
d4 q
|m2 }(x) = (2π) 3 δ(q − m )e
2 2 iqx
, |Y 3 | = d4 q δ(q 2 − 1) = ∞,
|Y 3 | m21 −m22
{m21 |m22 } = d4 x |m21 }(x) |m22 }(x) = (2π) 2 δ( m2
).
1
Relativistic SL(2, C)-fields for massless particles with SO(2)-polarization

pairs (±L3 ) contain, with the exception of scalar fields L3 = 0 and spinor
fields L3 = 12 , nonparticle degrees of freedom. The Feynman propagators are
2J
linear combinations with harmonic Lorentz group polynomials and β2J = 1:
4 $ (q) L
|L3 | = J = 0, 1, · · · : ρJ (0) πi (2π)
d q 1
3 q 2 +io
2J
βL 2
L e
iqx
,
(q +io) 2
4 $
L=0,2,...,2J
(q)L
|L3 | = J = 12 , 32 , · · · : ρJ (0) πi d q
(2π)3
1
q2 +io βL2J
L−1 eiqx .
(q2 +io) 2
L=1,3,...,2J
Only the q 2 -independent contributions give a Dirac distribution δ(q 2 )

(on-shell), e.g., for J = 12 and J = 1,
∼ ∼
1 2
i (q) (q)
π q2 +io = qa δ(q 2 ) + . . . , i
π (q2 +io)2 = − η4ca δ(q 2 ) + . . . .
In the Feynman propagators, there are also “gauge-dependent” contributions,
e.g., for the vector field with ν ∈ R,
d4 q −ηca +2ν qq2c+io
qa
ρ1 (0) πi (2π) 3 2
q +io eiqx .
The normalizations of the Poincaré group representations are the residues
of the Fourier-transformed Feynman propagators at the particles masses
q 2 = m2 in the complex q 2 -plane, e.g., for a massive Dirac spinor (with

left and right components) and a vector field or a massless Weyl spinor, a
vector, and a tensor field:
⎧ ρ(m2 )
⎪
⎪ 0|Ψ
Ψ (0)Ψ
Ψ(x)|0Feynman = i
(γq + m),
⎪
⎪
π q 2 +io−m2
⎪
⎪ 0|Zc (0)Za (x)|0Feynman = i ρ(m2Z)
( m2 − ηca ),
qc qa
⎪
⎨ π q 2 +io−m2
d4 q −iqx l
Z Z
2π
e
⎪ 0|l(0)l∗ (x)|0Feynman = i ρ (0)
2 σq,
⎪
⎪
π q +io
⎪
A
⎪
⎪ 0|Ac (0)Aa (x)|0Feynman = i ρ (0)
(−ηca + . . .),
⎪
⎩
π q 2 +io
E
ηca ηdb +ηda ηcb −ηcd ηab
0|Ecd (0)Eab (x)|0Feynman = i ρ (0)
π q 2 +io
( 2
+. . .).
For free fields (flat spacetime) with the on-shell contributions ρ(m2 )δ(q 2 −m2 )
for Poincaré group representation coefficients, the normalizations are free. For
interactions, as mediated by the causally supported off-shell contributions
ρ(m2 )
2 −m2 , the normalizations give the coupling constants.
qP
For a massive field, an expansion for small energy-momenta q = 0, e.g.,
ρ(m2Z )
0|Zc (0)Za (x)|0Feynman = 2i m2Z
ηca δ(x) + ...,
gives the local approximation of a field-mediated interaction as used in the

ρ(m2 )
pointlike description of weak interactions with coupling constant m2Z in-
Z
volving the inverse particle mass.
The normalization of the electromagnetic interaction is given by 4παS
with the dimensionless Sommerfeld constant. The Newton constant for the
gravitational interaction has the dimension of an area 2 = 8πGc3 or the
inverse Planck mass squared:
ρA (0) g2
A A ∼ q2 = q2 = 4παS
q2 ,
ρE (0)
2
E E ∼ q2 = q2 = 8π
m2P q2
.
With a mass unit M , dimensionless quantities (underlined) can be used,

leading to dimensionless Feynman propagators with dimensionless normal-
izations and coupling constants:
⎧
⎪
A
x
x =M , ⎪
⎨ A A ∼ ρ q(0) 2 , ρA (0) = g 2 ,
4
d4 q d q
q = M q, q2 = M 2 q2 , ⇒
⎪
⎪
⎩ E E ∼ ρ (0)
E
E 8πM 2
A = M A, E = E q 2 , ρ (0) = m2 . P
5.11 Renormalization of Gauge Fields

Quantum field theories for flat spacetime start from an action as the sum
of the free-field part and the interaction. Flat spacetime operations can-
not describe interactions. This shows up in a perturbative expansion of an
5.11 Renormalization of Gauge Fields 141
interaction with flat spacetime quantum fields. By giving up to calculate, e.g.,

the mass ratios and the coupling constants, the “divergencies” can be tamed
in “renormalizable” theories by the familiar regularization-renormalization
procedure, e.g., for quantum electrodynamics. This is no longer possible for
“nonrenormalizable” quantum gravity in the flat spacetime approach.
5.11.1 Perturbative Corrections of Normalizations

In a perturbative expansion of an interacting theory, the coupling constants
are renormalized as exemplified by the electromagnetic interaction Ja Aa : The
field equation for the gauge field,
1 2
g2 (−∂ Aa + ∂a ∂ b Ab ) = Ja ,
contains the current as position density of the U(1)-generator. The normal-

ization of the current-carrying fields, e.g., a massive Dirac field with charge
number z (electron-positron with z = ±1),
d4 q γa qa +m iqx
0|Ψ
Ψ(0)ΨΨ(x)|0Feynman = ρ(m2 ) πi (2π) 3 q 2 +io−m2 e ,
/
{Ψ / 2
Ψ(0), Ψ (x)} x =0 = ρ(m )γ0 δ(x),
0
a Ψ Ψ(x)|0Feynman = −2iρ(m2 )δ(x),
(iγa ∂ + m)0|Ψ (0)Ψ
has to be compatible with the U(1)-generator properties; i.e., the current has
to involve the inverse
normalization factor, e.g., a position translation Dirac
normalization d3 x δ(x) with ρ(m2 ) = 1:
⎧
⎨ Ψ −→ e Ψ ,
izα
3
Q = i d x J0 (x) ∈ log U(1), [Q, Ψ (x)] = izΨ Ψ(x),
⎩ z
Ja = ρ(m 2) Ψ γaΨ .
Green’s function of the free gauge field equation, up to gauge-dependent

terms,
d4 q g2 2
− g12 ∂ 2 κ0 (x) = δ(x) ⇒ κ0 (x) = (2π) 4 q 2 +io e
iqx
, κ̃0 (q) = q2g+io ,
can be made to coincide up to a factor 2i with the gauge field Feynman

propagator:
d4 q g2 (−ηab ) iqx
0|Aa (0)Ab (x)|00 = i
π (2π)3 q2 +io e = −2iηab κ0 (x) = 2iκ0ab (x).
In a perturbative approach, the Feynman progagator of the free electro-

magnetic field (subindex 0) is modified in the first order (subindex 1),
0|Aa (x1 )Ab (x2 )|01 = 0|A

a (x1 )Ab (x2 )|00
+ d4 y1 d4 y2 κ0ac (x1 − y1 )0|Jc (y1 )Jd (y2 )|0κ0db (y2 − x2 ),
Aa Ab 1 (q) = A
+ a Ab 1 (q) + κ̃ac (q)Π (q,
0 cd 2 0
, m )κ̃db (q)
g 2 (−ηac ) cd 2
i g (−ηdb )
= δad + q 2 +io
Π (q, m2 ) π q 2 +io
,
by the vacuum polarization as the ground-state value of the bilinear bilocal

product of currents:
d4 q 2
i g (−ηab )
0|Jc (0)Jd (x)|0 = i
π (2π)3 Πcd (q, m2 )eiqx , Aa Ab 0 (q) = π q2 +io .
The vacuum polarization involves the convolution of the energy-momen-

tum propagators (see Chapter 9), the normalization factors ρ2 (m2 ) drop out:
d4 p γ(q−p)+m
Πcd (q, m2 ) = −2iz 2 (2π)4 tr γ c (q−p) d γp+m
2 +io−m2 γ p2 +io−m2 .
Its quadratic “divergence,” formally illustrated by

d4 p p2 −2m2 d4 p
m2 2κ2
Πcc (0, m2 ) = z 2 iπ 4 (p2 +io−m2 )2 = z
2
iπ 4
2
∞ dκ (−p2 −io+κ2 )3
z2
m2 2 d 4 p 2κ2 2 m2
= π2 ∞
dκ iπ 2 (−p2 −io+κ2 )3 = πz 2 ∞ dκ2 ,
shows the failure to expand a spacetime interaction in terms of free fields

with a Fock ground-state vector |0.
The Pauli–Villars regularization, denoted with a ground-state vector |M
for a field theory with interactions, keeps the naive gauge invariance and cur-
rent conservation qc Πcd (q) = 0 by regularizing the vacuum polarization “as a
whole,” i.e., pairwise, not by a modification of the individual fermion prop-
agators. The regularization trivializes the “divergent” moments of a dilation
(mass) expansion:
$
N
d4 x −iqx
2i M|J (0)J (x)|Me
c d
= Πcd (q) = ρi Πcd (q, m2i ),
i=0
$
N
with ρi (m2i )k = 0, k = 0, 1, . . . , where ρ0 = 1, m20 = m2 ,
i=0
$
N

z2
Πcd (q) = cd 2
π 2 (η q − q c q d ) ρi dζ ζ(1 − ζ) log[−ζ(1 − ζ)q 2 + io + m2i ].
i=0
The gauge-invariant part of the renormalization of the free gauge field prop-
agator can be regularized by one regulator, k = 0:
Πcd (q, m2 ) − Πcd (q, M 2 ),

Πcd (q) =
2 2 1
−ζ(1−ζ)q2 −io+m2 2
i g (−ηab )
Aa Ab 1 (q) = 1 + gπz2 0 dζ ζ(1 − ζ) log −ζ(1−ζ)q2 −io+M 2 π q2 +io .
It is even possible to restrict the regularization to lightlike energy-momentum

q 2 = 0:
2 2 1
2
Aa Ab 1 (q) = 1 + gπz2 0 dζ ζ(1 − ζ) log −ζ(1−ζ)q −io+m2
2
i g (−ηab )
M 2 π q2 +io .
The vaccum polarization involves the characteristic convolution for

four-dimensional spacetime,
g2 (−ηac ) cd g2 2 d
q2 Π (q) = 6π 2 Π(q )δa ,
1 −ζ(1−ζ)q2 −io+m2
0 dζ 6ζ(1 − ζ) log 2
1 2
z 2 Π(q + io) = M2
m2 2 1 ζ (3−2ζ)(1−2ζ)
= log M 2 + q 0 dζ −ζ(1−ζ)q2 −io+m2
1
log M 2 − 3 − q2 − (q +2m 2q)(q2 −4m )
2 2 2 2
m2 5 4m2 dζ
= 0 −ζ(1−ζ)q2 −io+m2
,
with three parts: momentumlike, energylike over and under the two particle
threshold q 2 = 4m2 (see Chapter 9),
m2 4m2
1
Π(q 2 + io) = log − 5
−
z2 M2 3 q2 ⎡ / 2 2 √ 2 2 / ⎤
/ 2m −q + q (q −4m2 ) /
ϑ(−q 2
) log / /
q 2 −4m2 ⎢
⎢ 2m 2
√ 2 ⎥
2
− q +2m
2
| | q (4m2 −q 2 ) ⎥
q 2 q 2 ⎣ + ϑ(q )ϑ(4m − q ) arctan
2 2 2 ⎦
q 2 −2m2
+ iπ ϑ(q 2 − 4m2 )
m2 q2
2 − 5m2 + . . . for |q | m .
2 2
= log M
The three contributions in the gauge field Feynman propagator, on-shell,

off-shell momentumlike, and off-shell energylike,
2 2
i g (−ηab )
π q2 +io = g 2 (−ηab )δ(q 2 ) + [ϑ(−q 2 ) + ϑ(q 2 )] πi g (−ηab )
q2 ,
are modified differently. The Poincaré group representation coefficient (on-

shell contribution, real) is renormalized by
g2 m2
g 2 −→ gR
2
= (1 + 2
6π 2 Π(0))g , Π(0) = z 2 log M2 .
The imaginary contribution over the threshold for the creation of oppositely
2
charged particle pairs is multiplied with the off-shell πi g (−η
2
qP
ab )
and gives an
additional real two particle on-shell contribution, all proportional to −ηab :

g2 z 2 m2 2 2
4m2 2m2 g 2
g 2 δ(q 2 ) −→ (1 + 6π 2
log M2
)g 2 δ(q 2 ) + ϑ(q 2 − 4m2 ) g6πz2 1− q2
(1 + q2
) q2 .
The off-shell contribution (imaginary) with the principal value and

d4 q g2 iqx g2
the Coulomb interaction dx0 iπ(2π) 3 −q 2 −io e = 2iπr has an energy-
momentum–dependent modification, for the momentumlike ϑ(−q 2 ) contri-
butions:
2
2 2
+
m2 4m2
ϑ(−q 2 ) gq2 −→ ϑ(−q 2 ) 1 + g6πz2 log M 2 − 3 − q2
5
√ , 2
2 2
q2 −4m2 2m2 −q2 + q2 (q2 −4m2 )
− q +2m
q 2 q 2 log 2m 2
g
q2 ,
and for the energylike ϑ(q 2 )-contributions:

2
2 2
+
m2 4m2
ϑ(q 2 ) gq2 −→ ϑ(q 2 ) 1 + g6πz2 log − 5
−
M2 3 q2
√ ,
2
+2m2 4m2 −q 2 q 2 (4m2 −q 2 ) g2
−ϑ(4m2 − q 2 ) q q2 q2
arctan q 2 −2m2 q2
.
The first-order perturbative correction can be considered as the first term

of a geometrical series expansion for the gauge field propagator:
g2 (−ηab )
Aa Ab (q) = i
π (q2 +io)[1− g2 Π(q2 +io)]
2
= gR (−ηab )δ(q 2 ) + . . . .
6π 2
The renormalized coupling constant is the residue at q 2 = 0, i.e., the inverse

derivative there of the denominator function,
/
6π 2 ∂ 2 6π 2 2 / 2
m2
g2
= ∂q2 q [ g2 − Π(q )]/ g2 − z log M 2 .
= 6π 2
R q2 =0
5.11.2 Lie Algebra Renormalization

by Vacuum Polarization
Lie algebra L representations on finite-dimensional vector spaces V with their
nondecomposable parts on Vι ,

d
d
L lA −→ D(lA ) = Dι (lA ) ∼
= D Aβ
α , V = Vι ,
ι=1 ι=1
are implemented in field theories by charges as position integrals over the

timelike component of currents,

L lA −→ iQA = i d3 x JA A B AB
0 (x), [iQ , iQ ] = C iQ ,
C
/
e.g., the currents of massive Dirac fields, normalized by {Ψ Ψ(0), Ψ (x)}/x =0 =
0
γ0 δ(x), for electromagnetic U(1) or isospin SU(2) with Pauli matrices (see
Chapter 6):
⎧
⎪ $ d
⎨ J = zιΨ ι γaΨ ι , zι ∈ Z, for U(1),
a
iJAa = Ψ β D α γaΨ , e.g.,
Aβ α
⎪
⎩ A
ι=1
τ Aβ
Ja = Ψ β 2 α γaΨ α for SU(2).
The vacuum polarization for current-gauge field coupling JB a

a AB contains
the ground-state expectation value of the current square:
d4 x −iqx
ΠAB 2i M|Ja (0)Jb (x)|Me
A B
ab (q) = ,
g2 (−η ac )δAC CD g 2
2 a D
q2 Πcd (q) = 6π2 Π(q )δd δA .
It involves invariant symmetric bilinear forms of the Lie algebra represen-

tatations, which are orthogonally diagonalizable,
$
d $
d
L × L −→ R, κV (lA , lB ) = κAB ∼
= δ AB κι ,
ι
ι=1 ι=1
$
d A
iτ B AB
e.g., − zι2 for electromagnetic U(1) and tr iτ2 ◦ 2 = −δ 2 for isospin
ι=1
SU(2).
The perturbative corrections by the corresponding vacuum polarizations
are illustrated by Dirac field currents with one regulator:
g2+(−η ab )δAB 6d
AaA AbB (q) = i
Π(q 2 ) = − ι=1 κι Πι (q 2 ),
,
π (q2 +io) 1− g2 Π(q2 +io) ,
6π 2 6
6 m2
6π 2 6π 2 zι2 log Mι2 , U(1),
2
gR
= g2 − Π(0), Π(0) = − κι Π (0) =
ι
1 m2
2 log M 2 , SU(2).
Chapter 6
External and Internal

Operations
Basic physical theories involve both external or spacetimelike and internal or

chargelike degrees of freedom in complex representation vector spaces that are
acted on, respectively, by operations from the Poincaré group, i.e., by Lorentz
transformations and spacetime translations, and by electroweak and strong
operations from the hypercharge, isospin, and color groups. The external–
internal dichotomy goes with a noncompact–compact distinction of the rele-
vant groups.
The properties of all basic interactions and particles are determined and
characterized by invariants and eigenvalues for these operation groups. Al-
though the product of external and internal operations in the acting group
is direct, Gext × Gint , the internal “chargelike” operations are coupled to the
external “spacetimelike” ones: Any spacetime translation is accompanied by
a chargelike operation. This is implemented by the gauge fields and the cor-
responding covariant derivatives for the particle fields in the standard model
of electroweak and strong interactions. All interactions can be formulated, in
a classical geometrical language, by connections of bundles, by a Riemannian
connection for the tangent spaces of “horizontal” spacetime as the base, yield-
ing the external interactions, i.e., gravity, and by connections of “vertical”
complex vector spaces as fibers, yielding the internal interactions, e.g., the
electroweak and strong ones.
This chapter reviews the internal gauge symmetries and their mathemati-
cal formalization in the standard model and in the framework of fiber bundles
with connections.

148 Chapter 6 External and Internal Operations
6.1 Fiber Bundles

Chargelike operations can be described with bundles. Bundles generalize the
“infinitesimal” (tangent space) structures T(M) with the derivations of the
functions of a manifold (see Chapter 2). The local structure of a manifold M
can be enriched by an additional differential manifold F . The fiber F may
be a topological vector space, an operational Lie group, or an operational Lie
algebra. One specimen of the fiber F is “planted” at each point of the base
manifold M.
6.1.1 Fibers and Base

A fiber bundle F (M) ∈ dif K (top) with a topological space M ∈ top as base
and a manifold F ∈ dif K as typical fiber is characterized by a surjective
continuous projection from bundle to base whose local inverses are the local
fibers, all isomorphic to the typical fiber:
π : F (M) −→ M, ξ −→ π(ξ),
M x −→ π −1 (x) = Fx ⊂ F (M), Fx ∼
= F.
The continuity and “sufficient smoothness” of the mappings used will not be
discussed. The additional conditions for local triviality and the structural
group of a bundle are given in the next subsection. In the cases considered,
the base is also a manifold, M ∈ dif R .
With the equivalence relation to belong to the same fiber, the base char-
acterizes equivalence classes:
ξ, ξ ∈ F (M) : ξπξ ⇐⇒ π(ξ) = π(ξ ) ⇒ F (M)/π ∼

= M.
A bundle is the union of the local fibers with the base, the hedgehog with
prickles and skin,

F (M) = (x, Fx ) = π −1 [M].
x∈M
The trivial F -bundle for M is the set product F × M.

Properties related to the base M are called horizontal, later used as “exter-
nal” and “spacetimelike,” whereas properties related to the fiber F are called
vertical, later used as “internal” and “chargelike.”
The category from which the fiber comes gives the name for the bundle,
e.g., a topological vector space bundle tvecK (top) with a vector space fiber
F ∈ tvecK or a Lie group bundle lgrpK (top). Examples of vector space
bundles are the tangent and cotangent bundle T(M), TT (M) of a manifold
with the (co)tangent spaces as the local fibers Tx (M) ∼ = Rn ∼ = TTx (M). The
frame bundle AL(R )(M) with the linear tangent space mappings AL(Rn )
n
is both an algebra and a Lie algebra bundle.

6.1 Fiber Bundles 149
The subcategories dif K (M) are important, i.e., different typical fibers for
one fixed base space M, e.g., for spacetime M(t,s) .
The sections ψ of a bundle are projection-compatible mappings of the
base into the bundle, i.e., the mappings F M from base to fiber,
ψ : M −→ F (M), π ◦ ψ = idM ,M x −→ ψ(x) ∈ F,

⎪
⎪
ψ = {(x, ψ(x))⎪
⎪x ∈ M} ∈ (x, Fx ).
x∈M
Examples are vector and tensor fields.

The fiber bundle morphisms (f, ϕ) have to be compatible with the projec-
tions. The restrictions to the local fibers Fx have to yield morphisms in the
fiber category:
f
F (M)
⏐ −→ F (M
⏐ )

⏐ ⏐
π π , f |Fx : Fx −→ Fϕ(x) .

M −→ M
ϕ
Operations with the fiber category dif K , i.e., functors acting on dif K ,
can be transferred to the corresponding bundles dif K (M). In this way,
there arise associated bundles, e.g., dual bundles with vector space duality
(V (M), V T (M)) in tvecK (M) (T is the dual functor for vector spaces), direct
sum bundles V1 ⊕ V2 (M), and tensor product bundles V1 ⊗ V2 (M), and, cor-
respondingly,
! the transition from a vector space bundle to its tensor algebra
bundle V (M) ∈ aagK (M) (direct sum ⊕, tensor product ⊗, and tensor
'
algebra are functors for vector spaces), or from a Lie group bundle to
its Lie algebra bundle lgrpR (M) G(M) −→ log G(M) ∈ lagR (M) via the
functor log : lgrpR −→ lagR .
6.1.2 Structural and Gauge Groups

Now we have the additional condition for a fiber bundle with respect to
its operational structure — its local triviality up to its structural group:
A manifold is required to be locally like Rn ; its tangent bundle involves free
modules, T(U ) ∈ modC(U) for neighborhoods U x. Similarly, a bundle is
required to be locally trivial F × U ; i.e., a bundle F (M) comes with a covering
of the base M and a fiber isomorphism for each set from the covering, called
a trivialization (Uι , χι )ι∈I , leading to local trivializations,
χι
= F, π −1 [Uι ] ∼
x ∈ Uι : Fx ∼ = F × Uι .
If a base point is in the intersection of two covering sets, the two local trivi-
alizations define a diffeomorphism of the typical fiber:
o
x ∈ Uι ∩ Uκ , gικ (x) = χι ◦ χ−1
κ ∈ dif K (F, F ).
The group of all these F -isomorphisms from all local trivializations is required
to be equal for all manifold points. It is called the structural group of the
bundle, or also its gauge group (ahead):
⎪
⎪ o
for all x ∈ M : {gικ (x)⎪
⎪x ∈ Uι ∩ Uκ } = G ⊆ dif K (F, F ).
It has to act effectively on the fiber, i.e., in a faithful realization.

The definition of a fiber bundle has to specify the structural group. The
fiber with the structural group is a Klein space G • F , i.e., a differential
manifold F ∈ dif G with effective and “smooth” Lie group G action:
G • F (M) ∈ dif G (M), especially F ∈ tvecK , lgrpR , lagR .
The bundle morphisms have to include a compatible morphism ρ for the

structural groups G and G :
(f, ϕ, ρ) : G • F (M) −→ G • F (M ),

with ρ : G −→ G , ρ(gικ ) = g κι .
The structural group is closely related to the local trivialization, given

in the definition of the bundle. For example, the trivial bundle F × M with
trivialization (M, idF × idM ) has as structural group only the fiber identity
o
{ idF }; for the trivialization (M, dif K (F, F )× idM ), one obtains the full group
o
dif K (F, F ).
It is also possible to impose the structural group via fiber properties, e.g.,
maximality, a fiber metric, etc., with a correspondingly chosen trivialization.
For example, the maximal structural group of a tangent bundle T(M) is
the full linear group GL(n, R) of the typical fiber. For a Riemannian man-
ifold, the structural group can be restricted via the tangent space metric
and orthonormal bases to the tangent Lorentz group SO0 (t, s). In any case,
the structural group is realized faithfully in the diffeomorphism group of the
o
fiber, G −→ dif R (F, F ).
6.2 Nonrelativistic and Relativistic Bundles

Relativistic fields are mappings of a spacetime manifold into a real or complex
vector space, i.e., sections of a bundle V (M(1,3) ),
M(1,3) x −→ Φ (x) ∈ V.
Special relativistic fields (see Chapter 5) map the spacetime translations,

i.e., the flat manifold R4 ∼ = SO0 (1, 3) ×
R4 /SO0 (1, 3), into a vector space
with the action (a representation) of the Lorentz group. The fields carry
6.2 Nonrelativistic and Relativistic Bundles 151
the Lorentz group transformation behavior, which is induced by the Lorentz

group action on both spaces, as given in the commutative diagram
R⏐4
Λ(s)
−→ R⏐4 SL(2, C) s −→ Λ(s) ∈ SO0 (1, 3),
⏐ ⏐ SL(2, C) s −→ D(s) ∈ SL(V ),
Φ Φs , s
VR Φ −→ Φ s ∈ V R ,
4 4
V −→ V Φ s (x) = D(s).ΦΦ(Λ−1 (s).x).

D(s)
For example, a Dirac spinor field is transformed by the complex four-

dimensional Dirac representation,
→ Ψ (x) ∈ V ∼
R4 x − = C4 ,
R4 s
V Ψ − → Ψ s ∈ V , Ψ s (x) = (s ⊕ s−1 )Ψ
R4 Ψ(Λ−1 (s).x).
In addition to the external Lorentz transformations, the vector space V

with the field values can be acted on by an internal operation group U, which
acts trivially on spacetime:
u
U u −→ R(u) ∈ GL(V ), V R Φ −→ R(u).Φ
Φ ∈VR .
4 4
Summarizing: Special relativistic fields use trivial vector space bundles

with the Minkowski translations R4 as base and, as fiber, a representation
space V for the Lorentz group and for the chargelike operations U as struc-
tural group, e.g., for the hyperisospin group U(2) −→ U(V ),
[SL(2, C) × U] • V (R4 ), with V (R4 ) ∼

= V × R4 , V ∼
= CN .
If nonrelativistic mechanics is described by the same concepts, posi-

tions and momenta are mappings of the real one-dimensional time manifold
M1 ∼= R into a value space:
M1 t −→ (x, p)(t) ∈ V.
In contrast to the Lorentz group SO0 (1, 3) for spacetime translations R4 , the
external homogeneous group on time translations R is trivial, SO(1) = {1}.
The vector space with the position-momentum values can be acted on by
an internal operation group O, which acts trivially on time,
O
O O −→ R(O) ∈ GL(V ), V R (x, p) −→ (R(O).x, R(O).p) ∈ V R ,
e.g., the rotation group O(3) in three-dimensional mechanics, a, b = 1, 2, 3,
R t −→ (xa , pa )(t) ∈ V ∼
= R3 ,
R O
O(3) O, V (x , p ) −→ (Ob x , Ob p ) ∈ V R .
a a a b a b
In such a parallel interpretation of fields on spacetime x −→ Φ (x) and

“fields on time” t −→ x(t), the rotations in mechanics can be viewed as an
internal (“vertical”) operation group and the angular momenta viewed as an

internal quantum number with the vector space bundles
SO(3) • V (R), with SO(3) −→ SO(V ), V (R) ∼

= V × R, V ∼
= RN .
in comparison with the internal quantum numbers for hypercharge and
isospin and the special relativistic bundles U(2) • V (R4 ). In contrast to the
internal rotations in mechanics on time, the rotations SO(3) ⊂ SO0 (1, 3) in
spacetime relativity are an external (“horizontal”) group.
6.3 Connections of Vector Space Bundles

A connection of a vector space bundle connects the Lie algebra operations
as given by the manifold vector fields, i.e., derivations of manifold functions,
with the structural Lie algebra acting on the vector space fibers.
A vector space bundle G • V (M) ∈ vecK (M) with a base manifold M ∈
dif R and a finite-dimensional G-space V ∼ = KN as typical fiber is a module
V (M) ∈ modC(M) over the ring with the manifold functions. One has local
free modules V (U ) ∼ = C(U )N , U ⊆ M, with dual fiber bases (V -frames)
idV (U) = E ⊗ Ěα , α = 1, . . . , N , as familiar from the tangent bundle T(M) ∈
α
modC(M) with local dual tangent bases idT(U) = ei ⊗ ěi , i = 1, . . . , n and

maximal stuctural group e(x) ∈ GL(n, R) (see Chapter 2). Holonomic bases
∂ i ⊗ dxi are a specialty of the tangent bundle.
To connect, on the manifold, the structural group of the vector space
fibers to each other, one defines, in addition to and by generalizing an affine
connection for a tangent bundle ∇ : T(M) −→ T ⊗ TT (M), a connection of
the vector space bundle: It is given by a mapping of the manifold vector fields
(derivations) into R-linear mappings of the local vector spaces,
D : T(M) −→ V ⊗ V T (M), v −→ Dv ,
f, g ∈ C(M) : Df v+gw = f Dv + gDw ,
which defines actions on sections, in holonomic bases,
⎧
⎪
⎪ ψ −
→ Dv ψ,
⎨ Dv (f ψ) = v(f )ψ + f Dv ψ,
Dv : V (M) −→ V (M),
⎪
⎪ Dei E α = Di E α = Aiα β
βE ,
⎩ i
D (ψα E ) = (∂ ψβ + A β ψα )E β .
α i iα
The connection is determined by N 2 1-forms:
β = A β ěi ∈ T (M),
Aα iα T
D = A = A β E ⊗ Ěα ∈ V ⊗ V T (M),
i i iα β
D = A = Aiα β E ⊗ Ěα ⊗ ěi ∈ V ⊗ V

β T
⊗ TT (M).
More: The endomorphisms Dv have to be compatible with the structural

Lie group G ⊆ GL(V ) of the typical fiber vector space; i.e., they have to be
6.3 Connections of Vector Space Bundles 153
valued in the structural Lie algebra L = log G ⊆ V ⊗ V T . Therefore, D and A

are called the G-covariant derivation and G-gauge field on the base M. The
coefficients of an L-basis define R-linear local mappings from the Lie algebra
Tx (M) to the structural Lie algebra Lx (M) — in general, no Lie algebra
morphisms —
D : T(M) −→ L(M), L ⊆ V ⊗ V T ,
Rn ∼ −→ Lx (M) ∼
A(x)
= Tx (M) ⎧ = Rd ,
⎨ A = Aia la ⊗ ěi ∈ L ⊗ TT (M),
L-basis: {la ∈ V ⊗ V T )}da=1 ,
Ai = tr ľa ◦ Ai ,
L-frame: idL = la ⊗ ˇla , ⎩ iαa
A β = Aia laα β.
A connection is grading-compatibly extendable, by Leibniz’s rule, to the

full tensor algebra,
! !
Dv : ⎧ L(M) −→ L(M), L ∈ V ⊗ V T ,
⎪
⎪ Dv f = ∇v f = v(f ), f ∈ C(M),
⎨ D (a ⊗ b) = D a ⊗ b + a ⊗ D b,
v v v
with
⎪
⎪ Dv ψ̌, ψ = Dv ψ̌, ψ + ψ̌, Dv ψ, ψ̌ ∈ V T (M),
⎩ i β
D (ψ̌ Ěβ ) = (∂ i ψ̌ α − Aiα β
β ψ̌ )Ěα .
The local V -frames are covariant constant, Dv idV (U) = Dv (E β ⊗ Ěβ ) = 0.

The connection can be used as a G-covariant derivative for any represen-
tation L of the structural Lie algebra on a vector space W :
V ⊗VT ⊇Ll − → L(l) ∈ L ⊆ W ⊗ W T ,
D : T(M) −→ L(M).
W
The fiber endomorphisms (Ai )ni=1 are not C(M)-tensors. Their gauge trans-
formation, effected by G-changing the local V -frames E α ⊗ Ěα = E μ ⊗ Ěμ ,
involves derivatives of the N -bein from the structural group:
⎧ μ
⎪
⎪ E (x) = Eαμ (x)E α (x),
⎨ Aiμ = Eαμ Aiα −1β i μ −1α
βE ν + (∂ Eα )E ν,
E ∈ G(M) : ν
⎪
⎪ Ai −→ E ◦ A ◦ E + (∂ E) ◦ E −1 ,
i −1 i
⎩
∂ i Eβμ −Aiμν Eβν + Eαμ Aiαβ = 0.
With the exception of torsion, first Bianchi identity, and Ricci tensor,
which are specialties of the tangent bundle T(M), the tangent bundle anal-
ogous structures can be defined on vector space bundles in general: The
field strengths (curvature) of a G-connection on M involve the difference of
two tangent field commutators. They define endomorphisms of the vector
space fibers, which can be spanned by an L-basis, i.e., they are Lie algebra
operations:
F : T ∧ T(M) −→ L(M) ⊆ V ⊗ V T (M),

F(v ∧ w) = [Dv , Dw ] − D[v,w] = −F(w ∧ v),
F(ei ∧ ej ) = Fij = ei (Aj ) − ej (Ai ) − ij k i
k A + [A , A ]
j
αij β
= l Fa = Fβ E ⊗ Ěα ,
a ij
Fαij
β = ei (Aj βα ) − ej (Aiα ij kα iγ j α j γ iα
β ) − k A β + A β A γ − A β A γ
γ γ
= ∂ i Aj βα − ∂ j Aiα
β + A β A γ − A β A γ (holonomic bases),
i jα j iα
1 αij β
F = 2 Fβ E ⊗ Ěα ⊗ ěi ∧ ěj ∈ L ⊗ T ∧ TT (M).
T
[Ai , Aj ] is the L-bracket. The field strengths (curvature) are 2-forms

1 αij γ
β = 2 Fβ ěi ∧ ěj = dAβ + Aβ ∧ Aγ ∈ T ∧ T (M).
Fα α α T T
The field strengths (curvature) map the Lorentz Lie algebra, acting on the
tangent spaces Fαab
β : log SO0 (t, s) −→ log G, into the structural Lie algebra,
acting on the fiber (no Lie algebra morphism). The geometrical meaning of
“curvature” (local area change) applies only for the tangent bundle, where
the curvature is a bilinear form (operational metric) of the tangent Lorentz
Lie algebra (see Chapter 1).
In contrast to the gauge fields (connection) A : T(M) −→ L(M), the field
strengths (curvature) F : T∧T(M) −→ L(M) are C(M)-linear; i.e., they have
“homogeneous” local transformation behavior with respect to the structural
group, given by the adjoint action of the group on its Lie algebra,
G(M) × L(M) −→ L(M), Fij −→ E ◦ Fij ◦ E −1 .
Via the Jacobi identities for the brackets with the covariant derivatives
Dv , the curvature fulfills the (second) Bianchi identity:
Dz [F(v ∧ w)ψ] + Dv [F(w ∧ z)ψ] + Dw [F(z ∧ v)ψ]

= [Dz , [Dv , Dw ]]ψ + [Dv , [Dw , Dz ]]ψ + [Dw , [Dz , Dv ]]ψ = 0,
Dk Fαij
β + Di Fαjk
β + Dj Fαki
β = 0.
6.4 Pure Gauges, Distinguished Frames,

and Composite Gauge Fields
If there exists for a connection of a vector space bundle,
T(M) v −→ Dv ∈ log G(M) ⊆ V ⊗ V T (M), V ∼

= KN ,
a local V -frame E α ⊗ Ěα with trivial gauge fields for the structural group G,
the connection has the form of a pure gauge. Pure gauge fields have trivial
field strengths (curvature):
Aiμν = (∂ i Eαμ )E −1α

ν , A = l (E) = (∂ E) ◦ E
i i i −1
⇒ F = ∂ A − ∂ A + [A , A ] = 0.
ij i j j i i j
6.4 Pure Gauges, Distinguished Frames, and Composite Gauge Fields 155
It may occur, e.g., in the case of a degenerate ground state, that there
exist distinguished V -frames, which consist of transmutators from G to a
subgroup H (see Chapter 7):
GL(N, K) ∼
= GL(V ) E(x) ∼
= Eαμ (x) ∈∈G/H.
Transmutators transform from G-active vectors (here, indices μ = 1, . . . , N )
to only H-active ones (here, indices α = 1, . . . , N ).
Reducing gauge fields to the “little” group H ⊆ G: If there exist, in such
a case, basic gauge fields for the “large” group G with the affine G × log G
transformation behavior,
M −→ log G, x −→ Ai (x) = Aia (x)la = ∼ Ai (x)laμ ,
a ν
M −→ G, x− → g(x),
M −→ log G, x − → li (g(x)), li (g) = (∂ i g) ◦ g −1 ,
A −→ g ◦ Ai ◦ g −1 + li (g),
i
they can be “frozen” up to gauge fields for the “little” group H: They are
stripped of the G/H degrees of freedom, provided by the distinguished frames,
M −→ log H, x −→ Ai (x), Ai = E −1 ◦ Ai ◦ E − li (E),

−1α iμ ν −1α i μ
Ai α
β = E μ A ν Eβ − E μ ∂ Eβ .
The basic G-gauge fields are “muted” up to the H-gauge degrees of freedom:
M −→ H, x −→ h(x) = E −1 (x) ◦ g(x),
M −→ log H, x −→ li (h(x)),
Ai −→ h ◦ Ai ◦ h−1 + li (h).
An example is the reduction of the four gauge fields for hyperisospin U(2)
in the electroweak standard model to an electromagnetic U(1)-gauge with
massless gauge field (ahead).
Constructing gauge fields for the “little” group H ⊆ G: There may exist
distinguished frames (transmutators) for a coset space E(x) ∈∈G/H without
basic gauge fields for the group G. An example is a Riemannian manifold with
the distinguished frames e(x) ∈∈GL(n, R)/SO0 (t, s) yielding transmutators
from the full linear group to the tangent Lorentz group (see Chapter 2). The
Riemannian connection is given by composite Lorentz group gauge fields and
can be expressed by the n-bein Γ ∼ e−1 ∂e (Fock–Iwanenkov coefficients):
Γicb ηca = Γ ab = −Γ
i Γiba
= ee (ľab )cd (η df eek ∂ c + η ef edk ∂ c − η df eck ∂ e )ekf ,
i
(ľab )cd = ηac ηbd − ηbc ηad ∈ [log SO0 (t, s)]T .
It has nontrivial curvature; i.e., it is not a pure gauge.
In general: For vertical operations with cosets G/H, the distinguished
frames are determined up to local H-transformation and allow the construc-
tion of composite gauge fields for a nontrivial fixgroup H by log H-projection
of the pure log G-gauge:
⎪
⎪
log H-basis: {lA ∈ V ⊗ V T ∼ = AL(KN )⎪ ⎪A = 1, . . . , dimR H},
dual basis: ľA ∈ (log H) , log H-frame: Plog H = lA ⊗ ˇlA ,
T
Bi = Plog H (E −1 ◦ ∂ i E), Bi (x) = BiA (x)lA ,

BiA = tr ľA ◦ E −1 ◦ ∂ i E = ľA βα E −1α i μ
μ ∂ Eβ .
6.5 Chargelike Internal Connections

Gauge theories for flat spacetime (Lorentz indices j, k, l, . . . ) connect with
each other spacetime translations and internal Lie algebra operations.
6.5.1 Currents as Lie Algebra Densities

A Lie algebra with dual bases and Lie bracket,
lagR L ∼
= Rd : ľa , lb = δab , [la , lb ] = ab c
c l ,
internal indices a = 1, . . . , d,
is represented in the endomorphisms of a vector space and its dual,
W, W T ∼
= Cn , dual bases ěβ , eγ = δβγ = eγ , ěβ ,
α = 1, . . . , n, = ±1 for Fermi or Bose,
D : L −→ AL(W ), la −→ D(la ) = Daβγ eγ ⊗ ěβ ,
L × W −→ W, la • eβ = Daβγ eγ ,
Ď : L −→ AL(W T ), la −→ −D(la )T = −Daβγ ěβ ⊗ eγ ,
L × W T −→ W T , la • ěγ = −Daβγ ěβ ,
e.g., the Lie algebras of U(1) and SU(n),
log U(1) −→ AL(C), D(l0 ) ∼ = iz, z ∈ Z,

log SU(n) −→ AL(C ), n
D(la ) ∼ = 2i τ a (n)βγ ,
⎪
⎪ tr τ (n) = 0,
Pauli matrices: {τ a (n)⎪
⎪a = 1, . . . , n2 − 1}, with
tr τ a (n) ◦ τ b (n) = 2δ ab .
For flat spacetime fields, the Lie algebra representation is given by the
charges:

la −→ iQa = i d3 x Ja0 (x), [iQa , iQb ] = ab c
c iQ .
They are position integrals over the currents, which are defined with the
quantization opposite (anti)commutators (normalized with ρ(m2 ) = 1, see
Chapter 5), e.g.,
6.5 Chargelike Internal Connections 157
Field Quantization Current
Scalar γ 0 Φγ ,iΦ
{Φ Φkβ }
Φ kβ , Φ γ ]( k = D γ
iJa aβ
[iΦ x) = δβ δk δ(
x) 2
(Hermitian)
Scalar Φ
[iΦ γ
kβ , Φ ]( Φγ
x) = [iΦ
k , Φ β ](
x) Φγ ,iΦ
{Φ Φγ ,Φ
Φ }+{−iΦ Φ }
k = D γ
kβ β
γ 0 iJa aβ
2
k
(complex) = δβ δk δ(x)
j γ
Vector {A ,iG }
[iGγ l
kj , Aβ ](
γ 0 a l b
x) = δβ δk δj δb δa δ(
x) k = D γ
iJa aβ β
2
kj
(Hermitian)
Weyl γ 0 [lγ σ̌k ,l
β]
{l
β , l }(
γ
x) = δβ σ δ(
x) k = D γ
iJa aβ
2
(left)
Weyl γ 0 [rγ σk ,rβ]
{r
β , r }(
γ
x) = δβ σ̌ δ(
x) k = D γ
iJa aβ
2
(right)
γ 0 Ψγ γk ,Ψ
[Ψ Ψβ ]
Dirac {Ψ
Ψ β , Ψ γ }(
x) = δβ γ δ(
x) k = D γ
iJa aβ
2
The adjoint action on the fields reads

[iQa , Φ β ] = Daβγ Φ γ , [iQa , Φ γ ] = −Daβγ Φ β .
The simultaneous external–internal action (covariant derivatives),

(∂δγβ − Daβγ Aa )Φ
Φ γ , (∂δγβ + Daβγ Aa )Φ
Φ β ,
is implemented by gauge vertices (gauge interactions). They are Lorentz-

compatible spacetime distributions of the power-3 Lie algebra representation
tensor:
D= ˇ
la ⊗ D(la ) = ˇla ⊗ Daβ
γ eγ ⊗ ěβ ,
e.g. Ψ γ γk ,Ψ
[Ψ Ψβ ]
implemented by Aka Jak = Aka Daβγ 2 .
Gauge fields go with a Lie algebra and its currents: The number of gauge
fields is given by the Lie algebra dimension L ∼ = Rd : they transform under
the adjoint Lie algebra representation (for field strength Fb and the currents
jb , Jb ) and its dual coadjoint representation (for gauge fields Ac ):
ad : L −→ AL(L), ˇ
c l ⊗ lb ,
ad la = ab lb −→ ab
c c
c l ,
ˇ : L −→ AL(LT ),
ad ˇ
ad l = −c ľb ⊗ l , ľc −
a ab c ab ˇ
→ c lb .
The gauge field currents are products of the dual pairs (Aa , Fa ):
j ab c
jak = A b c Fkj ,
[iQa , (Fbkj , jb , Jb )] = ab c c c
c (Fkj , j , J ),
⇒
a 3 a
Q = d x (j0 + J0 ) a [iQ , Ac ] = −c Ab .
a k ab k
With the adjoint Lie algebra representation, the gauge field self-coupling is
nontrivial only for a nonabelian Lie algebra.
In the Lagrangian for the gauge field sector,
∂ k Ajc −∂ j Ak Fckj Fkjb
L(A, F) = Fckj 2
c
+ gcb 4 − 12 Aka jak ,
the statistical factor 12 in 12 Aka jak takes into account the tensor power 2 of
the gauge field A ∨ A in the interaction. The current arises by gauge field
derivation:
j
1 k a Ak
a Ab ∂ 12 Ak a
a jk j c
2 Aa jk = ab
c
c
2 Fkj , ∂Ak
= ab
c Ab Fkj .
a
In a second-order derivative formulation, derivatives of the gauge fields

and cubic gauge field products occur in the current:
jak = Ajb ab

c (∂k Aj − ∂j Ak +
c c cde d e
Ak Aj + δkj ∂ l Acl ).
The Lagrangian for the gauge field sector,
∂ k Ajc −∂ j Ak ab k j
c −c Aa Ab
Fckj Fkj
L(A, F) = Fckj 2 + g2 4
c
,
gives the field equations:

k j
∂ k Ajc − ∂ j Akc − ab 2 jk j b ab j c
c Aa Ab = g Fc , ∂ Fkj + c Aa Fkj = 0.
6.5.2 Normalizations of Gauge Fields

A gauge field is normalized by an invariant nondegenerate symmetric bilinear
form of the Lie algebra and its dual,
L × L −→ R, la |lb = κab = κba ,
e.g., by the Killing form κ(la , lb ) = ac bd

d c for a semisimple Lie algebra
like SU(n), n ≥ 2, or by a squared linear form l|l = [κ(l)]2 for an abelian
Lie algebra like U(1). In the following with compact gauge group U, a posi-
tive definite diagonal normalization κab = κ2U δ ab is possible with Lie algebra
bases for totally antisymmetric structure constants ab
d δ
dc
= −abc . The gauge
field coupling constant in the field strength square can be seen as the nor-
malization ratio of the represented internal Lie algebra L = log U with κ2U
and the external Lorentz Lie algebra log SO0 (1, 3) with its Killing form η ∧ η,
normalized by κ2SO0 (1,3) :
F|F = g 2 δab η kl η jm Fakj Fblm = g 2 Fakj Fkj

a ,
κ2SO0 (1,3)
g2 = κ2U
.
In a theory with interaction, the gauge field coupling constant is renor-

malized (see Chapter 5).
6.5.3 Gauge Interactions in the Standard Model

The standard model of the electroweak and strong interactions in Minkowski
spacetime is a theory of compatibly represented external and internal opera-
tions. The electromagnetic U(1) is embedded into the product of the abelian
hypercharge U(1) and the nonabelian isospin-color group SU(2) × SU(3):
U(1)×SU(2)×SU(3)
U(1) → U(2 × 3) = U(16 ) ◦ [SU(2) × SU(3)] = I(2)×I(3) .
6.5 Chargelike Internal Connections 159
The electromagnetic interaction for a Dirac electron field (quantum

electrodynamics) becomes a part of the electroweak and strong gauge in-
teractions of quark and lepton Weyl fields. The fields involved are acted on
by irreducible representations [L|R] of the Lorentz group SL(2, C) and irre-
ducible representations of the hypercharge group U(1) (rational hypercharge
number in [y]), the isospin group SU(2) (integer or half-integer isospin in
[T ]), and the color group SU(3), as given in the following table:
Field symbol SL(2, C) U(1) SU(2) SU(3)

[L|R] [y] [T ] [2C1 , 2C2 ]
Left lepton l [ 12 |0] − 12 [ 12 ] [0, 0]
Right lepton e [0| 12 ] −1 [0] [0, 0]
Left quark q [ 12 |0] 1
6 [ 12 ] [1, 0]
Right up quark u [0| 12 ] 2
3 [0] [1, 0]
Right down quark d [0| 12 ] − 13 [0] [1, 0]
Hypercharge gauge A0 [ 12 | 12 ] 0 [0] [0, 0]
Isospin gauge
A [ 12 | 12 ] 0 [1] [0, 0]
Color gauge G [ 12 | 12 ] 0 [0] [1, 1]
1
Higgs Φ [0|0] 2 [ 12 ] [0, 0]
The fields of the minimal standard model
With the exception of the Higgs field, the isospin SU(2)-representation is

a subrepresentation of the Lorentz group SL(2, C)-representation. This is a
characteristic structure of induced representations that start with the two-
sided regular representation of the doubled group (see Chapter 7).
The two factors in the internal group U(16 ) ◦ [SU(2) × SU(3)] are not
direct, but centrally correlated; i.e., the representations of hypercharge U(1)
are related to the representations of the SU(2)×SU(3)-center, the cyclotomic
group I(2)×I(3) = I(6) (hexality = two-triality, “star of David”). All U(2×3)-
representations [y||T ; 2C1 , 2C2 ] carried by the standard model fields with the
isospin and color multiplicities,
dSU(2) = 1 + 2T, dSU(3) = (1 + 2C1 )(1 + 2C2 )(1 + C1 + C2 ),
can be generated by the dual defining representations of U(2 × 3),
u = [ 16 || 12 ; 1, 0], ǔ = [− 16 || 12 ; 0, 1],
"
n "
m
as seen in the powers u ⊗ ǔ (all fermion fields are taken as left-handed),
Field U(2 × 3) (n, m) n−m 6y 6y
[y||T ; 2C1 , 2C2 ] = 6y mod 2 mod 3
l [− 12 || 12 ; 0, 0] (0, 3) −3 1 0
e [1||0; 0, 0] (6, 0) 6 0 0
q [ 16 || 12 ; 1, 0] (1, 0) 1 1 1
u [− 23 ||0; 0, 1] (0, 4) −4 0 −1
d [ 13 ||0; 0, 1] (2, 0) 2 0 −1
A0 [0||0; 0, 0] (0, 0) 0 0 0

A [0||1; 0, 0] (1, 1) 0 0 0
G [0||0; 1, 1] (1, 1) 0 0 0
Φ [ 12 || 12 ; 0, 0] (3, 0) 3 1 0
The central correlations of the internal symmetries are expressed by the

modulo relations
6y mod 2 = 2T mod 2, 6y mod 3 = 2(C1 − C2 ) mod 3,
y · dSU(2) · dSU(3) ∈ Z.
The nongauge fields with the free Lagrangians,
left fermions: L(l) = ilα ∂ σ̌lα + iqαc ∂ σ̌qαc , α = 1, 2; c = 1, 2, 3,

right fermions: L(r) = ie ∂σe + iuc ∂σuc + idc ∂σdc ,
α k ∼
k ∂ Φ α − Φ k Φ α = (∂ Φ α )(∂kΦ ),
Φ ) = Φ kα ∂ kΦ α + Φ α k k α
Higgs: L(Φ
for the electroweak interactions

interact with the four gauge fields A0 and A
and the eight gauge fields G for the strong interactions:
∂ k Aj −∂ j Ak Fkj Fkj
L(A0 ) = Fkj 0
2
0
+ g12 4 ,
ab k j b kj
= Fc ∂ Ac −∂ Ac −c Aa Ab
k j j k
2 Fkj Fb
L(A) kj 2 + g2 4 ,
∂ k GjC −∂ j Gk j
C −C GA GB
AB k FB Fkj
L(G) = FC
kj 2 + g32 kj4 B .
The indices differentiate between the different Lie algebras in the case of the
field strengths F. The structure constants are taken in a Pauli and Gell–Mann
basis:
log SU(2): { 2i τ aβγ |a = 1, 2, 3, β = 1, 2}, [ 2i τ a , 2i τ b ] = ab i c
c 2τ ,
log SU(3): { 2i λAbc |A = 1, . . . , 8; b = 1, 2, 3}, [ 2i λA , 2i λB ] = AB i C

C 2λ .
The gauge field coupling constants are the normalization ratios of the
internal Lie algebras and the Lorentz Lie algebra:

1 1 1 (κ2U(1) ,κ2SU(2) ,κ2SU(3) )
g 2 , g 2 , g 2 = κ2
.
1 2 3 SO0 (1,3)
The gauge interactions of the matter fields,

+ L(G) + L(l) + L(r) + L(Φ
L(A0 ) + L(A) Φ ) − (Ak0 Jk + Aka Jak + GkA JA
k ),
involve the currents for the nongauge fields:
log U(1): Jk = − 12 lσ̌k l + 16 qσ̌k q − eσk e + 23 uσk u − 13 dσk d

− 2i (ΦΦΦ k − ΦΦ k ),
τa a
log SU(2): Jk = lσ̌k 2 l + qσ̌k τ2 q
a
a a
+ (iΦ Φ τ2 Φ k − iΦΦk τ2 Φ ),
λA λA λA
log SU(3): JA
k = qσ̌k 2 q + uσk 2 u + dσk 2 d .
The gauge field strength equations are given with the corresponding
currents:
∂ j Fkj + Fkj × Aj = Jk .
6.6 Ground-State Degeneracy 161
The spacetime translations for the lepton and quark come with internal
gauge field actions
a
(∂ + 2i A0 − i τ2 Aa )lσ̌ = 0, (∂ + iA0 )eσ = 0,
τa λA A
(∂ − 6 A0 − i 2 Aa − i 2 GA )qσ̌ = 0, (∂ − 3 A0 − i λ2 GA )uσ = 0,
i 2i
A
(∂ + 3i A0 − i λ2 GA )dσ = 0,
as well as the Higgs field:

12 Ak a k
0 +τ Aa 12 Ak a k
0 +τ Aa
(∂ k − i 2
Φ
)Φ = Φ k , (∂ k − i 2
Φk
)Φ = 0.
For the scalar Higgs field, the second-order Lagrangian reads
a a
Φ , A) = [(∂ − i 12 A0 +τ
L(Φ 2
Aa
Φ][(∂ + i 12 A0 +τ
)Φ 2
Aa
Φ ].
)Φ
6.6 Ground-State Degeneracy

For an interaction with symmetry group G, e.g., hypercharge-isospin U(2),
the symmetry of a ground state may be characterized by a subgroup H ⊆ G
as the fixgroup, e.g., by electromagnetic U(1). Then, with the interaction
symmetry G, there exists a ground state degeneracy, effected by the opera-
tions from the coset manifold G/H, e.g., from the weak Goldstone manifold
U(2)/U(1), implemented by long-range (“massless”) scalar Nambu–Goldstone
fields. All particles, defined with respect to a ground state, are rearranged into
fixgroup H-multiplets. The interaction symmetry G is spontaneously broken,
leaving the particle symmetry H. If operations from the degeneracy manifold
are implemented, in addition, by Lorentz vector gauge fields, e.g., by four
U(2)-gauge fields, their ground-state–related rearrangement with the corre-
sponding Nambu–Goldstone degrees of freedom leads, in a particle analysis,
to massive vector fields, e.g., to three weak spin-1 massive bosons.
6.6.1 Electroweak Symmetry Reduction

In the electroweak standard model, the definition of particles requires the
transition from the hyperisospin interaction symmetry U(2) = U(12 )◦SU(2)
(no direct product) to an abelian electromagnetic subsymmetry for the
ground state and particles. Taking into account the nontrivial central corre-
lation U(12 ) ∩ SU(2) = {±12 }, a Cartan torus of U(2) is U(1)+ × U(1)−
12 ±τ 3
with e 2 α ∈ U(1)± (projective generators). For a ground-state fixgroup
U(1)+ , as anticipated in the hypercharge representation numbers, chosen
above, the electromagnetic charge number is the sum of the hypercharge
number and the third isospin eigenvalue:
z = y + T 3.
Asymmetric boundary conditions are implemented by choosing from

a ground-state manifold U(2)/U(1)+ (Goldstone degrees of freedom) one
representative and hence violating (stripping, rearranging) the U(2)/U(1)+ -
related transformations in hyperisospin U(2). The symmetry rearrangement
is implemented by an ad hoc scalar field, the Higgs field Φ in the defining
U(2)-representation, with a Higgs potential:
V(Φ
Φ) = g0
Φ
8 (Φ Φ − M 2 )2 , g0 > 0.
The minima of the potential,
V(Φ
Φ) = min ⇒ Φ
Φ Φ (x) = M 2 ,
give the breakdown-characterizing mass unit and, taking the appropriate rep-
resentative, leaves an electromagnetic U(1)+ (ground-state fixgroup) as the
remaining particle symmetry:

= 12 −τ
0 3
Φ
Φ(x) = M 2 M.
The U(2)-asymmetric effects in Weinberg’s original “model of leptons,”

Φ) = [(∂ − i A0 122+Aτ )Φ
L(Φ Φ ] − V(Φ
Φ ][(∂ + i A0 122+Aτ )Φ Φ)
− ge (eΦ
Φl + lΦ
Φ e ),
come in the particle structure of the U(2)-gauge fields via the covariant
derivative of the Higgs field and in the lepton particles via a Yukawa interac-
tion. The ground-state value of the Higgs field gives the mass contributions:
/
/
= M 2 tr 12 −τ
3
Φ )/
L(Φ 2 ( A0 122+Aτ )2 − M ge (el2 + l2 e ).
Φ
Φ =Φ
For the lepton fields, the U(1)+ -trivial component (up component in the
isospin doublet, neutrino) remains massless. The electron mass me for the
massive electron Dirac field Ψ e with the down component in the isospin
doublet as the left-handed part can replace the Yukawa coupling constant ge ;
both are theoretically undetermined parameters in the model:
massless neutrino: ν e = l1 , mν = 0,
massive electron: Ψ e = (eL , eR ) = (l2 , e), me = M ge .

In general, a Yukawa coupling for a left-handed isodoublet Q = U L
DL
with hypercharge number y and two corresponding right-handed isosinglets
(UR , DR ) with adapted hypercharge numbers y ± 12 generates a mass term
after the symmetry reduction:
/ /
/ /
Φ )/
LYuk (Φ = −gD (DR ΦQ + QΦ Φ DR )−gU (UR Φ Q + QΦΦUR )/
Φ
Φ =Φ Φ
Φ =Φ
= −mD (DR DL + DL DR ) − mU (UR UL + UL UR )
= −mDΨ DΨ D − mU Ψ U Ψ U .
Left and right components constitute two massive Dirac fields:

Ψ U = (UL , UR ), Ψ D = (DL , DR ), mD = M gD , mU = M gU .
An SU(2)-index notation for the Yukawa couplings looks like the following:
DRΦ Q = DRΦ α Qα , URΦ Q = URΦ β βα Qα .

Pairing the left-handed lepton fields l = νeLL not only with a right-handed
electron eR but also with a right-handed neutrino partner νR , a nontrivial
neutrino mass mν = M gν is possible. Such a right-handed isosinglet field
νR comes with trivial hypercharge y = 0, i.e., without any internal gauge
interaction (“sterile neutrino”).
The vector-field–related terms in the Higgs field coupling,
τ )2 2 2
M 2 tr 12 −τ = M 2 tr 12 −τ
3 (A0 12 +A 3
[(A0 ) +(A) ]12 +2A0 A
τ
2 4 2 4 2
(A ) 2
+(A ) 2
+(A −A )
= M2 1 2
4
3 0
,
contribute to the free theory of two massive charged vector fields W ∈ {A1,2 }:
k
Wj −∂ j Wk Fkj Fkj M 2 W k Wk g22
L(W) = Fkj ∂ 2 + g22 4 + 2 2 ⇒ m2W = 2
2 M .
The two neutral vector fields come with the free theory:
∂ k Aj0 −∂ j Ak ∂ k Aj3 −∂ j Ak
L(A0 , A3 ) = F0kj 2
0
+ F3kj 2
3
0 kj 3 kj
F F F F (A −Ak )2
k
+ g12 kj4 0 + g22 kj4 3 + M 2 3 4 0 .
The diagonalization from interaction to particle fields, required by the non-
diagonal mass term, is performed by the Weinberg SO(2)-rotation:
3
F2 F2 2 2 cos θ − sin θ g2 F
g12 40 + g22 43 = γ 2 G4 + g 2 F4 , with F0
= γG ,
sin θ cos θ
g1 1A gF
1

cos θ − sin θ g2 3 γZ
F0 ∂A0 + F3 ∂A3 = G∂Z + F∂A, with sin θ cos θ 1
A0 = 1
A . g1 g
1
It involves the Weinberg angle θ and dual normalizations (κ, for the cou- κ)
pling constants κ ∈ {g1 , g2 , γ, g}. The combination Z = A3 − A0 arises as
a massive vector field; the massless gauge field A carries the ground-state
fixgroup U(1)+ transformations; this defines the particle field normalizations
{g, γ} in terms of the gauge field normalizations and the Weinberg angle:
Z = A3 − A0 ⇒ cos θ
g2 =
sin θ
g1 = γ1 ,
A = sin2 θ A3 + cos2 θ A0 ⇒ sin θ = g
g2 .
The Weinberg rotation,

⎧
⎪
⎨ (A0 , A3 ) −
→ (Z, A),
electroweak: U(1 2 ) ◦ U(1)3 = U(1)+ × U(1)− ,
1
, 1|1, 1 ⎪
⎩ ei(γ0 12 +γ3 τ 3 ) = ei(γ0 +γ3 ) 12 +τ 3 12 −τ 3
g12 g22 γ 2 g2 2 × ei(γ0 −γ3 ) 2 ,
defines the electroweak orthogonal triangle with ( g11 , g12 ) the orthogonal sides.
The hypotenuse g1 is related to Sommerfeld’s fine structure constant αS =
g2
4π ∼ 1
137 for the electromagnetic U(1)+ -gauge field A:

1 1 1
g2 g2 = g12 + g22 ,
g1 g2 = γg, g1 = cot θ
γ 2 = g12 + g22 .
Multiplication by the area dilation factor γg = g1 g2 gives the similar dual

triangle (g2 , g2 ) with the squared lengths (g22 , g12 |g 2 , γ 2 ).
The Weinberg rotation diagonalizes the free theory with two neutral vec-
tor particle fields:
kj
7 2
L(A0 , A3 ) = Fkj ∂ A −∂
k j j
Ak 2 Fkj F mA = 0,
+ g
2 4
kj ⇒ 2
m2Z = γ2 M 2 .
k
∂ k Zj −∂ j Zk G G Z Z
+ Gkj 2 + G2 kj
4 + M2 4
k
The electroweak gauge field interactions are rearranged with the neutral
and charged vector particle fields:
(
A0 = A − sin2 θ Z
, A1 ∓ iA2 = W± .
A3 = A + cos2 θ Z

In general, one obtains for a left-handed isodoublet Q = U L
DL with hyper-
charge number y and two corresponding right-handed isosinglets (UR , DR )
with hypercharge numbers y ± 12 as the interaction with the vector particle
fields,
+ 2 )UR σUR + (y − 2 )DR σDR ]A0,+ Qσ̌ 2 Q A

1 1
τ
[yQσ̌Q + (y +
1
= (y + 2 )ΨΨU γΨΨU + (y − 2 )Ψ 1
ΨD γΨ ΨD A
+ ,
1−4(y+ 12 ) sin2 θ+iγ5 1+4(y− 12 ) sin2 θ+iγ5
+ ΨU γ 4 Ψ U − Ψ D γ 4 Ψ D Z

+ UL σ̌DL W− + DL σ̌UL W+ ,
where the parity combinations have been used in Dirac fields Ψ U,D , e.g.,
for Ψ U ,
ΨU = UL σ̌UL + UR σUR ,
Ψ U γΨ
ΨU γγ5Ψ U = UL σ̌UL − UR σUR .
iΨ
This leads for the leptons with y = − 12 to
(− 12 lσ̌l − eσe )A0 + lσ̌ τ2 l A

2
= − Ψ e γΨΨe A + 12 ν e σ̌νν e − Ψ e γ 1−4 sin4 θ+iγ5 Ψ e Z
+ ν e σ̌eL W− + eL σ̌νν e W+ .
1
A “sterile neutrino” remains “sterile.” The quark fields with y = 6 have the
electroweak interactions

6 qσ̌q + 3 uσu − 3 dσd
1 2 1
A0 + qσ̌ τ2 q A Ψu − 13 Ψ d γΨ
= 2 Ψ u γΨ Ψd A
3

1− 83 sin2 θ+iγ5 1− 43 sin2 θ+iγ5
+ Ψu γ 4 Ψu − Ψdγ 4 Ψd Z
+ uL σ̌dL W− + dL σ̌uL W+ .
The electroweak model contains many basically unknown parameters,

2
especially the gauge field normalizations g1,2 and the ground-state or elec-
2
troweak mass unit M . The weak breakdown mass can be replaced by the
experimentally determined Fermi constant for the four fermion interactions as
the low-energy limit of the charged weak interaction, i.e., for the propagator:
g2 g22
q 2 → 0 : − q2 −m
2
2 → m2W
= 2
M2 ,
W
experiment: M ∼ 169 GeV

c2 .
To determine the electroweak orthogonal triangle, one needs one constant in

addition to the experimentally determined fine structure constant, e.g., the
experimentally determined Weinberg angle:
g2
(
= αS ∼ 137
1
experiment: 4π ⇒ ( g12 , g12 | γ12 , g12 ) ∼ (8.4, 2.5|1.9, 10.9),
sin2 θ ∼ 0.23 1 2
from which the dual electroweak mass triangle can be computed:

2
2 2
(m2W , m21 |m20 , m2Z ) = (g22 , g12 |g 2 , γ 2 ) M2 = sin12 θ , cos12 θ |1, sin42 2θ g M
2 ,
(mW , m1 |m0 , mZ ) ∼ (2.1, 1.2|1, 2.4) 37 GeV

c2 .
The weak boson masses are in good agreement with the experimental results:
mW ∼ 80 GeV
c2 , mZ ∼ 91 GeV
c2 .
6.6.2 Transmutation from Hyperisospin

to Electromagnetic Symmetry
The electroweak symmetry “breakdown” (rearrangement) is a transmutation
from a hyperisopin U(2)-compatible framework for the interaction to a for-
mulation with the remaining electromagnetic fixgroup U(1)+ -symmetry for
the particles (see Chapter 7):
⎪
⎪
{t ∈ U(2)⎪
⎪Φ
0
Φ (x) = M =t M0
}∼= U(1)+ .
The U(2)/U(1)+ -isomorphic orbit of the Higgs field in the Hilbert space
C2 provides translation-dependent Lie parameters for the fixgroup classes
(symmetric space):
Φ (x) = e 2 γ⊥ (x) 12 −τ
i 3
2 R(x),
R x −→ 2 γ⊥ (x) = i −γ
4 i

3 (x)12 +
2
γ (x)τ
∈ log U(2),
Φ (x) i
R x −→ V (x) = v R(x) = e 2 (x) ∈ U(2).
4 γ ⊥
The representations of the electromagnetic orientation manifold (Gold-

stone or ground-state manifold) on a vector space W ∼
= C1+2T with U(2)-re-
presentation,
= G i −→ U(1 + 2T ), R −→ D(v( R )), R = |Φ

Φ|,
3 Φ Φ
(U(2)/U(1)
+ )repr
1
Φ2 Φ
Φ 1
v( R ) = R −ΦΦ Φ 2 = e 2 ∈∈U(2)/U(1)+ ,
γ ⊥
1
are products of the fundamental representation. A “left” hyperisospin U(2)

action gives the representation with the U(2)-transformed Higgs vector up
to a “right” action with the electromagnetic fixgroup U(1)+ :
γ0 12 +i
γτ
u = ei 2 ∈ U(2) ⇒ u ◦ v( Φ
R ) = v( R ) ◦ t(u),
Φ
u.Φ
12 +τ 3
with t(u) = ei2γ0 2 ∈ U(1)+ .
The G 3 -representations are decomposable into transmutators from U(2)-

vectors (boldface) to U(1)+ -active vectors with G 3 -frozen components (un-
derlined) α = 1, 2:
U(1)+
vecU(2) W ∼
= W ι , W ι ∈ vecU(1)+ ,
ι ⎫
W Eα = D(V )ααE
α
∈ W ι, ⎪
⎬
ι with Eα Eα = Eα E α ,
W T Eα α ∈
= Eα D(V )α W ιT ⎪
⎭
ι
especially for the Higgs field, where the dilation degree of freedom R consti-
tutes the frozen field:

(αβ Φ β , Φ α ) = (V1α , V2α )R ∼
= V ◦ 12 +τ
3
, 12 −τ
3
R,
2 2
2 ∼ 12 +τ 3 12 −τ 3
β 1
(αβ Φ , Φ α ) = R (Vα , Vα ) = R 2 , 2 ◦ V .
Hence, U(2)-invariants have the same form in the frozen U(1)+ -active fields
as in the boldface unfrozen U(2)-active ones. For example, the Yukawa cou-
pling above with a U(2)-doublet fermion field Q, like the left-handed lepton
fields, is written with the U(1)+ -fields Q1,2 = Vα1,2 Qα = (UL , DL ):
Φ) = −gD (DRΦ Q + QΦ
LYuk (Φ Φ DR ) − gU (URΦ Q + QΦ
ΦUR )
= −gD R(DR DL + DL DR ) − gU R(UR UL + UL UR ).

6.7 Lie Group Coset Bundles 167
Φ
The group element V (x) = v( Φ (x)
R(x) ) ∈
∈U(2)/U(1)+ defines a gauge
transformation with a translation-parametrized Lie algebra element as rep-
resentative of the fix-Lie algebra classes log U(2)/ log U(1)+ :
R4 −→ log U(2), x −→ l(V (x)),

γc γa γb
−(∂γ3 )12 +(∂γa )[δab sin γ
γ +abc γ
1−cos γ
+ (1− sin γ
γ )]τ
b
l(V ) = (∂V ) ◦ V = i 2
γ γ2
.
The Lie algebra-valued U(2)-gauge fields, in the (2×2) Pauli representation,

R4 −→ log U(2), x −→ A(x) = i A0 (x)122+A(x)τ ,
R4 −→ U(2), x −→ U (x),
R4 −→ log U(2), x −→ l(U (x)), l(U ) = (∂U ) ◦ U ,
with the transformation behavior
A −→ U ◦ A ◦ U + l(U ),
are stripped of the Higgs field-provided U(2)/U(1)+ -degrees of freedom:
R4 −→ log U(1)+ , x −→ A(x), A = l(V ) + V ◦ A ◦ V.
There remains only the electromagnetic U(1)+ -gauge degree of freedom,
A −→ V ◦ U ◦ A ◦ U ◦ V + l(V ◦ U )
parametrized with spacetime translations,
R4 x −→ V (x) ◦ U (x) ∈ U(1)+ ,

R4 x −→ l(V (x) ◦ U (x)) ∈ log U(1)+ .
The Higgs field derivative,
Φ = V (∂ − A) 12 −τ
3
(∂ − A)Φ 2 R,
leads with the ground-state-characterizing mass R = M to the mass terms

for the spin-1 particles (weak bosons) in the vector fields related to the three
ground-state degrees of freedom U(2)/U(1)+ .
6.7 Lie Group Coset Bundles

A principal bundle is a Lie group bundle H(M) ∈ lgrpK (M), where the typical
fiber H ∈ lgrpK coincides with its structural group. The structural group has
to act on itself either by left or by right translations. An example is the frame
bundle of a manifold M ∈ dif R with H = GL(n, R).
A Lie group with a closed subgroup defines the associated coset bundle,
a principal bundle with the subgroup as structural group:
G ⊇ H : lgrpR G −→ H(M) ∈ lgrpR (M), with M = (G/H)r ∈ dif R .

The base manifold is constituted by a representative for each coset

(H-equivalence class):
π : G −→ G/H −→ M = (G/H)r ⊆ G, g −→ gH −
→ gr = (gH)r ,
with gr H = gH.
The local fiber is the fixgroup Ggr ∼= H of the representative. As manifold, a
coset bundle is isomorphic to the full group:

G∼= H(M) = (gr , Ggr ), Ggr ∼
= H.
gr ∈M=(G/H)r
Examples are the spheres, hyperboloids, and Euclidean spaces with the
rotation group SO(s) as typical fiber and the corresponding manifold iso-
morphies. They can be visualized for s = 2, e.g., the antipodally identified
2-sphere with all diameters,
SO(1 + s) ∼
= SO(s)(Ωs ), with Ωs ∼
= SO(1 + s)/SO(s),
SO0 (1, s) ∼
= SO(s)(Y s ), with Ys ∼
= SO0 (1, s)/SO(s),
SO(s) ×
Rs ∼
= SO(s)(Rs ), with Rs ∼
= SO(s) × Rs /SO(s).
The coset bundle notation of a direct product group is H1 × H2 ∼ = H1 (H2 ).

It is a trivial bundle.
The left multiplication with a group element k ∈ G on the base with the
representatives gives the shifted representative up to the right multiplication
with a subgroup element (Wigner element ). It defines an isomorphism f of
the group coset bundle:
f Lk : M⎧ −→ M, gr −→ (kg)r
H(M)
⏐ −→ H(M)
⏐ ⎨ kgr = (kg)r h(k, gr ) ∈ G,
⏐ ⏐
π π, with (kg)r = (kgH)r ∈ M,
⎩
M −→ M h(k, gr ) ∈ H,
Lk f |Ggr : H ∼
= Ggr −→ G(kg)r ∼ = H,
hr −→ hr h(k, gr ).
(G/H)r = M,
The Lie algebra bundle for a Lie group coset bundle,
M = (G/H)r : lgrpR (M) H(M) −→ log H(M) ∈ lagR (M),
comes with the subgroup-H gauge transformations in a connection.
6.8 Electroweak Spacetime

Reissner spacetime with metric, curvature and Ricci tensor (see Chapter 2):

2 2
g = 1 − 2
ρm + ρz2 dt2 − 2dρm 2
z
− ρ2 dω 2 ,
1− ρ + ρ2
6.8 Electroweak Spacetime 169
⎛ ⎞
32
2m
− z 0 0 0
⎜ ρ2
⎟
ρ

⎜ 2 ⎟
⎜ ⎟
m
0 − − z 12 0 0
R ) ∼
= ρ2 ⎜
ρ2
⎟,
dabc 1 ρ
(t, ρ, ω
⎜ − 2ρm +
2 ⎟
⎝ ⎠
0 0 z 0
ρ2

m 2
0 0 0 ρ − z
ρ2
12
1 0 0

Rab ∼
2z
−1
• (t, ρ, ω
) = ρ4
0 0 ,
0 0 12
has a local SO(2)-invariance with the decomposition of the Lorentz Lie alge-
bra and tangent Minkowski spacetime:
log SO0 (1, 3) ∼

SO(2)
= R6 : 6 = 1 ⊕ 2 ⊕ 1 ⊕ 2,
SO(2)
R4 : 4 = 1 ⊕ 1 ⊕ 2.
The local SO(2)-invariance is reflected by the invariance of the energy-

momentum tensor of the charged matter fields with respect to a local
electromagnetic SO(2) ∼
= U(1)-transformation (see Chapter 5), e.g., for a
scalar field,
Tab (Φ Φ ∗ (∂ a − izAa )Φ
Φ ) = (∂ b + izAb )Φ Φ
− η [(∂ + izAc)Φ
ab c Φ ∗ (∂c − izAc )Φ
Φ − m2Φ ∗Φ ].
This suggests the identification of the local invariance group Hg of the

spacetime metrical tensor g with the internal (chargelike) operation group
for spacetime fields. If such an identification is taken as the guiding princi-
ple, there are some suggestive arguments to distinguish a spacetime mani-
fold, which shows an integrative symbiosis of spacetimelike (horizontal) and
chargelike (vertical) transformation groups.
An immediate argument against such an approach is the identification of
an external subgroup, i.e., a Lorentz subgroup Hg ⊂ SO0 (1, 3) ∼ SL(2, C)
with an internal operation group, i.e., a group action on the chargelike degrees
of freedom as given by hypercharge U(1), isospin SU(2), and color SU(3) in
the standard model of particles.
The situation is more subtle: There is a conceptual difference between a
group and the action of the group on itself: The left–right multiplications
of the group G define the bi-regular action with the group doubling G × G
(see Chapter 8). Such a doubling will be used for the dichotomy of external
and internal operations with the prominent example of the isomorphic but
action-different group SU(2) for spin and isospin in SU(2) × SU(2). This
doubling also occurs in the Cartan classification of the globally symmetric
Riemannian manifolds, used there for the compact and noncompact type II
manifolds (see Chapter 2).
If the local invariance group Hg ⊆ SO0 (1, 3) of a real four-dimensional
spacetime with causal signature (M(1,3) , g) is identified with internal gauge
operations involving the electromagnetic U(1) ∼ = SO(2) ⊂ Hg , it has to be
brought in connection with the real four-dimensional hyperisospin group U(2)

of the electroweak standard model with the centrally1 correlated groups U(1)
for hypercharge and SU(2) for isospin. The center classes of U(2) constitute a
rotation group SO(3), a Lorentz subgroup. It is taken as the local invariance
group of spacetime:
U(2) = U(12 ) ◦ SU(2) ∼

=
U(1)×SU(2)
I(2) ,
∼ ∼
U(2)/U(1) = SU(2)/I(2) = SO(3) = Hg ⊂ SO0 (1, 3).
A local invariance group SO(3) ∼ SU(2) characterizes position sub-

manifolds M(0,3) ⊂ M4 with maximal hyperbolic, flat, and spherical motion
groups, as used for Friedmann universes (see Chapter 1):
⎧ 3 ∼
⎨ Y = SO0 (1, 3)/SO(3) ∼ = SL(2, C)/SU(2),
(0,3) ∼ 3 ∼ R3 /SO(3) ∼
M = R = SO(3) × = SU(2) × R3 /SU(2),
⎩ 3 ∼ ∼ SU(2)×SU(2)
Ω = SO(4)/SO(3) = SU(2) .
The hyperbolic and spherical position manifolds constitute a type II

noncompact–compact pair in the Cartan classification. Such pairs use for
the compact IIc -partner the doubling K = diagK×K
(K×K) of a group K, here of
K = SU(2), and, in the doubling, the classes of the diagonal group, here
Ω3 ∼
∗
= SU(2), and, for the noncompact IInc -partner K×K K with the Weyl
complexification in the Lie algebra log K ⊕ log K ∗ ∼ = (iR ⊕ R)s , here
Y3 ∼
= SL(2, C)/SU(2). The three-dimensional nonflat position manifolds are
isomorphic to the type I orthogonal Riemannian spaces of subtype BD I.
A real four-dimensional operational manifold for spacetime as a transitive
homogeneous space consisting of one orbit Gg /SO(3) requires a real sevven-
dimensional motion group Gg ⊃ SO(3). An extension of the isotropy group
SU(2) to U(2), whereof SO(3) describes the U(12 )-classes, requires a real
eight-dimensional supgroup G ⊃ U(2):
= Gg /SO(3) ∼
M4 ∼ = G/U(2), Gg ∼
= G/U(1).
The nonflat position manifolds, hyperbolic Y 3 and spherical Ω3 , can be

extended with an additional hypercharge fixgroup U(1) as follows:
M(1,3) ∼
= GL(2, C)/U(2), M(0,4) ∼
=
U(2)×U(2)
U(2) .
Only noncompact electroweak spacetime D(2), i.e., the classes of the hy-
perisospin group in the extended Lorentz group, with a hyperbolic position
submanifold has a causal structure:
D(2) = M(1,3) ∼
= D(1) × Y 3 ,
(0,4) ∼
U(2) = M = U(1) × Ω3 .
1 The SU(n)-center is isomorphic to the multiplicative cyclotomic group I(n) = {z ∈
⎪
⎪ n
C⎪⎪ z = 1} and to the additive group Zn = Z/nZ. U(1n ) ⊂ GL(n, C) is the phase group
for a complex n-dimensional vector space.
6.8 Electroweak Spacetime 171
The definition of electroweak spacetime D(2) ∼ = D(1) × SO0 (1, 3)/SO(3)

does not involve the real eight-dimensional color group SU(3) as a constitu-
tive operation group.
Electroweak spacetime is the lowest-dimensional noncompact nonabelian
case n = 2 of the noncompact–compact manifold pairs,
D(n) = GL(n, C)/U(n) ∼

= D(1) × SL(n, C)/SU(n),
U(n) U(n)×U(n) ∼
= U(1) × SU(n)×SU(n) .
U(n) SU(n)
The abelian case n = 1 gives the groups D(1) = D(1) and U(1) = U(1). The
motion groups of these manifolds can be considered as coset bundles with
D(n) and U(n) as base manifolds and the local invariance group U(n) as the
typical fiber:
GL(n, C) ∼
= U(n)(D(n)), U(n) × U(n) ∼
= U(n)(U(n)).
D(n) is constituted by the classes of the maximal compact group U(n)

in the general linear group GL(n, C). It uses the polar decomposition of
GL(n, C) into nonabelian phases U(n) and positive moduli D(n). The com-
pact group has a Lie algebra log U(n) ∼
2
= (iR)n with n2 imaginary param-
eters; D(n) is the positive cone in the C ∗ -algebra AL(Cn ) of the complex
(n × n)-matrices: The real n2 -dimensional vector subspace with the hermi-
tian (n × n)-matrices,
⎪
⎪
Rn = {x ∈ AL(Cn )⎪
2
⎪x = x } ∈ vecR ,
has a causal order by a positive spectrum:
AL(Cn ) x # 0 ⇐⇒ x = x∗ , spec x ≥ 0 ⇐⇒ x ∈ Rn+ .

2
The hermitian matrices parametrize the translations of real n2 -dimensional

Minkowski spacetime and are isomorphic to the tangent spaces of D(n), which
itself is parametrizable by the spacetime translations of the open future cone
D(n) ∼
2
= Rn+ (see Chapter 11).
Chapter 7
Relativities and
Homogeneous Spaces
This chapter discusses the dichotomy and connection of external and internal
operations in real four-dimensional electroweak spacetime D(2) = D(1) ×
Y 3 = GL(2, C)/U(2) will be under the label unitary relativity. To see its
general and specific structures, unitary relativity will be considered as one
example in five relativities: “up-down” or perpendicular relativity as realized
after discovering the earth’s surface to be spherical; then rotation or space
and time relativity, as used in what we call special relativity; then Lorentz
group or flat Minkowski spacetime relativity or, also, with Wigner’s definition,
particle relativity as an important ingredient (local inertial systems) of general
relativity; then electromagnetic relativity as used for the particle definition
in the standard model of electroweak interactions [58]; and, finally, unitary
relativity with “spacetimelike” and “chargelike” operations.
Relativity will be defined by operation group classes, e.g., in special rel-
ativity, the distinction of your rest system determines a decomposition of
spacetime translations into time and position translations. Compatible with
this decomposition is your position rotation group SO(3) as a subgroup of
the orthochronous Lorentz group SO0 (1, 3). There are as many decomposi-
tions of spacetime into time and position as there are rotation groups in a
Lorentz group. The rotation group classes are parametrizable by the points
of a one-shell three-dimensional hyperboloid Y 3 ∼ = SO0 (1, 3)/SO(3) that
give the momenta (velocities) for all your possible motions, i.e., by your
mass hyperboloid q02 − q 2 = m2 . Another example: The perpendicularities
of mankind, if earthbound, are characterized by the axial rotation groups in
a rotation group and parametrizable by the two coordinates of the earth’s
surface Ω2 ∼ = SO(3)/SO(2).

174 Chapter 7 Relativities and Homogeneous Spaces
Now in general: The distinction of an “idolized” operation subgroup1 H

in a “general” operation group G picks one element in the G-symmetric
space G/H, which stands for the relativity of the “idolized” group, called
H-relativity. Relativity groups H, i.e., the local fixgroups in G/H, are can-
didates for gauge groups. An “idolization” [1] goes, negatively, with the
“narrow-minded” assumption of an absolute point of view, e.g., absolute up-
down, absolute time, absolute Minkowski spacetime (particle universality),
or, positively, with the distinction of a smaller operation symmetry, enforced,
e.g., by initial or boundary conditions. Important examples are degenerate
ground states (“spontaneous symmetry breakdown”), where, by distinguishing
one ground state from the symmetric space G/H as the -degeneracy man-
ifold, an “interaction-symmetry” G is reduced to a “particle-symmetry” H,
e.g., the ground states of superconductivity, superfluidity, ferromagnetism,
and the electroweak standard model.
All of this gives the first four columns of the following table, which to-
gether with the last one will be discussed with their representations in more
detail ahead
“General” “Idolized” Homogeneous Relativity

Relativity group G subgroup space M ∼
= G/H manifold
(rc + r, r) H(fiber) (base manifold) parameters
Axial rotation SO(3) 2-sphere
two spherical
(perpendicular) ∼ SU(2) SO(2) Ω2 ∼= SO(3)/SO(2)
∼ coordinates
relativity (1, 0) = SU(2)/SO(2)
Rotation SO0 (1, 3) 3-hyperboloid
SO(3)
(special) ∼ SL(2, C) Y3 ∼= SO0 (1, 3)/SO(3) three momenta
∼ SU(2) ∼
relativity (2, 1) = SL(2, C)/SU(2)
Lorentz group tetrad or metric manifold
GL(4, R) 10 components
(particle) O(1, 3) M10 ∼ = GL(4, R)/O(1, 3)
(4, 4) ∼ for metrical tensor
relativity = D(1) × SO0 (3, 3)/SO0 (1, 3)
Electro-
U(2) Goldstone manifold three weak
magnetic U(1)+
(2, 0) G3 ∼
= U(2)/U(1)+ coordinates
relativity
Unitary future 4-cone
GL(2, C) four spacetime
(electroweak) U(2) D(2) ∼
= GL(2, C)/U(2)
(4, 2) ∼ coordinates
relativity = D(1) × Y 3
Coset bundles H(M) for five relativities
Somewhat in accordance with the historical development, the “general”

operations of one relativity can constitute the “idolized” group of the next
relativity as seen in the two chains ending in full general linear groups, a
real one for spacetime concepts, from flat to spherical earth to special and
general relativity, and a complex one for interactions, from electromagnetic
to electroweak transformations and their spacetime (gauge) dependence:
SO(2) ⊂ SO(3) ⊂ SO0 (1, 3) ⊂ GL(4, R),
U(1)+ ⊂ U(2) ⊂ GL(2, C).
1 In general, H is assumed to be no normal subgroup. Otherwise, it “disappears” in the
quotient group G0 = G/H.

Chapter 7 Relativities and Homogeneous Spaces 175
There is a connection of the two chains on the level of the orthochronous

Lorentz group SO0 (1, 3) and its cover group SL(2, C) by
SO0 (1, 3) ∼
= GL(2, C)/GL(1, C) ∼
= SL(2, C)/{±12 }.
All groups in the five relativities considered are real Lie groups. All “gen-
eral” groups have reductive Lie algebras2; for perpendicular and rotation
relativity they are even simple. Perpendicular and electromagnetic relativ-
ity have a compact “general” group. With the exception of Lorentz group
relativity, all “idolized” groups are compact subgroups. The second column
contains the dimension rc + r of the maximal abelian subgroups, which is the
rank of the group G generating Lie algebra L = log G, and of the maximal
noncompact abelian subgroups, i.e., the real rank r. The rank rc + r gives the
number of independent invariants, rational or continuous, that characterize
a G-representation. The real rank r is the maximal number of independent
continuous invariants.
Unitary relativity GL(2, C)/U(2), i.e., the complex linear relativization
of the maximal compact subgroup with the internal “chargelike” hypercharge
and isospin operations U(2), is parametrized by a noncompact real four-di-
mensional homogeneous space, called electroweak spacetime D(2). Unitary
relativity is visible in the spacetime dependence of quantum fields, which
represent the internal operations. The representations of unitary relativity
D(2) are characterized by two continuous invariants, which, in appropriate
units, can be taken as two masses. The D(2)-representations determine the
spacetime interactions with their normalization, especially the gauge inter-
actions with their coupling constants, which are related to the ratio of the
two invariants, and, for the D(2)-tangent translations R4 , the particles and
their masses. The common language for interactions and elementary particles
is the representation theory and harmonic analysis of unitary relativity (see
Chapters 11 and 12).
There is a mathematical framework, almost tailored for relativities G/H:
the theory of induced representations, pioneered by Frobenius [25], used
for free particles by Wigner [62] and worked out for noncompact groups
especially by Mackey [43]. There, a subgroup H-representation induces a full
group G-representation leading to a G × H-representation as subrepresen-
tation of the two-sided regular G × G-representation. Relativities G/H are
acted on by the bi-regular subgroup G×H. Such a dichotomic transformation
property with a doubled group in G × G as group and “isogroup” is familiar,
with respect to the Lorentz and the isospin group, SU(2) × SU(2) as spin
and isospin, from the fields in the electroweak standard model. The theory
is not easy to penetrate, especially for noncompact nonabelian groups. All
the mathematical details are given in the textbooks by Helgason [36], Knapp
[41], Folland [24] and, especially for distributions, by Treves [56].
2 A Lie group of complex hermitian matrices is reductive. A finite-dimensional Lie
algebra L with semisimple “square” [L, L] is reductive. Then, L = A ⊕ S with abelian

and semisimple Lie subalgebras.
In the following, only some motivating and qualititive mathematical

remarks will be given with respect to this theory, which will be used in phys-
ical implementations. This chapter works with finite-dimensional relativity
structures, which may not be so familiar in such a conceptual framework.
After a parametrization of the relativity manifold G/H, its representations,
called transmutators and closely related to n-beins, will be given, which
mediate the transition from an idolized group H to the full group G. By
products of the fundamental transmutators, all finite-dimensional relativity
representations can be constructed as used, e.g., in the transition from the
fields for the electroweak interactions to the asymptotic particles.
7.1 Parametrization of Relativity Manifolds

For real homogeneous relativity manifolds M ∼ = G/H with Lie groups, there
are operation-induced parameters, e.g., the three momenta (velocities) for
rotation (special) relativity or the three weak coordinates of electromagnetic
relativity as used in the mass modes of the three weak bosons.
The action of a “general” group G on a set S, denoted by •, decomposes
S into disjoint orbits G • x for x ∈ S that are isomorphic to subgroup classes
G•x ∼ = G/H, where the “idolized” group H ∼ = Gx is the fixgroup (fixer,
“little” group, isotropy group) of the G-action. The elements of homogeneous
spaces gH ∈ G/H are group subsets (cosets). The cosets have representatives
gr = (gH)r ∈ gH ∈ G/H, denoted as gr ∈∈G/H, which can be characterized
by what will be called relativity parameters, a real parametrization of the
subgroup classes.
Relativity parameters can be obtained via orbit parametrizations. The
real Lie groups considered are linear groups H ⊆ G ⊆ GL(V ), acting on real
or complex vector spaces V . The orbit G • x parametrizes the homogeneous
space G/H by V -vectors and their components with respect to a basis:
⎪
⎪
x ∈ V, H ∼ = Gx = {g ∈ G⎪ ⎪g • x = x}
⇒ G/H ∼ = G • x ⊆ V.
7.1.1 Goldstone Manifold for Weak Coordinates

The hypercharge-isospin group U(2) acts, in the defining representation, on
a complex two-dimensional vector space:
iα3
cos θ2 −e−iϕ sin θ2
U(2) u = eiα0 eeiϕ sin θ
e−iα3 cos θ
.
2 2
Each nontrivial vector has a U(1)-isomorphic fixgroup, e.g., e2 , which defines

U(1)+ as an “idolized” electromagnetic subgroup:
2iα
C2 ∼ 0 0
= V e2 = 1 ⇒ 1 ∈ U(2)e = U(1)+ .
e 0
2
0
7.1 Parametrization of Relativity Manifolds 177
The orbit points for the chosen “startvector,” here u • e2 , and for its
U(2)-orthonormal partner, here (u • e2 )⊥ , give the two columns of the
matrix parametrization v ∈ U(2) of the real three-dimensional Goldstone
manifold G 3 :
⎪
⎪
G3 ∼ = U(2)/U(1)+ ∼ i(α•3e−α)0⊥) , u •θ e ) =
= {((u 2 2
v⎪⎪u ∈ U(2)},
−i(ϕ−α0 )
e cos 2 −e sin θ2
v = ei(ϕ−α0 ) sin θ e−i(α3 −α0 ) cos θ .
2 2
In the standard model of electroweak interactions, the vector space V ∼ = C2

desribes the chargelike degrees of freedom of the Higgs field Φ ∈ V . The three
weak parameters (α3 − α0 , ϕ − α0 , θ) parametrize electromagnetic relativity
G 3 . As manifold, not as homogeneous space, G 3 is isomorphic to Ω3 and
SU(2).
7.1.2 Orientation Manifolds of Metrical Tensors

With the exception of electromagnetic relativity in the last subsection, all
relativity parameters will be given by the “general” group G-orbit of a met-
ric that is invariant under the action of an “idolized” subgroup H. In this
context, Weyl [59] called the homogeneous space G/H for H-relativity ori-
entation manifold of the metric (bilinear or sesquilinear product), familiar
from the orientations of the metrical ellipsoid of 3-position. The transforma-
tions involved constitute n-beins.
The invariance of a vector space metric g with respect to the action of a
linear group H,
GL(V ) ⊃ H h, g(x, y) −→ g(h • x, h • y) = g(x, y) for all x, y ∈ V,
gives the parametrization of the fixgroup classes G/H by the G-orbit of the
metrical tensor g:
⎪
⎪
H = {h ∈ G ⊆ GL(V ⎪ )⎪
⎪h ◦ g ◦ h∗ = g}
⎪
⇒ G/H ∼ = {g ◦ g ◦ g ∗⎪
⎪g ∈ G}.
For V ∼
= Kn , G/H is parametrized by (n × n)-matrices.
7.1.3 Orientation Manifold of Metrical Hyperboloids

A real vector space V ∼= Rn has a causality structure by embedding the cone
of the positive numbers R+ −→ V+ ⊂ V into the “future” cone of the vector
space x # 0 ⇐⇒ x ∈ V+ . A nontrivial “future” cone V+ = {0} can be defined
by a bilinear symmetric form with “causal” signature (t, s) = (1, s), invariant
under the generalized Lorentz group SO0 (1, s). Such a causality structure for
V ∼ = R1+s is familiar for time R with total order and Minkowski spacetime
R with the special relativistic partial order.
4
Any metrical tensor of V ∼= R4 with causal signature (1, 3), e.g., an or-
thonormal Lorentz metrical tensor,

V ∼ 1 0
= R4 , η = 0 −13 ∈ V T ∨ V T ,
defines an “idolized” Lorentz group as its invariance group. Its GL(4, R)-orbit
leads to a parametrization of the orientation manifold for
the metrical hy-
perboloid or for Lorentz group relativity with dimension 52 = 10:
⎪
⎪
M10 ∼= GL(4, R)/O(1, 3) ∼ = {e ◦ η ◦ eT = g⎪ ⎪e ∈ GL(4, R)},
g ∼= g il
= e i
a η ab
e l
b = gli , with i, a = 0, 1, 2, 3.
In general, one has GL(1 + s, R)/O(1, s) ∼

2+s
= M( 2 ) for s = 0, 1, . . . .
7.1.4 2-Sphere for Perpendicular Relativity

With the local isomorphy of the rotation group to the spin group SO(3) ∼
SU(2), an “idolized” axial rotation subgroup SO(2) ⊂ SU(2) is given by the
invariance group of the hermitian and traceless Pauli matrix σ3 . Its SU(2)-
orbit leads to the 2-sphere parametrization of perpendicular relativity:
⎪
⎪
σ3 = 10 −1 0
, Ω2 ∼= SO(3)/SO(2) ∼ = {u ◦ σ3 ◦ u = xr ⎪
⎪u ∈ SU(2)},
xα

x ∼ β
= uα A B
r = r A σ3 B u β , with α, A = 1, 2.
The two angles (spherical coordinates) in the traceless hermitian matrix xr
can be parametrized by three position translations, with one condition for
the determinant:

x x
1 x3 x1 − ix2 cos θ e−iϕ sin θ
r = r = r x1 + ix2 −x3 = iϕ
e sin θ − cos θ
,
x2
with tr xr = 0 and − det xr = r2 = 1.
The restriction uses the rotation SO(3)-invariant product x2 = x23 + x21 + x22
in three dimensions.
7.1.5 Mass-Hyperboloid for Rotation Relativity

An “idolized” rotation group SO(3) in a Lorentz group SO0 (1, 3) is charac-
terized by a distinguished definite metric of a real three-dimensional vector
space (position), e.g., g = 13 . Similarly, one can work with a sesquilinear
scalar product δ of a complex two-dimensional space V ∼ = C2 invariant un-
der the locally isomorphic spin group SU(2) ∼ SO(3) in the special linear
group SL(2, C) ∼ SO0 (1, 3). The SL(2, C)-orbit of the metric parametrizes
rotation relativity by the points of an energy-momentum 3-hyperboloid:
⎪
q ⎪
= SO0 (1, 3)/SO(3) ∼
δ = 12 , Y 3 ∼ = {s ◦ δ ◦ s = m ⎪
⎪s ∈ SL(2, C)},
A
q ∼ Ḃ
q A α β
m= m = s δ s , with A, α = 1, 2.
α β Ḃ
7.1 Parametrization of Relativity Manifolds 179
q
The three real hyperbolic coordinates in the hermitian matrix m can be
chosen from four energy-momenta with one condition for the determinant:

q q 1 q0 + q3 q1 − iq2 cosh ψ + cos θ sinh ψ e−iϕ sin θ sinh ψ
m = m = m q1 + iq2 q0 − q3 = iϕ
e sin θ sinh ψ cosh ψ − cos θ sinh ψ
,
q q2
with det m = m2 = 1.
The restriction of the four energy-momenta to the three momenta uses the
SO0 (1, 3)-invariant bilinear form q 2 = q02 −
q2 .
7.1.6 Spacetime Future Cone for Unitary Relativity

An “idolized” unitary group U(2), called a hyperisospin group, is a maximal
compact subgroup of the general linear group GL(2, C), called the extended
Lorentz group. It is given by the invariance group of a scalar product δ of
a complex two-dimensional vector space. The GL(2, C)-orbit defines four
real parameters for unitary relativity, i.e., for the orientation manifold of the
U(2)-scalar product,
⎪
⎪
δ = 12 , D(2) ∼
= GL(2, C)/U(2) ∼
= {χ ◦ δ ◦ χ = x⎪⎪χ ∈ GL(2, C)},
x ∼ A A α β
= xḂ = χα δβ χḂ ,
x +x x − ix
x = x = x10 + ix32 x10 − x32 , with A, α = 1, 2.
These four real orbit parameters characterize the strictly positive elements
in the C ∗ -algebra of complex (2 × 2) matrices,
x = χ ◦ χ ⇐⇒ x = x and spec x > 0

⇐⇒ det x = x2 > 0 and tr x = 2x0 > 0.
They describe the absolute modulus set in the polar decomposition of

GL(2, C) into noncompact classes for the maximal compact group with the
unitary phases,
GL(2, C) χ = |χ| ◦ u ∈ √
D(2) ◦ U(2),
√
GL(2, C)/U(2) ∼
= D(2) |χ| = χ ◦ χ = x.
The positive matrices x are parametrizable by the points of the open future
cone in flat Minkowski spacetime:
⎪
⎪ 2
D(2) ∼
= R4+ = {x ∈ R4 ⎪⎪x > 0, x0 > 0}.
The cone manifold is embeddable into its own tangent space, the spacetime
translations R4 ⊃ D(2). The translations inherit the action x −→ χ ◦ x ◦ χ∗ of
the dilation-extended orthochronous Lorentz group GL(2, C)/U(1) ∼ = D(1)×
SO0 (1, 3), which constitutes the homogeneous part of the extended Poincaré
group [D(1) × SO0 (1, 3)] × R4 .
7.2 Relativity Transitions by Transmutators

Elements of a relativity, i.e., of a homogeneous space G/H, are related to
each other by the action of the full group G, e.g., different perpendicularities
by rotations of the earth’s surface or different nonrelativistic space–times by
Lorentz transformations of a mass hyperboloid.
With real parameters for H-relativity G/H, the “general” group G is
partly parametrized. Each coset can be given a defining representative gr ∈
gH ⊆ G. Such representatives have a characteristic two-sided G×H-transfor-
mation behavior in the group G × G, called relativity transition or transmu-
tation from the “general” group to the “idolized” group: A left multiplication
of the representative with k ∈ G hits the chosen representative (kg)r ∈ kgH
up to a right multiplication with an H-element,
k ∈ G, kgr = k(gH)r = (kgH)r h(k, gr ) = (kg)r h(k, gr ), with h(k, gr ) ∈ H.
The left group action k ∈ G is accompanied by a right action from the

“idolized” subgroup h(k, gr ) ∈ H, which depends on both k and the repre-
sentative gr . It is called Wigner element and Wigner subgroup-operation, in
generalization of the familiar Wigner rotation, which arises from a Lorentz
transformation of a boost.
7.2.1 From Interactions to Particles

An example in which both electromagnetic relativity with the transition from
hyperisospin to electromagnetic group U(2) → U(1)+ and rotation (special)
relativity with the transition from Lorentz to spin group SL(2, C) → SU(2)
play a role is the transition from relativistic electroweak interaction fields to
particles in the standard model,
SL(2, C) × U(2) −→ SU(2) × U(1)+ .

Lorentz hypercharge-isospin spin electromagnetism
The lepton field in the minimal electroweak model connects, for each
spacetime translation, the external Lorentz SL(2, C)-degrees of freedom with
the internal isospin SU(2)-degrees of freedom and a hypercharge U(1)-value
y = − 12 :
R4 x −→ lA
α (x), with A, α = 1, 2.
The transition from field to particles with respect to the internal degrees
of freedom uses the ground-state degeneracy, implemented by the U(2)-
ΦΦ (x) = M 2 > 0 of the Lorentz scalar Higgs field,
invariant condition Φ
R4 x −→ Φ α (x).
The Higgs field is an isospin dublet with hypercharge y = 12 . It transmutes

from “general” hyperisospin U(2)-properties of the interaction fields to “idol-
ized” electromagnetic U(1)+ -properties of the particles, e.g., from the lepton
7.2 Relativity Transitions by Transmutators 181
field isodublet to the electron-positron field, an isosinglet with electromag-

netic charge number z = −1:
Φ α (x) A
α (x) −→ e (x) =
U(2) −→ U(1)+ : lA A
Φ|(x) lα (x)
|Φ = lA
2 (x) + . . . .
The “idolization” of an electromagnetic U(1) comes with the distinction of

a ground state and the expansion of the Higgs transmutator (more ahead),
Φ α (x) 0 ∼ α

Φ |(x) = δ2 + . . . for e = 1 = δ2 with |Φ Φ|(x) = Φ Φ (x).
α 2
e.g., |Φ
With respect to external degrees of freedom, the transition from a left-
handed Weyl field with Lorentz group SL(2, C)-action to particles with mass
m2 > 0 and SU(2)-spin requires a rest system. The related harmonic ex-
pansion (particle analysis) of the spacetime field with respect to eigenvectors
involves the electron creation and positron annihilation operators ua (q) and
aa (
q ), respectively, for spin directions a = 1, 2 and momentum q as transla-
tion eigenvalues:
SL(2, C) −→ SU(2) : eA (x) −→ ua (
q ), aa (q )
3
where eA (x) = ⊕ 2q0d(2π)
q A q
3 sa ( m ) [e u (q ) + e−iqx aa (q )],
iqx a

with q0 = m2 + q2 .
q
The boost representation sA a ( m ), discussed ahead as a Weyl transmutator,
connects the Lorentz group SL(2, C)-action for fields with a rest system spin
SU(2)-action for massive particles.
Altogether in the field-particle transition, there are four types of transmu-
transmutator
tators G ↔ H involved with G × H-transformations for four different
group pairs H ⊂ G: the lepton field l with external–internal transformation
behavior, the Lorentz scalar Higgs field Φ as internal transmutator from in-
teraction to particles, the boost representation s as corresponding external
transmutator, and, finally, the creation and annihilation operators (u, a) with
the external–internal properties of the particles (spin and charge)
Extended Lorentz group GL(2, C) Hyperisospin group U(2)
Spin group SU(2) Electromagnetic group U(1)
interactions
lA
α (x)
GL(2, C) ←→ U(2)
external sa q
A( m ) $ $ Φ α (x) internal
SU(2) ←→ U(1)+
ua (
q ), aa (
q)
free particles
7.2.2 Pauli Transmutator

Perpendicular relativity is the base manifold of the coset bundle SU(2) ∼ =
SO(2)(Ω2 ) with the axial rotations as typical fiber. It is linearly represented
by the fundamental Pauli transmutator from rotations to axial rotations,
R3 ⊃ Ω2 xr −→ u( xr ) ∈∈SU(2)/SO(2),

u( xr ) ◦ σ3 ◦ = xr = σarxa with r 2 =
u ( xr ) x2 , √ 2 2
x1 +x2
u( xr ) = eiα = 12 cos α + i α α

sin α with tan 2α = tan θ =
x3
x3 + r −x1 + ix2
= x3 +r
2r [1 2 + i x⊥
x3 +r ] = √ 1
2r(x3 +r) x1 + ix2 x3 + r

cos θ2 −e−iϕ sin θ2
= u(ϕ, θ) = .
eiϕ sin θ2 cos θ2
A left action on the Pauli transmutator u( xr ) with the spin group SU(2)
gives the transmutator at the rotated point O.x on the 2-sphere up to a right
action with the axial group SO(2) (Wigner axial rotation):
o ∈ SU(2) : ⎧ r ) ◦ v(o, r )
o ◦ u( xr ) = u( O.
x
x
⎨ v(o, r ) ∈ SO(2),

x
with O.x = o ◦ x ◦ o ,
⎩ b
Oa = 12 tr σa ◦ o ◦ σb ◦ o ∈ SO(3).
The complicated explicit expression for the Wigner axial rotation can be
r ) ◦ o ◦ u( r ).
computed from v(o, xr ) = u ( O.
x
x
7.2.3 Weyl Transmutators

Special relativity, e.g., the energy-momentum hyperboloid for a mass m2 > 0,
is the base manifold for the coset bundle SL(2, C) ∼
= SU(2)(Y 3 ) with the spin
rotations as typical fiber. The related Weyl representations of the boosts are
a familiar example for a transmutator,
R4 ⊃ Y 3 mq
−→ s( m
q
) ∈∈SL(2, C)/SU(2),
σa qa
q
s( m ) ◦ 12 ◦ s ( m ) = m = m with m2 = q 2 ,
q q
and Weyl matrices, σ a = (12 , σ ) (left) and σ̌ a = (12 , −σ ) (right). The explicit
expressions involve the Lorentz dilations eβσ3 ∈ SO0 (1, 1) in addition to the
Pauli transmutator for the two spherical degrees of freedom:
q
= u( |qq| ) ◦ eψσ3 ◦ u ( |qq | ), ψ = 2β,
m
q 1 q0 + | q| 0 |
q|
e 2βσ3
= diag m = m 0 q0 − |
q| , tanh 2β = q0 = vc ,
q
s( m ) = u( |qq| ) ◦ eβσ3 = 12 cosh β + q

q| sinh β
+ , |
q1 − iq2
= q02m +m q

12 + q0 +m = √ 1 q0 + q3 + m
q1 + iq2 q0 − q3 + m .
2m(q0 +m)
The left-handed Weyl transmutator s( m q

) ∈ SL(2, C) together with its
right-handed partner ŝ = s , i.e., ŝ( m ) = u( |qq| ) ◦ e−βσ3 ∈ SL(2, C), are the
−1 q
two fundamental transmutators from Lorentz group to rotation subgroups.

The restriction of the energy-momenta from four to three parameters by the
7.2 Relativity Transitions by Transmutators 183
q2
on-shell hyperboloid Y 3 condition m2 = 1 is expressed by the Dirac equation
in energy-momentum space:
σa qa σa qa
⎫
q
s( m ) ◦ ŝ−1 ( m
q
) = m ⇒ s( m
q
) = m ◦ ŝ( m
q
)⎬
⇒ (γ a qa − m)s( m
q
) = 0,
a a ⎭
q
ŝ( m ) ◦ s−1 ( m
q
) = σ̌ qa
m ⇒ ŝ( m
q
) = σ̌ qa
m ◦ s( m
q
)
q

0 σa q s( m ) 0
with γ a = σ̌a 0 , s( m )= 0 q
ŝ( m ) .
The four columns of the (4 × 4) matrix s( m q

) are familiar as solutions of the
Dirac equation.
For the Pauli transmutator, the analogue to the Dirac equation is the
condition σ xu( xr )−u( xr )σ3 r = 0, which restricts the three rotation parameters
to two independent parameters for SU(2)/SO(2).
A left action with the Lorentz group SL(2, C) gives the Weyl transmutator
at the Lorentz-transformed energy-momenta Λ.q on the hyperboloid q 2 = m2 ,
up to a right action by a Wigner spin SU(2)-rotation:
λ ∈ SL(2, C) : ⎧
λ ◦ s( m
q
m ) ◦ u( m , λ),
) = s( Λ.q q
⎨ u( m , λ) ∈ SU(2),
q
with Λ.q = λ ◦ q ◦ λ ,
⎩ b
Λa = 12 tr σa ◦ λ ◦ σ̌ b ◦ λ ∈ SO0 (1, 3).
7.2.4 Higgs Transmutators

In the standard model of electroweak interactions, the three weak parameters
for the Goldstone manifold of electromagnetic relativity as base manifold of
the coset bundle U(2) ∼ = U(1)(G 3 ) are implemented
by three chargelike de-
α ∼ Φ1 ∼
grees of freedom of the Higgs vector Φ = Φ 2 ∈ V = C2 and its orthogonal
α

Φ2
Φ = αβ Φ β = −Φ
Φ̃ Φ :
1
C ⊃G
2 3
−→ v( M
Φ
M
Φ
) ∈∈U(2)/U(1)+ ,
i(α3 −α0 )
θ
−e−i(ϕ−α0 ) sin θ2
Φ
v( M ) = eei(ϕ−α0 ) sin
cos 2
θ
e−i(α3 −α0 )
cos θ = u(ϕ − α3 , θ) ◦ ei(α3 −α0 )σ3
2 2

) = |Φ | M+|Φ
1 Φ2 Φ1 Φ Φ 1 2 Φ 2 |2
=M −ΦΦ Φ 2 with det v( M 2 = 1.
1
The restriction from four to three real weak degrees of freedom uses the
U(2)-invariant scalar product ΦΦ|Φ
Φ = |Φ
Φ|2 = M 2 of the Higgs vector space.
A left hypercharge-isospin action on the fundamental Higgs transmutator
gives the transmutator at the U(2)-transformed Higgs vector on the Gold-
stone manifold, accompanied by a Wigner electromagnetic U(1)+ -transfor-
mation from the right:

u = eiγ0 u2 ∈ U(1) ◦ SU(2),
u ∈ U(2) : u ◦ v( M ) = v( M ) ◦ u+ , with
Φ Φ
u.Φ
ei2γ0
1 ∈ U(1)+ .
0
u+ = 0
7.2.5 Real Tetrads (Vierbeins)

Lorentz group or particle relativity is the 10-parametric base manifold of the
coset bundle GL(4, R) ∼= SO0 (1, 3)(M10 ) with the Lorentz group as typical
fiber. It is the GL(4, R)-orbit of the orthonormal O(1, 3)-“idolized” Lorentz
metric in a symmetric matrix η and diagonalizable to four principal axes with
a transformation from a maximal compact subgroup O(4) ⊂ GL(4, R) (six
parameters):
g = eT ◦ η ◦ e = O4T ◦ diag g ◦ O4 , with O4 ∈ O(4).
The diagonal part of the metrical hyperboloid, multiplied by the inverse met-
ric η −1 , displays the remaining four dilation transformations from the maxi-
mal noncompact abelian subgroup:
η −1 ◦ diag g ∈ D(1) × SO0 (1, 1)3 ∼

= D(1)4 ⊂ GL(4, R).
The diagonal elements are four directional units, one for time and three for
the metrical ellipsoid of the 3-position.
The operational decomposition of the metrical hyperboloid leads to the
parametrization of the 10-dimensional tetrad e as a basis of real four-dimen-
sional tangent spacetime R4 by four dilations and a six-dimensional rotation:
M10 g −→ e(g) ∈∈GL(4, R)/O(1, 3).
A general linear GL(4, R) left multiplication gives the tetrad for a trans-
formed metrical tensor and a Wigner right transformation by the idolized
Lorentz group O(1, 3):
h ∈ GL(4, R) : h ◦ e(g) = e(hT ◦ g ◦ h) ◦ Λ(h, g), with Λ(h, g) ∈ O(1, 3).
7.2.6 Complex Dyads (Zweibeins)

Electroweak spacetime D(2) is the orientation manifold of U(2)-scalar prod-
ucts (unitary relativity) as base manifold of the coset bundle U(2)(D(2)) with
the hyperisospin group as the typical fiber:
x ∈ D(2) ∼
= GL(2, C)/U(2) = D(12 ) × SL(2, C)/SU(2) ∼
= D(1) × Y 3 .
It is transformed to an “idolized” diagonal scalar product by a Weyl trans-

mutator s( √xx2 ) ∈ SL(2, C) for the three hyperbolic degrees of freedom and
√
a dilation D(1) = exp R ∼= R for eigentime e2β0 = x2 :

x0 + x3 x1 − ix2
x = x1 + ix2 x0 − x3 = s( √xx2 ) ◦ e2β0 12 ◦ s ( √xx2 ) = u( xr ) ◦ diag x ◦ u ( xr ),

x0 + r 0
diag x = 0 x0 − r = e2(β0 12 +βσ3 ) ∈ D(1) × SO0 (1, 1),
with e4β0 = x2 , tanh 2β = r
x0
.
7.3 Linear Representations of Relativities 185
The diagonalization of the scalar products gives the fundamental

transmutator from the extended Lorentz group GL(2, C) to the hyperisospin
subgroup U(2). It contains a basis of the complex two-dimensional space and
will be called, in analogy to a real tetrad or vierbein e(g(x)) ∈∈GL(4, R)/
SO0 (1, 3), a complex dyad (zweibein) χ(x) ∈∈GL(2, C)/ U(2). It is para-
metrized by the future cone spacetime points as orbit of the U(2)-scalar
product,
R4 ⊃ D(2) x −→ χ(x) ∈∈GL(2, C)/U(2),
χ(x) ◦ 12 ◦ χ (x) = x = s( √xx2 ) ◦ eβ0 12 = u( xr ) ◦ eβ0 12 +βσ3 .
The left action by the extended Lorentz group GL(2, C) as external trans-
formation gives the dyad χ at a Lorentz-transformed and dilated spacetime
point in the future cone, accompanied by an action from the right with an
internal spacetime-dependent Wigner hyperisospin U(2)-transformation:
g ◦ χ(x) = χ(e2δ0 Λ.x) ◦ u(x, g),

g ∈ GL(2, C) : ⎧
⎨ u(x, g) ∈ U(2),
with g = eδ0 +iα0 λ ∈ D(1) × U(1) ◦ SL(2, C),
⎩
g ◦ x ◦ g = e2δ0 λ ◦ x ◦ λ = e2δ0 Λ.x, Λ ∈ SO0 (1, 3).
7.3 Linear Representations of Relativities

In the foregoing section, the classes M ∼
= G/H of the five relativities with
linear groups H ⊂ G were represented by linear transformations of the group
G-defining vector spaces:
Relativity manifold Fundamental transmutator

fixgroup H M −→ G/H
2 2 Pauli transmutator
2-sphere
x =
axial rotations Ω2 −→ uk ( ) ∈

x α x
∈SU(2)/SO(2)
Goldstone manifold Φ Φ = M 2 Higgs transmutator
Φ α Φ
electromagnetic group G3 M −→ vk ( M ) ∈∈U(2)/U(1)+
3-hyperboloid q2 = m2 Weyl transmutators
rotations Y m −→ sα ( m ) ∈∈SL(2, C)/SU(2)
3 q A q
Future 4-cone dyad

unitary group D(2) x −→ χAα (x) ∈∈GL(2, C)/U(2)
Metric manifold tetrad
Lorentz group M10 g −→ eja (g) ∈∈GL(4, R)/O(1, 3)
As for the G-representations, the products of these fundamental transmu-

tators give the finite-dimensional representations of the homogeneous spaces
G/H. For a compact group G, the finite-dimensional transmutators are
complete for the harmonic analysis of the Hilbert spaces with the square-
integrable functions L2 (M, V T ) of the orientation manifold of the relativity
valued in a vector space V with H-action (more ahead).
7.3.1 Rectangular Transmutators

Representations of the “general” group G involve representations of the cosets
G/H representatives,
Gg − → D(g) ∈ GL(V ),
G/H gH gr −→ D(gr ),
e.g., for perpendicular relativity,
χ+ϕ χ−ϕ

i i
SU(2) u(ϕ, θ, χ) = e−i χ−ϕcos 2θ −e
θ
2 2 sin θ2
χ+ϕ
−i
e θ2 sin 2 −iϕe θ2 cos 2
θ
cos 2
−→ u(ϕ, θ, χ)r = u(ϕ, θ) = eiϕ sin θ −e sin 2
cos θ
∈∈SU(2)/SO(2) ∼
= Ω2 .
2 2
A G-representation by an (n × n) matrix, with an (8 × 8)-example for

SU(3),
⎛ • • • • • • • •⎞
• • • • • • • •
⎜• • • • • • • •⎟
⎜ ⎟
D(g) ∈ V ⊗ V T ∼ n e.g. ⎜•
= C ⊗ C ∼ ⎜•
n • • • • • • •⎟
• • • • • • •⎟ ,
⎝• • • • • • • •⎠
• • • • • • • •
• • • • • • • •
can be decomposed into H-representations, exemplified by the octet decom-

position 8 = 2 ⊕ 1 ⊕ 3 ⊕ 2 into SU(2)-representations,
⎛ ⎞
• • 0 0 0 0 0 0
⎜ • • 0 0 0 0 0 0 ⎟
⎜ 0 0 • 0 0 0 0 0 ⎟
H H e.g. ⎜ ⎟
k k
• • •
V ∼
= Vι , H • Vι ⊆ Vι , D(h) ∼
= dι (h) ∼ ⎜ ⎜
0
0
0
0
0
0 • • •
0
0
0
0
⎟.
⎟
⎜ • • • ⎟
ι=1 ι=1
⎝ 0 0 0 0 0 ⎠
0 0 0 0 0 0 • •
0 0 0 0 0 0 • •
Thus, the G-representation matrices can be decomposed “lopsidedly” into

rectangular (n × nι ) matrices, nι ≤ n,
⎛ • • • • • • • • ⎞
• • • • • • • •
⎜ • • • • • • • • ⎟

k / / / e.g. ⎜ ⎟
/ / / ⎜ • • • • • • • • ⎟
D(g) = Dι (g) = D1 (g)/D2 (g)/ · · · /Dk (g) ∼ ⎜ • • • • • • • • ⎟.
⎜ ⎟
ι=1 ⎝ • • • • • • • • ⎠
• • • • • • • •
• • • • • • • •
The rectangular matrices have a left-right G × H-action, e.g., the SU(3) ×

SU(2)-action on octet-dublet, octet-singlet, octet-triplet, and octet-dublet.
As mappings of the coset representatives (G/H)r , they are called
transmutators:
T ∼ n
r gr −→ Dι (gr ) ∈ V ⊗ Vι = C ⊗ C ,
nι
(G/H)⎧
⎨ Dι (gr h) = Dι (gr ) ◦ dι (h), h ∈ H,
with Dι (kgr ) = D(k) ◦ Dι (gr )
⎩
= D((kg)r ) ◦ dι (h(k, gr )), k ∈ G.
With bases of the G-vector space |n; j ∈ V and the H-vector spaces |nι ; a ∈
Vι , one has in a Dirac notation with kets for vectors | ∈ V and bras for
linear forms | ∈ VιT ,
V ⊗ VιT Dι (gr ) = |n; jDι (gr )ja nι ; a|,

j = 1, . . . , 8,
e.g., C ⊗ C D2 (gr ) =
8 2
|8; jD2 (gr )ja 2; a|, with
a = 1, 2.
The finite-dimensional transmutators are from (n×nι )-dimensional vector

spaces V ⊗ VιT with representations of the bi-regular subgroup G × H ⊆
G × G. Those representations are unitary, i.e., Hilbert representations, only
for the compact groups G, i.e., in the examples above, for perpendicular and
electromagnetic relativity.
7.3.2 Representations of Perpendicular Relativity

For perpendicular relativity, all transmutators from rotations to axial rota-
-
2J
tions arise by the totally symmetric products, denoted by , of the funda-
mental Pauli transmutator u( xr ) ∈ SU(2),
SU(2)/SO(2) ∼
= Ω2 −→ SU(1 + 2J),
-
2J

x
r −→ [J]( xr ) = u( xr ), x2 = r2 .
The irreducible spin SU(2)-representations [J] with dimensions 1 + 2J are

decomposable into axial rotation SO(2)-representations (±n) with dimension
2 for n = 0 and two polarizations ±n (left- and right-circularly polarized):
⎧
⎪
2J
⎪
⎪
⎪ (±n) for J = 0, 1, . . . ,
SO(2) ⎨
irrep SU(2) [J] ∼
=
n=0,2,...,
⎪
⎪
2J
⎪
⎪ (±n) for J = 12 , 32 , . . . ,
⎩
n=1,3,...,
e.g., for rotations acting on 3-position R3 :
[1]( xr ) ∼
= Ors ( xr ) = 1
2tr u( xr ) ◦ σ s ◦ u ( xr ) ◦ σ r
xα x r = 1, 2, 3,
δ αβ r − r+xβ xα
= 1
r −xβ
3 ∈ SO(3), with
x3 α = 1, 2,
SO(2)
[1] ∼
= (±2) ⊕ (0),
with the relations for the SO(3) and SO(2) metrical tensors:

δαβ 0
r
Oα,3 ( xr )δrs Oβ,3
s
( xr ) = 0 1 , Oαr ( xr )δ αβ Oβs ( xr ) = δ rs − xr xs
r2 .
The symmetric square u ∨ u of the Pauli transmutator, in a Cartesian and

a spherical basis,
⎛ iϕ 2 θ ⎞
αβ xα xβ e cos 2 −eiϕ sin2 2θ ieiϕ sin
√θ
δ − r(r+x3 ) xα
∼
2
O( xr ) = xβ
r
=⎝ √θ
i sin √θ
i sin cos θ ⎠,
− x3 2 2
r r −e−iϕ sin2 θ
2
e−iϕ cos2 θ
2
ie−iϕ sin
√θ
2
displays in the third column O3b ( xr ) the spherical harmonics Y1 (ϕ, θ) ∼ xr as a
basis for subspace C3 in the Hilbert space L2 (Ω2 ) with the square-integrable
functions on the 2-sphere. Its symmetric traceless products of power J =
1, 2, . . . give all spherical harmonics YJ (ϕ, θ) ∼ ( xr )Jtraceless , which arise as
the (1 + 2J)-entries in one column of the (1 + 2J) × (1 + 2J)-matrices for the
representation [J]. The spherical harmonics are bases for the Hilbert spaces
C1+2J ⊂ L2 (Ω2 ) acted on by the irreducible SO(3)-representations.
With respect to the bi-regular SU(2) × SO(2)-transformation behavior,
the four functions in the two columns of the (2 × 2)-Pauli transmutator,
⎛ ⎞
x3 +r
− √x1 −ix2
⎝ 2r
⎠ cos θ2 −e−iϕ sin θ2
∈ C2 ⊗ C2 ,
x
2r(x3 +r)
u( r ) = = iϕ θ θ
√ x +ix
1 2 x +r
3 e sin 2
cos 2
2r(x3 +r) 2r
and the six functions in the first and second columns O1,2 b
( xr ) ∼ b
= O+,− ( xr )
above in a rectangular (3 × 2) matrix constitute bases for finite-dimensional
Hilbert spaces C2 ⊗ C2 and C3 ⊗ C2 with SU(2)-representations, acting on
the left factors C2 and C3 , and nontrivial SO(2)-representations SO(2)
eiα3 σ3 −→ einα3 σ3 , acting on the right factor C2 . They span irreducible sub-
spaces for the harmonic analysis of the Hilbert space L2 (Ω2 , C2 ) with the
square-integrable mappings from the 2-sphere into a vector space with non-
trivial SO(2)-action.
In general, one has the Peter–Weyl decompositions [47] (see Chapter 8)
into irreducible subspaces for SU(2) × SO(2)-action:
dense
Vn ∼= C2−δn0 : L2 (Ω2 , Vn ) ∼ = C1+2J ⊗ Vn .
2J≥n
The orthogonal sum goes over all SU(2)-representation that contain the
SO(2)-representations on Vn ∼ = C, C2 (Frobenius’ reciprocity; more ahead).
This generalizes the case for the spherical harmonics with V0 ∼
= C.
7.3.3 Representations of Electromagnetic Relativity

For the standard model of electroweak interactions, the Higgs-parametrized
defining representation of the orientation manifold G 3 of the electromagnetic
group with hypercharge y = 12 and isospin T = 12 and its conjugate, i.e., the
two Higgs transmutators,

Φ2 Φ1
[ 12 || 12 ]( M
Φ Φ
) = v( M 1
)= M −ΦΦ
Φ 2 ∈∈U(2)/U(1)+ ,
1
[− 12 || 12 ]( M
Φ Φ
) = v ( M ),
give, via their products, all irreducible representations of the Goldstone

manifold:
U(2)/U(1)+ ∼= G 3 −→ U(1 + 2T ), M Φ
−→ [±n + T ||T ]( M
Φ
),
-
2T
∼
irrep U(2) [±n + T ||T ] = [±1||0] ⊗ [ 2 || 2 ],
n 1 1
irrep U(1) [±1||0] ∼

= [± 12 || 12 ] ∧ [± 12 || 12 ].
Because of the central correlation SU(2)∩U(12 ) = {±12 } in U(2), the U(2)-

representations have the correlation of the hypercharge- and isospin-invariant
y = T ± n with natural n, i.e., the two invariants (y, T ) for the rank-2 U(2)-
transformations are either both integer or both half-integer, as visible in the
colorless fields of the standard model.
The decomposition of a hyperisospin U(2)-representation into irreducible
representations of the electromagnetic group U(1)+ ei2γ0 −→ ezi2γ0 is
given with integer charge numbers z ∈ Z:
U(1)+ ±n+2T

U(2) [±n + T ||T ] ∼ = [z],
z=±n
[± 12 || 12 ] ∼
= [0] ⊕ [±1],
e.g.,
[0||1] ∼ = [−1] ⊕ [0] ⊕ [1].
v and v transmute from hyperisospin U(2)-dublets to isospin SU(2)-

singlets with nontrivial charge numbers z = ±1 for electromagnetic U(1)+ .
The antisymmetric squares v ∧ v and v ∧ v of the fundamental Higgs trans-
mutators have charge numbers z = ±1:
U(1)+
Φ
[1||0]( M ) =
Φ αΦ α
Φ̃
∈ U(1), with [1||0] ∼
= [1],
M2
α U(1)+
Φ
[−1||0]( M ) =
Φ Φ
αΦ̃
∈ U(1), with [−1||0] ∼
= [−1].
M2
In the hypercharge trivial product v ⊗v of both fundamental Higgs transmu-

tators, the columns transmute from a triplet to three isosinglets with charge
z ∈ {−1, 0, 1}:
Φ
[0||1]( M ) = 12 tr τ b ◦ v( M
Φ
) ◦ τ a ◦ v ( MΦ
)

= Φ τΦ̃

Φ+Φ̃
Φ τΦ
Φ Φ−Φ̃
τ Φ̃ Φ τΦ Φ
Φ̃ Φ−Φ
τ Φ̃ Φ
τΦ ∈ SO(3),
2M 2 2iM 2 2M 2
U(1)+
with [0||1] ∼
= [−1] ⊕ [0] ⊕ [1].
These three transmutators are used for the transition from the three isospin
gauge fields in the electroweak standard model to the weak boson particles:
Φ Φ Φ Φ
τ a Aa (x) = A(x) −→ v ( ΦM
(x)
) ◦ A(x) ◦ v( ΦM(x)
) − v ( ΦM
(x)
) ◦ ∂v ( ΦM
(x)
)
= (W− (x), W0 (x), W+ (x)) = (A− (x), A0 (x), A+ (x)) + . . .
= δia + . . . , with [0||1]( ΦM (x) a
)i .
For the definition of particles, which requires the additional transmutation

from Lorentz group to rotation group, the neutral field W0 is combined, in
the Weinberg rotation, with the hypercharge gauge field.
Similar to perpendicular relativity, the Hilbert spaces of the square-
integrable mappings of the compact Goldstone manifold L2 (G 3 , Vz ) into a
Hilbert space Vz ∼
= C with electromagnetic action U(1)+ ei2γ0 −→ ezi2γ0
have Peter–Weyl decompositions into finite-dimensional subspaces C1+2T ⊗C
with irreducible representations of the bi-regular subgroup U(2) × U(1),
where the isospin representations fulfill 2T ≥ |z|. The representation
spaces are given by the columns in the products of the fundamental Higgs
transmutators.
7.3.4 Representations of Rotation Relativity

For special relativity, all finite-dimensional representations of the 3-hyper-
boloid (boost representations), i.e., all finite-dimensional transmutators from
Lorentz group to rotation group, can be built by the totally symmetric prod-
ucts of the two fundamental Weyl transmutators s( m q q
), ŝ( m ) ∈ SL(2, C):
SL(2, C)/SU(2) ∼
= Y 3 −→ SL((1 + 2L)(1 + 2R), C),
-
2L -
2R
q
m −→ [L|R]( m
q
)= q
s( m )⊗ q
ŝ( m ), q 2 = m2 .
The finite-dimensional irreducible Lorentz group representations [L|R] with

dimensions (1 + 2L)(1 + 2R) can be decomposed into irreducible spin repre-
sentations (see Chapter 5):
SU(2)
L+R
irrep fin SL(2, C) [L|R] ∼
= [J].
J=|L−R|
For example, the vector representation s ⊗ s ∼

= Λ = [ 12 | 12 ] gives two
irreducible transmutators from Lorentz group to rotation group, the first
column for a spin-0 representation and the three remaining columns for a
spin-1 representation:
[ 12 | 12 ]( m
q q
) = Λab ( m ) ∼
= 1
tr s( mq
) ◦ σ a ◦ s ( m
q
) ◦ σ̌b

2
r = 1, 2, 3,
q0 qr
1
= m qs δrs m + qr qs ∈ SO0 (1, 3), with
m+q0 a = 0, 1, 2, 3,
SO(3)
[ 12 | 12 ] ∼
= [0] ⊕ [1].
q
The four columns of the matrix Λa0,r ( m ) relate to each other the metrical
tensors of SO0 (1, 3) and SO(3):
a b
1 0
q
Λa0,r ( m q
)ηab Λb0,s ( m q
) = 0 −δrs , Λar ( m q
)δ rs Λbs ( m ) = −η ab + qmq2 .
The transmutator from Lorentz group to rotation group in the rectangular

(4 × 3)-submatrix Λar ( m
q
) ∈ R4 ⊗ R3 is used for a massive spin-1 particle in
a Lorentz vector field, e.g., for the neutral weak boson and its Feynman
propagator,
d3 q
Za (x) =⊕ q
Λar ( m
2q0 (2π)3
q ) + e−iqx u r (
) [eiqx ur ( q )],
a b
ab q q
d4 q (−η + m2 ) iq(x−y)
0|{Zb (y), Za (x)} − (x0 − y0 )[Zb (y), Za (x)]|0 = πi (2π)3 q 2 +io−m2
e .
In contrast to compact perpendicular relativity with the Hilbert space

L2 (Ω2 ) for the 2-sphere functions, the Hilbert space with the square-
integrable functions on the 3-hyperboloid L2 (Y 3 ) has no finite-dimensional
Hilbert subspaces with irreducible SL(2, C)-representations. The monomials
in the columns of the fundamental Weyl transmutators are not in L2 (Y 3 ).
They give bases for finite-dimensional SL(2, C) × SU(2)-representations on
C(1+2L)(1+2R) ⊗ C1+2J , which are indefinite unitary for the noncompact
Lorentz group SL(2, C). The spin SU(2)-representation has to be contained
in the SL(2, C)-representation; i.e., |L − R| ≤ J ≤ L + R.
7.3.5 Representations of Unitary Relativity

All finite-dimensional representations of unitary relativity, i.e., of nonlinear
spacetime D(2),
D(2) ∼
= GL(2, C)/U(2) ∼
= D(1) × SO0 (1, 3)/SO(3)
−→ GL((1 + 2L)(1 + 2R), C),
use products of the two conjugated dyads, e.g.,
χ(x) = u( xr ) ◦ eβ0 12 +βσ3 ∈∈GL(2, C)/U(2),
χ(x) ◦ 12 ◦ χ (x) = u( xr ) ◦ e2(β0 12 +βσ3 ) ◦ u ( xr ) = x ∈ GL(2, C),
(χ(x) ◦ 12 ◦ χ (x))2 = e4β0 = x2 ∈ D(1).
The monomials in the dyads span finite-dimensional spaces with GL(2, C) ×
U(2)-representations, which are, because of the noncompact group GL(2, C),
not Hilbert spaces. Hilbert spaces with faithful representations of nonlin-
ear spacetime have to be infinite-dimensional. They will be considered in
Chapter 11.
7.3.6 Representations of Lorentz Group Relativity

All finite-dimensional representations of the general linear group GL(4, R),
the structure group of the tangent bundle of four-dimensional spacetimes,
GL(4, R) ∼
= D(1) × {±14 } × SL0 (R4 ), SL0 (R4 ) ∼ SO0 (3, 3),
contain products of the fundamental representations of the 15-dimensional
rank-3 special subgroup SL0 (R4 ), which, as a specialty of four spacetime di-
mensions and the isomorphy A3 ∼ = D3 in the Cartan classification, is locally
isomorphic to the indefinite orthogonal group SO0 (3, 3), the invariance group
of the Killing form of the Lorentz group SO0 (1, 3). It has the Iwazawa de-
composition with maximal compact and maximal abelian noncompact group
(see Chapter 8):
SO0 (3, 3) = [SO(3) × SO(3)] ◦ SO0 (1, 1)3 ◦ exp R6 .
The three fundamental representations of the cover group SO0 (3, 3)

are the two four-dimensional spinor representations, which coincide with
the defining SL0 (R4 )-representation on V ∼ = R4 and its dual, and a six-
dimensional self-dual one:

dimR [1, 0, 0] = 4, dimR [0, 1, 0] = 42 = 6, dimR [0, 0, 1] = 43 = 4.

The dimensions of the finite-dimensional irreducible representations are given
by the Weyl formula:
irrep fin SO0 (3, 3) [n1 , n2 , n3 ] ∼ = N3 ,
(n1 +1)(n2 +1)(n3 +1)(n1 +n2 +2)(n3 +n2 +2)(n1 +n2 +n3 +3)
dimR [n1 , n2 , n3 ] = 12 ,
dual reflection: [n1 , n2 , n3 ] ↔ [n3 , n2 , n1 ].
Spaces with self-dual representations [n, m, n] have an SO0 (3, 3)-invariant

symmetric bilinear form.
The finite-dimensional representations of Lorentz group relativity, para-
metrized by the metric manifold,
GL(4, R)/O(1, 3) ∼
= M10 g −→ [n1 , n2 , n3 ](g) ∈ GL(dn , R),
have decompositions with respect to an “idolized” Lorentz group. The three
fundamental SO0 (3, 3)-representations decompose into the two fundamental
SO0 (1, 3)-representations, i.e., the four-dimensional Minkowski representa-
tion and the six-dimensional adjoint representation:
SO0 (1,3)
[1, 0, 0], [0, 0, 1] ∼
= [ 12 | 12 ],
SO0 (1,3)
[0, 1, 0] ∼
= [1|0] ⊕ [0|1],
SO0 (1,3)
adjoint [1, 0, 1] ∼
= [1|0] ⊕ [0|1] ⊕ [1|1].
The Killing form of log SO(3, 3) has signature (9, 6).
The three fundamental SL(4, R)-representations act as a left factor in
SL(4, R) × SO0 (1, 3) on the tetrad and its antisymmetric products as funda-
mental transmutators from the general linear group to the Lorentz group:
[1, 0, 0] ∼
= eja ∈∈GL(4, R)/O(1, 3),
"2
[1, 0, 0] = [0, 1, 0] ∼
= jikl abcd eja eib ∈ GL(6, R),
"3
[1, 0, 0] = [0, 0, 1] ∼
= jikl abcd eja eib ekc ∈ GL(4, R),
"4
[1, 0, 0] = [0, 0, 0] ∼
= det e ∈ GL(1, R).
The antisymmetric tetrad square e ∧ e is a (6 × 6) transmutator acted on

by the self-dual fundamental SO0 (3, 3)-representation (from the left) and by
the adjoint Lorentz group representation (from the right). The inverse tetrad
contains the antisymmetric cube e ∧ e ∧ e, divided by the determinant with
power 4, which is an SO0 (3, 3)-scalar with nontrivial D(1)-dilation properties.
The metrical tensor is the symmetrical tetrad square e ∨ e:
-
2
[1, 0, 0] = [2, 0, 0] ∼
= gji = η ab eja eib ∈ GL(10, R),
SO0 (1,3)
[2, 0, 0] ∼
= [0|0] ⊕ [1|1].
The curvature is like the quartic tetrad product (e ∧ e) ∨ (e ∧ e) from a

20-dimensional representation:
[0, 1, 0] ∨ [0, 1, 0] = [0, 0, 0] ⊕ [0, 2, 0],

SO0 (1,3)
Rijkl ∼
= [0, 2, 0] ∼
= [0|0] ⊕ [1|1] ⊕ [2|0] ⊕ [0|2].
Obviously, all those real finite-dimensional representation spaces of the

noncompact bi-regular subgroup GL(4, R) × SO0 (1, 3) have no invariant
Hilbert product. A harmonic analysis of, e.g., square-integrable functions
L2 (M10 ) on the metric manifold M10 ∼ = GL(4, R)/SO0 (1, 3) with infinite-
dimensional Hilbert representations does not play a role in classical gravity.
Without rotation degrees of freedom in the abelian Lorentz group
SO0 (1, 1) ⊂ SO0 (1, 3), four-dimensional causal spacetimes M(1,3) contain
two-dimensional causal spacetimes M(1,1) with a three-dimensional metrical
manifold. The dyad ej=0,3
a=0,3 involves one compact rotation and two dilations
for the metrical hyperbola:
M3 ∼ = GL(2, R)/O(1, 1), GL(2, R) = D(1) × SL(2, R),

SL(2, R) ∼ SO0 (1, 2) ∼ SU(1, 1).
The group SL(2, R) has one fundamental two-dimensional spinor representa-

tion, e.g., the dyad, whose totally symmetric products of power 2J ∈ N give
all irreducible finite-dimensional representations with dimension 1 + 2J —
no Hilbert representations. The decomposition of SL(2, R)-representations
into Lorentz group SO0 (1, 1)-representations is parallel to the SO(2)-
decomposition for SU(2)-representations.
Classes of orthogonal groups constitute manifolds with q compact and p
noncompact dimensions:

c+d
p + q = a+b − 2 ≥ 0,
M(p,q) = SO0 (a, b)/SO0 (c, d), with 2
p = ab − cd ≥ 0,
e.g., M3 ∼= GL(2, R)/O(1, 1) ∼ D(1) × SO0 (1, 2)/SO0 (1, 1) = D(1) × Y (1,1) ,
10 ∼
M = GL(4, R)/O(1, 3) ∼ D(1) × SO0 (3, 3)/SO0 (1, 3) = D(1) × M(6,3) .
7.4 Relativity Representations by Induction

Finite-dimensional rectangular matrices, as discussed in the foregoing section,
give all Hilbert representation spaces of homogeneous spaces G/H (H-
relativity) only for compact groups G ⊇ H, e.g., for perpendicular and elec-
tromagnetic relativity. In general, the faithful G × H-Hilbert representations
of a locally compact relativity G/H, infinite-dimensional for noncompact G,
can be induced by representations of an “idolized” subgroup H.
7.4.1 Induced Representations

Induced G-representations are G × H-subrepresentations of the two-sided
regular G × G-representation. They are the extension of the left G-action
L
gH −k→ kgH on the right H-cosets in the form of linear transformations.
The vector spaces for G-representations, induced by subgroup H-repre-
sentations, consist of H-intertwiners on the group w : G −→ W with values
in a Hilbert space with a unitary action of the “idolized” subgroup d : H −→
U(W ). The G-action on the intertwiners is defined by left multiplication k L.
All of this is expressed in the followingcommutative diagram:
G
k L×Rh
−→ G g, k ∈ G, h ∈ H : k L × Rh (g) = kgh−1 ,
⏐ ⏐
⏐ ⏐ H-intertwiner: w(gh−1 ) = d(h).w(g),
w kw ,
G-action: w −→ k w,
W −→ W k w(g) = w(k
−1
g).
d(h)
An H-intertwiner on the group w ∈ W G/H maps H-cosets of the group into

H-orbits in the Hilbert space W . It is defined by its values on representatives
gr ∈ (G/H)r ⊆ G. The G-action comes with the representative-dependent
H-action of the related Wigner element h(k, gr ) ∈ H:
kL
(G/H)
⏐ r −→ (G/H)
⏐ r G × W (G/H)r −→ W (G/H)r ,
⏐ ⏐
w k w , k−1 w(gr ) = w(kgr ) = d(h−1 (k, gr )).w((kg)r ),
W −→ W with kgr = (kg)r h(k, gr ).
In fiber bundle terms: The vector space bundle H • W (M) with base
manifold M = (G/H)r represents the coset bundle H • M. The “idolized”
subgroup acts on the vector space H • W ∈ tvecC ; it is a “gauge group.” The
full group acts on the manifold base G • M ∈ dif R .
In general, the induced G-representation on the vector space W (G/H)r is
decomposable. Since the fixgroups for the left G-action on the right H-cosets
are conjugates of H,
⎪
⎪
GgH = {k ∈ G⎪ ⎪kgH = gH} = gHg −1 ∼ = H,
7.4 Relativity Representations by Induction 195
each G-representation on W (G/H)r and on its G-stable subspaces contains

the inducing H-representation d.
With a G left-invariant measure dgr = dkgr of the manifold M = (G/H)r ,
the intertwiners, in a bra vector notation w| : (G/H)r −→ W , have a direct
integral expansion with the cosets as natural distributive basis gr , a| and
complex coefficients w(gr )a ∈ C (examples ahead):

w| = ⊕ (G/H)r dgr w(gr )a gr , a| ∈ W (G/H)r for W ∼
= Cn , a = 1, . . . , n.
An orthonormal distributive basis is defined with a Dirac distribution

δ(gr , gr ), supported by the relativity manifold G/H and normalized with
respect to the invariant measure dgr :
gr , a |gr , a = δaa δ(gr , gr ),

with w|gr , a = (G/H)r dgr δ(gr , gr )w(gr )a = w(gr )a .
The G-invariant Hilbert product integrates the Hilbert product of the value
space W over the cosets:

W (G/H)r × W (G/H)r −→ C, % w %2 = (G/H)r dgr w(gr )a w(gr )a .
In the simplest case, the functions on the homogeneous space G/H for
H-relativity are valued in the complex numbers as one-dimensional space
W = C1| with trivial H-action d0 (h) = 1. They are expanded as a direct
integral over the cosets with the corresponding function values:

f | : (G/H)r −→ C, f | = ⊕ (G/H)r dgr f (gr )gr |.
They are matrix elements (coefficients) of G-representations D that contain

a trivial H-representation D ⊇ d0 .
7.4.2 Transmutators as Induced Representations

The transmutators above, given by finite-dimensional rectangular matrices
VD ⊗ VιT ∼
= Cn ⊗ Cnι , are acted on by G × H-representations. A left G-action
induces a right H-action by a Wigner element:
kL
(G/H)
⏐ r −→ (G/H)
⏐ r
⏐ ⏐ k−1 Dι (gr ) = Dι (kgr ) = D(k) ◦ Dι (gr )
Dι k Dι ,
= D((kg)r ) ◦ dι (h(k, gr )).
VD ⊗ VιT −→ VD ⊗ VιT
The transmutator bundle VD ⊗ VιT (M) has the G-space M ∼ = (G/H)r as base
manifold and, in the fiber, the linear transmutators with rectangular matrices
from the G-space VD to the H-space VιT .
Transmutators can be used for a decomposition of any G-representation

induced by a H-representation dι on a vector space with basis nι ; a| ∈ VιT :

wι : (G/H)r −→ VιT , wι | = ⊕ (G/H)r dgr wι (gr )a nι ; gr , a|.
There occur all G-representations D, which contain the inducing H-represen-

tation dι . With a basis |nD ; j ∈ VD , one obtains the harmonic D-components
w̃ι (D)j , which come with multiplicity nD ,

Dιj | =⊕ (G/H)r dgr Dι (gr )ja nι ; gr , a|,
wιfin |
= nD w̃ι (D)j Dιj |,
with
w̃ι (D)j = wι |nD ; j,

D⊇dι

fin
wι (gr )a = j fin
nD w̃ι (D)j D(gr )a , wι (kgr )a = nD w̃ι (D)j D(k)jk D(gr )ka ,
D⊇dι D⊇dι
e.g., the harmonic analysis of functions with the harmonic D-components

f˜(D)j ,

j D j
| = ⊕
dgr D(gr )j0 gr |,
f | =
fin
nD f˜(D)j D0 |, with 0 (G/H)r
D⊇d
f˜(D)j = f |D; j.
0
7.5 Hilbert Spaces of Compact Relativities

For a compact group G ⊇ H, the finite-dimensional rectangular transmuta-
tors are square-integrable on the manifolds M = (G/H)r . They are complete
for the harmonic analysis of the group G and its homogeneous spaces G/H;
i.e., they exhaust all square-integrable induced representations by orthogonal
direct Peter–Weyl decompositions:
⎧
⎪
⎪ wι | = wιfin |,
⎪
⎪ dense
⎪
⎨ L2 (M, V T ) ∼
ι = nD VD ⊗ VιT ,
compact G :
⎪
⎪ ι
D⊇d
⎪
⎪ id(VιT )G/H ∼ nD |nD ; jnD ; j|.
⎪
⎩
= nD idVD =
D⊇dι D⊇dι
The representation structure of the full group G is determined by that

of the inducing subgroup; for compact groups, there is Frobenius’ reciprocity
theorem: The number nD of equivalent irreducible G-representations of type
D on L2 (G/K, VιT ), induced by an irreducible subgroup K-representations
dι on Vι , is given by the number of dι -equivalent K-representations in D,
i.e., nD = ndι ⊆D . Therefore, there are n2D K-representations in G-repre-
sentations of type D. For example, an SU(2)-dublet representation induces
on L2 (SU(3)/SU(2), C2 ) no SU(3)-singlet, one SU(3)-triplet, one SU(3)-
antitriplet, two SU(3)-octet representations, etc. Therefore, for SU(2)-dublet
representations: 1 in SU(3)-triplets and 1 in antitriplets, 22 in octets, etc.
7.5 Hilbert Spaces of Compact Relativities 197
In the examples discussed earlier, all representation matrix elements of

the compact groups U(2) and SU(2) are square-integrable with the finitely
decomposable Hilbert spaces for electromagnetic relativity L2 (G 3 , Vz ) and
perpendicular relativity L2 (Ω2 , V|z| ): The complex functions for perpendicu-
lar relativity, i.e., L2 (Ω2 , V|z| ) = L2 (Ω2 ) for z = 0, V0 ∼
= C, are the spherical

harmonics as products of the three matrix elements ω −→
3 4π 1
ω) ∈ C
3 Ya (
in the middle column with trivial representations of SO(2) eiχ and a triplet
representation of SO(3):
i(χ+ϕ) 2 θ −i(χ−ϕ)

e cos iϕ sin
2
√θ
ie 2
2 θ
−e sin 2
ieiχ sin
√ θ
2
cos θ ∈ SO(3),
ie−iχ sin
√θ
2
−ei(χ−ϕ) sin2 θ
ie−iϕ sin
√ θ e−i(χ+ϕ) cos2 θ
2 2 2
SU(2)/SO(2) ∼
= Ω2 xr =
ω −→ [J]( ω )a0 = 1+2J 4π
YaJ (ω ) ∈ C
for J = 0, 1, 2, . . . with a = −J, . . . , J,
O ∈ SO(3) : [J](O)ab YbJ (ω ) = YaJ (O.ω ).
There is Schur orthogonality [52, 24, 41] with the Plancherel distribution (see
Chapter 8) given by the dimension 1 + 2J of the irreducible representation
space,
d2 ω Y J ( J
JJ
ω )a0 [J ](
ω ) Ya (
ω)
[J]( ω )a0 = d2 ω √a √ = 1
Ω2 4π Ω2 1+2J 1+2J 1+2J δ δaa .
It involves the rotation-invariant normalizable Haar measure and the dis-

tributive basis of the 2-sphere,
2π 1
Ω2
d2 ω = 0 dϕ −1 d cos θ = 4π,
ω |
ω = δ(
ω−ω ) = δ(ϕ − ϕ ) sin1 θ δ(θ − θ ).
The spherical harmonics exhaust the square-integrable 2-sphere functions:

∞

L2 (Ω2 ) f | = ⊕ Ω2 d2 ωf ( ω |, f (
ω) ω) = f˜(J)a YaJ (ω ).
J=0
The Haar measure can be rewritten with a 2-sphere-supported Dirac distri-

bution:

[J]a0 | ∼ ⊕ Ω2 d2 ω YJ ( ω | = ⊕ d3 x δ(x2 − 1)(x)Jtraceless x|,
ω )
J
with |xx| = [J](ω ).
traceless
There are no finite-dimensional faithful Hilbert representations of non-

compact Lie groups. For example, the three matrix elements in the middle
3 In the Euler angle parametrization, both the middle column and the middle row define
the SO(3)-action on the 2-sphere Ω2 ∼ = SO(3)/SO(2) ∼ = SO(2)\SO(3). The central element

θ −→ cos θ parametrizes the double-coset space, the 1-sphere Ω1 ∼
= SO(2)\SO(3)/SO(2) ∼ =
SO(2) and is a spherical Ω2 -function.
column with trivial representations of SO(2) eiχ , i.e., the noncompact

hyperbolic partners of the spherical harmonics Ya1 (ϕ, θ) for iθ = ψ,
i(χ+ϕ) −i(χ−ϕ) 2 ψ
e 2 ψ
cosh 2
iϕ sinh
e √ ψ
2
−e sinh 2
eiχ sinh
√ ψ
2
cosh ψ e−iχ sinh
√ ψ
2 ∈ SO0 (1, 2),
−ei(χ−ϕ) sinh2 ψ
2 e−iϕ sinh
√ ψ
2
e−i(χ+ϕ) cosh2 ψ
2
are complex functions on the 2-hyperboloid SO0 (1, 2)/SO(2) ∼ = Y 2 −→ C

with a triplet representation of SO0 (1, 2), not square-integrable.
7.6 Flat Spaces as Orthogonal Relativities

Representations of a Lie group can be induced from any subgroup. Sometimes
there are distinguished subgroups, e.g., for the infinite-dimensional Hilbert
representations of the affine groups SO0 (t, s) × Rn (semidirect products),
with translations or (energy-)momenta x, q ∈ R , x2 = x2t − x2s , and the ho-
n
mogeneous groups — rotations for ts = 0 or the causality-compatible Lorentz

groups for t = 1. They are used for flat spaces as orthogonal relativities:
Rn /SO0 (t, s) ∼
SO0 (t, s) × = Rn , n = t + s ≥ 2.
H × Rn -representations are inducible from representations of direct product

subgroups H0 × Rn with the homogeneous fixgroups H0 ⊆ H for the different
types of translations or (energy-)momenta,

which, for the nontrivial cases
q = 0, all have the real dimension n−1
2 :
⎧
⎪
⎪ SO0 (t, s) for q = 0,
⎪
⎪
⎪ ⎨ SO 0 (t − 1, s) for q 2 > 0; t ≥ 1,
⎪
⎪ ∼
SO0 (t, s)q ={g ∈ SO0 (t, s)⎪g.q = q}= SO0 (t, s − 1) for q 2 < 0; s ≥ 1,
⎪
⎪
⎪
⎪ SO0 (t − 1, s − 1) × Rn−2
⎩
for q = 0, q = 0; t, s ≥ 1.
2
For example, the Hilbert representations of the Euclidean group SO(3) ×

R3 for nonrelativistic scattering are inducible by the Hilbert representations
of the direct product SO(2) × R3 with the axial rotation subgroups SO(2)
around the momentum direction |qq| = ω ∈ Ω2 ∼
= SO(3)/SO(2). The transla-
tion representations R x −→ e
3 iP
ωx
∈ U(1) with invariant P > 0 use the
directions on the manifold Ω2 of SO(2)-subgroups in SO(3) (see Chapter 8).
7.6.1 Particle Analysis of Special Relativity

The Hilbert representations of the Poincaré group SO0 (1, 3) × R4 and its
cover group SL(2, C) × R (see Chapter 5) for free relativistic particles are
4
induced, for massive particles m > 0 (translation-invariant), from SU(2)×R4

with the spin subgroups SU(2). The translation representations R4 x −→
eimyx ∈ U(1) use the timelike direction to the point m q
= y ∈ Y3 ∼ =
7.6 Flat Spaces as Orthogonal Relativities 199
SO0 (1, 3)/SO(3) on the energy-momentum hyperboloid q 2 = m2 , which

parametrizes the SU(2)-subgroups in SL(2, C). For massless particles, the
induction starts from the fixgroup (axial rotations around the momentum)
in the fixgroup SO(2) ⊂ SO(2) × R2 ⊆ SO0 (1, 3) ×
R4 .
The massive particle analysis of free quantum fields with respect to the
eigenvectors for spin rotations involves non-Hilbert representations of the
Lorentz group via the rectangular transmutators acted on by SL(2, C) ×
SU(2)-representations. For example, the representation Λ = [ 12 | 12 ] of
the Lorentz group in the SO0 (1, 3) × SO(3)-representation on
C4 ⊗ VιT for a special relativistic vector field and the tensor representation
Λ ∧ Λ = [1|0] ⊕ [0|1] on C6 ⊗ VιT for its field strength,
3
Za (0) = ⊕ 2q0d(2π)
q
3
q
Λar ( m ) [ur (q ) + ur (q )],
3
iFab (0) = ⊕ 2q0 (2π)3 Λc0 ( m )ab
d q q d q r
cd Λr ( m )[u (q ) − ur (q )],

with q0 = m + 2 q , cd = δc δd − δd δc , r = 1, 2, 3, a = 0, 1, 2, 3,
2 ab a b a b
are both induced by an adjoint SO(3)-representation [1] on Vι ∼ = C3 and

T
its dual Vι . These spin representations act on the creation and annihilation
operators ur ( q ), ur (
q ) for a massive particle with momentum q and spin-1
directions r = 1, 2, 3. The action of the creation operators on the Fock ground
state |0 gives dual distributive bases of the special relativistic manifold,
i.e., of the energy-momentum hyperboloid Y 3 ∼ = SO0 (1, 3)/SO(3) for mass
m2 > 0. The distributive orthogonality is given by the Fock expectation value
0| . . . |0:
|1; q )|0 ∈ Vι (
q, r = ur ( q ), 1; q , r| = 0|ur (q ) ∈ VιT (q ),
1; p, s|1;
q, r = 0|us ( q )|0 = δ rs 2q0 (2π)3 δ(q − p ).
p )ur (
The four- and six-dimensional Lorentz group representations do not act on

Hilbert spaces. This can be seen at the transmutators from the Lorentz group
to the rotation group, e.g., { m
q
−→ Λja ( m
q
)}, which are not square-integrable
L2 (Y 3 ) on the energy-momentum hyperboloid.
The Lorentz invariant nonnormalizable measure of the 3-hyperboloid in
the momentum parametrization can be written as an integral with a Y 3 -
supported Dirac distribution:
3
m2 d3 y = √ d 2q = d4 q ϑ(q0 ) 2δ(q 2 − m2 ) = m2 |Y 3 | = ∞.
m q2
+
The finite-dimensional Lorentz group SL(2, C)-representations that con-

tain a trivial rotation group SU(2)-representation are the Minkowski rep-
resentations [ n2 | n2 ] (see Chapter 5). They act on vector spaces C(1+n)(1+n) ,
n = 0, 1, . . . , with the spin-representation decomposition:
⎧
SO(3) n ⎨ [0|0] ∼ = [0],
irrep Mink SO0 (1, 3) [ 2 | 2 ] =
n n ∼ [J], e.g., [ 12 | 12 ] ∼
= [0] ⊕ [1],
⎩
J=0 [1|1] ∼ = [0] ⊕ [1] ⊕ [2].
They are used for the expansion of the Minkowski representations in the
complex functions on energy-momentum hyperboloids,
[ n2 | n2 ]j01 ...jn | : Y 3 −→ C for n = 0, 1, 2, . . . ,

3
[ n2 | n2 ]j01 ...jn | = ⊕ 2q0d(2π)
q n n j1 ...jn
3 [ 2 | 2 ]0 q )q|, with q0 = m2 + q2
(
d4 q
= ⊕ (2π) 2 n n j1 ...jn
3 ϑ(q0 )δ(q − m )[ 2 | 2 ]0
2
(q)q|,
and arise as contributions of the Feynman propagators for trivial translations,

e.g., for a spin-0 particle in a scalar field Φ , a spin- 12 -particle in a Dirac field
Ψ , and a spin-1 particle in a vector field Z:
d4 q
Φ (0) : [0|0]| = ⊕ (2π)3 ϑ(q0 )δ(q 2 − m2 )q|,
d4 q γa q a
Ψ (0) : [0|0]| ⊕ γa [ 12 | 12 ]a0 | = ⊕ (2π)3 ϑ(q0 )δ(q 2 − m2 )(14 + m )q|,
⊕
d4 q a b
Za (0) : [1|1]ab 0 | = (2π)3 ϑ(q0 )δ(q 2 − m2 )(−η ab + qmq2 )q|.
The spacetime translation-dependent fields, e.g., a massive vector field,

c
R4 x −→ Za (x), Fab (x) = ab ∂ d
cd m Z (x),
q ) and e−iqx ur (

involve eiqx ur ( q ), which are the translation orbits R4 x −→
e±iqx ∈ U(1) for a representation of the Poincaré group SO0 (1, 3) × R4 . This
leads to the spacetime translation representation coefficients with q|x = eiqx
and x|q = e−iqx as the on-shell part of the Feynman propagator:
d4 q qa qb iqx
[1|1]ab
0 |x + x|[1|1]0 =
ab
(2π)3 δ(q − m )(−η
2 2 ab
+ m2 )e
= 0|{Z (0), Z (x)}|0.
a b
Chapter 8
Representation Coeffficients
Internal chargelike (“vertical”) operations come from compact groups, e.g.,

hypercharge U(1), isospin SU(2), and color SU(3). In the electroweak and
strong standard model, they are implemented as gauge transformations that,
via gauge fields in covariant derivatives, accompany the translations (see
Chapter 6). External spacetimelike (“horizontal”) operations come from non-
compact groups: For example, interaction-free vectors are acted on by Hilbert
representations of flat space operation groups, e.g., of the Euclidean group
SO(3) × R3 in nonrelativistic quantum mechanical scattering theory, or of
the Poincaré (cover) group SL(2, C) × R4 for elementary particles in rela-
tivistic quantum fields.
Each group determines its Hilbert spaces, finite- or infinite-dimensional,
where its action can be represented by definite unitary automorphisms. A
Hilbert space, e.g., the Fock space for translations and free particles, may
not be appropriate for another group, e.g., for bound states or for the imple-
mentation of interactions.
Each representation of a compact Lie group K is equivalent to a repre-
sentation D : K −→ U(V ) in the unitary group of a Hilbert space V . It is
decomposable into irreducible finite-dimensional ones, V ∼ = Cd . The invari-
ants and eigenvalues, such as quantum numbers for charge, isospin, color, etc.,
are rational and ultimately connected to integer winding numbers z ∈ Z from
the dual group of U(1), which characterize the irreducible representations
U(1) eiα −→ eizα ∈ U(1). A maximal abelian compact subgroup with di-
rect factors U(1) ∼
= SO(2) is called a Cartan torus, e.g., SO(2)r ⊂ SU(1 + r)
or, for hyperisospin, U(1)+ × U(1)− ⊂ U(2). The representations (char-
acters) of the Cartan tori are basic for the representations of a compact
group.
The theory of the Hilbert representations of noncompact Lie groups,
infinite-dimensional if faithful, is more complicated and difficult. A noncom-
pact semisimple group has an Iwasawa decomposition G = K ◦ A ◦ N into a
maximal compact subgroup K, a maximal noncompact abelian subgroup A,
202 Chapter 8 Representation Coeffficients
and a subgroup N with nilpotent Lie algebra. The factor subgroups are
unique up to isomorphy, e.g.,
1 ≤ t ≤ s : SO0 (t, s) = [SO(t) × SO(s)] ◦ SO0 (1, 1)t ◦ exp Rt(s−1) ,
1+r
SL(1 + r, R) = SO(1 + r) ◦ SO0 (1, 1)r ◦ exp R( 2 ) ,
SL(1 + r, C) = SU(1 + r) ◦ SO0 (1, 1)r ◦ exp R(1+r)r .
The group A is isomorphic to a translation group A ∼ = D(1)r ∼= SO0 (1, 1)r ∼=

R and called a Cartan plane. Its dimension r is the real rank of the noncom-
r
pact group G and gives the maximal number of representation characteriz-

ing invariants from a continuous spectrum, called continuous invariants. In
physics, they can be used as basic units for the related operations.
Hilbert representations of a Lie group contain, as basic substructures,
Hilbert representations of its Cartan tori and planes, which determine, respec-
tively, the rational and continuous representation invariants and eigenvalues,
e.g., one rational and one continuous invariant for a Cartan cylinder with
axial rotations and dilations SO(2) × SO0 (1, 1) ⊂ SO0 (1, 3) in the Lorentz
group.
The cyclic Hilbert representations cycrep + Rr (definition ahead) of a
Cartan plane with translations are characterized by essentially bounded
positive-type functions (scalar product or Hilbert metric-inducing functions)
Rr x −→ d(x), d ∈ L∞ (Rr )+ . Such positive-type functions are, in a
sense, the noncompact analogue to the representation-characterizing inte-
ger winding numbers Zr for Cartan tori. Scalar product–inducing functions
are Fourier-transformed positive Radon measures (distributions, densities)
M(Řr )+ of the group dual Řr , i.e., of the irreducible, but not faithful trans-
lation representations Rn x −→ eiqx ∈ U(1) with a continuous eigenvalue
iq for “energy-momentum” q ∈ Rn .
8.1 Finite-dimensional Representations

In this section, all vector spaces are assumed to be finite-dimensional over
the scalars K ∈ {R, C}.
8.1.1 Metrics of Representation Spaces

If a vector space is acted on by a group G or a Lie algebra L,
D : G −→ GL(V ), G × V −→ V, g • v = D(g)v,
D : L −→ V ⊗ V T , L × V −→ V, l • v = D(l)v,
the dual vector space V T has the dual representations G g −→ Ď(g) =

D(g −1 )T and L l −→ Ď(l) = −D(l)T . With a trivial action on the scalars,
g • α =! α ∈ K, l • α = 0, this defines the group action on the full tensor
algebra (V ⊕ V T ) by g • (a1 ⊗ a2 ) = (g • a1 ) ⊗ (g • a2 ) and the Lie algebra
action by Leibniz’s rule, l • (a1 ⊗ a2 ) = (l • a1 ) ⊗ a2 + a1 ⊗ (l • a2 ).
8.1 Finite-dimensional Representations 203
An (anti-)symmetric inner product of a vector space is invariant under

the corresponding action for

ζ(v, g • w) = ζ(g −1 • v, w), g ∈ G,
ζ( , ) : V × V −→ K,
ζ(v, l • w) = −ζ(l • v, w), l ∈ L.
A product representation on V1 ⊗ V2 with invariant forms has ζ1 ⊗ ζ2 with
the product matrix elements ζ1 ⊗ ζ2 (v1 ⊗ v2 , w1 ⊗ w2 ) = ζ1 (v1 , w1 )ζ2 (v2 , w2 )
as an invariant form.
Any invariant subspace has its invariant ζ-orthogonal subspace partner,
W ⊆ V : G⎪ • W ⊆ W, L • W ⊆ W,
⎪
Wζ⊥ = {v ∈ V ⎪
⎪ζ(W, v) = {0}} ⇒ G • Wζ⊥ ⊆ Wζ⊥ , L • Wζ⊥ ⊆ Wζ⊥ .
Therefore, if an irreducible nontrivial representation has a nontrivial invariant
form, the inner product must be nondegenerate:
V irreducible ⇒ ζ nondegenerate ⇐⇒ Vζ⊥ = {0}.
With Schur’s lemma, an irreducible, complex finite-dimensional representa-
tion of a group or Lie algebra can have, up to scalar multiples, only one
invariant linear or only one invariant conjugate linear dual isomorphism ζ,
possibly both or none. With two ζ1,2 , one has ζ1 ◦ ζ2−1 = α idV .
A nondegenerate inner product defines a (conjugate) linear dual isomor-
phism ζ : V −→ V T , v −→ ζ( , v) to the dual space and an invariant product
ζ −1 of V T with ζ −1 (ω, θ) = ζ(ζ −1 .ω, ζ −1 .θ). The representations are called
(conjugate) linear self-dual:
D(g), D(l)
V
⏐ −→ V
⏐
⏐ ⏐ D(g −1 ) = ζ −1 ◦ D(g)T ◦ ζ = D(g)∗ ,
ζ ζ ,
−D(l) = ζ −1 ◦ D(l)T ◦ ζ = D(l)∗ .
VT −→ VT
Ď(g), Ď(l)
With an inner product ζ of V and ζ −1 of its dual V T , and the correspond-

ing product of the scalars, linear αβ or !
antilinear!αβ, there is an invariant
inner product of all finite tensor powers k V ⊗ l V T .
The properties of nondegenerate inner products determine the groups,
⎧ ⎧
⎨ ζ(u, v) ⎨ O(V, ζ), orthogonally self-dual,
ζ(v, u) = −ζ(u, v) ⇒ D[G] ⊆ Sp(V, ζ), symplectically self-dual,
⎩ ⎩
ζ(u, v) U(V, ζ), unitarily self-dual,
and the Lie algebras, e.g., D[L] ⊆ log O(V, ζ). The diagonalization of orthog-
onal and unitary ζ displays its signature (p, q).
A complex representation of a real Lie group GR or Lie algebra LR has
to come with a conjugation of the representation space, i.e., it has to be
unitarily self-dual:
D[GR ] ⊆ U(p, q), with D(g)∗ = D(g −1 ),
D[LR ] ⊆ log U(p, q), with D(l)∗ = −D(l).
A real Lie algebra representation in compact O(n) or U(n) is positive

self-dual, i.e., a Hilbert representation. Then the double trace is strictly
positive and the associated inner product is strictly negative:
tr D(l) ◦ D(l) = − tr D(l) ◦ D(l) ≥ 0,

notation: ζ(v, w) = v|w with v|v =% v %2 ≥ 0 ⇐⇒ v = 0.
Some examples: The irreducible SU(n)-representations have an invariant

antilinear scalar product as the product of the group-defining scalar product
ζn (v, w) = v|w on V ∼= Cn . The SU(2)- and SL(2, C)-spinor representa-
∼
tions on V =C have
2
the invariant indefinite antisymmetric linear “metric”
A B 0 1
(e , e ) = −1 0 , given by the volume elements. There is no invariant
bilinear form for the (anti)quark representation of SU(3). The SU(3)-octet
representation has the Killing form κ8 ∼
= ζ3 ⊗ ζ3−1 as invariant definite sym-
metric linear metric — more generally, κn2 −1 ∼ = ζn ⊗ ζn−1 for SU(n)/I(n)
∼
on V = R n2 −1
. The Minkowski SO0 (1, 3)-representation has the invariant
indefinite symmetric linear Lorentz “metric” η ∼ = ⊗ −1 as the product of
the dual volume elements.
The natural isomorphisms (dual representations) for transposed endomor-
phism Lie algebras, AL(V ) ∼= AL(V T ), l ↔ −lT , and automorphism groups,
∼
GL(V ) = GL(V ), g ↔ g −1T , do not imply that these isomorphisms have
T
ζ
to arise as products ζ ⊗ ζ −1 with a dual isomorphism V ∼
= V T.
8.1.2 Metrics of Lie Algebras

Now, the special case of linear metrics for a Lie algebra with adjoint action:
An invariant linear form ω ∈ LT is trivial on its commutator ω([L, L]) =
{0}. Since [L, L] = L for a semisimple Lie algebra, such an L has no nontriv-
ial invariant linear form. Each representation of a semisimple Lie algebra is
traceless, e.g., the adjoint one (structure constants):
D : L −→ AL(V ),
L −→ K, l −→ tr D(l),
tr D(l) = 0,
semisimple L ⇒
tr ad la = ab
b = 0.
Any representation of a Lie algebra has as associate inner product κD the

symmetric “double trace,” invariant under the adjoint action,
κD ( , ) : L × L −→ K, κD (l, m) = tr D(l) ◦ D(m).
Associated to the adjoint Lie algebra representation is the Killing form

κ(l, m) = tr ad l ◦ ad m. An associate inner product can be diagonalized
with κD (la , lb ) = ±δ ab , 0.
Precisely for semisimple Lie algebras, the Killing form is nondegenerate.
For a complex, simple Lie algebra, the Killing form is, up to a scalar factor,
the unique invariant inner product.
8.2 Group Algebras and Representation Spaces 205
An abelian Lie algebra has a trivial Killing form. However, the associ-
ated inner products κD are not neccessarily trivial for all its representations.
Nonsemisimple Lie algebras, e.g., the abelian Lie algebra K, can have repre-
sentations with nondegenerate inner products κD .
An invariant linear inner product γ of a complex vector space with a
faithful irreducible representation of a simple Lie algebra defines an invariant
inner product of the endomorphisms AL(V ) ∼ = V ⊗ V T:
Γ = γ ⊗ γ −1 : AL(V ) × AL(V ) −→ C.
It has to coincide, up to a nontrivial factor, on the image of the Lie algebra

D[L] ⊆ AL(V ) with the associated inner product κD . In addition, γ has to
be either symmetric or antisymmetric:
κD (l, m) = tr D(l) ◦ D(m) = eα Γ(l, m), α ∈ C,

Γ = ΓT ⇐⇒ γ = γ T or γ = −γ T .
A semisimple real Lie algebra is compact with strictly negative Killing

form. Compact Lie algebras have bases with totally antisymmetric structure
constants. Also, noncompact Lie algebras, e.g., R, can have Hilbert represen-
tations. All complex representations of compact Lie algebras are semisimple
and decomposable into irreducible Hilbert representations, all of which are
finite-dimensional (theorem of Weyl).
8.2 Group Algebras and Representation Spaces

The equivalence classes of the irreducible complex Hilbert representations of
a group G −→ U(V ) constitute the group dual Ǧ = irrep + G. Its deter-
mination, especially for noncompact nonabelian groups, may be a difficult
task. Group representation matrix elements are group functions, called re-
presentation coefficients. For an orientation, it is useful to recapitulate in
this section, in general, the relevant vector spaces and algebras for group
functions [24].
8.2.1 Finite Groups

For a finite group, compact with the8discrete topology, the complex linear
combinations1 of group elements f = g∈G gf (g) ∈ CG can be seen as group
functions g −→ f (g). They constitute the group algebra CG = {G −→ C}
with the group elements as natural basis CG ∼= C card G . The group product
induces an algebra structure with convolution product f1 ∗ f2 ∈ CG . The
minimal two-sided ideals of ⎪
the group algebra are generated by the conju-
⎪
gacy classes G(k) = {gkg −1 ⎪
⎪g ∈ G} of the group [40]. They characterize
1 Sometimes, for distinction, the addition of linear independent vectors is written as
|v ⊕ |w.
the irreducible representations with dimension dι of the group G by finite

unitary groups in U(dι ) ⊆ U( card G), and of the group algebra CG by the
corresponding full matrix ideals with dimensions d2ι :

n $
n
finite G : CG ∼
= Cdι ⊗ Cdι , card G = d2ι .
ι=1 ι=1
A characteristic and familiar example is the Young tableaux–related irre-

ducible representations of the permutation group G(n) with cardinality n!.
8.2.2 Algebras and Vector Spaces for Locally

Compact Groups
More general: For a real finite-dimensional Lie group G, complex-valued map-
pings (functions, distributions, measures) of the group have to be considered
[57,28]. Summation over group elements is expressed by invariant integration
with Haar measure, a counting measure for finite groups.
A (direct)
integral over group elements uses a left or right Haar measure
G dg = dg — both unique up to a scalar and related to each other by the
modular function in the dilation representation G g −→ Δ(g) ∈ D(1). A
non-unimodular group has to have a normal subgroup with dilation classes
G/N ∼ = D(1).
The group
functions contain all representation matrix elements. The map-
pings f = ⊕ dg gf (g) constitute group representation spaces; they inherit the
group multiplication in the left-right regular (bi-regular) representation of
G × G,
(G × G) • G −→ G, (g1 , g2 ) • g = g1 gg2−1 ,
f −→ (g1 , g2 ) • f = g1 fg2 = ⊕ dg gf (g1−1 gg2 ).
With a Haar measure basis, all group measures can be characterized by (gen-
eralized) functions μ(g)dg. Distributions have, via duality to their functions,
the dual action

(g1 , g2 ) • μ, f = μ, (g1 , g2 ) • f , dg μ(g1 gg2−1 )f (g) = dg μ(g)f (g1−1 gg2 ).
As exemplified by the finite groups with the convolution algebra CG , also

in the general case, all group representation spaces come from a convolution
algebra with group functions.
Two products are important for the (generalized) group functions μ = ⊕
dg gμ(g) of a locally compact group (unimodular Δ(g) = 1 if necessary). The
group functions inherit the group multiplication g1 g2 = g ∈ G as convolution
product (where defined),

μ1 ∗ μ2 = [ ⊕ dg gμ1 (g)] ∗ [ ⊕ dg gμ2 (g)] = ⊕ dg g(μ1 ∗ μ2 )(g),

μ1 ∗ μ2 (g) = dg1 dg2 μ1 (g1 ) δ(g1 g2 g −1 )μ2 (g2 ) = dg1 μ1 (g1 )μ2 (g1−1 g).
8.2 Group Algebras and Representation Spaces 207
The associative convolution is abelian if, and only if, the group multipli-
cation is abelian. From the complex numbers, the group functions inherit
the abelian pointwise multiplication (where defined), important for product
representations,
μ1 · μ2 (g) = μ1 (g)μ2 (g).
With respect to the convolution and pointwise product, the Lebesgue
Banach spaces Lp (G), 1 ≤ p ≤ ∞, with the classes of Haar measure al-
most
everywhere defined absolute p-integrable functions, i.e., (% f %p )p =
dg |f (g)|p < ∞, are related to each other as follows:
p
L (G) ∗ Lr (G) ⊆ Ls (G), with p1 + 1r − 1s = 1,
1 ≤ p, r, s ≤ ∞ :
Lp (G) · Lr (G) ⊆ Ls (G), with p1 + 1r − 1s = 0.
They are left-right modules, respectively, for the absolute integrable group
function classes L1 (G), a convolution algebra, and for the essentially bounded
group functions L∞ (G), i.e., |f (g)| < ∞ almost everywhere, a pointwise
product algebra,
L1 (G) ∗ L1 (G) ⊆ L1 (G), L∞ (G) · L∞ (G) = L∞ (G).
They are left-right convolution modules even for the Radon measures M(G)
of the group (definition ahead), a unital convolution Banach algebra,
M(G) ∗ M(G) = M(G),

1 ≤ p ≤ ∞ : M(G) ∗ Lp (G) ∗ M(G) = Lp (G).
The Radon measures, in the form of Radon distributions with Haar measure
ω(g)dg, embed the group by Dirac measures:

G k −→ δk ∈ M(G) with δk , f = δk (g)dg f (g) = f (k),
δk ∗ δl = δkl , δ1 = δ,
δk (g) = δ(gk −1 ).
Dirac distributions and Haar measures are “inverse to each other,” especially
their normalizations:

δk , 1 = δk (g)dg = 1 = δ(g)dg.
The left-right action of a group on itself is embedded in the left-right

convolution module property of the function spaces for the Radon group
measures,
M(G) ∗ Lp (G) ∗ M(G) −→ p

L (G),
f −→ μ1 ∗ f ∗ μ2 ,
μ1 ∗ f ∗ μ2 (g) = dg1 dg2 μ1 (g1 )f (g1−1 gg2 )μ2 (g2−1 ),
e.g., δk ∗ f ∗ δl (g) = f (k −1 gl−1 ).
The Radon measures can be defined as the dual M(G) = Cc (G) of the
compactly supported continuous functions Cc (G); these functions are dense
in all Lp (G), 1 ≤ p < ∞. M(G) contains the function algebra L1 (G) as

two-sided ideal. The involutive convolution algebra Cc (G) is a subspace of the
bounded continuous functions Cb (G), which, in their turn, can be considered
as a closed, in general, proper subspace of the essentially bounded functions,
L1 (G) ⊆ M(G) ⊃ G,
Cc (G) ⊆ Cb (G) ⊆ L∞ (G).
As suggested by the measure dn x in the integration with a distribution,

every Radon distribution of an open real set T ⊆ Rn is a finite sum of
derivatives up to order n of locally essentially bounded functions [56],
7
$
n
T ⊆ Rn : M(T ) ⊆ αN ∂ N L∞ (T ) ,
N =0
e.g., the Dirac distribution as derivation of the step and sign functions
L∞ (R) ϑ, ∈
/ M(R) or the Yukawa potential,
d (x)
R with dx : d
dx ϑ(x) = dx 2 = δ(x),
SO(3) × R3 ∼
= R3 with d3 x : d −mr m −mr
− 2r
SO(3) dr 2 e = e ,
SO0 (1,3) ×
R4 ∼
= R4 4
with d x : ( dxd 2 )N ϑ(x2 ) = δ (N −1) (x2 ), N = 1, 2.
SO0 (1,3)
All the (generalized) function vector spaces and algebras considered have
an involution:
μ ↔ μ̂ with μ̂(g) = μ(g −1 ) for unimodular G.
With a group representation D : G −→ GL(V ), there may be a representa-

tion of the three convolution group algebras A(G) ∈ {Cc (G), L1 (G), M(G)}
in the endomorphism algebra AL(V ), for a function or a Radon distribu-
tion μ,

D : A(G) −→ AL(V ), μ −→ D(μ) = dg D(g)μ(g),
D(μ1 ∗ μ2 ) = D(μ1 ) ◦ D(μ2 ),
D(δk ) = D(k).
In the following, the unital convolution algebra M(G) with the Radon
measures, and its ideal the Lebesgue functions L1 (G), and the unital abelian
pointwise product algebra L∞ (G) with the essentially bounded functions play
the most important roles. With the two-sided convolutive action of M(G) and
8.3 Schur Product of Group Functions 209
L1 (G) and the pointwise action of L∞ (G), these spaces will be used for the
group G and the group dual Ǧ,
∗ M(G) L1 (G) L∞ (G) · L∞ (G) L1 (G) M(G)
M(G) M(G) L1 (G) L∞ (G) L∞ (G) L∞ (G) L1 (G) M(G)
L1 (G) L1 (G) L1 (G) L∞ (G) L1 (G) L1 (G) − −
L∞ (G) L∞ (G) L∞ (G) − M(G) M(G) − −
Convolution product Pointwise product

μ1 ∗ μ2 (g) μ1 · μ2 (g)
from group product for product representations
G × G −→ G rep G × rep G −→ rep G
The “divergences” of quantum field theory have their origin in undefined
pointwise multiplications of Radon measures M(R4 ) · M(R4 ) of spacetime
translations.
8.3 Schur Product of Group Functions

For a vector space with group functions, the value at the neutral element, if
defined, is a linear form:

F (G) f −→ f (1) = dg δ(g)f (g) ∈ C.
Two vector spaces are put in duality by a bilinear form W ×V (w, v) −→
d(w, v) ∈ K. It defines a mapping to the linear forms W w −→ wd =
d(w, ) ∈ V T with wd , v = d(w, v). For finite-dimension, all duality struc-
tures are described by the algebraic dual V T and the dual product V T × V
(θ, v) −→ θ, v ∈ K, for dual bases ěa , eb = δba . That is more complicated
for infinite-dimension and continuous linear forms from the topological dual
V ⊆ V T.
For bilinear and sesquilinear forms of vector spaces with (generalized)
group functions, the Schur product is used. It is the convolution product at
the neutral element, if defined,

F1 (G) × F2 (G) (μ1 , μ2 ) −→ μ1 ∗ μ2 (1) = dg μ1 (g −1 )μ2 (g),

μ̂1 ∗ μ2 (1) = dg μ1 (g)μ2 (g).
It does not have to be a complex number and may be a distribution, e.g., on
the group dual Ǧ with the Plancherel measure (more ahead).
8.3.1 Duality for Group Function Spaces

The duality for the Lebesgue spaces is given by the Schur product, valued in
the essentially bounded functions, at the neutral group element,
Lp (G) = Lr (G), with Lp (G) ∗ Lr (G) ⊆ L∞ (G)
for 1p + 1r = 1, 1 < p, r < ∞,

Lp (G) × Lr (G) −→ C, fp , fr = fp ∗ fr (1) = dg fp (g −1 )fr (g).
The space L2 (G) with the square-integrable functions is self-dual.
The essentially bounded functions L∞ (G) constitute the dual space for
the Lebesgue function algebra L1 (G) and the Radon distributions for the
compactly supported functions, M(G) = Cc (G) . The Radon distributions
are put into duality also with the essentially bounded functions by μ, d =
μ ∗ d(1):
L1 (G) = L∞ (G), with L1 (G) ∗ L∞ (G) ⊆ L∞ (G),
L (G) ⊆ M(G) ⊆ L∞ (G) ,
1
with M(G) ∗ L∞ (G) ⊆ L∞ (G).
8.3.2 Hilbert Metrics of Cyclic Representation Spaces

For a cyclic (“one orbit-based”) Hilbert representation of a locally compact
group G, the representation Hilbert space is the closure of the C-span of the
G-orbit V = C(G•|c) of a vector |c, which is called a cyclic vector. With
the fixgroup H of a cyclic vector, G • |c ∼ = G/H, one has V ∼ = C(G/H) . All
vectors |c ∈ G • |c of the orbit are cyclic. An invariant vector G • |c = |c
gives a one-dimensional trivial representation space C|c ∼ = C. For example,
the Fock state vector, cyclic for translations, is nondegenerate Rn • |0 = |0.
Ground-state vectors in physics are cyclic; they determine cyclic G-re-
presentations. With a proper subgroup H = G as fixgroup, e.g., U(1)+ ⊂
U(2) in the electroweak standard model (see Chapter 6) or {1} ⊂ U(1)
in superconductivity and the chiral model of Nambu and Jona-Lasinio (see
Chapter 9), the ground-state is degenerate; all vectors of the ground-state
orbit, called degeneracy manifold, |c ∈ G • |c ∼
= G/H, are possible ground-
state vectors.
The wave function conditions in quantum mechanics concerning their be-
havior for infinity and near by singular points are more generally considered
in the framework of Lebesgue function spaces. All cyclic Hilbert spaces can be
constructed from the Lebesgue function algebra L1 (G) with its dual L∞ (G):
A positive-type function [21, 24] or metric- (scalar product) inducing func-
tion is an essentially bounded function d ∈ L∞ (G), which defines a positive
product of the convolution algebra L1 (G):

d ∈ L∞ (G)+ ⇐⇒ f |f d = dg1 dg2 f (g1 )d(g1−1 g2 )f (g2 ) ≥ 0
for all f ∈ L1 (G).
A positive-type function is locally almost everywhere a continuous bounded
function,
dg
L∞ (G)+ = Cb (G)+ .
Connected to each positive-type function d ∈ L∞ (G)+ is a cyclic Hilbert
space: The induced scalar product is the value of the convolution L1 (G) ∗
L∞ (G) ∗ L1 (G) ⊆ L∞ (G) at the neutral group element 1 ∈ G:
L1 (G) × L1 (G) −→ C, f |f d = fˆ ∗ f d = fˆ ∗ d ∗ f (1),

with fˆ(g) = f (g −1 ).
8.3 Schur Product of Group Functions 211
With this product, the algebra functions define a pre-Hilbert space and, by
canonical Cauchy completion of the nontrivial norm classes |f d , a Hilbert
space |L1 (G)d . There exists a cyclic vector |c ∈ |L1 (G)d whose positive-
type function is the expectation value d(g) = c|g • |c.
All diagonal matrix elements of a Hilbert representation D : G −→ U(V )
define continuous positive-type functions,
0 = |v ∈ V, G g −→ dv (g) = v|D(g)|v, dv ∈ Cb (G)+ ,
e.g., R x −→ diq (x) = eiqx ∈ U(1). Compact group examples are the
diagonal elements in the SU(2)- and SO(3)-matrices:
ϕ+χ ϕ−χ

ei 2 cos θ2 iei 2 sin θ2
u= −i
ϕ−χ
θ −i
ϕ+χ
θ
∈ SU(2),
ie 2 sin 2 e 2 cos 2

ei(ϕ+χ) cos2 θ
2 ieiϕ sin
√ θ
2
−ei(ϕ−χ) sin2 θ
2
u∨u= ieiχ sin
√ θ
2
cos θ ie−iχ sin
√ θ
2 ∈ SO(3).
−e−i(ϕ−χ) sin2 θ
2 ie−iϕ sin
√θ
2
e−i(ϕ+χ) cos2 θ
2
The representation normalization of a positive-type functions at the neu-

tral element, then called a state, is related to a quantum theoretical proba-
bility normalization,
G 1 −→ dv (1) = v|D(1)|v = v|v.
According to Gel’fand and Raikov [27], there is a surjection from the
positive-type functions (representation metrics) to the equivalence classes of
cyclic Hilbert representations:
L∞ (G)+ −→ cycrep + G ⊇ Ǧ.
In general, an essentially bounded function d ∈ L∞ (G) defines a bilinear
form on the Radon measures and its subspaces, M(G) ∗ L∞ (G) ∗ M(G) =
L∞ (G):

M(G) × M(G) −→ C, ω ∗ d ∗ ω (1) = dg1 dg2 ω(g1−1 )d(g1−1 g2 )ω (g2 ).
The essentially bounded functions L∞ (G) are ordered. The cone L∞ (G)+
with the positive-type functions is convex. A positive-type function is
reflection-symmetric, i.e., unitary, and bounded by the neutral element
value:
d, d ∈ L∞ (G)+ , α, α ≥ 0 ⇒ αd + α d ∈ L∞ (G)+ ,
d = dˆ ⇐⇒ d = d− , with d− (g) = d(g −1 ), d(g) = d(g),
|d(g)| ≤ d(1).
The unit function G g −→ d1 (g) = 1 characterizes the trivial group re-

presentation. Functions of the inverse, i.e., conjugate functions, are for dual
(inverse-transposed) representations:
duality: d ↔ d− = d, self-dual: d = d− = d.
The pointwise product of two positive-type functions gives a positive-type

function for the product representation (see Chapter 9). The extremal con-
tinuous states, i.e., there exist only trivial cone combinations with states,
d = αd + (1 − α)d , α ≥ 0, with α ∈ {0, 1}, are called pure states. They
give the equivalence classes of the irreducible Hilbert representations. For an
irreducible Hilbert representation, every nontrivial vector |v = 0 is cyclic.
Positive-type functions are the “continuous” extension of positive sesquilin-
ear forms (scalar products) of finite-dimensional vector spaces with v|w =
ζ vw = ζ wv = v i ζij wj . Finite-dimensional scalar products can be unitarily
diagonalized ζ = ξ ∗ ◦ diag ζ ◦ ξ with ξ ∈ U(n).
The familiar Hilbert spaces with square-integrable functions are included
as follows: An absolute square of an L2 -function is a positive-type function:
L2 (G) ∗ L2 (G) ⊆ L∞ (G), L2 (G) ξ −→ d = ξ̂ ∗ ξ ∈ L∞ (G)+ .
If, and only if, a positive-type function is the absolute square of a square-
integrable one, the Hilbert space can be constructed with square-integrable
group functions:
f |f d = fˆ ∗ ξˆ ∗ ξ ∗ f (1) = ξ ∗ f |ξ ∗ f δ with ξ ∗ L1 (G) ⊆ L2 (G).
In general, the mapping above L2 (G) −→ L∞ (G)+ is not surjective; i.e.,

not all cyclic representations d can be characterized by a square-integrable
G-function ξd .
In analogy to positive-type functions, a positive-type Radon distribution
defines a positive product of functions from the convolution algebra with
compact support and the Hilbert space |Cc (G)ω with function classes, e.g.,
for compact groups G,
ω ∈ M(G)+ : Cc (G) × Cc (G) −→ C,

f |f ω = fˆ ∗ f ω = fˆ ∗ ω ∗ f (1) = dg1 dg2 f (g1 )ω(g1−1 g2 )f (g2 ).
Positive-type functions yield only cyclic representations; positive-type mea-

sures yield more general representations. For example, the Dirac distribution
at the unit δ leads to L2 (G) = |Cc (G)δ .
8.3.3 Induced Positive-Type Measures

The embedding of a positive-type Radon distribution of a closed subgroup
of a locally compact group H ⊆ G defines a positive-type Radon G-distribu-
tion [24]:
M(H)+ ωH −→ ωG ∈ M(G) +,
with ωG , f = ωH , f |H = H ωH (h)dhf (h) for f ∈ Cc (G).
For
non unimodular groups, the embedded measure has to be multiplied
ΔG (h)
by ΔH (h) with the modular functions. If the positive-type distribution
8.4 Harmonic Analysis of Representations 213
ωH characterizes the class [dH ] of the regular Hilbert representation of the

subgroup, then ωG characterizes the class [indG H
H d ] of the induced Hilbert
representation of G. The induced scalar product comes from

f, f ∈ Cc (G):f |f ωG =fˆ ∗ f |H ωH = G dg H dh f (g)ωH (h)f (gh)

= G/H dgH H×H dh1 dh2 f (gh1 ) ωH (h−1
1 h2 ) f (gh2 ).
Examples are given by the functions on a (semi)direct product group:

H : f |f ωG =
G=K × K
dk H×H dh1 dh2 f (k, h1 ) ωH (h−1
1 h2 ) f (k, h2 ),
in the simplest case for the abelian product group Rk+s and their Fourier
transforms, e.g., for time and position translations with (X, x) → (x0 , x) and
(Q, q) = (q0 , q):

G=Rk ⊕ Rs : f |f ωk+s = dk X ds x1 ds x2 f (X, x1 ) ωs (x2 − x1 ) f (X, x2 )
dk Q ds q
= (2π)t (2π)s f˜(Q, q) ω̃s (q) f˜ (Q, q).
8.4 Harmonic Analysis of Representations

9
A set with group action is the disjoint union of its orbits, ι G/Hι , each
characterized by a fixgroup G • xι ∼ = G/Hι . The linear extension: A Hilbert
8
representation space is a direct sum ι C(G/Hι ) of cyclic ones. An irreducible
representation is cyclic, but the converse is not rtrue:
rep G ⊃ cycrep G ⊃ irrep G.
Any cyclic representation and, a fortiori, any irreducible Hilbert representa-

tion of G distinguish a subgroup H ⊆ G.
A cyclic representation of a locally compact group is a direct integral ⊕ Ǧ
dǧ of irreducible ones. Corresponding to the Haar measure of a locally com-
pact group, unique up to a constant factor, there is the positive Plancherel
measure ⊕ Ǧ dǧ = ⊕ dǧ of the group dual Ǧ = irrep + G with the equivalence
classes of the irreducible Hilbert representations, defined up to normaliza-
tion and unique for a given Haar measure. For a unimodular group, the
renormalization are inverse to each other: M(G)+ × M(Ǧ)+ (dg, dǧ) →
(eλ dg, e−λ dǧ). The Plancherel measure is a measure of the group invariants,
which, for a Lie group, are generated by multilinear Lie algebra forms. Topolo-
gies of Ǧ are not discussed here.
Familiar examples are the translation groups Rn and Haar measure dn x
with the energy-momenta Řn as dual groups and associated Plancherel mea-
q
sure dn 2π with the Fourier transformation from translation functions L1 (Rn )
to energy-momentum functions C0 (Řn ), vanishing at infinity (more ahead):

f˜(q) = dn x e−iqx f (x) and f (x) = dn 2π q
eiqx f˜(q).
In general, the unitary Fourier transformation or harmonic analysis

associates functions of the group dual, called harmonic components, to
appropriate group function spaces F (G):

F : F (G) f −→ Ff = f˜ ∈ ⊕ dǧ Vǧ ⊗ Vǧ .
The unitary irreducible G-representation, G g −→ Dǧ (g) ∈ U(Vǧ ), Dǧ ∼ =

ǧ ∈ Ǧ, acts on the Hilbert space Vǧ . The Fourier transform f˜(ǧ) gives the
coefficients of this representation:

Ǧ ǧ −→ f˜(ǧ) = dg Dǧ (g −1 )f (g) = ǧ|f = Dǧ ∗ f (1).
The function f can be decomposed with its Fourier integral. Its normalization
is the integral with a trace over the irreducible components:

G g −→ f (g) = dǧ tr Dǧ (g)f˜(ǧ), f (1) = dǧ tr f˜(ǧ).
The use of representation equivalence classes makes all this rather trouble-
some for nonabelian groups.
The Fourier transform is a conjugation-compatible algebra morphism. It
is injective, i.e., invertible on the image F(F (G)):
F(f + f ) = Ff + Ff , F(αf ) = αFf,

F(f ∗ f ) = Ff · Ff, F(f ) = Ff .
For compact groups, one has as Fourier transformable functions F (G) =

L1 (G), for noncompact groups (second countable, unimodular, type I), a
restriction to F (G) = L1 (G) ∩ L2 (G).
The Plancherel measure of a group dual is defined by and allows an
orthogonal direct integral decomposition of the both-sided regular G × G-
representation, and the left- and right-regular G-representation,

L2 (G) ∼
= ⊕
dǧ Vǧ ⊗ Vǧ , idL2 (G) ∼
= ⊕
dǧ Dǧ ⊗ Dǧ ,
⊕
⊕

f = dg gf (g) = f= dǧ Dǧ ⊗ f (ǧ) (where
˜ ˜
defined),
= ⊕ dǧ Dǧ ⊗ 1, L ∼
right-, left-regular: R ∼ = ⊕ dǧ 1 ⊗ Dǧ .
The Parseval formula for the scalar product and the Fourier inversion are

f1,2 ∈ F(G) : dg f1 (g)f2 (g) = dǧ tr f˜1 (ǧ)f˜2 (ǧ) = f1 |f2 ,
f (g) = dǧ tr Dǧ (g)f˜(ǧ).
For noncompact groups, the inversion holds only for a subspace of functions.
A direct decomposition of a Hilbert representation D displays the nor-
malizations (cardinal multiplicities for compact groups) of the irreducible
components Dǧ by a positive spectral distribution ρD of Ǧ:

D = ⊕ dǧ Dǧ ρD (ǧ).
8.5 Schur Orthogonality for Compact Groups 215
A positive-type function d for a cyclic representation is an integral of

positive-type functions dǧ for its irreducible components, i.e., a matrix el-
ement for a cyclic vector dǧ (g) = c|Dǧ (g)|c, with a positive Plancherel
distribution d˜ of the group dual,

˜
d(g) = dǧ dǧ (g) d(ǧ).
For a compact group with normalized Haar measure, the Plancherel mea-

sure is the counting measure dι with the dimensions dι = dimC Vι of the
ι∈Ǧ
irreducible representation spaces, i.e., the number of columns or lines in the
full matrix algebra Cdι ⊗Cdι . The Plancherel measure for a noncompact group
has continuous support. For noncompact nonabelian groups, determining the
Plancherel measure is difficult. The support of the Plancherel measure is
the reduced dual; it must not be the full group dual; if not, the group is
nonamenable. There may be irreducible Hilbert representations, not in the
support of the Plancherel measure.
The Schur product for cyclic representations integrates their positive-type
functions over the group:

d1,2 ∈ L∞ (G)+ : {d2 |d1 } = dg d2 (g)d1 (g) = dǧ tr d˜2 (ǧ)d˜1 (ǧ)
= dˆ2 ∗ d1 (1) ∈ M(Ǧ).
It is a Plancherel distribution, i.e., a distribution of the group dual with the
Plancherel measure. The inverse Plancherel measure is the Plancherel density.
8.5 Schur Orthogonality for Compact Groups

The Lebesgue function spaces for a compact group K are all subspaces of the
convolution algebra with the absolute integrable function classes,
L∞ (K) ⊆ Lp (K) ⊆ Lq (K) ⊆ L1 (K) for ∞ ≥ p ≥ q ≥ 1.
All representations can be formulated as acting on square-integrable func-
tions L2 (K). For example, the central element in the SO(3)-matrix above
and the corresponding elements for the irreducible (1 + 2L)-dimensional re-
presentations are metric-inducing functions, the Legendre polynomials PL ∈
L1 (SO(3))+ :
L; 0|O(χ, ϕ, θ)|L; 0 = 1 ∂

sin2L θ ⎧
2L L! (− ∂ cos θ )
L
= PL (cos θ)
⎪ 1,
⎪
⎪
⎨ 1 − sin2 θ = cos θ,
$L
2
(L+k)! (− sin θ2 )
k
2
= =
(L−k)! (k!)2 ⎪
⎪ 1 − 6 sin2 θ 4 θ
2 + 6 sin 2
k=0 ⎪
⎩
= 32 (cos2 θ − 13 ), . . . .
They define the scalar product for the Hilbert spaces |L1 (SO(3))PL ∼
=
C1+2L .
There is the Peter–Weyl decomposition [47] of the algebra with the

square-integrable functions into the countably many finite-dimensional full
matrix algebras as irreducible group algebra representations with the columns
(or lines) the irreducible Hilbert spaces:
∞

dense
compact K: L2 (K) = Cdι ⊗ Cdι , ξι |ξι = 1
dι διι .

ι=1
It is a special case of Frobenius’ reciprocity (see Chapter 7) for the K-repre-

sentation induced by the trivial representation of the trivial subgroup,
∞

dense
L2 (K) = L2 (K/{1}, C) ∼
= dι Cdι .
ι=1
The algebras are Schur-orthogonal to each other with respect to the Schur
scalar product, defined with normalized group measure,
Dι : K −→ U(Vι ), Vι ∼ = Cdι ,
{Dι |Dι } = Dι Dι (1) = K dk Dι (k) ⊗ Dι (k) = διι d1ι 1d2ι .
ˆ
More explicit: Two representation matrix elements of inequivalent irreducible

Hilbert representations are orthogonal, with vι , wι ∈ Vι :
∼ Dι ,
0, Dι =
K dk vι |Dι (k)|wι vι |Dι (k)|wι =
1
v
dι ι ι|v wι |wι , D ι = Dι .
The Plancherel measure of an irreducible representation Dι of a compact

group with normalized Haar measure is a counting measure, given by the
number dι of equivalent irreducible representations Cdι in one algebra Cdι ⊗
Cdι , which coincides with its dimension, ΠK (Dι ) = dι . The Schur orthogo-
nality involves the Plancherel density (inverse Plancherel measure).
The Schur orthogonality is conceptually different from a possible or-
thonormality
⎪ of the vectors in the representation space, e.g., for a Hilbert
⎪
basis {ej ⎪
⎪j = 1, . . . , ι},

ei |ej = δ ij ⇒ K dk Dι (k)ji Dι (k)ji = διι d1ι δii δ jj .
The Schur orthogonality is exemplified for SU(2) ∼ SO(3) with Plancherel
density dJ = 1 + 2J by the positive-type square-integrable functions arising
as diagonal matrix elements of the 2- and three-dimensional representation:
: : 2π : :
dχ 2π dϕ 1 d cos θ
d3 k = = 1,
−2π 4π 0 2π −1 2
⎧ :
⎪
⎪
χ+ϕ θ
d3 k |ei 2 cos |2 = 12 ,
⎪
⎪
⎪
⎨ : 2
θ
d3 k |ei(χ+ϕ) cos2 |2 = 13 ,
⎪
⎪ : 2
⎪
⎪ θ χ+ϕ θ
⎪
⎩ d3 k e−i(χ+ϕ) cos2 ei 2 cos = 0.
2 2
8.6 Translation Representations 217
Positive-type functions, normalized as states (representation normalization),

do not have to be Schur-normalized. A familiar example is given by the
Legendre polynomials PL , which are normalized positive-type functions for
irreducible representations [L] of the rotations SO(3). They are the (m, m ) =
(0, 0)-components in the rotation group representing matrices, PL = [L]00 .
The Legendre polynomials are Schur-orthogonal for different angular mo-
menta L and have as Schur norm the inverse of the 1 + 2L dimension of the
representation space:
|L}(θ) = PL (cos θ), PL (1) = 1,

1
{L |L} = −1 d cos
2
θ 1
PL (cos θ)PL (cos θ) = δ LL 1+2L .
The transition to the spherical harmonics Ym L

= |L; m involves the multipli-
cation with the inverse square root of its Schur norm:

|L; 0 = Y0L = 1+2L L
|Ω2 | P ,
L (ϕ, θ)YL (ϕ, θ) = δ LL δ
L ; m |L; m = d2 ω Ym m mm .
⎪ the 2-sphere area |Ω | = 4π, equal for

2
With the additional normalization with
L⎪⎪
all L, the spherical harmonics {Ym ⎪m = −L, . . . , L} are an orthonormal ba-
sis for the irreducible SO(3)-representation space with angular momentum L.
8.6 Translation Representations

In the following, the structures for locally compact groups, as given above,
are specialized to the noncompact abelian translation groups Rn , e.g., to a
Cartan plane A in a semisimple Lie group G = K ◦ A ◦ N or to translations
in an affine group G × Rn . The (energy-)momenta constitute the group dual
with the equivalence classes of the irreducible Hilbert representations x −→
eiqx = x|q ∈ U(1). The translation group has the distinction of being
isomorphic to its group dual Řn , an abelian group,
x ∈ Rn ∼
= Řn = (iR)n iq.
All faithful Hilbert representations of the translations R are infinite-

dimensional.
8.6.1 Fourier Transformation

The harmonic components of a translation distribution, its Fourier compo-
nents, are given by the Fourier transform:
n n
q|μ = μ̃(q) = dNx e−iqx μ(x) = dNx q|xx|μ,
N dn q dn x
(2π)n |μ̃(q)| = N |μ(x)| , where defined.
2 2
n
If the Haar measure of the translations Rn is normalized as dNx , the
dn q
corresponding Plancherel measure of the (energy-)momenta Řn is N (2π)n with
n
a free normalization factor N > 0, e.g., N = 1 or N = (2π)n or N = (2π) 2 .
The Fourier transforms of the three relevant spaces L1 (Rn ), M(Rn ), and
∞
L (Rn ) are as follows: According to a theorem of Lebesgue, the Fourier trans-
formation of the convolution algebra L1 (Rn ) is an injective algebra morphism,
with a dense range, but not surjective, into the continuous functions C0 (Řn ),
vanishing at infinity. The Fourier transformation can be extended to the
Radon measure algebra M(Rn ) with values in the bounded continuous func-
tions Cb (Řn ). Positive Radon measures with spectral energy-momentum dis-
tributions and the continuous positive-type functions are bijective (Bochner’s
theorem [8]):
⎧
⎪ L
⎪
⎪
p (Rn ) ⊆ Lr (Řn ), 1 + 1 = 1, 1 ≤ p ≤ 2, ∞ ≥ r ≥ 2,
⎪
⎨
p r
L1 (Rn ) = C˙0 (Řn ) dense in C0 (Řn ) (Lebesgue),
Fourier:
⎪
⎪
M(R
dn q ∞
+ = Cb (Ř )+ = L (Ř )+ (Bochner),
n) n n
⎪
⎪
⎩ n
M(R ) = C˙b (Řn ) = complex span of Cb (Řn )+ .
The convolution and pointwise product are Fourier-compatible (with con-
venient measure normalization N = 1),
dn q dn q iqx
μ(x) = (2π) n μ̃(q)e
iqx
, μ̃(q) = dn x μ(x)e−iqx , δ(x) = (2π) ne ,
dn q
μ1 · μ2 (x) = (2π)n μ̃1 ∗ μ̃2 (q)eiqx ,
˜ = μ, μ
μ̃ 1 · μ2 = μ̃1 ∗ μ̃2 , dn p
μ̃1 ∗ μ̃2 (q) = (2π) n μ̃1 (p)μ̃2 (q − p),
for the representation relevant spaces are exchanged in the spaces with the
harmonic components:
∗ M(Rn ) L1 (Rn ) C˙b (Rn ) · C˙b (Rn ) L1 (Rn ) M(Rn )

M(R ) n
M(R ) n 1 n
C˙b (Rn ) C˙b (Rn ) C˙b (Rn ) L1 (Rn ) M(Rn )
L (R )
G=R n
L1 (Rn ) L1 (Rn ) L1 (Rn ) C˙b (Rn ) L1 (Rn ) L1 (Rn ) − −
C˙b (Rn ) C˙b (Rn ) C˙b (Rn ) − M(Rn ) M(Rn ) − −
μ1 ∗ μ2 (x) F μ1 · μ2 (x)
$ Fourier $
· C˙b (Řn ) Ċ0 (Řn ) M(Řn ) ∗ M(Řn ) Ċ0 (Řn ) C˙b (Řn )
C˙b (Řn ) C˙b (Řn ) Ċ0 (Ř )
n
M(Ř ) n
M(Řn ) M(Řn ) C˙0 (Řn ) Ċb (Řn )
Ǧ = Řn
Ċ0 (Řn ) Ċ0 (Řn ) Ċ0 (Řn ) M(Řn ) C˙0 (Řn ) C˙0 (Řn ) − −
M(Řn ) M(Řn ) M(Řn ) − Ċb (Řn ) Ċb (Řn ) − −
μ̃1 · μ̃2 (q) μ̃1 ∗ μ̃2 (q)

from group product for product representations
The Fourier transformation defines algebra isomorphisms:
F n F
M(Rn ) ∼
= C˙b (Řn ) = L∞ (Řn ), L1 (Rn ) ∼
d q
= C˙0 (Řn ).
Dual and sesquilinear products use the convolution product at the trivial
translation or at the trivial (energy-)momentum as integrals over all transla-
tions or over all (energy-)momenta. For a sesquilinear form, the conjugated
(generalized) function is used:

μ1 , μ2 = dn x μ1 (−x)μ2 (x) = μ1 ∗ μ2 (0)
dn q
= (2π) n μ̃1 (q)μ̃2 (q) = μ̃− ∗
1 (2π)n μ̃2 (0),
dn q
μ1 |μ2 = μ1 , μ2 , with μ̂(x) = μ− (x) = μ(−x) = (2π) n μ̃(q)e
iqx
.
8.6.2 Cyclic Translation Representations

The continuous translation positive-type functions, isomorphic to the positive
(energy-)momentum Radon measures, characterize all cyclic translation re-
presentations in the bijection
Cb (Rn )+ ∼
= M(Řn )+ ↔ cycrep + Rn .
The Hilbert product induced by a function positive-type function,

dn q
Cb (Rn )+ d ↔ d˜ ∈ M(Řn )+ , d(x) = (2π) ˜
n d(q) e
iqx
,
can be transformed into an integration of the pointwise product of the har-

monic components with a representation characteristic positive (energy-)mo-
˜ dn q
mentum measure d(q) (2π)n , which involves a positive Radon distribution as
spectral measure,
F
L1 (Rn ) ∗ L1 (Rn ) −→ C ↔ C˙0 (Řn ) ·˜ C˙0 (Řn ) −→ C,
d
n d
f |f d = d x1 dn x2 f (x1 )d(x2 − x1 )f (x2 ) = fˆ ∗ d ∗ f (0)
dn q
˜ ˜ ˜ ˜ ˜ ˜ n
= (2π)n f (q) d(q) f (q) = f · d · f (Ř ).
˜ f˜ of the (energy-)momenta.

A finite product gives a finite Radon measure f˜·d·
The product can be extended to the Radon distributions and their
Fourier-transformed bounded (energy-)momentum functions
F
M(Rn ) ∗ M(Rn ) −→ C ↔ C˙b (Řn ) ·˜ C˙b (Řn ) −→ C,
d
dn q
d
ω|ω d = (2π)n ω̃(q) ˜
d(q) ω̃ (q).
The transition to a Hilbert space with square-integrable Rn -functions

requires the positive-type function to be an absolute square of a square-
integrable function:
˜ 2 with L2 (Rn ) ξ ↔ ξ˜ ∈ L2 (Řn ),
d = ξ̂ ∗ ξ ⇐⇒ d˜ = |ξ|
1 n 2 n ˜
ξ ∗ L (R ) ⊆ L (R ) ↔ ξ · C0 (Řn ) ⊆ L2 (Řn ).
˙
The Hilbert space H = |L1 (Rn )d with the cyclic translation representation
has a distributive basis, labeled by the eigenvalues (energy-momenta),
⎧ n
⊕ d q
⎪
⎪
⎪ (2π)n |qd q|d = idH ,
⎪
⎪ ˜ ⊕ dn q ˜
⎪
⎪ H |f d = |d; f = (2π) n f (q)|qd ,
⎪ ⎨
⎪ ˜
with f (q) = q|f d ,
{|qd = |d; q⎪
⎪q ∈ Rn } :
⎪
⎪ f |f d = d; f˜ |d; f˜,
⎪
⎪
⎪
⎪ q |qd = d; q |d; q = d(q)(2π)
˜ n
δ(q − q ),
⎪
⎩
D(x)|qd = e |qd .
iqx
If normalizable, the integration of a distributive basis over the eigenvalues

gives a cyclic vector:
dn q dn q dn q iqx
|1d = |d; 1 = ⊕ (2π) n |qd : d(x) = 1|D(x)|1d = (2π) ne q |qd
dn q (2π) iqx
n
˜
= (2π)n d(q) e .
The Schur product of two translation representations, characterized by

positive-type functions, integrates over the translations
dn q
{d2 |d1 } = dn x d2 (x)d1 (x) = (2π) ˜ ˜
n d1 (q)d2 (q)
= dˆ2 ∗ d1 (0) = d˜− ∗ ˜

1 (2π)n d2 (0).
It is a Plancherel distribution of the (energy-)momenta.
8.6.3 Spherical and Hyperbolic Positive-Type Functions

The positive Radon measures of the dual group iq ∈ Ř with the momenta (or
energies) are characterized by their support. The extremal states for the irre-
ducible translation representations in U(1) acting on one-dimensional Hilbert
spaces are the Dirac measures δP ∈ M(Ř)+ supported by the real momentum
P for the invariant imaginary eigenvalue iP ∈ iR:
. dq 1 iqx
R x −→ diP (x) = eiP x = dq δ(q − P )eiqx = 2iπ q−P e .
The Dirac distribution restricts the induced scalar product to one component:

f, f ∈ L1 (R) : f |f d = dx1 dx2 f (x1 )eiP (x2 −x1 ) f (x2 )

˜(q) δ(q − P ) f˜ (q) = f˜(P ) f˜ (P ),
= dq f⎧
⎨ P |P = 1,
|L1 (R)diP ∼= C, with basis {|P } D(x)|P = eiP x |P ,
⎩
cyclic vector: |P ; 1 = |P .
The Schur product of the irreducible Hilbert representations gives the

Plancherel density (inverse Plancherel measure):

−P
Schur product: {P |P } = dx e−iP x eiP x = δ( P 2π ).
The self-dual representations on two-dimensional Hilbert spaces are cyclic

with the reflection E ↔ −E invariant positive Radon distributions, supported
by two reflected eigenvalues ±iE ∈ iR and positive invariant E 2 , as used for
time translations in the harmonic oscillator,
2
R t −→ dE (t) = dq |q|δ(q 2 − E 2 )eiqt = cos Et,
|q|δ(q 2 − E 2 ) = δ(q−E)+δ(q+E) ,
⎧2
⎪ ±E| ± E = 1,
⎪
⎪
⎪
⎨ E| − E = 0,
|L1 (R)dE2 ∼= C2 , D(t)| ± E = e±iEt | ± E,
with basis {|E, | − E} ⎪ ⎪
⎪
⎪ cyclic vector: |E 2 ; 1 = |E ⊕
√
|−E
,
⎩ 2
E ; 1|E ; 1 = 1,
2 2

Schur product: {E 2 |E 2 } = dt cos E t cos Et = 12 [δ( E−E
2π ) + δ( E+E
2π )]
|E|−|E |
= δ( π ).
Irreducible faithful translation representations x −→ eQx with real eigen-

value Q ∈ R are not unitary. From the residual form of the positive-type
functions for the circle SO(2) and its finite two-dimensional representations,
2 .
R x −→ dP (x) = cos P x = dq q
iπ q2 −P 2 e
iqx ,
one obtains by the transition P → iQ the positive-type functions for the

hyperbola SO0 (1, 1) ∼
= R and its cyclic Hilbert representations with negative
invariant −Q2 :
dq |Q|
R x −→ d−Q (x) = = e−|Qx| ,
2
iqx
πq2 +Q2 e
1 |Q|
2iπ q−i|Q| − q+i|Q| .
1 1 1
π q2 +Q2 =
They characterize a scalar product for an infinite-dimensional Hilbert space:

f, f ∈ L1 (R) : f |f d = dx1 dx2 f (x1 )e−|Q(x2 −x1 )| f (x2 )
|Q|
= dq ˜ ˜
π f (q) q2 +Q2 f (q), ⎪
⎪
|L (R)d−Q2 ∼
1 = L (R) with distributive basis {| − Q2 ; q⎪
2
⎪q ∈ R},
⎧
⎪
⎪ −Q2 ; q | − Q2 ; q = q2 +Q2 δ( q−q
2|Q|
2π ),
⎪
⎪
⎨ D(x)| − Q ; q = e | − Q ; q,
2 iqx 2
dq
⎪
⎪ cyclic vector: | − Q2 ; 1 = ⊕ 2π | − Q2 ; q,
⎪
⎪
⎩
−Q2 ; 1| − Q2 ; 1 = 1.
In contrast to the spherical positive-type functions, the hyperbolic ones for

the noncompact group are not orthogonal for different invariants:

{−Q2 | − Q2 } = dx e−|Q |x e−|Q|x = |Q|+|Q
2
| ,
Schur product:
{−Q | − Q } = |Q| .
2 2 1
The singularities of the function in the Radon measure (imaginary

momenta ±iQ) are not eigenvalues for unitary translation representa-
tions. The positive-type function is an absolute square of L2 -functions
d˜−Q = |ξ|
2
˜ 2 ∈ L2 (Ř).
8.6.4 Breit–Wigner Functions

Time translations together with their reflection t −→ −t constitute a non-
abelian semidirect group:

R (, t) −→ 0 1t ∈ GL(2, R),
I(2) ×
∈ I(2) = {1, −1}, (1 , t1 ) ◦ (2 , t2 ) = (1 2 , t1 + 1 t2 ).
The reflection group I(2) is discrete. Therefore, and with Wigner, time
reflection can be represented in quantum theories by an antilinear transfor-
mation. For the complex energies q + iΓ ∈ C as eigenvalues for all irreducible
time translation representations (not only unitary), the reflection group is
implemented by the identity and the complex conjugation:

I(2) × C −→ C,
I(2) = {1, −1} −→ { idC , ∗} :
−1 • (q + iΓ) = (q + iΓ)∗ = q − iΓ.
A complex energy has as reflection orbit {E + iΓ, E − iΓ} with fixgroup I(2)
for real energies Γ = 0 and trivial fixgroup {1} ⊂ I(2) for nontrivial width
Γ = 0.
The characteristic functions for the future and past have an energy pole
in the upper and lower complex energy plane, respectively,

dq 1 iqt 1, ±t > 0,
ϑ(±t) = ± 2iπ e =
q∓io 0, ±t < 0,
ϑ(t) + ϑ(−t) = 1, ϑ(t) − ϑ(−t) = (t) = |t|t
.
Summing the time representations on a nontrivial I(2)-orbit gives the
Breit–Wigner states, which are used for unstable particles. They are both
the sum and the convolution product of its future and past contributions:
dq 1 dq
Γ > 0 : R t −→ 2iπ q−E−iΓ − q−E+iΓ e
1 iqt
= Γ
π (q−E)2 +Γ2 e
iqt
= ϑ(t)ei(E+iΓ)t + ϑ(−t)ei(E−iΓ)t
√ √
= [ϑ(t) 2Γei(E+iΓ)t ] ∗ [ϑ(−t) 2Γei(E−iΓ)t ] = eiEt−Γ|t| .
For a nontrivial width, the energy-distribution supporting pair E±iΓ with the
representation characterizing two reals (E, Γ) are eigenvalues of irreducible
time translations, however not of unitary ones.
The Breit–Wigner states are Fourier-transformed positive energy densi-
ties. The trivial width limit gives the Dirac distribution for stable states,

Γ ≥ 0 : |E, Γ}(t) = eiEt−Γ|t| = dq δΓ (q − E)eiqt ,
δΓ (q − E) = π1 (q−E)Γ2 +Γ2 = δ(q − E) ∗ q22Γ
+Γ2 ,
δ0 (q − E) = δ(q − E).
Breit–Wigner states are products of spherical and hyperbolic positive-type

functions. In the language of residues in the complex energy plane, time trans-
lation representations without reflection are obtained in the distributive limit
of vanishing width 0 < Γ = o → 0, i.e., in the Breit–Wigner approximation
of the Dirac distribution,
. dq 1 iqt
Γ = 0 : R t −→ 2iπ e = dq δ(q − E)eiqt = eiEt ,
q−E
q−E−io − q−E+io = δ(q − E).
1 1 1
2iπ
The Breit–Wigner energy densities characterize the product of infinite-

dimensional Hilbert spaces:
⎧ ⎪
⎪
⎪ distributive Hilbert space basis: {|E, Γ; q⎪ ⎪q ∈ Ř},
⎪
⎨translation action: D(t)|E, Γ; q = eiqt |E, Γ; q,
R t −→ eiEt−Γ|t| : time reflection: α|E, Γ; q ↔ E, Γ; q|α, α ∈ C,
⎪
⎪
⎩ q−q
Hilbert product: E, Γ; q |E, Γ; q = (q−E)
2Γ
2 +Γ2 δ( 2π ).
The time reflection −1 I(2) acts antilinearly with the exchange of bra . . . |
for in-going or past and ket | . . . for out-going or future.
The proper Hilbert vectors use energy functions, square-integrable with
the Breit–Wigner functions,

|E, Γ; f˜ = dq ˜
2π f (q)|E, Γ; q, E, Γ; f˜ |E, Γ; f˜ = dq
2π f˜ (q) 2Γ
(q−E)2 +Γ2 f˜(q).
The sum over the distributive basis, i.e., with f˜(q) = 1, gives a cyclic vector:
dq
|E, Γ; 1 = 2π |E, Γ; q, E, Γ; 1|t • |E, Γ; 1 = eiEt−Γ|t| .
Two Breit–Wigner states are not Schur-orthogonal for a nontrivial width

sum Γ + Γ > 0. The Schur norm is the lifetime,
⎧
⎪ i(E−E )t−(Γ+Γ )|t|
⎨{E , Γ |E, Γ} = dt e

Schur product: = (E−E2(Γ+Γ )
)2 +(Γ+Γ )2 = 2πδΓ+Γ (E − E ),
⎪
⎩
{E, Γ|E, Γ} = Γ1 .
8.6.5 Gaussian Functions

With the Fourier transformation–effected isomorphy of the square-integrable
functions of translations and its dual momentum group,
L2 (R) f ↔ f˜ ∈ L2 (Ř),
there exists a Fourier self-dual representation, where the scalar product–

inducing function d1 ∈ L2 (R) and its position-momentum expectations
coincide, in appropriate normalization, with their Fourier transforms. This

determines the Gaussian positive-type function:
⎫
d1 = d˜1 : ⎪
⎪
⎪
⎪
d1 (x) = √dq2π d1 (q) eiqx , ⎪
⎬ q2
d1 (q) = e− 2 ,
xd1 (x) = √dq2π (−iq)d1 (q) eiqx ⇒
⎪
⎪ or e−πx = dq e−πq e2iπqx .
2 2
⇒ dq
d
d1 (q) = −qd1 (q), ⎪
⎪
⎪
⎭
d1 (0) =1
The Gaussian positive-type function is used for the position representation

coefficients (Schrödinger functions) of the harmonic oscillator,
x2
R x −→ d1 (x) = e− 2 .
Its Schur normalization is

√
dx e−x =
2
{d1 |d1 } = π.
The cyclic Hilbert representations for Fourier self-dual representations of

Euclidean translations Rs ∼
= SO(s) × Rs /SO(s) use products of the basic
Gaussian positive-type function:
2 s
Rs x −→ ds (x) = e− √d
x
2 = d1 (x1 ) · · · d1 (xs ) = q
ds (q) eiqx .
(2π)s
The additional rotation operations SO(s) are Fourier self-dually repre-

sented with the homogeneous position translation polynomials,
r2 ds q 2
s ≥ 2 : Rs x −→ x e− q ) e−
q
2 = s (−i 2 eiqx ,
(2π) 2
2
-
k
eiqx ∈ V ∼
r2 ds q
x −→ P k (x) e− q ) e−
q
2 = s P k (−i 2 = Cs ,
(2π) 2
-
k
with P k (x) = xa1 · · · xak ∈ Rs .
The irreducible SU(s)-representations [k, 0, . . . , 0] (s − 2 zeros) act on

the totally symmetric products P k (x ) of the

defining representation space
[1, 0, . . . , 0] with complex dimension k+s−1
k .
The SU(s)-representations are decomposable into SO(s)-representations
(see Chapter 4 for s = 3). Their representation coefficients contain the har-
monic SO(s)-polynomials, which are generated by vectors of the defining
SO(s)-representation on Rs (next section). By complexification, one has a
representation on a complex vector space Cs where the Cartan subgroup
operations are diagonalizable as familiar from the rotations SO(3) with the
SO(2) eigenvectors (x1 ± ix2 = re±iϕ sin θ, x3 = r cos θ).
8.7 Harmonic Representations of Orthogonal Groups 225
8.7 Harmonic Representations

of Orthogonal Groups
Product representations of the orthogonal group SO0 (t, s) act on the totally
-
L
symmetric tensor products V of a vector space, e.g., with (energy-)momenta
q∈V ∼
= Rn , n = t + s = 2, 3, . . . , and its complexification C ⊗ V ∼
= Cn :
-
L
n+L−1
Γ(n+L)
dimR V = L = Γ(1+L)Γ(n) .
-
L -
L
Vand its complexification C ⊗ V are irreducible, e.g., for the action of
SL(n, R) and SU(n), respectively, and in general, with irreducible represen-
tation [L, 0, . . . , 0] (n − 1 entries), for the Lie algebra An−1 and its real forms.
-
L
In general, V is decomposable for the action of orthogonal groups. The
irreducible subspaces are acted on by the harmonic SO0 (t, s)-representa-
tions, denoted by (SO0 (t, s))L . They have the harmonic polynomials (totally
symmetric, homogeneous, and “traceless”) as bases (q)L ,
⎧
⎪
⎪
0
⎪(q) = 1,
⎪ ⎪
⎪
⎪ ⎪
⎪ ⎪
⎨(q) = q = {qa ⎪a = 1, . . . , n},
1
⎪
⎪
(q) = {(q)a1 ...aL ⎪ak = 1, . . . , n},
L L
(q)2 = (q ∨ q) = {qa qb − ηnab q 2 },
L = 0, 1, . . . , ⎪
⎪
⎪(q)3 = {qa qb qc − ηab qc +ηac qb +ηbc qa q 2 },
⎪
⎪
⎪ n
⎩. . . .
Since “traceless” harmonic polynomials are translation-invariant, i.e., they

have a trivial action of the Laplacian,
∂ 2 = ηab ∂ a ∂ b , ∂ 2 (q)L = {0}.
Because of the quadratic metrical invariant q 2 = η ab qa qb , the decompo-

sition of the totally symmetric vector space powers into orthogonally irre-
ducible subspaces has to distinguish between even and odd powers L — for
the dimensions, independent of the signature:
⎧ $
⎪
⎪ dimR (SO0 (t, s))k , L = 0, 2, . . . ,
n+L−1
⎨k=0,2,...,L
t+s=n: = $
L ⎪
⎪ dimR (SO0 (t, s))k , L = 1, 3, . . . .
⎩
k=1,3,...,L
This gives the dimensions of the harmonic representations,

n+L−3
n−2+L
n−2+2L
dimR (SO0 (t, s))L = n+L−1
L − L−2 = n−2 n−2+L ,
t+s=n 2 3 4 5
(1+L)(2+L)(3+2L)
dimR (SO0 (t, s))L 2 1 + 2L (1 + L)2 6
For example, the rotation group SO(3) leads to irreducible (1 + 2L)-

dimensional representations (SO(3))L = [L], the harmonic Lorentz group
polynomials to irreducible (1 + L)2 -dimensional Minkowski representations
(SO0 (1, 3))L = [ L2 | L2 ], and its compact partner to the irreducible (1 + L)2 -
dimensional Kepler representations (SO(4))L = ( L2 , L2 ).
Each polynomial is a linear combination of powers of the metrical invari-
ant (q 2 )L ∈ I(V ) with harmonic polynomials hL ∈ H(V ):
- -
L
R[q1 , . . . , qn ] = V= V = H(V )I(V ),
L≥0
$k
P (q) = hL (q)(q 2 )L , e.g., 3qa qb = 3(qa qb − ηab 2
n q ) + 3 ηnab q 2 .
L=0
For compact groups, the SO(s)-representations are Hilbert. There, the spher-
ical harmonics are defined by ( √q 2 )L = (ωs−1
)L .
q

The vector space V to be considered may be a Lie algebra x ∈ L ∼ = Rn or
its linear forms q ∈ L , with, respectively, the adjoint and coadjoint action
T
of the corresponding Lie group Ad G ⊆ GL(n, R), Ad G ∼ = G/ centr G. For

a semisimple Lie algebra, the adjoint group is a subgroup of the Killing form
invariance group SO0 (t, s). The harmonic polynomial SO0 (t, s)-representa-
tions are decomposable into Ad G-representations.
The harmonic O(t, s)-units (projectors), related to the corresponding har-
monic polynomials, arise by their harmonic derivation, i.e., derivations and
polynomials are dual to each other,
⎧
⎪ (1n )0 = 1,
⎪
⎨(1 )1 = 1 ∼ a
( ) ⊗(q)
∂ L L
n n = δd ,
(1n )L = ∂qΓ(1+L) : ∼ d e e δd − ηab ηde ,
a b a b
⎪
⎪ (1 )2
=
δ δ +δ
⎩ n 2 n
....
8.8 Hilbert Metrics for Flat Manifolds

The representations of flat manifolds with translations describe interaction-
free structures, asymptotic scattering for Euclidean position, and free parti-
cles for Minkowski spacetime.
Cyclic Hilbert representations of translations arise in irreducible Hilbert
representations of affine groups SO0 (t, s) × Rn . The induced representations
collect, via direct integrals over the orbits of a compact translation fix-
group in the homogeneous group, e.g., over spheres Ωs ∼ = SO(1 + s)/SO(s)
or hyperboloids Y s ∼ = SO0 (1, s)/SO(s), the translation representations
with Dirac distributions of the (energy-)momentum invariants. This will be
given for Euclidean groups SO(1 + s) × R1+s , isomorphic to Galilei groups
as contractions of Lorentz groups SO0 (1, 1 + s), and for Poincaré groups
SO0 (1, s) ×
R1+s .
8.8 Hilbert Metrics for Flat Manifolds 227
8.8.1 Euclidean Position for Nonrelativistic Scattering

The position translation representations in the free scattering structures
have a momentum P 2 > 0 as translation invariant for the Euclidean group
SO(3) × R3 . The spherical Bessel functions are positive-type functions for
an irreducible representation with P = |P |, normalized by j0 (0) = 1:
d3 q
R3 x −→ d3 (P x) = j0 (P r) = sin P r
Pr = 2πP δ(q 2 − P 2 )eiqx , d3 ∈ Cb (R3 )+
The L2 (Ω2 )-isomorphic Hilbert spaces for scattering states are infinite-
dimensional.
More general, the irreducible infinite-dimensional Hilbert representations
of the Euclidean groups SO(1 + s) × R1+s , s ≥ 0, are, for a nontriv-
ial translation invariant and for s ≥ 1, inducible with translation fix-
group SO(s). The scalar representation coefficients for the Euclidean spaces
SO(1 + s) × R1+s /SO(1 + s) ∼= R1+s ,

P > 0 : R1+s x −→ D1+s (P 2 r2 ) = d1+s q δ(q2 − P 2 )e−iqx
s−1
= P2 ds ω e−iP ωs x
s−1
= P2 ds ω cos P ωs x,
use the Fourier-transformed measure of the momentum direction sphere ωs =

s ∼
q| ∈ Ω = SO(1 + s)/SO(s). The scalar representation coefficient can be
q

|
normalized as a positive-type function for a cyclic translation representation
where the momentum sphere has the invariant as the intrinsic unit:
1+s
state: d1+s (P x) = |Ω2ds |P s−1 q
δ(q 2 − P 2 )e−iqx
ds ω
= |Ωs | cos P ωs x, d1+s (0) = 1,
⎪
⎪
L (Ω ) with distributive basis {|P 2 ; ωs ⎪
2 s
⎪ωs ∈ Ωs },
⎧
⎪
⎪ P 2 ; ωs |P 2 ; ωs = |Ωs |δ(ωs − ωs ),
⎪
⎨ D(x)|P 2 ; ωs = eiP ωs x |P 2 ; ωs ,
ds ω 2
⎪ cyclic vector: |P 2 ; 1 = ⊕ |Ω
⎪ s | |P ; ω
s ,
⎪
⎩ P 2 ; 1|P 2 ; 1 = 1.
For s ≥ 2, the scalar representation coefficients arise by a 2-sphere spread

via derivations − dr2 ,
d 4π
πJ s−1 (r)
D1+s (r2 ) = − d
r2
Ds−1 (r2 ) = 2
s−1 ,
d 4π r
( 2π ) 2
and embed the self-dual R-representation matrix element:

R x −→ D1 (r2 ) = dq δ(q 2 − 1)e−iqx = cos r.
For odd-dimensional spaces with SO(1 + 2R), they involve half-integer-index

(spherical) Bessel functions, whereas integer-index Bessel functions are used
for even-dimensional spaces with SO(2R), both with rank R:

⎧ R
⎪
⎪ 2R−1 πjR−1 (r)
− d
⎪
⎪ 2 r R−1 = 2 cos r,
⎪ ( ) r
d 4π
⎨ 2π
for 1 + s = 1 + 2R = 1, 3, . . . ,
D1+s (r2 ) = R−1
⎪
⎪ πJR−1 (r)
− d
⎪
⎪ r R−1 = πJ0 (r),
⎪ ( 2π ) r 2
d 4π
⎩
for 1 + s = 2R = 2, 4, . . . .
The integrals sum over the embedded R-representation coefficients:
π
D2 (r2 ) = 0 dχ cos(r cos χ) = πJ0 (r),
1
D3 (r2 ) = π −1 dζ cos(rζ) = 2πj0 (r) = 2π sinr r = − dr2 cos r.
d 4π
The 2-sphere spread from one to three dimensions is illustrated by the

scalar integral formula with the spherical Bessel function j0 :
3 ∞
q 2 )eiqx = − 2π
d q f ( ∂
r ∂r dq f (q 2 ) eiqr |r=|x| = 4π 0 q 2 dq f (q 2 ) j0 (qr).
SO(1 + s)-nontrivial degrees of freedom R1+s ∼ = R+ × Ωs use derivations,
q )L ∼ (i ∂
( ∂ L
x ) , L = 0, 1, . . . ,
∂
∂x =
x ∂
r ∂r = 2x ∂r∂ 2 .
The related coefficients,
2d1+s q
Rs x −→ |P 2 , L}(x) = |Ωs |P s−1 ( Pq )L δ(q 2 − P 2 )eiqx ,
contain products for the rotations SO(1 + s) (harmonic polynomials) and
translations R1+s . There is Schur orthogonality for different translation and
rotation invariants. The Schur products involves as Plancherel distribution the
1+s |Ω |:
P dP s
inverse of the translation measure (2π)
s P −P
4d1+s q
{P 2 , L |P 2 , L} = ( 2π
P ) δ( π ) q )L ⊗ (q)L δ(q 2 − 1)
|Ωs |2 (
1+s
2 2π s P −P LL Γ(1+L) Γ( 2 )
= |Ωs | ( P ) δ( π )δ 2L Γ( 1+s
(11+s )L .
2 +L)
and the multiplicity factors for the harmonic polynomials and units (see
Chapter 10):
2d1+s q Γ( 1+s
2 )
q )L ⊗ (
|Ωs | ( q 2 − 1) = δ LL Γ(1+L)
q )L δ( 2L Γ( 1+s +L)
(11+s )L .
2
The harmonic analysis for representations of the Euclidean group is famil-

iar from the nonrelativistic scattering SO(3) ×
R3 with the angular momen-
tum decomposition of a plane wave with z = r cos θ and translation-invariant
momentum P , involving Legendre polynomials, i.e., spherical harmonics, and
spherical Bessel functions jL :
⎧ 1 dζ L
⎪
⎪ jL (r) = −1 2iL
P (ζ)eiP rζ
$∞ ⎨
eiP z = (1 + 2L)iLPL (cos θ)jL (P r), = rL (− 1r dr
d L sin r
) r ,
⎪
⎪
L=0 ⎩ PL (cos θ) = 4π
YL (0, θ). 1+2L 0
8.8 Hilbert Metrics for Flat Manifolds 229
8.8.2 Minkowski Spacetime for Relativistic Particles

The cyclic spacetime translation representations for free particles with mass
m2 ≥ 0 as translation invariant of the Poincaré group SO0 (1, 3) × R4 with
the positive on-shell distribution of the Feynman propagator are similar,
4
R4 x −→ d(1,3) (mx) = Vd3 mq 2 δ(q 2 − m2 )eiqx ,
normalized with a constant V 3 > 0, e.g., V 3 m2 = (2π)3 . The L2 (Y 3 )-isomor-

phic Hilbert spaces for free particles are infinite-dimensional.
The irreducible Hilbert representations of the Poincaré group SO0 (1, s)
×
R1+s , s ≥ 1, for positive translation invariant, e.g., massive particle repre-
sentations for m > 0, are inducible with translation fixgroup SO(s). They
come as the Fourier-transformed measure of the energy-momentum hyper-
s ∼
boloid with the directions |q| q
= ±ys ∈ Y± = SO0 (1, s)/SO(s) as eigen-
values. The scalar representation coefficients for the Minkowski translations
read, with |x| = |x2 |,

m2 > 0 : R1+s x −→ D(1,s) (m2 x2 ) = d1+s q δ(q 2 − m2 )eiqx

= ms−1 ds y cos mys x,
−ϑ(x2 )πN− s−1 (|x|)+ϑ(−x2 )2K s−1 (|x|)
d
D(1,s) (x2 ) = 2 D(1,s−2) (x2 ) = 2
s−1
2
.
dx
4π | 2π
x
| 2
The hyperbolic invariant m is used as intrinsic unit in the representation

coefficients:
1+s q s
d(1,s) (mx) = Vds ms−1 δ(q 2 − m2 )eiqx = dV sy cos mys x.
⎪
⎪
L2 (Y s ) with distributive basis {|m2 ; ys ⎪ ⎪ys ∈ Y s }

m2 ; ys |m2 ; ys = V s δ(ys − ys ),
D(x)|m2 ; ys = eimys x |m2 ; ys .
Since the hyperboloid volume is infinite, the positive-type function is not in

L∞ (R1+s ):
1+s s
d(1,s) (0) = d V s q δ(q 2 − 1) = dV sy = |Y |
s
Vs .
A formal analogy to the Euclidean case above with the sphere volume is
obtained with an “infinite normalization” V s = |Y s | = ∞.
In particle physics, the momentum parametrization (geodesic polar coor-
dinates; see Chapter 2) with the manifold isomorphy Y s ∼ = Rs is more familiar
than the hyperbolic parametrization with the rapidity ψ = arsinh |m q|
:
⎧ q|2 q2
2 2
⎨ dy2s = md| 2 +
q
2
2 + m2 dωs−1 ,
cosh ψ
ys = m 1 m + q
= sinh ψ ω s−1 , s
q
⎩ ds y = s d q q 2 ,
⎪ m 1+ m2
⎪
⎪
{|m ;
2
q ⎪q ∈ R } with m ;
s 2
q |m ;
2
q = m 1+ m
s q2 s

2 V δ(q − q ).
The “cyclic vector” e.g., for a free scalar field (see Chapter 5),
ds y 3
m2 ⊕ 2 (|m2 ; ys ⊕ m2 ; ys |) = ⊕ 2q0d(2π)
q 2
q ⊕ m2 ; q |)
3 (|m ;
= Φ(0)(|0 ⊕ 0|),
is not normalizable and, therefore, not in the Hilbert space. There exist cyclic
vectors as limits of normalizable functions on the momentum hyperboloid Y s .
The representation coefficients embed time and 1-position representations

R t −→ dq δ(q 2 − 1)eiqt = cos t,
−iqz
R z −→ dq 1
π q2 +1 e = e−|z| ,
for s ≥ 2 as 2-sphere spreads. For odd dimension and SO0 (1, 2R), they involve
half-integer-index functions, hyperbolic Macdonald functions, and spherical
Bessel functions:
−|x|
D(1,2) (x2 ) = 2π −ϑ(x ) sin |x|+ϑ(−x
2 2
)e
|x| ,
R
D(1,2R) (x2 ) = ∂
x2
[ϑ(x2 ) cos |x| + ϑ(−x2 )e−|x| ]
∂ 4π
ϑ(x2 )(−1)R πJR− 1 (|x|)+ϑ(−x2 )2KR− 1 (|x|)
= 2
x R− 2
1
2
for
| 2π |
1 + s = 1 + 2R = 3, 5, . . . .
For even spacetime dimension and SO0 (1, 2R − 1) they start with the
rank-R = 1 Poincaré group by integrating R-representation coefficients on a
hyperbola, leading to timelike oscillations and a spacelike hyperbolic fall-off:

D(1,1) (x2 ) = dψ [ϑ(x2 ) cos(|x| cosh ψ) + ϑ(−x2 )e−|x| cosh ψ ]
= −ϑ(x2 )πN0 (|x|) + ϑ(−x2 )2K0 (|x|).
A 2-sphere spread gives the Hilbert representation of the rank-2 Poincaré

group with Minkowski translations and, in general, the integer-index functions:
R−1
∂
D(1,2R−1) (x2 ) = x2
[−ϑ(x2 )πN0 (|x|) + ϑ(−x2 )2K0 (|x|)]
∂ 4π
ϑ(x )(−1)R πNR−1 (|x|)+ϑ(−x2 )2KR−1 (|x|)

2
= | 2π |
x R−1 for
1 + s = 2R = 2, 4, . . . .
SO0 (1, s)-nontrivial coefficients use derivations
(q)L ∼ (−i ∂x
∂ L
) , L = 0, 1, . . . , ∂
∂x
∂
= 2x ∂x 2
and the products for the two factors — Lorentz group with harmonic poyno-
mials and translation group,
d1+s q
R1+s x −→ |m2 , L}(x) = V s ms−1 ) δ(q 2 − m2 )eiqx .
q L
(m
8.9 Parabolic Subgroups 231
For different translation and Lorentz group invariants, the representation

coefficients are Schur-orthogonal with a Plancherel distribution containing
ms dm
the inverse of the measure (2π) 1+s for the translation invariants,
1+s
|Y s | 2π s m−m LL Γ(1+L) Γ( 2 )
{m2 , L |m2 , L} = (V s )2 ( m ) δ( π )δ 2L Γ( 1+s
(11+s )L ,
2 +L)
and the multiplicity factors for the harmonic polynomials and units,
1+s
Γ(1+L) Γ( 2 )
d1+s q (q)L ⊗ (q)L δ(q 2 − 1) = |Y s |δ LL 2L Γ( 1+s
(11+s )L .
2 +L)
The on-shell contributions of Feynman propagators for particle fields con-

tain, in general, linear combinations of harmonic SO(1, 3) energy-momentum
polynomials (see Chapter 5).
8.9 Parabolic Subgroups

For a noncompact semisimple group G with known representations of its
maximal compact group K, there remains the representation of the compact
group classes G/K. G-representations are inducible from those of parabolic
subgroups.
With an Iwasawa decomposition G = K ◦A◦N = Ň ◦A◦K, the K-classes
can be parametrized by the triagonal group A ◦ N ∼ = G/K with maximal
noncompact abelian A and nilpotent Lie algebra log N or by Ň ◦ A with the
negative transposed log Ň . A minimal parabolic subgroup P = [K0 × A] ◦ N
extends the Cartan plane A by its centralizer ZG (A) = K0 ⊆ K in the
compact group. One can visualize, in appropriate bases, parabolic subgroups
as block-triagonal,
K ◦ A ◦ N = Ň ◦ [K0 × A]
G =⎛ ⎞ ◦ N,
⎛ log Ň = −(log N ) , ⎞
T
× × n n n n × n n n n n
× × n n n n ×
⎜ ⎟ ⎜ 0 n n n n
⎟
[K0 × A] ◦ N = ⎜
⎝
0
0
0
0
×
×
×
×
×
×
n
n
⎟⊇⎜
⎠ ⎝
0
0
0
0
×
0
n
×
n
n
n
n
⎟
⎠
0 0 × × × n 0 0 0 0 × n
0 0 0 0 0 × 0 0 0 0 0 ×
= A ◦ N,
dimR G = dimR K0 + dimR A + 2 dimR N, dimR K
= dimR K0 + dimR N,
with K0 × A block-diagonal (entries ×) and N with diagonal 1 and entries n

strictly above and to the right of the blocks, with the examples
SO0 (1, s) ⊃ [SO(s − 1) × SO0 (1, 1)] ◦ exp R

s−1
, s = 2, 3, . . . ,
eiα3 +β3 n1 + in2
SL(2, C) ⊃ [SO(2) × SO0 (1, 1)] ◦ exp R2 0 −(iα3 +β3 )
e
,
1+r
SL0 (1 + r, R) ⊃ SO0 (1, 1)r ◦ exp R( 2 ) ,
SL(1 + r, C) ⊃ [SO(2) × SO0 (1, 1)]r ◦ exp Rr(1+r) .
A parabolic subgroup of G is a closed subgroup P ⊇ P containing a minimal

parabolic subgroup. It has the Langlands decomposition
P = [K0 × A] ◦ N , log P = log K0 ⊕ log A ⊕ log N (as vector space),
with the properties for the mutually orthogonal Lie subalgebras:
noncompact abelian: log A, nilpotent: log N , ⎪

⎪
log A-centralizer in log G: log K0 ⊕ log A = {l ∈ log G⎪
⎪[l, log A] = {0}},
N -normalizing: [log K0 ⊕ log A, log N ] ⊆ log N .
The Lie algebra can be decomposed with upper and lower triagonal matri-
ces into eigenspaces for the adjoint action of a commuting family of hermitian
operators involving log A with the roots R+ ∪ (−R+ ). The adjoint action de-
composes the nilpotent log N into an orthogonal sum of common eigenspaces
of log A:
Ň ⊕ log K0 ⊕ log

log G = log A ⊕ log N ,
log N = log Nω , log Ň = log N−ω ,
ω∈R+ ⎪
⎪
ω∈R+
log Nω = {n ∈ log N ⎪
⎪[a, n] = ω(a)n for all a ∈ log A}.
The nilpotent Lie algebra defines the nilpotent root sum ρN , which involves
the multiplicities nω of the positive roots (eigenvalues collection) of log A on
log N :
$
ρN = 12 nω ω ∈ (log A)T .
ω∈R+
The induction of representations of affine groups H × Rn (flat spaces)

by representations of direct product subgroups H0 × Rn (see Chapter 7)
is a good introduction to the parabolic induction for representations of a
semisimple noncompact Lie group G (also for a reductive Lie group with
suitable interpretations). In comparing K ◦ A ◦ N with affine groups H ×
Rn ,
the subgroups K0 × A are the analogues to the direct product subgroups
H0 × Rn with translation fixgroups H0 . The Hilbert representations of the
parabolic subgroups with trivial N -representation,
P = [K0 × A] ◦ N uea n −→ eiqa D(u) U(W ),
induce Hilbert representations of the full group G acting by left translations

on the P-intertwiners,
⎪
⎪
{w : G −→ W ⎪
⎪w(kuea n) = e−(ρN +iq)a D(u−1 ).w(k)}.
They involve a modular function ea −→ eρN a for the non-unimodular measure

of the parabolic subgroup.
8.9.1 Discrete and Continuous Invariants

For a semisimple Lie group, the irreducible discrete series Hilbert represen-
tations nontrivially and discretely support the Plancherel measure. They
correspond to representations (characters) of a Cartan torus (determined
up to conjugacy) and have square-integrable coefficients — therefore also
called square-integrable representations. There are countably many equiva-
lence classes of them. For a compact group, there are only discrete series
representations, and there is only one Cartan torus class. For noncompact
groups, there are finitely many classes of Cartan tori, e.g., SO(2) for SL(2, R),
sometimes none, e.g., for SL(2, C). Coefficients of square-integrable represen-
tations Dι display Schur orthogonality,

{Dι |Dι } = D̂ι Dι (1) = G dg Dι (g) ⊗ Dι (g) = διι c1ι 1d2ι ,

0, Dι ∼
= Dι
dg v |D (g)|w v |Dι (g)|wι =
G ι ι ι ι 1
v
cι ι ι|v wι |wι , D ι = Dι .
A noncompact group has a formal finite degree cι for the representation Dι ,

which, for a compact group, is the quotient of its dimension dι and the group
volume, both finite,
cι = dι for a compact group G.
G
dg
For a noncompact group, both volume and dimension are infinite.

Noncompact groups have Hilbert representations that continuously and
nontrivially support the Plancherel measure or have trivial Plancherel mea-
sure (nonamenable representations). The continuously contributing ones in
the reduced group dual can be induced from (conjugacy classes of) parabolic
subgroups [K0 × A] ◦ N ⊂ G.
8.10 Eigenfunctions of Homogeneous Spaces

(Spherical Functions)
Spherical functions, defined in the following, are positive-type, i.e., scalar
product–inducing functions for a special class of irreducible G-representa-
tions (definition ahead). The
representation normalized Legendre polynomials
4π
in the spherical harmonics 1+2L Y0L (θ, ϕ) = PL (cos θ) as the name-giving
example are given ahead.
For a compact subgroup K ⊆ G of a connected Lie group, the spherical
functions [36] are defined as nontrivial normalized K-invariant joint eigen-
functions (continuous functions of homogenous spaces G/K with compact
group classes),
⎧
⎨ Φ(K) = 1, representation-normalized,
Φ : G/K −→ C, Φ(kgK) = Φ(gK), K-left invariance,
⎩
DΦ = I(D)Φ, eigenfunction for each D ∈ D(G/K),
or with the class projection dI = Φ ◦ πK :

⎧
⎨ dI (1) = 1, representation-normalized,
d : G −→ C,
I dI (kgk ) = dI (g), K-bi-invariance,
⎩
DdI = I(D)dI , eigenfunction for each D ∈ DK (G).
In this definition, D(G) denotes the left-invariant differential operators on G,

DK (G) the subspace of those that are also right-K-invariant, and D(G/K)
the algebra of the left-G-invariant differential operators on the homogeneous
space G/K (see Chapter 10). For a semisimple Lie algebra, the invariants
constitute the center of the enveloping algebra.
The joint eigenfunctions are from the complex vector space
⎪
⎪
V I = {f ∈ C ∞ (G/K)⎪ ⎪Df = I(D)f for each D ∈ D(G/K)}
with characterizing representation invariants I(D) ∈ C. Each space of joint

eigenfunctions contains exactly one spherical function dI . The harmonic func-
tions (harmonic polynomials for a compact group G) f ∈ V 0 with triv-
ial invariants, i.e., Df = 0, contain the spherical function d1 = 1. A joint
eigenfunction is characterized by the product property with its spherical
function,

f, dI ∈ V I : f (g1 K)dI (g2 K) = K dk f (g1 kg2 K).
With their class K-property and with the integral notation for the
projection,
πK Φ
dI : G −→ G/K −→ C, g −→ gK −→ ⊕ K dk dI (gk) = Φ(gK),
spherical functions can be
written as G-representation coefficients integrated
over the compact group K dk = 1, as exemplified earlier.
Spherical functions are defined equivalently by
their G/K-coset-represen-
tation property, with normalized Haar measure K dk = 1,

dI ∈ C(G), dI = 0 : g1,2 ∈ G : dI (g1 )dI (g2 ) = K
dk dI (g1 kg2 ),
which reflects the representation property D(g1 K) ◦ D(g2 K) = D(g1 Kg2 K).
Now the connection to positive-type functions: A Hilbert representation
D : G −→ U(V ) is called K-spherical if there exists a vector with compact
fixgroup K. A spherical function on G/K defines a scalar product for an
irreducible spherical representation, and vice versa: The diagonal elements of
a normalized vector with fixgroup K (therefrom the bi-invariance) in an irre-
ducible spherical Hilbert representation constitute a positive-type spherical
function:
dv (g) = v|D(g)|v, with Gv = K, v|v = 1,
bijection: spherical L∞ (G)+ ∼
= irrep + G spherical.
8.10.1 Simple Examples

The simplest spherical functions are the unitary exponentials for the one-
dimensional abelian Lie groups with trivial compact subgroup K = {1}:
dz (α) = eiαz
diq (x) = eiqx
R∼
iαz
U(1),
d
diα e = zeiαz = D(1), d iqx
π izα −iz α dx e = iqeiqx
z ∈ Z, dα
e e = δzz iq ∈ iR, x
z −π 2π
$ iq dx eiqx e−iq = δ( q−q
2π )
d (0) = 1, izα −izα d (0) = 1,
e e = δ( α−α
2π ) iqx −iqx
dq
2π e e = δ(x − x )
z∈Z
Here and in the following examples, the left-invariant differential operator

stands in the second line. In the third and fourth lines, the Schur orthog-
onality and the completeness are given, involving the Haar and Plancherel
measures.
Rank-1 nonabelian examples with compact subgroup K = SO(2) are the
Legendre polynomials (spherical harmonics, analysis on Ω2 ) with the Casimir
invariant L(1 + L) as spherical functions for compact symmetric spaces
π dϕ
Ω2 ∼= SO(3)/SO(2), PL (cos θ) = −π 2π
(cos θ + i cos ϕ sin θ)L
2 dPL
$L
(L+k)! (−z)k
ξ = cos θ, dξ (1 − ξ ) dξ = −L(1 + L)PL
d
L
P (cos θ) = (L−k)! (k!)2 ,
1 dξ L
k=0 −1 2
P (ξ) PL (ξ) = 1+2L 1
δLL
$∞
L = 0, 1, . . . , z = sin2 θ
2,
PL (1) = 1, (1 + 2L)P (ξ) P (ξ ) = δ( ξ−ξ
L L
2 )
L=0
and the Legendre functions with continuous invariants as bounded spherical

functions for noncompact symmetric spaces as used for the harmonic analysis
(Fock–Mehler transformation) of the hyperboloid Y 2 (more later):
Y∼
2
= SO0 (1, 2)/SO(2),
∞
$ 1 π 1
1 Γ(iQ+ 1 +k) (−z)k PiQ− 2 (cosh ψ) = dϕ
(cosh ψ + cos ϕ sinh ψ)iQ− 2
PiQ− 2 (cosh ψ)
= 2
, −π 2π
1
Γ(iQ+ 2 −k) (k!) iQ− 1 1
k=0 2
ζ = cosh ψ, d
(ζ 2 − 1) dP 2
= −(Q2 + 1 )PiQ− 2
Q ∈ R, =
z − sinh2 ψ
, ∞ iQ− 1
dζ
−iQ − 1
dζ 4
iQ− 1
2
1
dζ P 2 (ζ) P 2 (ζ) =
1
δ(Q − Q )
Π2 (Q2 )
/P
2 (1)=
/ 1, ∞ 2
Π (Q2 )dQ PiQ− 2 (ζ) P−iQ− 2 (ζ ) = δ(ζ − ζ )
1 1
/ Γ(iQ+ 1 ) /2
Π (Q ) = //
2 2 2 / Q tanh πQ,
= 0
Γ(iQ) /
7
f (ζ) = 0∞ Π2 (Q2 )dQ PiQ− 2 (ζ)f(Q), ∞ ∞
1
˜
∞ −iQ− 1 dζ |f (ζ)|2 = Π2 (Q2 )dQ |f˜(Q)|2 .
˜
f (Q) = 1 dζ P 2 (ζ)f (ζ),
1 0
The spherical functions for the analysis of the Euclidean plane are the Bessel
functions with one momentum invariant (considered already, more ahead):
R2 ∼= SO(2) × R2 /SO(2), π 2
$∞ J0 (|P
x|) = dϕ
−π 2π
ei|P x| cos ϕ = dπq δ( q 2 − P 2 )ei
q
x
(P z)2k
J0 (|P |r) = (k!)2
, J0 (|P
∂ 2
x|) = −P 2 J0 (|P
x|)
2 |
x|)J0 (|P δ( |P |−|P
k=0
2 d x J0 (|P x|) = 1
|P |
)
P 2 ≥ 0, z 2 = − r4 , ∞ 2π
J0 (0) = 1, 0
dP 2
J0 (|P |r)J0 (|P |r ) = r δ(r − r )
1
8.10.2 Eigenfunctions of Compact Groups

All irreducible representations of a compact group U are spherical, i.e., char-
acterized by a spherical function for a compact fixgroup K. The spherical
functions on a compact homogeneous space U/K are precisely the character
class functions of the irreducible spherical U -representations,

U u −→ dI (u) = tr D(Ku−1 ) = tr K dk D(ku−1 ).
They are positive-type functions with the bijection
spherical L∞ (U )+ ∼
= irrep U.
8.10.3 Eigenfunctions of Euclidean Groups

The spherical functions on an affine group with compact homogeneous group
K and translations V ∼ = R1+s , considered earlier, are characterized by the
K-invariant momentum P ∈ V T ,

K × R1+s /K ∼= R1+s x −→ diP (x) = eiP,K•x = K dk eiP,k•x ,
e.g., on the Euclidean groups with real rank 1:
2 ds ω iP ω
dP ,1+s (|
x|) = |Ω s| e
sx
R1+s ∼
= SO(1 + s) × R
1+s
/SO(1 + s),
J s−1 (|P x|)
∞ 2d1+s q
2 $ Γ( 1+s ) (P z)2k = q 2 − P 2 )ei
δ( q
x
= Γ( 1+s ) 2
dP ,1+s (r) = 2 , |Ωs ||P |s−1 2 |P
x|
s−1
1+s
Γ( k!
+k) ( ) 2
2 2
k=0 2 2
2 2 dP ,1+s = −P 2 dP ,1+s
∂
P 2 ≥ 0, z 2 = − r4 ,
2 2
2π )s δ( |P |−|P | )
2
dP ,1+s (0) = 1, d1+s x dP ,1+s (|
x|)dP ,1+s (|
x|) = |Ω1s | ( |P | 2π
∞ P )1+s dP dP 2 ,1+s (r)dP 2 ,1+s (r ) =
0 |Ωs |( 2π 1
|Ωs |rs
δ(r − r )
8.10.4 Eigenfunctions of Noncompact Groups

For a noncompact semisimple Lie group G with Iwasawa decomposition
β(g)
1 0 e 0
G = N ◦ A ◦ K g = en(g) ea(g) k(g) = n(g) 1 ◦ 0 e−β(g) ◦ k(g),
the Hilbert representations, which contribute continuously to the Plancherel

measure, can be characterized, according to Harish–Chandra, by the spherical
functions of the noncompact homogeneous space G/K ∼ = N ◦ A. They have
r = dimR A complex invariants Q from the complexified weight space,
⎪
⎪
{diQ+ρN ∈ C(G/K)⎪ ⎪Q ∈ C ⊗ (log A)T = (log AC )T ∼= Cr },
G g = en(g) ea(g) k(g) −→ diQ+ρN (g) = eiQ+ρ

N ,a(gK)
= K dk eiQ+ρN ,a(gk) .
The dilation factor eρN ,a(gK) contains the nilpotent root sum ρN .
An example is the bounded Legendre functions on the hyperboloids Y s ,

s ≥ 2, with real rank 1:
⎛ ⎞
cosh ψ
SO0 (1, s)/SO(s) ∼= Y s ys = ⎝sinh ψ cos ϕ ⎠ with ωs−2 ∈ Ωs−2 ,
sinh ψ sin ϕ ωs−2

ys −→ eiQ− 2 , SO(2)•ψ = PiQ− 2 (cosh ψ)
s−1 s−1
π dϕ s−1
= −π 2π (cosh ψ + cos ϕ sinh ψ)iQ− 2 .
s − 1 is the dimension of the nilpotent Lie algebra Rs−1 and, therefore, the
multiplicity nω of the one positive root ω ∈ (log SO0 (1, 1))T . The compact
group integration goes over SO(2).
In the general orthogonality and L2 -completeness,
∞
$ π
Γ(λ+k+1) (−z)k Pλ (cosh ψ) = dϕ
(cosh ψ + cos ϕ sinh ψ)λ
Pλ (cosh ψ) = Γ(λ−k+1) (k!)2
, −π 2π
λ
k=0 ζ = cosh ψ, d
(ζ 2
− 1) dP
dζ = λ(1 + λ)P ,
λ
λ ∈ C, z = − sinh2 ψ
2, ∞ s−1
dζ
s−1
−iQ −
Pλ (1) = 1, 1
dζ P iQ−
2 (ζ) P 2 (ζ) = Πs (Q2 ) δ(Q − Q )
1
s ∼ ∞ s 2
for Y = SO0 (1, s)/SO(s), iQ− s−1 −iQ− s−1
Π (Q )dQ P 2 (ζ) P 2 (ζ ) = δ(ζ − ζ )
λ = iQ − s−1
2 , Q ∈ R,
0
the Plancherel measure for the harmonic analysis of hyperboloid functions

[55] has no discrete series contributions. It has to distinguish between even
and odd dimensions:
⎧
⎪
⎪ ;
R−1
4Q2

⎪
⎪ Γ(R − 1 2
× Q2 tanh πQ
⎪
⎪ 2) πQ 1+ (2k−1)2
,
/ / ⎪
⎪
/ Γ(iQ+ s−1 ) /2 ⎨ k=1
s = 2R = 2, 4, 6, . . . ,
Πs (Q2 ) = // 2 / =
/
Γ(iQ) ⎪
⎪ ;
R−2
⎪
⎪ Γ(R − 1)2 × Q2 1+ Q
2
⎪
⎪ k2
,
⎪
⎪
⎩ k=1
s = 2R − 1 = 3, 5, 7, . . . .
For the minimal nonabelian odd-dimensional

∞ case Y 3 with the proper
Lorentz group and Plancherel measure 0 dQ Q2 , the principal series
SO0 (1, 3)- representations are used.
Spherical functions are equal if, and only if, their Cartan invariants are
connected by the Weyl group; the r invariants are characterized by Weyl
group orbits:
S ∈ Weyl(G) : diQ1 +ρN = diQ2 +ρN ⇐⇒ Q1 = S.Q2 ,

Weyl(G) .Q ⊂ (log AC )T .
Spherical functions of noncompact spaces do not have to be bounded. The

s−1
Legendre functions above for real invariant Q ∈ R are bounded, PiQ− 2 ∈
Cb (Y s ). This illustrates the general property: A spherical function is bounded
if, and only if, the invariants are in the tube i(log A)T + C(ρN ), where C(ρN )
is the real convex hull of the Weyl orbit of the root sum ρN .
Spherical functions of noncompact spaces G/K have a tangent space

parametrization: The translations log G/K ∼
= log N ◦ A = V ∼
= Rs are acted
on by the adjoint G-representation, which defines the semidirect tangent
group,
log G/K (k, x), with k • x = k ◦ x ◦ k −1 .

Gtan = K ×
Examples are the Euclidean groups SO(s) × Rs for the hyperboloids, which
are manifold-isomorphic to vector spaces Y = SO0 (1, s)/SO(s) ∼
s ∼
= Rs . With
the Killing form q |x, restricted to the tangent space log A◦N = V , there ex-
ists[36], for each spherical function of G/K, a unique K-invariant function in
a tangent parametrization with r real invariants m ∈ log A, using translations
and momenta x, q ∈ V ,
⎪
⎪ ⎪
⎪
{diQ+ρN ⎪ ⎪Q ∈ (log AC )T ∼ = Cr } ∼
= {dim,J ⎪⎪m ∈ log N ◦ A ∼ = Rs },

G g = en ea k = ex k −→ diQ+ρN (g) = eiQ+ρN ,a(gK) = K dk eiQ+ρN ,a(gk)
q |k•
x
q |K•
x dk ei

= dim,J (x) = e √ √
i

= K
.
J(
x) J(
x)
The local normalization of the exponent involves the ratio J(x) of the volume
elements in V and G/K.
8.11 Hilbert Metrics for Hyperboloids

and Spheres
In contrast to the representations of flat spaces above (scattering waves,
free particles), Hilbert representations of hyperboloids and spheres with
nonabelian degrees of freedom use higher-order momentum poles. This will
be exemplified first by the nonrelativistic hydrogen atom bound waves,
which represent the noncompact nonabelian group SO0 (1, 3) and start with
momentum dipoles.
8.11.1 Hyperbolic Position in the Hydrogen Atom

2
The nonrelativistic dynamics H = p2 − 1r with the Coulomb–Kepler potential
has a rotation and a Lenz–Runge “perihelion” invariance (see Chapter 4).
The measure of the unit 3-sphere as the manifold of the orientations of the
rotation group SO(3) in the invariance group SO(4) for bound waves has a
momentum parametrization by a dipole (see Chapter 2):
3 3
√ 12 1
q ∈ Ω ⊂ R ⇒ |Ω | =
i
3 4 3
d ω = (q2d q 2
2 +1)2 = 2π .
q +1

1
2d3 q 2π 2
A sphere radius |Q| with curvature Q2 is implemented by q2 +Q 2 )2
( = |Q| .
8.11 Hilbert Metrics for Hyperboloids and Spheres 239
The Fourier-transformed Ω3 -measure, gives the hydrogen ground-state

function as a scalar representation coefficient of 3-position space. It is a
positive-type hyperbolic function:
d3 q
SO0 (1, 3)/SO(3) ∼
= Y3 ∼
|Q|
= R3 x −→ π2 (q 2 +Q2 )2 e−iqx = e−|Q|r .
In the bound waves, position as a noncompact hyperboloid is represented in

L2 (Y 3 ) in the form of Fourier-transformed Ω3 -measures with a continuous
invariant Q2 for the imaginary “momenta” q 2 = −Q2 on a 2-sphere Ω2 and
a rational rotation invariant 2J ∈ N. Hyperbolic position is isomorphic as
manifold, not as symmetric space, to the translations Y 3 ∼ = R3 .
The Kepler bound waves are coefficients of infinite-dimensional cyclic re-
presentations of the Lorentz group SO0 (1, 3). With the Cartan subgroups
SO(2) × SO0 (1, 1), the irreducible representations are characterized by one
integer and one continuous invariant. In the language of induced representa-
tions, the bound waves of the hydrogen atom are rotation SO(3)-intertwiners
on the group SO0 (1, 3) (Y 3 -functions) with values in Hilbert spaces with
SO(3)-representations in (1 + 2J)2 -dimensional SO(4)-representations.
∂
The rotation dependence x is effected by momentum derivation i ∂ q of the
3
Ω -measure:
3
xe−r = dπ2q (1+
4i
q
q 2 )3 e
−i
qx

with (1+q2 )3 = − ∂
4
q ∂ 1
q 2 )2 .
q (1+
2
q q

The 3-vector factor 1+ q 2 = | q| sin χ is uniquely supplemented to a parame-
trization of the unit 3-sphere by a normalized 4-vector:

cos χ 1−q2
q

i sin χ = 1
1+q2 2i
q = pip0 = p ∈ Ω3 ⊂ R4 , p20 + p 2 = 1.
|
q|
1 1

p0
The unit 4-vector Y( 2 , 2 ) (p) ∼ i
p ∈ Ω3 is the analogue to the unit 3-
vector Y1 ( |qq | ) ∼ q

|
q| ∈ Ω2 used for the buildup of the 2-sphere harmonics
YL ( |qq | ) ∼ ( |qq| )L . Analoguously, the Ω3 -harmonics are the totally symmetric
traceless products Y(J,J) (p) ∼ (p)2J , e.g., the nine independent components
SO(3)
in the (4 × 4)-matrix and the decomposition 9 = 1 ⊕ 3 ⊕ 5:
2 2
3p0 −
δjk ∼ p
Y (1,1)
(p) ∼ (p)jk = pj pk − 4 =
2 4
ip0 pa
δab ,
ip0 pb pa pb − 4
δab 3p0 −
2
p2
with pa pb − δab
4 = pa pb − δab 2
3 p
− 3 4 for p2 = 1.
The Kepler bound waves in (1 + 2J)2 -multiplets for SO(4) come with
momentum poles of order 2 + 2J:

1−q2
3 p = 1
,
x −→ dπ2q (1+1q 2 )2 (p)2J e−iqQx with
Q = 1 + 2J : Y
1 3 1+q 2 2i
q
Q2
E =− 2 .
1
The radius of the 3-sphere is the SU(2)-multiplicity Q = 1 + 2J with cur-
1 1
vature Q = (1+2J)2 , or, in the hydrogen atom ( (1+2J)
R )2 with the Bohr
2
length R = mecα2 ∼ 6.8 × 10−8 m as unit.

S
The Fourier transformations with the 3-sphere measure
μ(
x) μ̃(
q)
1 1
3−q2
r 1+q2
d3 q

x 2i
q
μ(x)e−r = q )e−iqx q2
1 2 1+
q 2 )2 π 2 μ̃(
(1+ : r2 4(1−q 2)
3 (1+q 2 )2
2i
q (5−q 2)
r
x (1+q 2 )2
q2

2 q ⊗
6( q−13 )
x⊗
x − 13 r3 − (1+q 2 )2
3
are used for the ground-state with a dipole,

−Qr
3 −i
d3 q −i
1
Q = 1: e = dπ2q (Q2 +
Q
q 2 )2 e
q
x
= 1
q 2 )2 e
π 2 (1+
qQ
x
,
1 1
the bound-state quartet with Y( 2 , 2 ) (p) and tripole, leading to Laguerre
polynomials,
d3 q 3
1
Q = 2:
1 p0
p e
−i
qQ
x
= dπ2q (1+1q 2 )3 1 − q2

e−iqQx
π 2 (1+
q 2 )2 i Qr−1 2i
q
1 1
2 − L (2Qr)
= e−Qr = Qx4 L10 (2Qr) e−Qr ,
Qx 2 2
2
and the bound-state nonet with Y(1,1) (p) and quadrupole:

2 1−q 2 2
d3 q 3p0 − p
2
−i
d3 q 3( 2 ) − q2
1
Q =3 :
1
π 2 (1+
q 2 )2
ip0 p
e q Q
x 4
= π2 (1+q 2 )4 q 1−
i 2
q 2
e−iqQx
2
p⊗p
3 − 13 p
q⊗
3 q − 13 q
2
2 2 1 2

1 − 2Qr + 2Q3 r 3 L1 (2Qr)
= Qr−2 Q
3 2
x
e−Qr = − Q
2
x 1 1
6 L3 (2Qr) e−Qr .
Q2 r2 Q2 2
2 (13 3 − x⊗ x) 2 (13 r3 −
x⊗ x)L05 (2Qr)

The Schur product for the wave functions involves the harmonic SO(3)-
momentum polynomials with the corresponding multipoles at the invariants:
d3 q q )L q )L

{−Q2L , L | − Q2L , L}3 = π 2 (
(2
q 2 +Q2L )2+L
(2
⊗ (q 2 +Q 2 )2+L
L
Γ(1+L)Γ( 3 ) d3 q q 2 )L
(2
LL
= δ Γ( 32 +L)
2
π 2 (q 2 +Q2L )4+2L
(13 )L

1 1
= δ LL 1+L 23+L |QL |5+2L (13 ) .
L
8.11.2 Representations of Hyperboloids and Spheres

q ∈ Rs with the action of the
Distributions of s-dimensional momenta
rotation group SO(s) are used for representations [55, 54] of the
hyperboloids Y s and spheres Ωs . Flat spaces and hyberboloids are isomorphic

as manifolds:
2
Rs ∼= Y s with R1+s q +1

q

= cosh ψ
sinh ψ ω s−1
∈ Ys
s s−1
√d 2q =√
q dq
ds−1 ω = ds y = (sinh ψ)s−1 dψ ds−1 ω.
q +1
q2 +1
The residual representations of nonabelian noncompact hyperboloids and

compact spheres with s ≥ 2 have to embed the nontrivial representa-
tions of the abelian groups with continuous and integer dual invariants,
respectively:

SO0 (1, 1) ∼
= Y 1 x −→ dq
|Q|
2e
−iqx
= e−|Qx| ,
⎧ π q2 +Q
dq −iqx
⎨ P
iπ q2 −io−P 2 e = eiP |x| ,
∼
SO(2) = Ω e −→
1 ix
dq |q|δ(P − q )e
2 2 −iqx
= cos P x,
⎩
P = 0, 1, 2, . . . .
The invariant poles {±iP } and {±Q} on the discrete sphere Ω0 = {±1}
are embedded, for the nonabelian case, in singularity spheres Ωs−1 ∼ =
SO(s)/SO(s − 1), whose action groups arise in the Iwasawa decomposition,
SO0 (1, s) = SO(s) ◦ SO0 (1, 1) ◦ exp Rs−1 ,
parabolic subgroup: [SO(s − 1) × SO0 (1, 1)] ◦ exp Rs−1 .
The Lorentz groups for even dimension have a unique Cartan subgroup
type,
SO0 (1, 2R − 1) ⊇ SO(2)R−1 × SO0 (1, 1), s = 2R − 1 = 1, 3, 5, . . . ,
in contrast to odd dimensions, e.g., SO0 (1, 1) and SO(2) for SO0 (1, 2).
The real rank 1 of the orthogonal groups SO0 (1, 2R − 1) gives the real
(noncompact) rank 1 for the odd-dimensional hyperboloids, i.e., one contin-
uous noncompact invariant. The noncompact–compact pairs (Y 2R−1 , Ω2R−1 )
with odd-dimensional hyperboloids and spheres will be considered as a
generalization of the minimal and characteristic nonabelian case (Y 3 , Ω3 )
with nontrivial rotations as used for the nonrelativistic hydrogen atom
above.
The coefficients of residual representations of hyperboloids Y 2R−1 use
the Fourier-transformed measure⎪ 2of the momentum sphere Ω2R−1 with
2R−1 ⎪
⎪ ∼
singularity sphere {
q ∈ R ⎪q = −Q < 0} = Ω2R−2 with continu-
2
2
ous noncompact invariant Q and for imaginary “momenta” as eigenvalues.

2
−
q2
SO(2R)-multiplets arise via the sphere parametrization q 2 +Q2 Q2i|Q|
1
q ∈
Ω2R−1 ⊂ R2R ,
Γ(R)
for Y 2R−1 , R = 1, 2, . . . , |Ω2R−1
2
| = πR ,
⎧ 2R−1
⎨ 2d 2R−1q 2 |Q| 2 R e−iqx = e−|Q|r ,
x −→
|Ω
2d2R−1 q
| (
q +Q ) 2
⎩ Q − 2
−i 1 − R + |Q|r
|Ω2R−1 | ( 2
R
q +Q ) 2 R+1
q
2i|Q|
q e q
x
=
x e−|Q|r .
The order of the singularity is related to the rank R of the acting group
SO0 (1, 2R − 1), i.e., the dimension of the Cartan subgroups SO(2)R−1 ×
SO0 (1, 1):
2 R
d2R−1 q q
d|q| 2(R−1)
q 2 +Q2 )R
( = q 2 +Q2
q| d
| ω.
Each state {x −→ e−|Q|r } ∈ L∞ (SO0 (1, 2R − 1))+ with invariant Q2 > 0
characterizes an infinite-dimensional Hilbert space with a faithful cyclic re-
presentation of SO0 (1, 2R − 1) as familiar for R = 2 from the principal series
representations of the Lorentz group SO0 (1, 3). The positive-type function
defines the Hilbert product:
distributive basis: {| − Q2 ; q |
q ∈ R2R−1 },
|Q| |Ω2R−1 |
scalar product distribution: −Q2 ; q | − Q2 ;
q = (q 2 +Q 2 )R 2 δ(q − q ),
2R−1
Hilbert vectors: | − Q2 ; f = ⊕ |Ω2R−1 | f (q )| − Q2 ; q ,
2d q
2R−1 q |Q|
−Q2 ; f |−Q2 ; f = 2d |Ω2R−1 | f (
q)
q 2 +Q2 )R f (
( q ).
There is a representation of each abelian noncompact subgroup in the Cartan

decomposition Y 2R−1 ∼ SO0 (1, 1) × Ω2R−2 with the action on a distributive
basis and hence on the Hilbert vectors:
SO0 (1, 1)-representations for all ω ∈ Ω2R−2 : e−ωx −→ e−i|q |ωx
= e−iqx ∈ U(1),
action of all SO0 (1, 1) : | − Q2 ; q −→ e−iqx | − Q2 ; q ,
2R−1 q
cyclic vector: | − Q2 ; 1 = ⊕ 2d |Ω2R−1 |
| − Q2 ; q
4d2R−1 q d2R−1 q
with |Ω2R−1 |2 q |e−iqx | − Q2 ; q = e−|Q|r .
−Q2 ;
The scalar product of L1 (Y 2R−1 ) involves the positive-type function, e.g.,

for three-dimensional position R = 2 with intrinsic unit Q2 = 1,
d3 q
−1; f | − 1; f = π2 f (
q) 1
q 2 +1)2
( d3 x1 d2 x2 f˜ (x2 )
f (q ) =
−|x1 −x2 | ˜
e 3 f (x1 ),
with f (q ) = d x f˜(x)eiqx .
It can be put in the form of square-integrability L2 (Y 3 ) by absorption of the

square-integrable square root of the positive-type function:
L1 (Y 3 ) −→ L2 (Y 3 ), f˜(x) −→ ψ̃(x) = ξ˜√∗ f˜(x),

f (q ) −→ ψ(q) = q 28π
+1 f (
q ),
d3 q −i −r
with 8π
2 +1)2 e
q
x ˜ ˜
= e = d(x) = ξ ∗ ξ̃(x),
(2π)3 (
d3 q q√8π −i −r
q
x √e
(2π)3 q 2 +1 e = = ξ̃(x)
3 d3 q
2πr
⇒ −1; f | − 1; f = d q q)
π 2 f (
1
q 2 +1)2 f (
( q) = q ) ψ(
(2π)3 ψ ( q)
3
=
d x ψ̃ (x) ψ̃(x).
Therefore, all infinite-dimensional Hilbert spaces for different continuous

invariants Q2 > 0 can be transformed to subspaces of one Hilbert space
L1 (Y 2R−1 ) −→ L2 (Y 2R−1 ) ∼
= L2 (R2R−1 ).
Hyperboloid representations are characterized by an integer invariant L ∈
N for the harmonic O(2R − 1)-momentum polynomials and a continuous
invariant Q > 0 related to SO0 (1, 1) ⊆ SO0 (1, 2R − 1):
2d2R−1 q q )L
| − Q2 , L}2R−1 (x) = |Ω2R−1 | (
(2
q 2 +Q2 )R+L
e−iqx .
The Schur product displays orthogonality for different rotation invariants (see
Chapters 9 and 10):
2d2R−1 q q )L q )L

{−Q2 , L | − Q2 , L}2R−1 = |Ω2R−1 | (
(2
q 2 +Q2 )R+L ⊗ (2
q 2 +Q 2 )R+L
(
Γ(1+L)Γ(R− 1 )
= δ LL Γ(R− 1 +L)2
2
2d2R−1 q (2q2 )L L
|Ω2R−1 | ( q 2 +Q 2 )R+L (12R−1 )
q 2 +Q2 )R+L (
Γ(1+L)Γ(R) 2L
= δ LL L
Γ(R+L) QQ (Q+Q )2R+2L−1 (12R−1 ) .
States with equal rotation invariants L = L and different continuous invari-

ants Q = Q are not orthogonal. The orthogonality of the Y 3 -representation
coefficients with different invariants in the hydrogen atom is a consequence
of the different rotation invariants. The Schur normalization of the hydrogen
wave functions is used for the probability interpretation.
The positive-type functions for L = 0 are representation-normalized with
a factor |Q|:
2d2R−1 q |Q|
q 2 +Q2 )R = 1.
|Ω2R−1 | (
The representation normalization of the L = 0 functions will be discussed in

Chapter 10.
The corresponding matrix elements of representations of odd-dimensional
spheres are obtained by a real–imaginary transition. They involve multipole
Feynman distributions (derived Dirac distributions) with supporting singu-
larity sphere Ω2R−2 for real momenta with, for appropriate normalization,
q 2 = L2 , L = 0, 1, 2, . . . ,
integer invariants
for Ω2R−1
,R = 1, 2, . . . ,
2R−1 2d q
i|Ω2R−1 | (
L
q 2 −io−L2 )R
e−iqx = eiLr ,
x −→ 2R−1
2 −i
d
π R−1
q (R−1)
Lδ 2
(L − q )e qx
= cos Lr.
In contrast to hyperboloids Y s ∼= Rs , spheres are compact. Therefore, the

Fourier-transformed positive-type functions on the spheres for SO(1 + s)-
representations do not have to give positive Radon measures of the momenta.
The irreducible SO(1 + s)-representation spaces used for the sphere Ωs

are finite-dimensional. A basis is given by the spherical harmonics (ωs )L ,
L = 0, 1, . . . , with r ∼ θ:

ω0 = 1, ωs = cos θ
sin θ ωs−1
∈ Ωs ⊂ R1+s , s = 1, 2 . . . .
In the nontrivial case, L = 0, Ω1 = SO(2)

is acted on by the two-dimensional
cos Lθ − sin Lθ
SO(2)-representations sin Lθ cos Lθ , Ω2 by the (1 + 2L)-dimensional har-
monic SO(3)-representations [L], and Ω3 by the (1 + L)2 -dimensional har-
monic SO(4)-representations ( L2 , L2 ). Different invariants L characterize
Schur-orthogonal subspaces of the infinite-dimensional Hilbert space L2 (Ωs ).
Mathematical Tools
8.12 Spherical, Hyperbolic, Feynman,
and Causal Distributions
n
d q
The Lebesgue measure (2π) n is the Plancherel measure for the irreducible
translation representations Rn x −→ eiqx ∈ U(1) and Haar measure dn x.

The circle has different parametrizations, e.g.,
π 1 ∞
θ 2 1 − q2 √ 1 2 ip
Ω1 qiq0 , for semicircle: icos
sin θ = iq
= 1
.
−2
π
−1 1+p −∞
Therefore, the compact classes of orthogonal groups SO(1 + s)/SO(s) ∼ = Ωs

(unit sphere) have volume
s
|Ωs | = ds ω = d1+s q 2δ(q02 + q 2 − 1) = q 2 ≤1 √d q 2
π 2ds p
1− q
= 0 (sin θ)s−1 dθ ds−1 ω = 1+s
(4π)R Γ(R)
p 2) 2
(1+
2
1+s
|Ω | = Γ(2R) = 2, 4π, 8π3 , . . . , |Ωs−2 |
2R
|Ω | = Γ( 1+s ) :
s 2π 2
2π R
s−1
|Ωs | = 2π ,
2 |Ω2R−1 | = Γ(R) = 2π, 2π 2 , . . . ,
polar decomposition: q = |q|ωs with |q|2 =q02 + q 2 , ω s ∈ Ωs ,
∞
d q = 0 |q| d|q| d ω.
1+s s s
The Gamma function has the following properties, where defined,

⎧ ∞
⎪
⎨ Γ(z) = 0 dx e−x xz−1 = Γ(z),
z∈C: Γ(1 + z) = zΓ(z), Γ(1 + N ) = N !, N = 0, 1, . . . ,
⎪
⎩ Γ(z)Γ(1 − z) = π , Γ( 2 +z)
1
sin πz Γ( 1 )
= 21−2z Γ(2z)
Γ(z) .
2
The spherical–hyperbolic, i.e., compact–noncompact partner transition,

for SO(1 + s)/SO(s) ∼
= Ωs : (iθ, i p) ↔ (ψ, q, p) for Y s ∼
q, i = SO0 (1, s)/SO(s),
8.12 Spherical, Hyperbolic, Feynman, and Causal Distributions 245
gives the parametrizations and volume elements of the unit hyperboloids with
noncompact classes of orthogonal groups,
∞ ∞ 1
Y 1 ϑ(qq0 )q0 , cosh ψ
sinh ψ = 1 + q2
= √ 1 1
,
q 1−p2 p 0
s 0 0 s
q 2 − 1) = √d q 2
d y = ϑ(q0 ) d1+s q 2δ(q02 −
∞ 1+ q
= 0 (sinh ψ)s−1 dψ ds−1 ω
s
2d p
= p 2 ≤1 1+s ,
p 2)
(1− 2
“polar” decomposition: q = |q|ys with |q|2 = q02 − q 2 , ys ∈ Y s ,

∞
ϑ(q0 )ϑ(q 2 )d1+s q = 0 |q|s d|q| ds y.
The measures of the momentumlike hyperboloids with the noncompact

classes of noncompact groups SO0 (1, s)/SO0 (1, s − 1) ∼ = Y (1,s−1) are
s s ∞
q 2 +1) = 2 q 2 ≥1 √d 2q = −∞ (cosh ψ)s−1 dψ ds−1 ω,
d s = d1+s q 2δ(q02 −
q −1

“polar” decomposition: q = |q|ss with |q|2 = −q02 + q 2 , ss ∈ Y (1,s−1) ,

∞
ϑ(−q 2 )d1+s q = 0 |q|s d|q| ds s.
Distributions with real poles have real–imaginary decompositions with

the principal value aP :
a ∈ R, ν ∈ R, ν = −1, −2, · · · : 1
(a−io)ν = 1
|a|ν [ϑ(a) + ϑ(−a)e
iνπ
],
Γ(1+N ) Γ(1+N )
N = 0, 1, 2, · · · : (a−io)1+N = a1+N
(N )
+ iπδ (−a).
P
2
The Dirac “on-shell” and the principal value (with qP ) “off-shell” distribu-
tions are imaginary and real part of the (anti-)Feynman distributions:
log(q 2 ∓ io − μ2 ) = log |q 2 − μ2 | ∓ iπϑ(μ2 − q 2 )
Γ(1+N )
(q2 ∓io−μ2 )1+N
= −(− ∂q∂ 2 )1+N log(q 2 ∓ io − μ2 )
= (− ∂q∂ 2 )N q2 ∓io−μ
1
2
Γ(1+N )
= (q2 −μ2 )1+N ± iπδ (N ) (μ2 − q 2 )
P
for μ ∈ R and N = 0, 1, . . . .
2
Feynman distributions are possible for any signature O(t, s) with positive or
negative invariant μ2 .
Characteristic for and compatible only with the orthochronous Lorentz
group SO0 (1, s) are the advanced (future) and retarded (past) causal ener-
gy-momentum distributions with positive invariant m2 only. They are distin-
guished by their energy q0 behavior:
log((q ∓ io)2 − m2 ) = log |q 2 − m2 | ∓ iπ(q0 )ϑ(m2 − q 2 )
Γ(1+N )
((q∓io)2 −m2 )1+N = −(− ∂q2 ) log((q ∓ io)2 − m2 )
∂ 1+N
= (− ∂q∂ 2 )N (q∓io)12 −m2

= (q2Γ(1+N )
2 1+N ± iπ(q0 )δ
(N )
(m2 − q 2 )
P −m )
for m ≥ 0 and (q ∓ io) = (q0 ∓ io) − q 2 .
2 2 2
The principal value integration is off-shell:

d1+s q Γ(1+N )
π 2 −μ2 )1+N e
(qP
iqx
= i(x0 ) d1+s q (q0 )δ (N ) (μ2 − q 2 )eiqx .
For the advanced and retarded integrations, one obtains the Fourier
transforms:
d1+s q Γ(1+N )
[(q∓io)2 −m2 ]1+N e
iqx
= d1+s q [ π1 (q2Γ(1+N )
P −m )
π 2 1+N
±i(q0 )δ (N ) (m2 − q 2 )]eiqx

1+s
= 2ϑ(±x0 ) d π q (q2Γ(1+N )
eiqx
1+s P −m2 )1+N
= ±2iϑ(±x0 ) d q (q0 )δ (N ) (m2 − q 2 )eiqx .
8.13 Residual Distributions

The causal structure of the reals and its unitary representations occur in
the Fourier-transformed causal measures,
dq Γ(1−ν) iqx = ϑ(x) eimx .
m, ν ∈ R : 2iπ (q−io−m)1−ν e (ix)ν
Here and in the following, the integrals hold wherever the Γ-functions are
defined.
8.13.1 Macdonald, Bessel, and Neumann Functions

The scalar distributions for the definite orthogonal groups in general dimen-
sion with real and imaginary singularities on spheres Ωs with q 2 = ±1 define
the Macdonald functions Kν and, for imaginary argument, the corresponding
Hankel (Bessel with Neumann) functions Hν1,2 = Jν ± iNν :
⎧ d1+s q Γ( 1+s
⎪
⎪ 2 −ν)
eiqx = (Γ(ν)
O(1 + s), ⎪
⎪ 1+s 1+s r 2ν ,
2)
⎪
⎨ π 2 (q ) 2 −ν
2
d1+s q Γ( 1+s −ν)
s = 0, 1, 2, . . . , 2
eiqx = ( rν)(r)
2K
ν ,
⎪
1+s 1+s
2 +1) 2 −ν
√ ⎪
⎪
π 2 (
q 2
⎪ d1+s q
r = x 2 , ν ∈ R, ⎪ ⎩ 1+s
2 −ν)
Γ( 1+s
1+s eiqx = − i 2Kνr(−ir)
ν = π[Jν +iN
r ν
ν ](r)
.
−ν (2) (2)
iπ 2 q 2 −io−1)
( 2
The angle integration is different for odd and even dimensions with dθ
and d cos θ = dζ:
d1+s q
∞ π
s≥1: q 2 )eiqx = 0 q s dq μ(q 2 ) 0 (sin θ)s−1 dθ eiqr cos θ
| μ(
s−1
|Ω
∞ π
0q
2R−1
dq μ(q 2 ) 0 (1 − cos2 θ)R−1 dθ eiqr cos θ , s = 2R − 1,
= ∞ 2R 1
− 0 q dq μ(q 2 ) −1 (1 − ζ 2 )R−1 dζ eiqrζ , s = 2R.
The integrals can be obtained by 2-sphere spread from the values for R = 1.
8.13 Residual Distributions 247
All (half-)integer-index functions can be obtained by derivation (“2-sphere

spread”) drd2 = 2r
1 d
dr :
⎧ N √
⎪
⎪ −2 d
2π(e−r , cos r, sin r),
⎪
⎪ dr 2
⎨
ν + 1 = N = 0, 1, 2, . . . ,
R+ r −→ (2Kν , πJrνν, πNν )(r) = N 2
⎪
⎪ −2 d
⎪
⎪ 2K 0 (r), πJ0 (r), πN0 (r) ,
⎩ dr 2
ν = N = 0, 1, 2, . . . ,
(2Kν+1 , πJν+1 , πNν+1 )(r) d (2Kν , πJν , πNν )(r)
r ν+1
(2) = −4 dr 2 ( r2 )ν ,
(K−N , J−N , N−N )(r) = (KN , (−1) JN , (−1) NN )(r),
N N
(K−N − 12 , J−N − 12 , N−N − 12 )(r) = (KN + 12 , (−1)N +1 NN + 12 , (−1)N JN + 12 )(r).
The half-integer index functions for ν > 0 start from ν = − 12 with R-

representations, whereas the integer-index functions start from ν = 0, which
involves a finite integration of R-representations:
1
' sin r
= 0 dζ cos ζr = −2 drd2 cos r for ν = 12 ,
for ν = − 21 : cos r r π
( J0 (r) = 0 dθπ cos(r cos θ) for ν = 0.
For ν ≥ 0, only the Bessel functions are regular at r = 0:

⎧ ∞
⎪
⎪ $ 2
(− r4 )k
$∞
2 ⎪
⎪ −πN (r) = r2
2 [2χk − log 4 ],
r ⎨
2
( r4 )k 0 (k!)
2K0 (r) = (k!)2 [2χk − log 4 ] k=0
$∞
k=0 ⎪
⎪
2
(− r4 )k
= H0 (ir) = iπ[J0 (ir) + iN0 (ir)], ⎪
1
⎪
⎩ J0 (r) = (k!)2 ,
k=0
Euler’s constant χ0 = Γ (1) = limk→∞ [log k − (1 + 12 + . . . + k1 )]
= −0.5772 . . . ,
χk = Γ (1) + 1 + 12 + . . . + k1 , k = 1, 2, . . . .
For integer ν, the index N characterizes the small distance behavior,

∞
$ 2
(− r4 )k
JN (r) = rN (−2 drd2 )N J0 (r) = ( r2 )N k!(N +k)! .
k=0
The half-integer-index functions start from the exponentials. The noncom-

pact and compact self-dual representations of the reals come with imaginary
and real poles in the complex plane:
−|x|
dq 1 iqx e , poles at q = ±i (hyperbolic),
O(1) : π q2 −io ±1 e =
iei|x| , poles at q = ±1 (spherical).
They involve the positive-type function for the basic self-dual spherical repre-
sentation R x −→ cos x and the basic self-dual hyperbolic one R x −→
e−|x| .
The integer-index functions begin with two-dimensional momentum inte-

grals, which integrate over the R-representation coefficients:
⎧ d2 q 1 iqx
⎪
⎪ q 2 +1 e = 2K0 (r) = dψ e−r cosh ψ ,
⎪
⎪ d2 q π

⎪
⎪ 1 i
q
x
= π[J0 + iN0 ](r) = −i dψ eir cosh ψ ,
O(2) : ⎨ iπ q 2 −io−1 e
√ d2 q δ(q 2 − 1)eiqx = πJ0 (r) = dψ sin(r cosh ψ)
r = x 2 ⎪
⎪ π
⎪
⎪ = dθ cos(r cos θ),
⎪
⎪ d2 q 1 0
⎩ e i
q
x
= −πN0 (r) = dψ cos(r cosh ψ),
π 2 q P −1
Those functions are combined for the indefinite case with two-dimensional
energy-momentum integrals:
⎧ d2 q √
⎪
⎪
1
iπ −q2 −io+1 e
iqx
= 2K0 ( −x2 + io)
⎪
⎪ ∞
⎪
⎪ $ 2
⎪
⎪
(− x4 )k −x2 +io
⎪
⎪
= (k!)2 [2χk − log 4 ]
⎪
⎪
⎪
⎪
k=0
⎪
⎪ = ϑ(−x2 )2K0 (|x|) − ϑ(x2 )
⎪
⎪
⎪
⎪ π[N0 + iJ0 ](|x|),
⎪
⎪
⎪
⎪
⎪
⎪ ∞
⎪
⎪ 2 $ 2
⎪ (− x )k
[2χk − log |x4 | ]
2
⎪
⎪ d q δ(q 2 − 1)eiqx = 4
⎪
⎪
(k!)2
O(1, 1) : ⎪
⎪ k=0
⎨ = ϑ(−x2 )2K0 (|x|) − ϑ(x2 )πN0 (|x|),
x = x0− x3 ,
2 2 2 d2 q 1
⎪
⎪ 2 +1 e
iqx
= −i(x0 ) d2 q (q0 )δ(q 2 − 1)eiqx
|x| = |x |, ⎪
2
⎪
π −qP
⎪
⎪
∞
$ 2
⎪
⎪ = ϑ(x2 )π
(− x4 )k 2
⎪
⎪ (k!)2 = ϑ(x )πJ0 (|x|),
⎪
⎪
⎪
⎪
k=0
⎪
⎪
⎪
⎪ d2 q √
⎪
⎪ 1 iqx
= 2K0 ( x2 + io)
⎪
⎪ iπ q2 −io+1 e
⎪
⎪ $∞
⎪
⎪
2
( x4 )k x2 +io
⎪
⎪ = (k!)2 [2χk − log 4 ]
⎪
⎪
⎪
⎪ k=0
⎪
⎪ = ϑ(x2 )2K0 (|x|)
⎪
⎩
−ϑ(−x2 )π[N0 + iJ0 ](|x|).
By analytic continuation and orthogonally invariant embedding, one ob-

tains for indefinite orthogonal groups
⎧ n
⎪ d q 2 −ν)
Γ( n Γ(ν)
⎪
⎪ n n eiqx = ,
(−q2 −io) 2 −ν
2
⎪ it π 2 ( −x 4+io )ν
O(t, s) : ⎪
⎪
t ≥ 1, s ≥ 1, ⎪
⎪
⎪
⎪ n 2 −ν)
Γ( n
n = t + s, ⎨ d qn n eiqx
it π 2 (−q2 −io+1) 2 −ν
n = 2, 3, . . . , ⎪
⎪ = ϑ(−x2 )2Kν (|x|)−ϑ(x2 )π[N−ν +iJ−ν ](|x|)
⎪
⎪ |x
2|
ν
x2 = x t−
2
x2s , ⎪
⎪
⎪
⎪ $N
|x| = |x |, 2 ⎪
⎪ −δνN iπ 1 (k−1) 2
(− x4 ).
⎩ (N −k)! δ
k=1
Equivalent formulas, also for O(t, s), are obtained by the exchange (t, s, q 2 , x2 )
↔ (s, t, −q 2 − x2 ). For integers N = 1, 2, . . . , there arise, via the phase of the
logarithm, x2 = 0 supported Dirac distributions
log(−x2 − io) = log |x2 | − iπϑ(x2 ),

k 2
− ∂x2 ϑ(x2 ) = δ (k−1) (− x4 ), k = 1, 2, . . . .
∂ 4
The residual normalizations for positive and negative invariants a are

⎧ dn q dn q
⎪ 2 −ν)
Γ( n 2 −ν)
Γ( n
⎪
⎪ n n −ν = n n
(q −io+a) 2 −ν
⎪
⎪
t
i π 2 (−q −io+a)
2 2 s
i π 2 2
O(t, s) : ⎨ Γ(−ν)
= (a−io)−ν ,
n = 1, 2, . . . , 2dn q 2dn q
⎪
⎪
1
−ν =
1
a ∈ R, ⎪
⎪ it |Ωn−1 |
(−q −io+a)
2
n
2 is |Ωn−1 | (q2 −io+a) n
2
−ν
⎪
⎩ Γ( n2 )Γ(−ν) 1
= Γ( n −ν) (a−io)−ν .
2
Hyperbolically invariant distributions are used for (1, s)-spacetime with

the general Lorentz groups,
⎧ 1+s
⎪
⎪ d q 2 −ν)
Γ( 1+s
eiqx = −xΓ(ν) 2 +io ν ,
⎪ iπ 1+s
⎪ 1+s
(−q2 −io) 2 −ν
⎪
⎪
2 ( 4 )
⎪
⎪
⎪
⎪ d1+s q
⎪
⎪ 2 −ν)
Γ( 1+s 2 2
⎪
⎪ 1+s −ν e
iqx
= ϑ(−x )2Kν (|x|)−ϑ(x )π[N−ν +iJ−ν ](|x|)
⎪ | |
1+s x ν
⎪
⎪ iπ 2 (−q2 −io+1) 2 2
⎨ $N
2
O(1, s): −δνN iπ 1
(N −k)! δ
(k−1)
(− x4 ),
⎪
⎪
⎪
⎪ 1+s k=1
⎪
⎪ 2 −ν)
Γ( 1+s ϑ(x2 )2Kν (|x|)−ϑ(−x2)π[N−ν +iJ−ν ](|x|)
⎪
⎪
d q
e iqx
=
⎪
⎪
1+s 1+s
(q2 −io+1) 2 −ν
|x2|
ν
⎪
⎪
is π 2
⎪
⎪ $N
⎪
⎪ −δνN iπ 1 (k−1) x2
⎪
⎩ (N −k)! δ ( 4 ).
k=1
With respect to hyperbolic differential equations with SO0 (1, s), Huygens’
principle with spherical SO(s)-boundary conditions holds for odd dimensions
1 + s, but not, however, for even spacetime dimensions [36].
8.13.2 Some Special Cases

Now special cases: For ν = − 21 , there are no singularities:
⎧
⎪
⎪
2d1+s q 1
eiqx = −r,
⎪
⎪ |Ω1+s | 2+s
⎨ q 2)
( 2
2d1+s q
O(1 + s) : |Ω1+s |
1
2+s eiqx = e−r ,
⎪
⎪ q 2 +1)
( 2
⎪
⎪ 2d1+s q
⎩ i|Ω1+s |
1
2+s eiqx = eir ,
q 2 −io−1)
( 2
⎧ 2d1+s q
⎪
⎪ 1
eiqx = −|x|[ϑ(−x2 ) + iϑ(x2 )],
⎪
⎪ i|Ω1+s | 2+s
⎨ 1+s (−q2 −io) 2
O(1, s) :
2d q
i|Ω1+s | 2+s e
1
iqx
= ϑ(−x2 )e−|x| + ϑ(x2 )e−i|x| ,
⎪
⎪ (−q2 −io+1) 2
⎪
⎪ 2d1+s q
⎩ is |Ω1+s |
1
2+s e
iqx
= ϑ(x2 )e−|x| + ϑ(−x2 )e−i|x| .
2 (q −io+1) 2
For ν = 0, there is a logarithmic singularity in K0 and N0 :

⎧ 2d1+s q
⎨ |Ωs |
1
1+s eiqx = 2K0 (r),
q 2 +1)
( 2
O(1 + s) :
⎩ 2d1+s q 1
eiqx = π[J0 + iN0 ](r),
i|Ωs | 1+s
q 2 −io−1)
( 2
⎧ 2d1+s q
⎪
⎪
1
eiqx
⎪
⎪
i|Ωs |
(−q2 −io+1)
1+s
⎨
2
= ϑ(−x2 )2K0 (|x|) − ϑ(x2 )π[N0 + iJ0 ](|x|),

O(1, s) :
⎪
⎪
2d1+s q 1
1+s e
iqx
⎪
⎪ is |Ωs |
(q2 −io+1)
⎩ 2
= ϑ(x2 )2K0 (|x|) − ϑ(−x2 )π[N0 + iJ0 ](|x|).
The Fourier-transformed Dirac part of the simple poles is used for repre-
sentations of the affine groups SO(1 + s) ×
R1+s and SO0 (1, s) × R1+s :
⎧
⎪ d1+s q Γ( s−12 )
⎪
⎪
1 i
2 e
q
x
= ( r2 )s−1 ,
⎪
1+s q
⎪
⎨
π 2
2K s−1 (r)
d1+s q 1 i
q
x
q 2 +1 e = ,
2
O(1 + s) : 1+s s−1
⎪
⎪ π 2 ( r2 ) 2
⎪
⎪ π[J s−1 +iN s−1 ](r)
⎪
⎩ d1+s q 1 i
q
x
q 2 −io−1 e = ,
2 2
1+s s−1
iπ 2 ( r2 ) 2
⎧ 1+s
Γ( s−1 )
⎪
⎪
d q 1
1+s −q 2 −io e
iqx
= −x2 +io2 s−1 ,
⎪
⎪
⎪
⎪
iπ 2 ( 4 ) 2
⎪
⎪
⎪
⎪
⎪
⎪ d1+s q ϑ(−x2 )2K s−1 (|x|)−ϑ(x2 )π[N− s−1 +iJ− s−1 ](|x|)
⎪
⎪ 1
e iqx
= 2 2 2
⎪
⎪ 1+s −q 2 −io+1 s−1
⎪
⎪ iπ 2 |x | 2
⎪
2
⎪
⎨ $
R−1
2
O(1, s) : −δ1+s iπ
2R 1
(R−1−k)! δ
(k−1)
(− x4 ),
⎪
⎪ k=1
⎪
⎪
⎪
⎪
⎪
⎪ d1+s q ϑ(x2 )2K s−1 (|x|)−ϑ(−x2 )π[N− s−1 +iJ− s−1 ](|x|)
⎪
⎪ 1
⎪
⎪ 1+s q 2 −io+1 e iqx
= 2 2 2
⎪
s−1
⎪ | |
sπ 2 x
i 2
⎪
⎪
2
⎪
⎪ $
R−1
⎪
⎪ −δ1+s iπ
2R 1 (k−1) x2
⎩ (R−1−k)! δ ( 4 ).
k=1
The lightcone-supported Dirac distributions arise for even-dimensional space-

time with nonflat position, i.e., for (1, s) = (1, 3), (1, 5), . . . .
The one-dimensional pole integrals are spread to odd dimensions starting

with 1 + s = 3 and with a singularity at |x| = 0:
⎧ d3 q 1 i
⎪
⎪ q2e
q
x
= 2r ,
⎨ 3
π2
−r
−r
O(3) :
d q 1
π2 q 2 +1 e
i
q
x
=− ∂
2 e = 2 er ,
∂ r4
⎪
⎩
⎪ d3 q 1 i
q
x ∂ ir ir
= 2 eir ,
q 2 −io−1 e
iπ 2 =
∂ r4
2 e
⎧ 3 2 2
⎪
⎪
d q 1 iqx
= 2 ϑ(−x )−iϑ(x )
⎨ 3iπ2 −q2 −io e |x|
2 −|x|
,
d q 1 iqx ϑ(−x )e −ϑ(x 2
)ie−i|x|
O(1, 2) : 2 −q 2 −io+1 e =2 ,
⎪ iπ
⎪ d3 q
|x|
⎩ − π2 q2 −io+11
eiqx = 2 ϑ(x 2 −|x|
)e −ϑ(−x 2
)ie−i|x|
.
|x|
The dipoles in three dimensions are without singularity:
⎧
⎪
⎪
d3 q 1
2 2e
i
q
x
= −r,
⎨ 2
d3 qπ (1q ) iqx
O(3) : q 2 +1)2 e = e−r ,
⎪ π 2 (
⎩
⎪ d3 q 1 i
q
x
= eir ,
q 2 −io−1)2 e
iπ 2 (
⎧
⎪
⎪
d3 q 1
2 (−q 2 −io)2 e
iqx
= |x|[−ϑ(−x2 ) + iϑ(x2 )],
⎨ d3iπ
O(1, 2) : q 1
2e
iqx
= ϑ(−x2 )e−|x| + ϑ(x2 )e−i|x| ,
⎪
⎪ iπd3 q(−q −io+1)
2 2
⎩ − π2 (q2 −io+1)
1 iqx
= ϑ(x2 )e−|x| + ϑ(−x2 )e−i|x| .
2e
The two-dimensional integrals are spread to even dimensions, starting

with 1 + s = 4:
⎧ d4 q 1 iqx
⎪
⎪ 2e = r42 ,
⎨ d4 qπ 1q iqx
2
2K1 (r)
O(4) : q 2 +1 e
π2 = − ∂r2 2K0 (r) = r ,
⎪ ∂ 4
⎩ d42q 2 1
2
⎪ π[J1 +iN1 ](r)
iπ q −io−1 e
i
q
x
= − ∂r2 π[J0 + iN0 ](r) = r ,
∂ 4 2
⎧ d4 q
⎪
⎪ 1
2 −q 2 −io e
iqx
= 4
⎪
⎨ d4iπ −x2 +io ,
ϑ(−x2 )2K1 (|x|)+ϑ(x2 )π[N1 +iJ1 ](|x|) 2
O(1, 3) :
q 1
iπ 2 −q2 −io+1 e
iqx
= |x| − iπδ( x4 ),
⎪
⎪ 2
⎪ 4
⎩ − diπ2q q2 −io+1
1
eiqx = ϑ(x2 )2K1 (|x|)+ϑ(−x2 )π[N1 +iJ1 ](|x|) 2
− iπδ( x4 ).
|x|
2
Dipoles for four spacetime dimensions lead to maximally logarithmic

singularities.
8.14 Hypergeometric Functions

A hypergeometric function,
⎧ $∞
⎪
⎪ Γ(γ) z k
⎪
⎪ 0 F1 (γ; z) = Γ(γ+k) k! ,
⎪
⎪
⎪
⎪ k=0
⎨ $∞
Γ(α+k) Γ(γ) z k
α, γ, z ∈ C : 1 F1 (α; γ; z) = Γ(α) Γ(γ+k) k! ,
⎪
⎪
⎪
⎪ k=0
⎪
⎪ $∞
⎪
⎪ Γ(α1 +k) Γ(α2 +k) Γ(γ) z k
⎩ 2 1 1 2
F (α , α ; γ; z) = Γ(α1 ) Γ(α2 ) Γ(γ+k) k! ,
k=0
is in the case of at least one nonpositive integer αj , e.g., α1 ∈ {−1, −2, . . . },

a polynomial of degree |α1 |. Functions of type 1 F1 were used for quantum
mechanical wave functions (see Chapter 4).
The spherical Bessel functions for the Euclidean spaces Rs are hypergeo-
metric functions of type 0 F1 :
J s−2 (|P |r) dω s−1
Γ( 2s ) 2
s−2 = |Ωs−1 | ei|P |ωs−1
x
= 0 F1 ( s2 ; z)
2
|P |r
2
∞
$ Γ( s2 ) z k 2 2
= Γ( 2s +k) k! , z = − P 4r .
k=0
The Legendre functions for the hyperboloids Y s are hypergeometric functions

of type 2 F1 with λ = iQ − s−1 2 :
π
Pλ (cosh ψ) = −π dϕ 2π (cosh ψ + cos ϕ sinh ψ)
λ
∞
$ Γ(λ+k+1) (−z)k
= 2 F1 (1 + λ, −λ; 1; z) = Γ(λ−k+1) (k!)2 , z = − sinh 2 .
2 ψ
k=0
The Legendre polynomials for the 2-sphere Ω2 are the compact partners with
the transition λ → L = 0, 1, . . . and ψ → iθ:
π
PL (cos θ) = −π dϕ 2π (cos θ + i cos ϕ sin θ)
L
$L
(L+k)! (−z)k
= 2 F1 (1 + L, −L; 1; z) = 2 θ
(L−k)! (k!)2 , z = sin 2 .
k=0
Precisely for the minimal nonabelian case s = 2, the Legendre Y 2 -

functions are real:
2 , Q ∈ R : λ(1 + λ) ∈ R
λ = iQ − s−1 ⇐⇒ s = 2 ⇒ −λ(1 + λ) = Q2 + 14 ,
R 2 F1 ( 2 + iQ, 2 − iQ; 1; z) = PiQ− 2 (cosh ψ) = P−iQ− 2 (cosh ψ)
1 1 1 1
$∞
zk
= αk (Q) (k!) 2, z = − sinh2 ψ2 ,
k=0
Γ(iQ+ 1 +k)
;k
(2n−1)2
with αk (Q) = (−1)k Γ(iQ+ 21 −k) = [Q2 + 4 ].
2
n=1
8.14 Hypergeometric Functions 253
In the Inönü-Wigner contraction SO0 (1, 1) → R (“flattening”),

cosh ψ Q sinh ψ
SO0 (1, 1) Λ(ψ) = 1 sinh ψ cosh ψ ,
Q
limQ→∞ Λ(ψ) = 10 P1x ∈ eR ∼ = R with ψ = PQx ,
similarly limQ→∞ SO0 (1, s) = SO(s) × Rs , with ψ = PQr ,
all hyperbolic Legendre functions lead to Bessel functions for trivial index:
Γ(iQ+ρ+k) P 2 r2
limQ→∞ 1
(−Q2 )k Γ(iQ+ρ−k)
= 1, limQ→∞ Q2 sinh2 ψ
2 = 4 for ψ = Pr
Q ,
limQ→∞ P iQ+ρ
(cosh ψ) = 0 F1 (1; P r) = J0 (P r)
∞
$
π (P r)k
= −π dϕ2π e iP r cos ϕ
= (k!)2 .
k=0
There is the connection of Macdonald (Bessel, Neumann) functions for

ν ∈ R with exponentially multiplied hypergeometric functions,
√
C z −→ 2Kν (z)= + 4π(2z)ν 1 G1 ( 12 + ν; 1 + 2ν; 2z)e−z
= Γ(−ν)( z2 )ν 1 F1 ( 12 + ν; 1 + 2ν; 2z)
,
+ Γ(ν) ( z2 )−ν 1 F1 ( 12 − ν; 1 − 2ν; 2z) e−z
< ∞ ∞
=
√ $ Γ( 1 +ν+k) (2z)k+ν $ Γ( 12 −ν+k) (2z)k−ν
π
= sin πν − 2
Γ(1+2ν+k) k! + Γ(1−2ν+k) k! e−z ,
k=0 k=0
where Γ(ν)Γ(1 − ν) = π
sin πν and
Γ(1−γ) Γ(γ−1) 1−γ
1 G1 (α; γ; z) = Γ(α−γ+1) 1 F1 (α; γ; z) + Γ(α) z 1 F1 (1 + α − γ; 2 − γ; z)
1
∞ −Qz α−1
= Γ(α) 0 dQ e Q (1 + Q)γ−α−1 if Re α > 0.
The functions 1 F1 (α; γ; z) = 1 F1 (γ − α; γ; z)ez and 1 G1 (α; γ; z) are lin-

early independent solutions of the confluent hypergeometric equation
[zd2z + (γ − z)dz − α]f (z) = 0.

Chapter 9
Convolutions and Product

Representations
The Feynman integrals in special relativistic quantum field theories involve

convolutions of energy-momentum distributions. The on-shell parts for trans-
lation representations give product representation coefficients of the Poincaré
group, i.e., energy-momentum distributions for free states (multiparticle mea-
sures, discussed ahead). The off-shell interaction contributions (“virtual par-
ticles”) are not convolutable; this is the origin of the “divergence” problem in
quantum field theories with interactions. With respect to Poincaré group rep-
resentations, the convolution of Feynman propagators embedding the point-
−mr
wise product of interactions, e.g. ( e r )2 , makes no sense.
The pointwise product algebra of the essentially bounded complex func-
tions of a real Lie group, L∞ (G) = L∞ (G) · L∞ (G), characterizes its rep-
resentations. The cone of positive-type functions d = dˆ ∈ L∞ (G)+ induces
the scalar products for cyclic Hilbert representations (see Chapter 8). Its
conjugation property d ↔ d− = d connects dual representations. The point-
wise product of two positive-type functions is a positive-type function for the
product representation:
d1 · d2 (g) = c1 |D1 (g)|c1 c2 |D2 (g)|c2 = c1 , c2 |D1 ⊗ D2 (g)|c1 , c2 .
With the trivial representation and its constant positive type function d1 = 1
as unit for the pointwise product, the cyclic Hilbert representation classes
L∞ (G)+ constitute a monoid.
The energies for time translations R and the momenta for position trans-
lations R3 are, as eigenvalues, the characters q ∈ Řn (representation classes,
dual group) of the additive group Rn . The Radon (energy-)momentum mea-
sures are a convolution algebra, which reflects the pointwise multiplication
256 Chapter 9 Convolutions and Product Representations
property of the essentially bounded function classes:

L∞ (Rn ) · L∞ (Rn ) = L∞ (Rn ), M(Řn ) ∗ M(Řn ) = M(Řn ),
dn q
d1 · d2 (x) = (2π) ˜ ˜
n d1 ∗ d2 (q)e
iqx
.
Product representations of Rn come with the product of representation coef-

ficients, i.e., with the convolution ∗ of (energy-)momentum distributions:
∗ ∼ δ(q1 + q2 − q).
The convolution adds (energy-)momenta of singularity manifolds, which sup-

port imaginary and real eigenvalues for compact and noncompact represen-
tation invariants.
After exemplifying a composite structure with spacetime product rep-
resentations by a Nambu–Goldstone field in the case of a chirally degener-
ate ground-state, this chapter considers the (energy-)momentum convolution
structure of time, position, and spacetime representations.
9.1 Composite Nambu–Goldstone Bosons

Electrodynamics and gravity, in the flat spacetime approach, are implemented
by massless fields. Also, a degenerate ground-state (“spontaneous symmetry
breakdown”) comes with long-range interactions, characterized, qualitatively,
by the “broken” symmetries of a Lie group G, i.e., by the degeneracy-effecting
operations in the classes G/H with respect to the remaining “unbroken” sym-
metry for a distinguished subgroup H ⊆ G, and, quantitatively, by a dilation
scale, a “breakdown mass.” The long-range interactions come in representa-
tions of the “unbroken” local group H.
As illustrated, e.g., by superfluid helium III, a dynamics, invariant under
a group G, can have different ground-states (phases) with degeneracy orbits
G/Hι for different invariance groups Hι ⊆ G. This is in some analogy to dif-
ferent boundary conditions for a classical dynamics or different cosmological
models (see Chapter 1).
A “dynamical breakdown” establishes the degeneracy-implementing mass-
less fields as product representations of spacetime. In the model of Nambu
and Jona-Lasinio, the massless chiral Goldstone boson for a U(1)-degeneracy
(“breakdown”) is a Lorentz pseudoscalar bound state of self-interacting mas-
sive Dirac fermions.
9.1.1 Chirality
The relative phase of the SL(C2 )-irreducible left- and right-handed Weyl
spinors in a Dirac spinor is affected by the chiral transformations,
α α
1e4 +iγ5Ψ, Ψ −
U(1) : Ψ −→ γ5 2
→
γ5 2
Ψe , i α
l e 2l
Ψ = 14 −iγ 5 Ψ = r −→ .
2
−i α
2 e 2r
9.1 Composite Nambu–Goldstone Bosons 257
The 16 = 1+4+6+4+1 elements of the Clifford algebra {γa , γb } = 2ηab 14

for Minkowski spacetime, represented by bilinear Dirac spinors, involve five
different couplings:
abcd
14 , γa , γab = − 41 [γa , γb ], γ5 γa , γ5 = 4! γa γb γc γd ,
with {γa , γ5 } = 0, γ52 = −14 ,
with vector γa and axial vector γ5 γa as chiral invariants, and two chiral
SO(2)-dublets, given by scalar and pseudoscalar (14 , γ5 ), and by the tensor
via (γab , γ5 γab = 12 abcd γ cd ), e.g.,
∗ iα ∗
U(1) : Ψ 14 Ψ
Ψ γ5 Ψ
−→ −cos α sin α
sin α cos α
Ψ 14 Ψ
Ψ γ5 Ψ
, Ψ 14 ±iγ
2
5
Ψ = lr∗ rl −→ ee−iα rl∗ rl .
The five quartic Lorentz scalar couplings,

⎛ ⎞ ⎛ ⎞
14 ⊗ 14 s
⊗ γa
⎜ γa
⎟ ⎜v ⎟
⎝ γab ⊗ γ ab ⎠ = ⎝ t ⎠,
γ5 γa ⊗ γ5 γ a a
γ5 ⊗ γ5 p
have, by the involutive Fierz recoupling1 (via index exchange M N ↔ N M ),

two Fierz-symmetric and three Fierz-antisymmetric linear combinations as
eigenvectors:
⎛ ⎞MN ⎛ ⎞N M
s s
⎜vt ⎟ Fierz⎜v ⎟
⎝ ⎠ = F ⎝t⎠ ,
a a
p p
KL ⎛ KL ⎞
1
4
1
4 − 18 1
4 − 14
− 12 1
⎜ 1 0 2 1
⎟
F = ⎝ −3 0 − 12 0 3 ⎠ , F 2 = 15 ,
1 1
2 0 − 12 1
− 14 1 1 1 1
⎛ ⎞ ⎛ 4 8
⎞4 4
s − p − 6t s − p − 6t
s − p + 2t −(s − p + 2t )
⎜ ⎟ Fierz ⎜ ⎟
⇒⎝ s + p + v+a
2 ⎠ ↔ ⎝ s + p + v+a
2 ⎠.
s + p − v+a
2 −(s + p − v+a
2 )
v−a −(v − a)
With the chiral invariants,

U(1) : (v, a, s + p) −→ (v, a, s + p),
one can combine the chiral invariant 4-fermion couplings. The two Fierz anti-
symmetric ones {s + p − v+a2 , v − a} are nontrivial for Fermi Dirac fields. The
chiral invariant coupling with a scalar term s + p − v+a2 is used in the nonlin-
ear interaction of the model of Nambu and Jona-Lasinio with the equation
of motion:
γa ⊗γ a +γ5 γa ⊗γ5 γ a
Ψ = 12 ΓΨ
iγ∂Ψ Ψ(Ψ Ψ), with Γ ⊗ Γ = 14 (14 ⊗ 14 + γ5 ⊗ γ5 −
ΨΓΨ 2 ).
1 The numbers in the 5 × 5-matrix are Lorentz group recoupling coefficients analogous
to 9j-symbols for the rotation group.

This quartic interaction is distinguished as the square of the radial part

R2 = ΦΦ ∗ in the scalar and pseudoscalar Dirac field product, which yield
the analogue of the U(1) × D(1)-factorization of a basic chiral Higgs field
Φ = r∗ l = eiα
α
R:
⎧
⎪ 2
⎨ R = (Ψ Ψ 14 +iγ
2
5
Ψ 14 −iγ
Ψ )(Ψ 2
5
Ψ ) = (r∗ l)(l∗ r)
14 ±iγ5
Ψ 2 Ψ = e±iα R ⇒α 1
ΨΓΨ
= 2 (Ψ Ψ)(Ψ ΨΓΨ Ψ),
⎪
⎩ ±iα α
1 ±iγ
Ψ 4 2 5Ψ
e = R .
It arises in the corresponding classical Lagrangian:
Ψ) = iΨ
L(Ψ Ψ − 14 (Ψ
Ψγ∂Ψ ΨΓΨ
Ψ)(Ψ
ΨΓΨ
Ψ).
9.1.2 Chiral Degeneracy

An interaction-free Dirac field has a Fock ground-state vector |0 for its
Feynman propagator:
d4 q γq+m iq(x−y)
0|Ψ Ψ(x)|0Feynman = ρ(m2 ) πi (2π)
Ψ(y)Ψ 3 q 2 +io−m2 e = dm (x − y).
An interaction determines the normalization of the Poincaré group represen-
tation: A coupling constant g0 Γ ⊗ Γ can be absorbed into the normalization
ρ(m2 ) −→ g0 ρ(m2 ).
A nontrivial mass term m = 0 characterizes a chirally U(1)-degenerate
ground-state. The local group (unbroken subgroup, fixgroup of the ground-
state orbit) is trivial, G/H = U(1)/{1}. Particles are chiral singlets. The non-
linear field equation yields, in a first-order approximation, a self-consistency
condition (“gap equation”) for the mass m as the chiral breakdown parameter:
Ψ = −ΓΨ
iγ∂Ψ Ψ tr Γdm (0) + 12 : ΓΨ
Ψ(Ψ Ψ) := −mΨ
ΨΓΨ Ψ + ...
d4 q ρ(m2 )
⇒ m = 4 tr 14 d (0) = m π (2π)3 q2 +io−m2 .
1 m i
Here, the distribution dm of the spacetime translations, used for a free par-
ticle, is taken for the trivial translation x = 0. With the volume of the mass
hyperboloid d4 q δ(q 2 − 1) = |Y 3 |, it is “divergent”; dm (0) does not make
sense. A perturbative approach with free fields is inappropriate for a bound-
state problem. The model of Nambu and Jona-Lasinio is nonrenormalizable.
The Fock ground-state vector |0 for free fields in flat spacetime has to
be replaced by a ground-state vector |C from a chiral ground-state manifold
U(1) for fields with an interaction. Such a ground-state will be implemented
in the form of a regularization, e.g., by a dipole at mass M 2 in the regularized
Feynman propagator:
2
C|Ψ Ψ(x)|CFeynman = dM ,m (x − y)
Ψ(y)Ψ
d4 q (m2 −M 2 )2 γq+m
= ρ(m2 ) πi (2π) 3 (q 2 +io−M 2 )2 q 2 +io−m2 e
iq(x−y)
.
Other regularizations are possible.
9.1 Composite Nambu–Goldstone Bosons 259
With respect to the spacetime translation representations, the dipole reg-

ularization involves a ghost pair (Witt pair) with an indefinite U(1, 1)-metric
(see Chapter 4). It cannot be interpreted as a particle:
(m2 −M 2 )2 m −M 2 2
2 − q 2 −M 2 − (q 2 −M 2 )2
1 1
(q2 −M 2 )2 (q2 −m2 ) = q2 −m
→ q2 −m
1
2 for M
2
→ ∞.
The modification of the free-field propagator for flat spacetime by the regular-
2
ization, here by the dipole with the “flattening” (contraction) dM ,m → dm for
M 2 → ∞, will be related to the transition to representations of curved space-
time. This connects a nontrivial curvature and interaction (see Chapter 11).
The dipole-regularized Feynman propagator can be used as a function
2
of the spacetime translations R4 x −→ dM ,m (x), defined for the trivial
translation in the modified consistency equation,
2
m14 = C|Ψ Ψ (x)|CFeynman = dM ,m (0)
Ψ(x)Ψ
d4 q (m2 −M 2 )2 ρ(m2 )
= m14 πi (2π) 3 (q 2 +io−M 2 )2 q 2 +io−m2
2 2
= m14 ρ(m )
8π 2 (M − m − m log m2 ).
2 2 2 M
For a chiral degeneracy, the consistency equation is a representation nor-

malization. It determines the ratio of the chiral breakdown mass and the
regularization mass in terms of the normalization factor ρ(m2 ):
1 M 2 ,m M2 2
8π 2
m = 0 : md (0) = 14 ⇒ m2 − 1 − log M
m2 = m2 ρ(m2 ) .
The normalization factor is trivial for infinite regularization mass.
9.1.3 Massless Chiral Boson

The Green’s distribution of the free Dirac equation with mass,
d4 q γq+m iqx
(iγ∂ + m)κm (x) = δ(x) ⇒ κm (x) = − (2π) 4 q 2 +io−m2 e ,
is not a function. Only for a free theory (flat spacetime) can the Feynman
propagator be identified, up to a constant factor, with the Green’s distri-
bution dm = −2iρ(m2 )κm . In the regularization used, it is related to the
Feynman propagator of the interacting Dirac field by a convolution with a
dipole regulator:
2 2 2
dM ,m (x) = −dM ∗ κm (x) = − d4 y dM (x − y)κm (y),
2 d4 q (m2 −M 2 )2 iqx
dM (x) = ρ(m2 ) πi (2π) 3 (q 2 +io−M 2 )2 e .
The Green’s distribution inverts the free-field differential operator:

1
Ψ =
(iγ∂ + m)Ψ 2
Ψ(Ψ
: ΓΨ ΨΓΨΨ) :,
Ψ (x) = 1
2 d4 y κm (x − y) : ΓΨ
Ψ(Ψ
ΨΓΨ
Ψ)(y) : .
The double-dot prescription : · · · : is defined by the subtraction of the mass

term as the leading bilinear ground-state contribution. The bilinear local
products of the Dirac field for bosonic matrix elements with a state vector |B,
Γ(x) = C|Ψ Ψ(x)|B for Γ ∈ {14 , γa , γ5 γa , γ5 },

ΨΓΨ
have the equations of motion

Γ1 (x) = 12 d4 y C|Ψ
Ψ(x)Γ1 κm (x − y) : ΓΨ
Ψ(Ψ Ψ)(y) : |B.
ΨΓΨ
They can be linearized with the Feynman propagator of the interacting Dirac
field to yield first-order eigenvalue equations,

Γ1 (x) = κΓΓ21 ∗ Γ2 (x) = d4 y κΓΓ21 (x − y) Γ2 (y),
with the tangent kernel matrix κ (see 10):
2 2
κΓΓ21 (x) = − tr (Γ1 ⊗ Γ2 ) ◦ (κm ⊗ dM ,−m
)(x) = − tr Γ1 κm (x)Γ2 dM ,−m
(x).
The corresponding convolution product for the energy-momenta (translation
eigenvalues) distributions gives the distributions for eigenvalues of the prod-
uct representations:
Γ̃1 (q) = κ̃ΓΓ21 (q)Γ̃2 (q), [δΓΓ12 − κ̃ΓΓ21 (q)]Γ̃2 (q) = 0,

2
with κ̃ΓΓ21 (q) = − tr (Γ1 ⊗ Γ2 ) ◦ (κ̃m d˜M ,−m )(q)
d4 p γ(p−q)+m (m2 −M 2 )2 γp+m
= ρ(m2 ) πi tr (2π) 3 Γ1 (p−q)2 +io−m2 Γ2 (p2 +io−M 2 )2 p2 +io−m2 .
The convolution contains a tensor product for Clifford algebra elements.

The product κm (x) ⊗ dm (−x) = −2iρ(m2 )κm (x) ⊗ κ−m (x) of the Green’s
distribution and the free-field propagator is not defined because of the dis-
tributional off-shell contributions (more ahead). With the regularization for
the quantization of the interacting Dirac field, the product is defined.
The scalar product and pseudoscalar product (Ψ Ψ14Ψ , Ψ γ5Ψ ) constitute a
chiral dublet. The degenerate ground-state is characterized by the particle
mass,
C|Ψ
Ψ 14 +iγ
2
5
Ψ 14 −iγ
Ψ |CC|Ψ 2
5
Ψ |C = 4m2 ;
i.e., the degeneracy manifold is a circle Ω1 = SO(2) ∼ = U(1) with radius 2m.
A scalar ground-state, defined by
C|Ψ
Ψ14Ψ |C = 4m,
C|r∗ l|C = C|l∗ r|C = 2m,
C|Ψ
Ψγ5Ψ |C = 0,
strips the particles of the chiral degree of freedom. In a first-order approxi-
mation, the pseudoscalar is the infinitesimal chiral field, i.e., the Goldstone
degree of freedom,
Ψ
Ψ (14 +iγ5 )Ψ
α
eiα = 1 + iαα + ··· = 2R = 1 + i Ψ4m
γ5 Ψ
+ . . . , with
⇒ Ψ γ5 Ψ
α = 4m + . . . ,
9.2 Convolutions for Abelian Groups 261
for the degeneracy transformation of the ground-state vector |C. The equa-
tion of motion for the composite Nambu–Goldstone field as the corresponding
matrix element with the “pion” |π is given as follows:
γ5 (x) = C|Ψ Ψγ5Ψ (x)|π,
γ̃5 (q) = κ̃γγ55 (q 2 )γ̃5 (q), [1 − κ̃γγ55 (q 2 )]γ̃5 (q) = 0,
2 d4 p (m2 −M 2 )2
with κ̃γγ55 (q 2 ) = ρ(m
4
) i
π tr
γ(p−q)+m γp+m
(2π)3 γ5 (p−q)2 +io−m2 γ5 (p2 +io−M 2 )2 p2 +io−m2 .
With γ5 (γp + m)γ5 = γp − m, it has a mass zero q 2 = 0 solution κ̃γγ55 (0) = 1

if compared with the consistency equation for the chiral breakdown, m = 0:
d4 q M 2 ,m
d4 q (m2 −M 2 )2 ρ(m2 )
1
4m tr (2π)4 d
˜ (q) = 1,
i
(2π)3 (q2 +io−M 2 )2 q2 +io−m2 = d4 q −m 2
π 1
tr 4 κ̃
4 (q)d˜M
(2π)
,m
(q) = κ̃γ5 (0). γ5
The consistency equation for the fermion mass as the degeneracy parameter
and the eigenvalue equation for the massless field coincide.
The expansion at the mass zero solution determines the normalization
ρ(0) of the corresponding spacetime translation representation as the residue
at the pole:
q2 ρ(0) ∂κ̃γ5
γ
1
1−κ̃γ55 (q2 )
= 1
q2 1−κ̃γ5 2 = q2 + . . . with 1
ρ(0) =− γ5
∂q2 (0).
γ5 (q )
The characteristic structures of the chiral model do not depend on a “per-

turbative” expansion. They are given by the relation between the normaliza-
tion, in the “gap” equation, of the spacetime representation by the regularized
Feynman propagator and the ensuing massless solution via the singularity of
the resolvent 1−κ̃γ15 (q2 ) with the normalized kernel κ̃γγ55 . This will be discussed
γ5
in more detail in Chapter 10.
9.2 Convolutions for Abelian Groups

Product representations of translations Rn with sum and difference of the
energy-momentum invariants arise by the pointwise product of positive-type
dn x
functions L∞ (Rn )+ = Cb (Rn )+ or the convolution of positive energy-
momentum Radon measures M(Řn )+ .
The simplest case is given for one-dimensional translations, e.g., for time
translations t ∈ R with an addition of the energy invariants in the irreducible
and self-dual representations:

eim1 t · eim2 t = eim+ t ,
Cb (R)+ · Cb (R)+ = Cb (R)+
2 cos m1 t · 2 cos m2 t = 2 cos m+ t + 2 cos m− t
with m± = m1 ± m2 ,
⎧
⎨ δ(q − m1 ) ∗ δ(q − m2 ) = δ(q − m+ ),
M(Ř)+ ∗ M(Ř)+ = M(Ř)+ 2|q|δ(q 2 − m21 ) ∗ 2|q|δ(q 2 − m22 )
⎩
= 2|q|δ(q 2 − m2+ ) + 2|q|δ(q 2 − m2− ).
9.2.1 Convolutions with Linear Invariants

The convolution product for the two causal function algebras, conjugate and
orthogonal to each other, and the Dirac convolution algebra is summarized
with the residually normalized representation functions and the integration
contours:
Causal time D(1) and energies R
1 ∗
(∗, q) = (± 2iπ , q ∓ io) causal, orthogonal
ϑ(±t)eimt = ± dq 1 iqt
2iπ q∓io−m e , 1
q−m1 ∗ q−m2
1 1 1
= q−m+
δ(q − m1 ) ∗ δ(q − m2 ) = δ(q − m+ )
The normalization factor for the convolution product (residual normalization)

is the 1-sphere measure as used in the residue:
. dq ∗ ∼ 1
2iπ = res, 2π = |Ω1 | δ(q1 + q2 − q).
There is the more general convolution

Γ(1+ν1 ) 1 Γ(1+ν2 ) Γ(1+ν1 +ν2 )
(q−m1 )1+ν1
∗ (q−m 2)
1+ν2 = (q−m+ )1+ν1 +ν2
,
∂ N
which generalizes the integer-power derivatives ( ∂m ) for nontrivial nildimen-
sions N = 1, 2, . . . to real powers ν ∈ R wherever the Γ-functions are defined.
9.2.2 Convolutions with Self-Dual Invariants

The causal distributions with compact dual invariants

± iπ
1
(q∓io)2 −m2 = |m|δ(q − m ) ± iπ q2 −m2 = ± 2iπ q∓io−|m| +
q 2 2 1 q 1 1 1
q∓io+|m|
P
keep the property of constituting orthogonal convolution algebras, conjugate

to each other:
Causal time D(1) and energies R

1 ∗
ϑ(±t)2 cos mt = ± dq q
iπ (q∓io)2 −m2
eiqt , (∗, q ) = (± iπ
2
, (q ∓ io)2 ) causal, orthogonal
1
q
∗ 2 q 2 = q
+ q
q2 −m2
1 q −m2 q2 −m2+ q2 −m2−
Since the Feynman energy distributions combine advanced and retarded

distributions,

|m| |m|
± iπ
1
q2 ∓io−m2 = |m|δ(q − m ) ± iπ q2 −m2 = ± 2iπ q∓io−|m| − q±io+|m| ,
2 2 1 1 1 1
P
they constitute convolution algebras, conjugate to each other, however not

orthogonal, e+i|m1 t| · e−i|m2 t| = 0:
Bicone time R+ R− and energies R
±i|mt|
|m| 1 ∗
e =± dq iqt
iπ q2 ∓io−m2 e ,
(∗, q2 ) = (± iπ , q2 ∓ io) Feynman, not orthogonal
|m+ |
|m1 | 1 |m |
∗ 2 2 2 =
q2 −m2 1 q −m2 q2 −m2
+
9.3 Convolutions for Position Representations 263
The faithful Hilbert representations of Y 1 ∼ = SO0 (1, 1) ∼

= R (one-
dimensional abelian position) with Fourier-transformed Ω1 -measures and
noncompact dual invariants constitute a convolution algebra:
Position Y 1 and “momenta” R

dq |m|
e−|mz| = −iqz 1 ∗
|Ω1 | = 2π, ∗
π q2 +m2 e , = π
|m1 | 1 |m2 | |m+ |
∗ =
q2 +m2 1 q2 +m2 2 q2 +m2
+
9.3 Convolutions for Position Representations

Representations of Euclidean, spherical, and hyperbolic spaces are charac-
terized by singularity spheres with real momenta (imaginary eigenvalues)
for free and scattering structures, and imaginary “momenta” (real eigenval-
ues) for bound structures. The convolution of the related “momentum” func-
tions reflect pointwise multiplications of Macdonald and Hankel (Bessel with
Neumann) functions (see Chapter 8):
⎧ s
Γ( 2s −ν)
⎪ d q
⎪ iπ s2 (q 2 −io−1) s −ν e
i
q
x
= π[Jν(+iN ν ](r)
, spherical,
⎪
⎪
r ν
2)
⎨
2
ds q J s−2 (r)
O(s) : s δ( q 2 − 1)eiqx = r 2 s−2 , Euclidean,
⎪
⎪ π2 ( 2)

2
⎪
⎪ s
−ν)
⎩
s Γ(
d q
s
2
2
s −ν e
i
q
x
= 2K ν (r)
( r )ν , hyperbolic.
2 π (
q +1)
2 2
9.3.1 Convolutions for Euclidean Spaces

Interaction-free product structures convolute Dirac distributions for cyclic
translation representations. In contrast to the convolution of Dirac distribu-
tions for self-dual invariants with basic spherically self-dual two-dimensional
representations,
abelian R : 2|q|δ(q 2 − P12 ) ∗ 2|q|δ(q 2 − P22 )

= 2|q|δ(q 2 − P−2 ) + 2|q|δ(q 2 − P+2 ),
with P± = |P1 | ± |P2 |,
the convolution of Dirac distributions for the infinite-dimensional represen-

tations of the Euclidean groups, s ≥ 1, with the sphere radii as momentum
q 2 = P 2 > 0 leads to position translation representations with the
invariants
momentum sphere radii between the invariants, P−2 ≤ q 2 ≤ P+2 ,
∗
SO(1 + s) × q 2 − P12 ) |Ωs−1
R1+s : δ( q 2 − P22 )
| δ(
s−2
[−Δ(q 2 )] 2
s = 1, 2, . . . , = q |)s−1
(2| ϑ(P+2 − q 2 )ϑ(q 2 − P−2 ).
The convolution product is normalized with the (s − 1)-sphere. There arises a

momentum-dependent normalization factor that contains the characteristic
two-particle convolution function:
q 2 ) = Δ(
Δ( q 2 − P+2 )(q 2 − P−2 ).
q 2 , P12 , P22 ) = (
It is symmetric in the three invariants involved:
Δ(a, b, c) = a2 + b2 + c2 − 2(ab + ac + bc) = (a + b − c)2 − 4ab.
The minimal cases s = 1, 2 are characteristic for even- and odd-dimensional
positions:
Scattering in two dimensions involves the pointwise product of Bessel
functions J0 ,
SO(2) × q 2 − P12 ) ∗2 δ(
R2 : δ( q 2 − P22 )
= √ 2 2 2 2 ϑ(P+2 − q 2 )ϑ(q 2 − P−2 ),
2
(P+ − q −P− )
q )(
d2 q
d2 x
J0 (r) = π q 2 − 1)eiqx ,
δ( 4π J0 (r)e
i
qx
= δ(q 2 − 1),
π
J0 (|P1 |r)J0 (|P2 |r) = dθ
0 π2 0
J (|P |r) with P 2 = P12 + P22 − 2P1 P2 cos θ

1 P+ √ dP 2
= π P− 2 2 −P 2 )(P 2 −P 2 )
J0 (|P |r) for P1 P2 = 0.
(P+ −
One obtains for the SO(3) ×

R3 scattering representations in three di-
mensions:
Cb (R3 )+ · Cb (R3 )+ = Cb (R3 )+ : sin P1 r
r ·sin P2 r
r = cos P− r−cos
2r 2
P+ r
,
∗
M(Ř3 )+ ∗ M(Ř3 )+ = M(Ř3 )+ : δ(
q 2
− P1 ) 2π δ(q − P2 )
2 2 2
q| ϑ(P+ − q 2 )ϑ(q 2 − P−2 ).

2 2
= |
The square of a representation is a normalized positive-type function:

∗
s = 3 : ( sinP Pr r )2 = 1−cos 2P r
2(P r)2 , q 2 − P 2)
δ( 2π δ(q 2 − P 2 ) = q | ϑ(4P −
2
|
2
q 2 ).
9.3.2 Convolutions for Odd-Dimensional Hyperboloids

Cyclic representations of the even-dimensional Lorentz group SO0 (1, 2R−1),
2R = 2, 4, . . . , for the hyperboloid Y 2R−1 ∼= SO0 (1, 2R − 1)/SO(2R − 1) with
real rank 1 and noncompact invariant q 2 = −m2 < 0 are characterized by
d2R−1 x
continuous positive-type functions Cb (Y 2R−1 )+ ∼ = L∞ (Y 2R−1 )+ :
2R−1 q
Y 2R−1 ∼= R2R−1 x −→ 2d |Ω2R−1 | (
|m|
q 2 +m2 )R e
i
qx

= e−|m|r .
Nontrivial properties for the maximal compact group, the rotations SO
(2R − 1), R ≥ 2, arise by derivations ∂∂
x ∼ i
q . These Lorentz group represen-
tations start from the characteristic hyperbolic exponentials for the maximal
noncompact abelian subgroup with imaginary singularities q = ±im:

SO0 (1, 1) ∼
= R x −→ dq
|m|
π q2 +m2 e
−iqx
= e−|mx| .
9.3 Convolutions for Position Representations 265
The representations are faithful and cyclic, but not irreducible. They
are square-integrable and not Schur-orthogonal for different invariants
m21 = m22 .
Product representations e−|m1 |r · e−|m2 |r = e−|m+ |r convolute the positive
momentum measures. The measure of the associated compact unit sphere
Ω2R−1 is used for the residual normalization (more on the normalization
ahead). The representations of three-dimensional hyperbolic position Y 3 use
the Fourier-transformed Ω3 -measure, familiar from the nonrelativistic hydro-
gen Schrödinger functions (see Chapter 8). The radii of the “momentum”
spheres as invariants are added up in the convolution
Position Y 3 ∼
= SO0 (1, 3)/SO(3)
d3 q |m|
and “momenta” R3 with SO(3)
e−|m|r = q 2 +m2 )2 e
π 2 (
i
qx

, |Ω3 | = 2π 2 ,
3
∗ = ∗
π2
|m1 | 3 |m2 | |m+ |
q 2 +m2
( 2 ∗ (
q 2 +m2 2 =
q 2 +m2 2
1) 2) ( +)
In general, the representations of odd-dimensional hyperboloids Y 2R−1

come with Fourier-transformed Ω2R−1 -measures and imaginary singularity
sphere Ω2R−2 for the “momentum” eigenvalues. The sphere measures can be
obtained by invariant momentum derivatives:
(− ∂∂q 2 )R−1 q 2|m|

+m2 = Γ(R) (
|m|
q 2 +m2 )R
, R = 1, 2, . . . .
Product representations arise by the convolution with the sphere volume as

residual normalization:
2R−1 q |m|
e−|m|r = 2d q 2 +m2 )R e
|Ω2R−1 | (
i
q
x
,
Position Y 2R−1 ∼
= SO0 (1, 2R − 1)/SO(2R − 1), 2R − 1 = 1, 3, . . . ,
and “momenta” R2R−1 with SO(2R − 1)
2π R 2R−1 ∗ 2
|Ω2R−1 | = Γ(R) , ∗ = |Ω2R−1 |
|m1 | 2R−1 |m2 | |m+ |
∂ L1
( ∂
q) ∗ ∂ L2
( ∂
q)
∂ L1 +L2
= ( ∂
q)
q 2 +m2
( 1)
R q 2 +m2
( 2)
R q 2 +m2
( +)
R
for L = 0, 1, . . . .
∂
Via momentum derivatives ∂ q , the convolutions may involve tensor products
for SO(2R − 1)-representations.
9.3.3 Convolutions for Odd-Dimensional Spheres

The representations of odd-dimensional spheres use a singularity sphere
Ω2R−2 with real momentum eigenvalues in the convolutions
2d2R−1 q |m|
e±i|m|r = ± q 2 ∓io−m2 )R e
i|Ω2R−1 | (
i
q
x
,
∼ SO(2R)/SO(2R − 1), 2R − 1 = 1, 3, . . . ,
Sphere Ω2R−1 =
and momenta R2R−1 with SO(2R − 1)
2π R 2R−1 ∗ 2
|Ω2R−1 | = Γ(R) , ( ∗ ,
q 2 ) = (± i|Ω2R−1 |
2 ∓ io) not orthogonal
,q
|m1 | 2R−1 |m2 | |m+ |
∂ L1
( ∂
q) ∗ ∂ L2
( ∂
q)
∂ L1 +L2
= ( ∂
q)
q 2 −m2
( 1)
R q 2 −m2
( 2)
R q 2 −m2
( +)
R
for L = 0, 1, . . . .
9.4 Residual Normalization

The abelian convolutions ∗ ∼ = δ(q1 +q2 −q) of the energy-momentum measures
above with the Rn -Lebesgue measures dn q as basis and the invariants as
|m|
pole singularities in additional factors, e.g., dq
π q2 +m2 , are “rationalized” with
respect to the product representations in such a way that the spherical degrees
do not show up (no π’s). This representation normalization (see Chapter 8)
results from spheres and related rotation groups in the definition of higher-
dimensional residues.
Since the convolution with a Dirac distribution amounts to a residue
(where defined),
. dp f (p)
f (q) = dp δ(p − q)f (p) = 2iπ p−q ,
the convolution normalization for the time representation coefficients is given

by the normalization of the residue of the real pole q = m,
∗
dq
D1 : 2iπ from 2iπ 1
q−io−m e
iqx0
= ϑ(x0 )eimx0 ;
2π is the length of the unit circle Ω1 ∼ = U(1), the compact representation

dq
image of D(1). It normalizes the energy Plancherel measure 2π for the time
translation Haar measure dx0 .
The convolution normalization for Y 1 ∼ = SO0 (1, 1) is determined by the
residual normalization in the faithful cyclic representation coefficient whose
self-duality causes the factor 2 for the two imaginary poles q = ±i|m|:
∗
|m| −iqx
Y1 : π from dq
π q2 +m2 e = e−|mx| .
For nontrivial invariant m2 > 0, one can replace

dq |m| dq |q| dq2 1
π q2 +m2 = π q2 +m2 = 2π q2 +m2 .
In general for odd-dimensional hyperboloids, the self-dual residual nor-

malization of the rotation scalar positive-type functions uses half the measures
of the corresponding spheres,
2R−1 ∗ 2
2d2R−1 q |m|
Y 2R−1 : ∗ = |Ω2R−1 | from q 2 +m2 )R e
|Ω2R−1 | (
i
q
x
= e−|m|r ,
9.5 Convolution of Feynman Measures 267
and, analogously, the residual normalization for odd-dimensional spheres:

2R−1 2d2R−1 q |m|
Ω2R−1 : ∗ = ± i|Ω∗2R−1
2
|
from ± i|Ω2R−1 | (
q 2 ∓io−m2 )R
eiqx = e±i|m|r .
The momentum eigenvalues lie on a sphere {q ∈ Ř2R−1 | q 2 = ∓m2 } ∼ =

Ω2R−2 .
Γ(R)
For the nonabelian case R ≥ 2, the normalization |Ω2R−12
| = π R for the
curved spaces differs from the “flat” normalization ( |Ω21 | )2R−1 = π2R−11
for
self-dually represented translations R 2R−1
−→ SO(2)2R−1
.
9.5 Convolution of Feynman Measures

The convolution of (energy-)momentum distributions adds the space(-time)
$
k
translation eigenvalues to the eigenvalue q = qj of the product represen-
j=1
tation:
$
k ;
k
μ1 ∗ · · · ∗ μk (q) = d1+s q1 · · · d1+s qk δ qj − q μj (qj ).
j=1 j=1
ν1 −1
1
In generalizing the formula Γ(ν 1 )Γ(ν2 )
Γ(ν1 +ν2 ) = 0 dζ ζ (1 − ζ)ν2 −1 for the beta
function, a convolution with denominator singularities can be performed by
joining first the denominators,
1 1
Γ(ν1 )Γ(ν2 ) = 0 dζ1 0 dζ2 δ(ζ1 + ζ2 − 1)ζ1ν1 −1 ζ2ν2 −1 Γ(ν1 + ν2 ),
Γ(ν1 )···Γ(νk ) 1 1 ν −1 ν −1
ζ 1 ···ζk k Γ(ν1 +···+νk )
ν ν
R 1 ···R k
= 0 dζ1 · · · 0 dζk δ(ζ1 + · · · + ζk − 1) (ζ1 1 R1 +···+ζ R )ν1 +···+νk
,
1 k k k
νj ∈ R, νj = 0, −1, −2, . . . ,
e.g., for multipole distributions,
2q
1 ν−1
1
−q2 +m2 (−q2 +M 2 )ν = ∂
∂q 0
dζ [−q2 +ζMζ2 +(1−ζ)m2 ]ν
∂
m2 dκ2 m2 −κ2 ν−1 1
= ∂q M 2 m2 −M 2 ( m2 −M 2 ) (−q2 +κ2 )ν .
Here and in the following, the convolutions exist only for pole orders where
the involved Γ-functions are defined. Elsewhere “divergences” arise.
The product of two (energy-) momentum distributions,
Γ(ν1 ) Γ(ν2 ) 1 ζ ν1 −1 (1−ζ)ν2 −1 Γ(ν1 +ν2 )
(−q12 +m21 )ν1
× (−q22 +m22 )ν2
= 0
dζ [ζ(−q12 +m21 )+(1−ζ)(−q22 +m22 )]ν1 +ν2
,
can be written with the center of mass (energy-)momentum q and relative

(energy-)momentum p:
⎧
⎨ q1 + q2 = q,
ζq12 + (1 − ζ)q22 = (p + 2ζ−1 q)2
+ ζ(1 − ζ)q 2
with q1 − q2 = 2p,
2 ⎩ n n
d q1 d q2 = dn q dn p.
The convoluted distributions must all be of the same type, e.g., either all
Feynman q 2 + io or all anti-Feynman q 2 − io. For two equal-type Feynman
1
distributions −q2 ∓io+m 2 with positive or negative invariants m
2
∈ R, the
product inherits this Feynman type for both the center-of-mass and the rel-
ative (energy-)momentum distributions:
q12 + io, with q22 + io ⇒ ζq12 + (1 − ζ)q22 + io ⇒ q 2 + io, p2 + io.
This is in contrast to the causal integration prescriptions (q ± io)2 =

(q0 ± io)2 −
q 2 = q 2 ± ioq0 for SO0 (1, s)-spacetimes, where there is no definite
integration prescription for the relative energy-momenta integration,
ζ(q1 + io)2 + (1 − ζ)(q2 + io)2 = (p + 2ζ−1

2 q) + io(2ζ − 1)p0
2
+ ζ(1 − ζ)q 2 + ioq0 ,

with (2ζ − 1) ∈ {±1} for ζ ∈ [0, 1].
The convolution product for causal measures will be defined by using the
Feynman integration for the relative energy-momenta, i.e., p2 ±io. The center-
of-mass prescription inherits the prescription of the factors (q ± io)2 .
The convolution product of two Feynman distributions, normalized by
1
n ,
t 2
i π
Γ( n
2 +ν1 ) ∗ Γ( n
2 +ν2 )
n n n
(−q2 −io+m21 ) 2 +ν1 it π 2 (−q2 −io+m22 ) 2 +ν2
1 dn p n +ν −1
1
n
(1−ζ) 2 +ν2 −1 Γ(n+ν1 +ν2 )
= 0 dζ t n2 [−p2 ζ−io−ζ(1−ζ)q
2
2 +ζm2 +(1−ζ)m2 ]n+ν1 +ν2 ,
i π 1 2
q1 −q2
is the residue with respect to the relative energy-momentum p = 2
dependence,
n
2π 2
O(t, n − t) with n = 1, 2, . . . ; t = 0, 1, . . . , |Ωn−1 | = Γ( n ) for m ∈ R
2
2
dn q n
Γ( +ν)
2 Γ(ν)
n n +ν =
(m2 −io)ν
it π 2 (−q2 −io+m2 ) 2
∂ L1 Γ( n +ν1 ) ∗ ∂ L Γ( n +ν2 )
2 2
( ∂q ) n +ν n ( ∂q ) 2 n +ν
2 2
(−q −io+m1 ) 2 1 t
i π 2 (−q −io+m2
2 2 2
n +ν −1 n +ν2)
−1
∂ L1 +L2 1 ζ2 1 (1−ζ) 2 2 Γ( n +ν1 +ν2 )
= ( ∂q ) dζ 2
0 n
[−ζ(1−ζ)q2 −io+ζm2 2 2 +ν1 +ν2
1 +(1−ζ)m2 ]
Nontrivial O(t, n − t)-properties are effected by the convolution-compatible

(energy-)momentum derivatives,
∂q ⊗ q = 1n + q ⊗ q 2 ∂q∂ 2 ,
∂
∂q = 2q ∂q∂ 2 , ∂
∂q ⊗ ∂q = (1n + q ⊗ q 2 ∂q∂ 2 )2 ∂q∂ 2 , . . . ,

∂ ∂
∂ L
e.g., in the form of harmonic polynomials ( ∂q ) , L = 1, 2, . . . . The derivatives,
acting on multipoles, raise the pole order,
∂ Γ(R) 2q Γ(1+R)
∂q (−q2 +m2 )R = (−q2 +m2 )1+R .
9.6 Convolutions for Even-Dimensional Spacetimes 269
By derivations with respect to the (energy-)momentum invariants,

Γ(ν)
( ∂q∂ 2 )k (− ∂m
∂
2)
k1 ∂
(− ∂m 2)
k2
[−ζ(1−ζ)q2 +ζm2 +(1−ζ)m2 ]ν
1 2 1 2
Γ(ν+k+k1 +k2 )ζ k+k1 (1−ζ)k+k2
= [−ζ(1−ζ)q2 +ζm21 +(1−ζ)m22 ]ν+k+k1 +k2
,
the integrals can be reduced for (half-)integer powers to a (cubic root)

quadratic polynomial in the denominator:
⎧
⎨ N ⇒ P (ζ)
1
,
n √
+ ν1 + ν2 = d2 P (ζ)
2 ⎩ 12 + N ⇒ 1 3 = (−q2 −io+m2 )(−q
4
2 −io+m2 ) dζ 2 ,
P (ζ) 2 − +
with N = 1, 2, . . . and P (ζ) = −ζ(1 − ζ)q 2 − io + ζm21 + (1 − ζ)m22 .
The integration is completely different for integer and half-integer powers:

For half-integer powers, the ζ-integration compensates the m2− -pole from the
discriminant:
1 dζ √ /
d P (ζ) /1 4|m+ |
= 4
(−q −io+m− )(−q −io+m+ ) / = m2 −m 1
2 ,
− −q −io+m+
3
0 P (ζ) 2 2 2 2 2 dζ 2 2
0 +
1 ζ N −1
(1−ζ) N −1 1
Γ(N + 2 ) 4|m+ | Γ(N )
0
dζ 3 2 N+
1 = m2 −m2 (−q2 −io+m2 )N
.
Γ( 2 )[−ζ(1−ζ)q −io+ζm1 +(1−ζ)m2 ]
2 2 2 + − +
This has been used above for odd-dimensional

1 spaces like time R and positions
Y 3 , R3 , and Ω3 . The integral 0 −ζ(1−ζ)q2 −io+ζm
dζ
2 +(1−ζ)m2 for integer powers,
1 2
characteristic for even-dimensional spaces, has more structure. It is discussed
next.
9.6 Convolutions for Even-Dimensional

Spacetimes
The convolution product for even-dimensional spacetimes SO0 (1, 2R−1) will
be given for Feynman measures that hold for real m2 . The convolutions of
Cartan energy-momentum pole distributions,
SO0 (1, 1)-spacetime, |Ω1 | = 2π
2 ∗
(∗, q2 ) = (± iπ , q2 ± io)
2 1
1
∗ 1
= dζ 1
−q2 +m2
1
2
−q +m2 2 0 −ζ(1−ζ)q +ζm2
2 2
1 +(1−ζ)m2
involves the definition of the convolution product of the causal (advanced

and retarded) measures, restricted to m2 ≥ 0, from the pointwise product
of Bessel functions ϑ(x)JN (|mx|) by analytic continuation of the convolu-
tion product of the Feynman measures as given above, i.e., with Feynman
integration for the relative energy-momenta.
In contrast to the factors for time and position, both odd-dimensional
with real rank 1, the convolutions for minimal two-dimensional spacetime
produces no pole singularities (zero-dimensional) for product representations.
The residual products of even-dimensional spaces display pole distributions

only before the finite ζ-integration over an invariant singularity line (one-
dimensional). They can be written with spectral functions, e.g., for one van-
ishing mass,
1 ∞ 2
dζ −ζq21+m2 = m2 dκ 1
κ2 −q2 +κ2 ,
1
0 ∞ dκ κ −m2
2 2
0
dζ −ζq1−ζ
2 +m2 = m2 κ 2 κ2
1
−q2 +κ2 .
After ζ-integration logarithms arise. The logarithm is typical for a finite inte-
gration [4], e.g., for a function holomorphic on the integration curve (where
defined),
∞ 6
α 6 β dz f (z) = − 6 res[f (z) log(z − β)],
z−α
β dz f (z) = res[f (z) log z−β ], ∞
dz f (z) = 2iπ resf (z),
−∞
with the sum of all residues in the closed complex plane, cut along the inte-
gration curve, e.g.,
1 6 + , log(1− mq 2
dζ 2)
0 −ζq2 +m2 = res 1
−ζq2 +m2 log ζ−1
ζ = −q2 ,
1 6 + , (1− m
2 q
) log(1− m
2
1−ζ 1−ζ 2 )−1
0
dζ −ζq2 +m2 = res −ζq2 +m2 log ζ−1
ζ = q2
−q2 .
In the second case, there is a nontrivial residue at the holomorphic point

ζ = ∞.
The corresponding structures for Minkowski spacetime as a minimal case
with nontrivial rotation degrees of freedom are as follows:
SO0 (1, 3)-spacetime, |Ω3 | = 2π 2

4
(∗, q2 ) = (± iπ∗2 , q2 ± io)
4 1
∂ 1
∂q2 −q2 +m2
∗ ∂ 1
∂q2 −q2 +m2
= ∂q ∂ 1
2 0 dζ −ζ(1−ζ)q2 +ζm2 +(1−ζ)m2
1
4
2
ζ(1−ζ)
1 2
= 1
∗ 1
= 0 dζ1
(−q2 +m2
1)
2 2
(−q +m2 )2 2 2 2
[−ζ(1−ζ)q +ζm1 +(1−ζ)m2 ]2 2
In general, one obtains the even-dimensional spacetime distributions of

energy-momenta by relativistically compatible 2-sphere spread. Measures for
higher-dimensional spacetimes are obtained by derivation of order R − 1, the
rank of the maximal compact group SO(2R − 1),
R = 1, 2, · · · : 1 ∂ R−1
Γ(R) ( ∂q2 )
1
−q2 +m2 = 1
(−q2 +m2 )R ,
with the convolution

2π R
SO0 (1, 2R − 1)-spacetime, 2R = 2, 4, . . . , |Ω2R−1 | = Γ(R)
2R ∗ 2
( ∗ , q2 ) = (± i|Ω2R−1 |
, q2 ± io)
2R
R ∗ ( ∂q )
∂ L1 1 ∂ L2 1
( ∂q )
(−q2 +m2 (−q2 +m2 R
∂ L1 +L2 1 ∂
1)
R−1 1
2)
1
= ( ∂q ) (
Γ(R) ∂q2 ) dζ
0 −ζ(1−ζ)q2 +ζm2 2
1 +(1−ζ)m2
9.7 Feynman Propagators 271
In contrast to the residual normalization of odd-dimensional eigenvalue

spaces (energy-momenta), the residual normalization for even-dimensional
spacetime (energy-momenta) uses the volume of the odd-dimensional unit
sphere Ω2R−1 , the compact partner of the embedded position hyperboloid
Y 2R−1 .
9.7 Feynman Propagators

A Feynman propagator for flat spacetime, e.g., for a free massive scalar field
(see Chapter 5),
d4 q ρ(m2 )
0|{Φ
Φ (y), Φ (x)}|0 − (x0 − y0 )0|[Φ
Φ (y), Φ (x)]|0 = iπ(2π)3 −q 2 −io+m2
eiq(x−y) ,
is the sum of a representation coefficient of the Poincaré group SO0 (1, 3) × R4

for particles, supported by all translations R , and the causally ordered quan-
4
tization, supported by the causal bicone R4+ ∪ R4− .

The Fourier-transformed energy-momentum distributions in Feynman
propagators give Macdonald K and Neumann N functions for the on-
shell contribution (real) and, in addition to Bessel functions J in the
off-shell contribution (imaginary), lightcone-supported Dirac distributions δ
for even-dimensional spacetime with nonabelian position rotations, i.e., for
1 + s = 2R = 4, 6, . . . . The Dirac distributions arise by derivations of the
characteristic causal function ϑ(x2 ):
R−1 > ?
d2R q
iπ
1
−q 2 −io+1 e
iqx
= ∂
x2
ϑ(−x2 )2K0 (|x|) − ϑ(x2 )π(N0 + iJ0 )(|x|)
∂
4π R−1
= ∂
2 dψ [ϑ(−x2 )e−|x| cosh ψ + ϑ(x2 )e−i|x| cosh ψ ].
∂x
4π
The projection of a free-field Feynman propagator to time and position by

position and time integration, respectively, displays a translation represen-
tation coefficient cos mx0 only for the on-shell part δ(q 2 − m2 ), whereas the
1 1
principal value P off-shell part iπ −q2 +m2
with causal support in spacetime is
P
position-projected to the exponential potential e−|m|r for Cartan spacetime
−|m|r
R = 1 and to the Yukawa potential e r for Minkowski spacetime R = 2:

d2R−1 x d2R q 1 iqx
dq0 1
iπ −q2 −io+1 e
iq0 x0
(2π)2R−1
dx0 −q2 −io+1 e = 0
2R−1
2π
iπ d
iπ
q

1
q 2 +1
e−iqx
(x0 ) sin x0

cos x0 R−1
= 0 −i − ∂r2 e−r .
∂
4π
The off-shell contributions (“virtual particles”) are not coefficients of Poincaré

group representations,
d2R q

π
1
2 +1 e
−qP
iqx
= −i(x0 ) d2R q (q0 )δ(q 2 − 1)eiqx
R−1
= ∂x2 ϑ(x2 )πJ0 (|x|),
∂ 4π
2
d2R q 1 iqx 1 x
(∂ + 1) π 2 +1 e
−qP
= π δ( 2π ).
They are the relevant part of the Green’s distributions associated with free-
field equations.
9.7.1 Convolutions of Feynman Propagators

Feynman integrals involve convolutions of energy-momentum distributions
for pointwise products of spacetime distributions. There arise undefined (“di-
vergent”) products of generalized functions. In general, they do not make
sense since the tempered distributions S (Rn ) do not constitute a convolu-
tion algebra.
In a schematic notation for Feynman propagators with the real on-shell
contribution, supported by all translations R1+s , and the imaginary off-shell
contribution, causally supported and, for even-dimensions 2R = 4, 6, . . . , with
Dirac distributions,
d2R q
iπ
1
−q2 −io+m2 e
iqx
= ϑs K + ϑt N + iϑt (J + δ), ϑs = ϑ(−x2 ), ϑt = ϑ(x2 ),
the convolution gives as real and imaginary contributions:
[ϑs K + ϑt N + iϑt (J + δ)] · [ϑs K + ϑt N + iϑt (J + δ)]

∼ iπ1
−q2 −io+m2, ∗ +
1 1 1
+ iπ −q2 −io+m2 ,
= δ(q 2 − m2 ) + iπ
1 1
−q2 +m2
∗ δ(q 2 − m2 ) + iπ
1 1
−q2 +m2
P P
+ ,
= δ(q 2 − m2 ) ∗ δ(q 2 − m2 ) + q2 −m
1
2 ∗ q 2 −m2
1
+ δ(q 2 − m2 ) ∗ iπ
1
−qP
1
2 +m2
P P
∼ [ϑs K + ϑt (N + J + J · δ + δ )] +
2 2 2 2
iϑt (N · J + N · δ).
The “divergent” parts δ 2 with the pointwise product of S (R4 )-distributions,

as familiar from a perturbation expansion quantum electrodynamics, do not
make sense:
+ , + ,
δ(q 2 − m21 ) + iπ1 1
−qP +m1
2 2 ∗ δ(q 2
− m 2
2 ) + 1 1
iπ −qP +m2
2 2
+ , + ,
R4
∼ x
1
2 + iδ(x 2
) + · · · · x
1
2 + iδ(x 2
) + · · · .
P P
The “divergent” parts δ(q 2 − m21 ) ∗ −q2 1+m2 and q2 −m1

2 ∗ −q2 1+m2 con-
P 2 P 1 P 2
tain the causally supported off-shell contribution from the energy-momentum
R4
principal value −q2 1+m2 ∼ δ(x2 ) + · · · . They are not coefficients of transla-
P
tion product representations like the meaningful on-shell convolution δ(q 2 −
m21 ) ∗ δ(q 2 − m22 ).
9.7.2 Convolutions for Free Particles

The on- and off-shell contributions are convoluted as follows:
O(1, s), m2 ≥ 0 : (Fm2 , Fm
2 ) = ± iπ −q 2 ∓io+m2 = δm2 ± iPm2 ,
1 1

Fm2 ∗ Fm2
F 2 ∗ F 2 = δ1∗2 ± iP1∗2 ,
1 2
m1 m2
with the Dirac and principal value contribution

δ1∗2 = δm21 ∗ δm22 − Pm21 ∗ Pm22 ,
P1∗2 = δm21 ∗ Pm22 + Pm21 ∗ δm22 .
The Fourier-transformed principal part of a Feynman distribution for an

orthochronous group SO0 (1, s) can be written as an order function times an
on-shell part:
⎧
⎪ δ +δ
⎨ δm2 = |m| 2 −|m| ,
1 1
Fm2 = iπ −q2 −io+m2 = δm2 + iPm2 : Pm2 ∼ i(x0 )m2 ,
⎪
⎩ 2 = δ|m| −δ−|m| .
m 2
In the principal value convolution contribution of two Feynman propagators

for spacetime R1+s the order function drops out for (x0 )2 = 1:
>
± iπ
1 1
∗ ± iπ
1 1
= 2 ϑ(+q0 )δ(q 2 − m21 ) ∗ ϑ(+q0 )δ(q 2 − m22 )
−q2 ∓io+m21 −q2 ∓io+m22 ?
+ + ϑ(−q0 )δ(q −m1 ) ∗ ϑ(−q0 )δ(q −m2 ) ,
2 2 2 2
± iπ
1
δ(q 2 −m21 ) ∗ −qP
1
2 +m2
1
+ −q2 +m 2 ∗ δ(q − m2 ) .
2 2
2 P 1
The principal value square is also an on-shell convolution only. The con-
volution of translation representation coefficients from the real part of the
propagator (free particles) gives corresponding coefficients for product repre-
sentations (product of free particles):
δ1∗2 = δm21 ∗ δm22 − Pm21 ∗ Pm22

δ|m1 | ∗δ|m2 | +δ−|m1 | ∗δ−|m2 |
= δm21 ∗ δm22 + m21 ∗ m22 = 2 .
The set with all (1+s)-dimensional “filled-up” forward (backward) energy-

momentum hyperboloids is an additive cone. Therefore, the distributions sup-
ported by positive and negative energy-momentum are convolution algebras,
however, not orthogonal to each other:
{q |m1 |} + {q |m2 |} = {q |m+ |},

with δ±|m| ∼ 2|m|ϑ(±q0 )δ(q 2 − m2 ) ∈ D (R1+s
± ) ∈ aagC (convolution product).
The convolution for abelian time with self-dual invariants m21,2 > 0,
abelian R : ϑ(±q0 )|q0 |δ(q02 − m21 ) ∗ ϑ(±q0 )|q0 |δ(q02 − m22 )

= ϑ(±q0 )|q0 |δ(q02 − m2+ ),
is embedded into the convolution of nonabelian hyperboloids for product

representations of the translations (the real part δ1∗2 for simple pole Feynman
propagators). With the hyberboloid “radii” as energy-momentum invariants
q 2 = m2 ≥ 0,
∗
SO0 (1, s) ×
R1+s : ϑ(±q0 )δ(q 2 − m21 )
|Ωs−1 | ϑ(±q0 )δ(q 2 − m22 )
s−2
Δ(q2 ) 2
= (2|q|) s−1 ϑ(±q0 )ϑ(q 2 − m2+ ), s = 1, 2, . . . ,
they involve the two-particle threshold factor:
Δ(q 2 ) = Δ(q 2 , m21 , m22 ) = (q 2 − m2+ )(q 2 − m2− ).
For nontrivial position, the convolution (phase space integral) of s-dimension-

al on-shell hyperboloids (particle measures) does not lead to s-dimensional
on-shell hyperboloids δ(q 2 − m2+ ). It leads to translation representations with
energy-momenta over the free particle threshold at the mass sum q 2 ≥ m2+ ,
q 2 = q02 − m2+ ≥ 0, with m± = |m1 | ± |m2 |. Here, the energy is enough to
i.e.,
produce two free particles with masses m1,2 and momentum (q1 + q2 )2 ≥ 0.
The minimal cases s = 1, 2 are characteristic for even- and odd-
dimensional spacetime:
R2 : ϑ(±q0 )δ(q 2 − m21 ) ∗2 ϑ(±q0 )δ(q 2 − m22 )

SO0 (1, 1) ×
= √ 2 21 2 2 ϑ(±q0 )ϑ(q 2 − m2+ ),
(q −m+ )(q −m− )
R3 : ϑ(±q0 )δ(q 2 − m21 ) ∗ ϑ(±q0 )δ(q 2 − m22 )
SO0 (1, 2) × 2π
1
= 2|q| ϑ(±q0 )ϑ(q 2 − m2+ ).
The Poincaré group SO0 (1, 3) ×

R4 is the minimal case with nonabelian
rotations:
∗
M(Ř4 )+ ∗ M(Ř4 )+ = M(Ř4 )+ : ϑ(q
√0 )δ(q
2
− m21 ) 4π ϑ(q0 )δ(q 2 − m22 )
(q2 −m+ )(q2 −m− )
2 2
= q2 ϑ(q0 )ϑ(q 2 − m2+ ).
Such convolutions arise, e.g., as the nondivergent on-shell contribution in the

quantum electrodynamical vacuum polarization by electron-positron pairs.
9.7.3 Off-Shell Convolution Contributions

Energy-momentum convolutions combine the points on the hyperbolic-
spherical singularity surfaces for particle-interaction structures, determined
by the invariants. The characteristic new feature is the on-shell–off-shell
convolution, i.e., of compact with noncompact invariants. The convolution
contribution in the mixed terms is not for product representations of the
spacetime translations,
P1∗2 = δm21 ∗ Pm22 + Pm21 ∗ δm22 .

The divergences in Minkowski space arise from the mixed terms (mathemat-
ically meaningless)
R4 1
δ(q 2 − m21 ) ∗ 2 +m2 ∼ x2 · δ(x2 ) + · · · .
1
−qP 2 P
Only for trivial position does the principal value part also add the invariant
poles:
δ ∗δ −δ ∗δ
s = 0 : P1∗2 ∼ i(t) |m1 | |m2 | 2−|m1 | −|m2 | ∼ Pm2+ .
The characteristic effect of a convolution of noncompact with compact in-
variant comes in the principal value part for nontrivial position degrees of
freedom:
δ(q 2 − m2 ) ∼ ϑ(q 2 − m2 ),
1
−q2 +m2
∼ ϑ(q 2 − m2 ) + ϑ(−q 2 + m2 )
P
∪ ∪
compact (free) + noncompact.
eimt e−|mz|
The denominator polynomial in the above convolution square has two ener-
gy-momentum-dependent zeros,
P (ζ) = −ζ(1 − ζ)q√

2
+ ζm21 + (1 − ζ)m22 = q 2 [ζ − ζ1 (q 2 )][ζ − ζ2 (q 2 )],
q2 −m+ m− ± Δ(q2 )
ζ1,2 (q 2 ) = 2q2 with m± = |m1 | ± |m2 |,
which are either both real or complex conjugate to each other according to
the sign of the discriminant Δ(q 2 ) (two-particle threshold factor):

ϑ(Δ(q 2 )) = ϑ(q 2 − m2+ ) + ϑ(m2− − q 2 ),
Δ(q 2 ) = (q 2 − m2+ )(q 2 − m2− ) :
ϑ(−Δ(q 2 )) = ϑ(m2+ − q 2 )ϑ(q 2 − m2− ).
Furthermore, real zeros, in the case of Δ(q 2 ) ≥ 0, are in the integration

interval ζ ∈ [0, 1] only for energy-momenta over the threshold ϑ(q 2 − m2+ ),
|m2 |
ζ1,2 (m2+ ) = |m1 |+|m2 | ∈ [0, 1],
−|m2 |
ζ1,2 (m2− ) = |m1 |−|m2 | ∈
/ [0, 1],
and graphically:
Δ(q2 ) ≥ 0 Δ(q2 ) ≤ 0 Δ(q2 ) ≥ 0
R ζ1,2 (q2 ) ∈
/ [0, 1] ζ1 (q2 ) = ζ2 (q2 ) ∈
/R ζ1,2 (q2 ) ∈ [0, 1]
• • • q2
0 m2− m2+
Therefore, the convolution of two energy-momentum Feynman pole dis-

tributions contains the following as relevant contributions, valid for general
signature O(t, s):
1 dζ
1 1
0 P (ζ)
= 0 dζ −ζ(1−ζ)q2 −io+ζm2 +(1−ζ)m2
1 2
1 + ,
2 +(1−ζ)m2 + iπδ −ζ(1 − ζ)q + ζm1 + (1 − ζ)m2
1 2 2 2
= 0 dζ −ζ(1−ζ)q2 +ζm
P 1 2
⎡ / 2 √ 2 / ⎤
/ Σ(q )−2 Δ(q ) /
ϑ(m2− − q 2 ) log / /
⎢ √+ −m−2
m2 2
⎥
= √ 2 ⎢ 2 −Δ(q ) ⎥,
|Δ(q 2 )| ⎣ + ϑ(q 2 − m2− )ϑ(m2+ − q 2 ) arctan Σ(q2 ) ⎦
+iπ ϑ(q 2 − m2+ )
with Δ(q 2 ) = (q 2 − m2+ )(q 2 − m2− ), Σ(q 2 ) = (q 2 − m2+ ) + (q 2 − m2− ).
The convolution product depends on the two variables {q 2 − m2+ , q 2 − m2− }.

The part with ϑ(q 2 − m2+ ) (here imaginary) is the on-shell convolution,
adding positive and negative energy contributions (above). The convolution of
compact with noncompact invariants, original for indefinite signature space-
time, (here real part) shows up for energy-momenta under the threshold
ϑ(m2+ − q 2 ) and for ϑ(m2− − q 2 ) containing the momentumlike contribution
ϑ(−q 2 ). This is illustrated by two equal masses:
⎡ / 2 2 √ 2 2 /⎤
/ 2m −q + q (q −4m2 ) /
ϑ(−q 2 )log / /
1 ⎢ √ 2
2m2
⎥
dζ −ζ(1−ζ)q12 −io+m2 = √ 2 ⎢ 2) ⎥ .
⎣ + ϑ(q 2 )ϑ(4m2 − q 2 ) arctan q 2(4m −q
2
0 |q 2 (q 2 −4m2 )| ⎦
q −2m 2
+iπ ϑ(q − 4m )
2 2
Chapter 10
Interactions and Kernels
It is physically obvious that interactions and bound-states have a close

connection. Both are described, in distinction to free objects, by nonabelian
operations of nonflat position. In contrast to position functions, which are
representation coefficients of position groups, position interactions are for-
malized by linear mappings of position functions and are, in general, position
distributions. The relation between bound-states and interactions will be
seen in parallel to the relation between Lie groups and Lie algebras.
Examples for the Lie algebra–interaction relationship are given by Hamil-
p2

tonians in mechanics, H = 2m + V (x), where a potential for an interaction
contributes to the time translation Lie algebra H ∼ dt d
. For example, the
convolutive action of the Yukawa and Kepler potential on the ground-state
function x −→ e−mr , m > 0, of the nonrelativistic hydrogen atom, a positive-
type L∞ (Y 3 )-function, representing hyperbolic 3-position Y 3 ∼= R3 , leads to
∞ 3
L (Y )-functions:
−mr
∗ e−M r e−M r −e−mr
−e r 4π e−Mr = 2M
M 2 −m2 + (M 2 −m2 )2 r ,
∗ e−M r 2 e−M r −1
− r1 4π e−Mr = M2 + M2 Mr .
Also, gauge interactions represent Lie algebras by charges, i.e., position-

integrated currents L l −→ iQ = i d3 x J0 (x) ∈ W ⊗ W T , which occur in
gauge vertices Aa Ja with gauge fields (see Chapter 6). The gauge field–
induced interactions show up in the off-shell contribution of their propagators,
d4 q αS
e.g., for electrodynamics dx0 π(2π) 2 −q 2 −io e
iqx
= αrS (see Chapter 5).
The transition from bound-states (functions) to interactions (distribu-
tions) is illustrated for three-dimensional position, L∞ (Y 3 ) −→ M(Y 3 ),
where derivations of the ground-state wave function of the nonrelativistic
hydrogen atom lead to the Yukawa potential and the Yukawa force, which
are Radon measures of the position hyperboloid,
−mr
∂ ∂ −mr x 1+mr −mr
∂x ∂r 2 e = −m ∂
∂ e
x 2r = m r 2r 2 e .
278 Chapter 10 Interactions and Kernels
As will be discussed ahead, there is an interpretation in terms of group

representations for the fact that the pointwise product of the Kepler potential
1 −mr
r with a bound-state wave function e for representation-characterizing
invariant m gives a Yukawa potential 1r · e−mr . The Kepler potential is the
2
inverse Laplacian, 1r = − 4π
2 , and the Yukawa potential and the bound-state
∂
function are inverse Laplacians of order 1 and 2 with invariant m2 :
−mr
(−∂ 2 + m2 ) e r = 4πδ(x),
(−∂ 2 + m2 )2 e−mr = 2m 4πδ(x).
Inverted differential operators (derivations) define kernels. The transition

from bound-states to interactions is expressible by the action of a Green’s
kernel on Lie group representation coefficients.
In a residual representation, the (energy-)momentum distributions for
bound state functions and interaction distributions have singularities at the
invariants, in the example above where the derivative of the bound-state
function is the Yukawa potential:
−mr
d3 q −i
(1, 2m
r )e = 2π 2m
q 2 +m2 )2−N e
2 (
q
x
for N = (0, 1),
2 2 −mr 2m −mr
with (−∂ + m )e = r e .
The structures of interactions as implemented by tangent kernels, con-

sidered in this chapter mainly for time D(1) ∼= R and position Y 3 , both with
real rank 1, and exemplified by the nonrelativistic hydrogen atom and the
model of Nambu and Jona-Lasinio (see Chapter 9), will be used for the more
complicated spacetime operations D(2) with real rank 2 in Chapter 10.
The dual of a real finite-dimensional Lie algebra L = log G, i.e., its linear
forms LT , is more easily accessible than the group dual Ǧ = irrep G, which,
in general, is not a group, but, e.g., a cone or a direct sum of cones. C ⊗
LT contains all eigenvalues of the Lie algebra action. The invariants that
characterize the group dual are given by!multilinear forms of the eigenvalues
in the complexified tensor algebra C ⊗ LT .
A Lie group and Lie algebra act on themselves and coadjointly on the
dual Lie algebra, with the actions denoted by • in the schema:
• G L LT ∗ M(G) L1 (G) L∞ (G)

G G L L T
M(G) M(G) 1
L (G) L∞ (G)
L L L LT L1 (G) L1 (G) L1 (G) L∞ (G)
LT LT LT − L∞ (G) L∞ (G) L∞ (G) −
Both a Lie group and the enveloping algebra of its Lie algebra can be realized
in one algebra: A matrix group G ⊆ GL(V ) for a finite-dimensional vector
space V ∼= Cn has a minimal endomorphism subalgebra A(G, L) ⊆ AL(V ),
which embeds both the algebra C(G) , generated by the group, and the al-
gebra C(L) , generated by its Lie algebra L = log G ⊆ AL(V ). C(L) is
the endomorphism image of the enveloping algebra E(L) −→ AL(V ). The
10.1 Invariant Differential Operators 279
algebra A(G, L) ⊇ C(G) ∪ C(L) is generalized by the group measure algebra

M(G) ⊇ L1 (G). The convolution algebra M(G) embeds the group G by
Dirac measures. It is in duality with L∞ (G). Tangent (Lie algebra) kernels
of a Lie group G are Radon measures M(G).
10.1 Invariant Differential Operators

Differential operators [36] acting on functions of a Lie group and its sym-
metric spaces M ∼ = G/H are related to its Lie algebra log G ∼ = Rn with the
left-invariant vector fields, i.e., to the derivations v ∈ T(M) from its tangent
spaces (see Chapter 2), built, e.g., by (∂ a )na=1 in holonomic bases.
The enveloping algebra E(L) with the associative products of the Lie
algebra elements, modulo the identification of the Lie bracket with com-
mutator, is mapped into the composition algebra, built by derivatives, e.g.,
αkj ∂ k ⊗ ∂ j −→ αkj ∂ k ∂ j . The invariant differential operators are determined
by the center of the enveloping algebra.
For simple Lie algebras, the invariant Laplace–Beltrami operator (see
Chapter 2) corresponds to the invariant Casimir operator κ−1 = κab la ⊗ lb
as inverse Killing form κ(l, m) = tr ad l ◦ ad m, represented on vector spaces
with manifold functions. It defines eigenfunctions or eigendistributions with
eigenvalues and invariants.
The Laplace–Beltrami operator of a Riemannian manifold (M, g), g =
gkj dxk ⊗ dxj , g−1 = gkj ∂ k ⊗ ∂ j , is invariant under a diffeomorphism if, and
only if, it is an element of the motion (global invariance) group ϕ ∈ Gg , i.e.,
an isometry, g(ϕ.x, ϕ.y) = g(x, y),
∂g2 : C(M(t,s) )−→ C(M(t,s) ),
∂g2 = √1 ∂ k |g| gkj ∂ j = gkj (∂ k ∂ j − Γkj i
i ∂ )
|g|

= ∂ k gkj ∂ j + gkj (∂ k log |g|)∂ j .
As the inverse metric g−1 , it has a degree-2 contribution, and for | det g| =
|g| = constant an additional contribution for the dilation group in GL(Rn ) =
D(1)×SL(Rn ), which contains the derivative of the dilation degree of freedom
of the metric. The D(1)-factor also arises in the invariant measure |g|dt+s x.
Locally, there exist stationary coordinates, where the Laplace–Beltrami op-
erator can be brought to the simple flat space form ∂g2 |x0 = ηab ∂ a ∂ b .
For a scalar spacetime field, the gravitative interaction is described by
∂g2Φ = −m2Φ (see Chapter 5). The negative mass −m2 is the invariant of
the Laplace–Beltrami operator, also for nonflat spacetime.
For translations, the algebra D(Rn ) of the invariant differential operators
consists of the operators C[(∂ a )na=1 ] with constant coefficients, generated by
the derivatives. The joint eigenfunctions are the exponentials; the eigenspace
representations are one-dimensional and irreducible.
The joint eigenfunctions on the cosets G/K of a Lie group for a compact
subgroup {Df = I(D)f for each D ∈ D(G/K)} are characterized by spherical
functions (see Chapter 8). The algebra of the invariant differential opera-
tors D(G/K) on cosets of a maximal compact subgroup K is a commutative
polynomial ring with algebraically independent generators C[(DA )rA=1 ] whose
degrees dA are canonically determined. r is the real rank of G/K.
For manifolds with constant curvature (maximal symmetry) R = kg ∧
g, the polynomials C[∂g2 ] of the Laplace–Beltrami operator constitute the
invariant differential operators, i.e., for the general hyperboloids Y (t,s) ∼ =
SO0 (t, 1 + s)/SO0 (t, s) with k = 1 and Y (s,t) ∼
= SO0 (1 + t, s)/SO0 (t, s) with
k = −1, and for the flat manifolds SO0 (t, s) × Rn /SO0 (t, s) with k = 0 and
Laplace–Beltrami operator ∂t2 − ∂s2 .
The Laplace–Beltrami operator for spheres, Euclidean spaces, and time-
like hyperboloids with dimension s ≥ 1 can be decomposed into a radial part
and a spherical part,
s
Ω SO(2)
Rs ∼ R × Ωs−1 ,
s
Y SO0 (1, 1)
p2 = p2r + Lr2 ∼
2
as familiar from the flat 3-position = −∂ 2 with radial and
angular momentum invariant L 2 ∼
= − ∂ω∂ 2
2 on Ω
2 ∂2
and ∂ω ∂2
2 = ∂ϕ2
∼
= −L23 on
2 1
Ω1 (see Chapter 4). The metrics and the operators in the Euler parametri-
zation display the general polar decomposition,
2 2 2
dωs dθ sin θ
2
g = dx2s = dr2 + r2 dωs−1 ,
dy2s dψ 2 sinh2 ψ
⎛ ∂2
⎞ ⎛ ⎞
∂2 1 ∂
∂ωs 2 ∂θ2 sin2 θ cot θ ∂θ
∂g2 =⎝ ∂s2 ⎠ =⎝ ∂2
∂r2
⎠+ 1
r2
1
∂2
2
∂ωs−1
+ (s − 1) 1 ∂
r ∂r
∂
,
∂2 ∂2 coth ψ ∂ψ
sinh2 ψ
∂y2s ∂ψ 2
where the (s − 1)-proportional degree-1 contribution contains the logarithm

of the dilation factor in the measure:
⎛d ⎞
cot θ dθ log sin
s−1
θ s sins−1 θdθ
d ω
⎝ ⎠
d s−1
dr log r
(s − 1) 1
r = d log sinhs−1 ψ , |g| d x = dssx =
s r s−1 dr ds−1 ω.
coth ψ dψ d y sinhs−1 ψdψ
They are explicitly in two and three dimensions:

∂2 ∂2
2,3 ∂ω22
= sin12 θ [(sin θ ∂θ
∂ 2
) + ∂ϕ 2 ],
Ω : ∂2 ∂2 ∂2
∂ω 2
= sin θ ( ∂θ2 − 1) sin θ + sin12 θ ∂ω
1
2,
3 2

∂22 = 1 ∂ 2 ∂2
r 2 [(r ∂r ) + ∂ϕ2 ],
R 2,3
:
∂32 = 1 ∂2 1 ∂2
r ∂r 2 r + r 2 ∂ω22 ,
⎧
⎨ ∂2
= 1 ∂ 2
[(sinh ψ ∂ψ ∂2
) + ∂ϕ 2 ],
∂y22 sinh2 ψ
Y 2,3
:
⎩ ∂2
= 1 ∂2 1 ∂2
∂y23 sinh ψ ( dψ 2 + 1) sinh ψ + sinh2 ψ ∂ω22 .
10.2 Kernels 281
For a three-dimensional flat space, the spherical-hyperbolic analogue formu-

∂2 1 ∂2
lation is 1r ( ∂r 2 − k)r = r ∂r 2 r with k = 0.
In geodesic polar and Cartesian coordinates, the metric, measure, and

Laplace–Beltrami operator look like the following:

1 sin θ
for k = 0 : r = ρ = 1+rk r2 ,
−1 sinh ψ
2
4
dωs
dρ2 dr 2 +r 2 dωs−1
2
metric: dx2s = 1−kρ2 + ρ2 dωs−1
2
= 2 ,
2 (1+k r4 )2
dys s
d ω s−1
measure: ds x =√
ρ dρ
ds−1 ω = r s−1 dr
2 ds−1 ω,
1−kρ2 (1+k r4 )s−1
⎛ d y⎞
s
∂2
∂ωs2
Laplacian: ⎝ ∂s2 ⎠ = (1−kρ2 ) ∂ 22 + 12 ∂22 d

+ dρ log[ρs−1 e−
ks 2
2 ρ
∂
] ∂ρ
∂ρ ρ ∂ω
∂2 s−1
∂y2
s
2
∂2
= (1 + k4 r2 )2 ( ∂r
∂
2 +
1
2
r 2 ∂ωs−1 )
k 2 s−3 2
d s−1 4 r (1−k 16 r ) ∂
+ dr log[r e ] ∂r .
10.2 Kernels
Kernels are defined as linear mappings of vector spaces that contain (gener-
alized) functions on measure spaces:

κ : F1 −→ F2 , f −→ κ(f ), κ(f )(y) = dx κ(y, x)f (x).
They generalize the structure of mappings⎪and tensor products of finite-

⎪
dimensional vector spaces {κ : V1 −→ V2 ⎪ ⎪linear} ∼ = V2 ⊗ V1T with κ =
κja ea ⊗ ěj for bases.
A kernel κ : F1 −→ F2 is semiregular if its action leads to functions F2 .
A regular kernel has to be semiregular for both κ and its transposed,

κT : F2 −→ F1 , ω −→ κT (ω), κT (ω)(x) = dy ω(y)κ(y, x),
e.g., a semiregular (anti)symmetric kernel κ : F −→ F , κ(x, y) = ±κ(y, x)

for (x, y) ∈ Rn × Rn or a semiregular diagonal kernel κ(y − x).
A kernel is regularized if (y, x) −→ κ(y, x) is a function. Then it is regular.
The inverse is not true, as seen at the Dirac distribution δ(y − x).
10.2.1 Kernels for Bilinear Forms

With dual finite-dimensional vector spaces, a linear mapping κ : V −→ V T
gives a bilinear V -form κ = κij ěi ⊗ ěj ∈ (V ⊗ V )T ∼
= V T ⊗ V T . As a gen-
eralization, the kernel theorems of L. Schwartz for nuclear vector spaces [56]
establish the isomorphy of, on the one hand, linear continuous mappings
κ : F −→ F of function spaces on a real set into a dual space with distribu-

tions and, on the other hand, the same distribution type F on the product
set, which can be “factorized" in the form of a completized tensor product:
⎪
⎪
{κ : F (T n ) −→ F(T m ) ⎪
⎪linear, continuous}
∼ n ∼ m n
= Fn(T × T ) = F (T ) ⊗F(T
m
),
fn −→ κ(fn ), κ(fn )(y) = T n d x κ(y, x)fn (x) for y ∈ T ,
m
e.g., for F = S, C ∞ , Cc∞ with F = S , E , D

and open T n,m ⊆ Rn,m .
The transposed kernel is given by
κT : F (T m ) −→ F(T n ) ,
fm −→ κT (fm ), κT (fm )(x) = T m dm y fm (y)κ(y, x) for x ∈ T n .
In a basis notation, one can write κ = κ(y, x) dm y dn x ∈ F(T m × T n ) .

Kernels with κ(x, y) = ±κ(y, x) for T n = T m are (anti)symmetric.
κ
Kernels F −→ F for dual spaces define a bilinear form:
f1 ⊗ f2 ∈ F ⊗ F : κ, f1 ⊗ f2 = f
1 , κ(f2 ) = κ (f1 ), f2
T
= dydx f1 (y)κ(y, x)f2 (x) ∈ C.
10.2.2 Group Kernels

The group product–induced convolution product defines group kernels as lin-
ear mappings of group functions or distributions:
κ
κ ∈ F(G) : F1 (G) −→ F2 (G), μ − →
κ ∗ μ, μ ∗−1κ,
with κ ∗ μ(g) = dg1 κ(gg1 )μ(g1 ).
The left–right action of the Radon distributions M(G) on group functions

and on itself F (G) ∈ {Lp (G), M(G)} involves the left and right multiplica-
tions with the group elements G k −→ δk ∈ M(G):
M(G) ∗ F (G) ∗ M(G) −→ F(G), e.g., μ −→ δk ∗ μ ∗ δl ,

δk ∗ μ ∗ δl (g) = μ(k −1 gl−1 ).
Group kernels, depending on the quotient of group elements, i.e., on the

diagonal group, G × G (g1 , g2 ) −→ κ(g2 g1−1 ) (see Chapter 2), are called
diagonal. For translations, they depend on the difference Rn ×Rn (y, x) −→
κ(y−x), e.g., the symmetric Dirac distribution δ(y−x) or the Kepler potential
γ0
|
y−x| .
For a manifold with cosets G/K ∼= M ⊆ Rn , one has as convolution prod-
ucts and as pointwise products for the Fourier transforms with the crossover
10.3 Green’s Kernels 283
F
Fourier correspondence (C˙b (Rn ), M(Rn )) ↔ (M(Řn ), C˙b (Řn )) of bounded
functions and Radon distributions (see Chapter 8):
∗ M(Rn ) L1 (Rn ) L∞ (Rn ) · C˙b (Řn ) Ċ0 (Řn ) M(Řn )

M(Rn ) M(Rn ) L1 (Rn ) L∞ (Rn ) F C˙b (Řn ) C˙b (Řn ) Ċ0 (Řn ) M(Řn )
L1 (Rn ) L1 (Rn ) L1 (Rn ) L∞ (Rn )
↔ Ċ0 (Řn ) Ċ0 (Řn ) Ċ0 (Řn ) M(Řn )
L∞ (Rn ) L∞ (Rn ) L∞ (Rn ) − M(Řn ) M(Řn ) M(Řn ) −
F dn q
n ∼ ˙ n ∼ ∞
μ
1 ∗ μ2 = μ̃1 · μ̃2 , M(R ) = Cb (Ř ) = L (Ř ).
n
This defines group kernels (distributions and functions):

κ κ
κ ∈ M(M) : M(M) −→ M(M), Lp (M) −→ Lp (M), 1 ≤ p ≤ ∞,
f f
f ∈ L1 (M) : M(M) −→ L1 (M), Lp (M) −→ Lp (M),
d d
d ∈ L∞ (M) : M(M) −→ L∞ (M), L1 (M) −→ L∞ (M).
Convolution algebras A(M) ⊆ M(M), L1 (M) with Radon distributions or

absolute integrable functions give algebras with diagonal kernels A(M × M).
10.3 Green’s Kernels

The inverse (dual) of derivations Lp (M) f −→ Df ∈ Lp (M) for functions on
a manifold M with invariant measure dM x and associated Dirac distribution
δM define Green’s kernels κ = D−1 , assumed as diagonal Radon distributions:

x, y ∈M : Dy κ(y, x) = δM (y, x)
D −→ D−1 = κ ∈ M(M),
with dM x δM⎧ (y, x)f (x) = f (y),
⎨ f (y) = κ ∗ g(y)
M(M) ∗ Lp (M) ⊆ Lp (M), Df (x) = g(x) ⇒ = dn x κ(y − x)g(x),
⎩ ˜
f (q) = κ̃ · g̃(q).
Such kernels are familiar from differential equations of motion, as used for
the composite massless Nambu–Goldstone fields (see Chapter 9). Green’s
kernels are determined up to a solution of the homogeneous equation Dκ0 = 0,
which may be related to a trivial eigenvalue or a trivial invariant, i.e., for
a harmonic function, e.g., the harmonic polynomials of the groups O(t, s).
Invariant differential operators D(M) lead to invariant kernels D−1 (M).
Kernels generalize the dual structure of finite-dimensional endomor-
phisms:
f = fab ea ∈ ěb ∈ V ⊗ V T ∼
= AL(V ), V ∼= Kn ,
−1
D, κ ∈ AL(V ) : κ = D ⇐⇒ D, κ = tr D ◦ κ = Dba κab = 1.
The trace-inverse (dual) endomorphism is determined up to κ0 ∈ AL(V )

with tr D ◦ κ0 = 0 (more ahead).
The composition of two differential operators, e.g., for the product in the
enveloping algebra of a Lie algebra, leads to the convolution product of the
corresponding Green’s kernels, which, for a translation parametrization, is
the pointwise product of the Fourier transforms, if defined,
(D1 D2 )−1 = D−1 −1
2 ∗ D 1 = κ2 ∗ κ1 :
Dj κj (x) = δ(x), κ12 (x) = κ2 ∗ κ1 (x),
⇒ D2 κ12 (x) = κ1 (x) ⇒
D1 D2 κ12 (x) = δ(x) κ̃12 (q) = κ̃2 · κ̃1 (q).
In correspondence to the composition algebra D(M) with the invariant
differential operators, generated by fundamental ones, there are fundamental
invariant kernels that generate by convolution or, for the Fourier transforms,
by pointwise product the algebra D−1 (M) with the invariant kernels.
10.3.1 Linear Kernels for Spacetime

Linear kernels are defined as inverse derivatives of degree 1 ∂1 . They are
Fourier-transformed simple (energy-)momentum poles at the trivial invariant,
to be taken from the following expressions (see Chapter 8):
dq Γ(1−ν) iqx
R: 2iπ (q−io)1−ν e
1
= ϑ(x) (ix) ν , ν ∈ R,
⎛ ⎞
Γ(ν)
1 Γ( s2 −ν)
( r )2ν
e−iqx = ⎝ ⎠, s = 1, 2, . . . ,
s 2
d q
SO(s) : s s
π2 i
q q 2 ) 2 −ν
(
x Γ(1+ν)
2 ( r )2(1+ν)
2
⎛ ⎞
Γ(ν) ν
⎜
−x2 +io
⎟
d1+s q
1
2 −ν)
Γ( 1+s iqx ⎜ ⎟
4
SO0 (1, s) : e =⎜ ⎟, s = 0, 1, . . . .
⎝ 2 −x2 +io 1+ν ⎠
1+s 1+s x Γ(1+ν)
iπ 2 iq (−q2 −io) 2 −ν
4
d
The inverse of the invariant derivative dx ∈ D(R) is half the sign function
for principal value integration and the characteristic function for the future
and past for advanced and retarded integration:
⎧ (x)
⎪
⎪ d
⎨ dx ±ϑ(±x) = δ(x),
2
( d )−1 (x) 1
R: dq
⎪ dx qP
2 q
⎪
⎩ d −1
= = 2iπ eiqx , q1P = q2 +o 2.
( ) dx ± ±ϑ(±x) 1
q∓io
The difference between the R-kernels is constant: 2ϑ(±x) ∓ (x) = 1.

The inverse derivatives of Euclidean spaces are not invariant for s ≥ 1:
s
∂ 1
x
x) with |Ωs−1 | = Γ(
∂ xs |Ωs−1 | r s = δ(
2π 2
s ,
SO(s) : ds q iq −iqx 2)
( ∂∂xs )−1 = |Ωs−11
x
| rs = (2π)s
q2 e .
An example is the Kepler force 1

x
4π r 3 in R3 -position.
10.3 Green’s Kernels 285
The Feynman measures of Minkowski spacetimes, not invariant for s ≥ 2,

d1+s q q iqx 1 x
SO0 (1, s) : (2π)1+s −q2 −io e = |Ωs | 1+s ,
(−x2 +io) 2
give as linear kernels with principal value integration and odd spacetime
dimensions 1 + 2R = 1, 3, . . . :
d1+2R q q iqx 1 x 2
SO0 (1, 2R) : (2π)1+2R −q2 −io e = |Ω2R | |x|1+2R [ϑ(−x ) + i(−1)R ϑ(x2 )],
d1+2R q q iqx (−1)R
∂
( ∂x1+2R )−1 = 2 e
i(2π)1+2R qP
= |Ω2R |
x
ϑ(x2 ) |x|1+2R .
∂ −1
They start with half the sign distribution ( ∂x 1
x
) = 2|x| = (x)
2 . The kernels
for even spacetime dimensions 2R = 2, 4, . . . involve a Dirac distribution on
the lightcone:
+ ,
d2R q
SO0 (1, 2R − 1) : q
(2π)2R −q2 −io e
iqx
= x
|Ω2R−1 | − Γ(R)
iπ
δ (R−1) (x2 )+ (−x12 )R ,
2R
P
( ∂x∂2R )−1 = i(2π) d q q iqx

2R q 2 e = x
2π R−1 δ
(R−1) 2
(x ).
P
∂ −1
They start with ( ∂x 2
) = x2 δ(x2 ).
The linear kernels with support by the future and past are

∂
( ∂x1+2R )−1
± =
d1+2R q q
i(2π)1+2R (q∓io)2 e
iqx
R
= 2ϑ(±x0 )( ∂ )−1 = 2ϑ(±x0 ) (−1) 2 x
|Ω2R | ϑ(x ) |x|1+2R ,
d2R q ∂xq1+2R iqx
( ∂x∂2R )−1
± = i(2π)2R (q∓io)2 e
= 2ϑ(±x0 )( ∂x∂2R )−1 x
= 2ϑ(±x0 ) 2πR−1 δ (R−1) (x2 ).
The difference between the kernels with Feynman and causal integration
prescription is an on-shell contribution, i.e., a solution of the homogeneous
equation (∂ 2 + m2 )1+N f0 = 0,
+ ,
d1+s q
2iπ
Γ(1+N)
(q 2 −io−m2 )1+N
+ Γ(1+N)
[(q−io)2 −m2 ]1+N
eiqx = d1+s q ϑ(q0 )δ (N) (m2 − q 2 )eiqx
10.3.2 Laplace–Beltrami Kernels

The inverse ∂12 ∈ D−1 (M) of the Laplace–Beltrami operator of a Riemannian
g
manifold (M, g) is a kernel, invariant under the global group Gg : For dimen-
sion s ≥ 2, the spheres, Euclidean spaces, and timelike hyperboloids with real
rank 1 have a parametrization with geodesic polar coordinates: There the ra-
dial function R −→ ρ(R) of the invariant measure is used in the characterizing
radial-spherical decomposition of the Laplace–Beltrami operator:
1
Ωs
for k = 0 ∼ Rs :
−1 Ys
2 s−1
2 ∂ d log ρ ∂ 1 ∂2 1 ∂ s−1 ∂ 1 ∂2
∂k,s = ∂R2 + ∂R + ρ2 ∂ωs−1
2 = ρs−1 ∂R ρ ∂R + ρ2 2 ,
dR ∂ωs−1
|g|ds x = ρ s−1
θ dRd
s−1
ω,
sin θ
with R = r , ρ(R) = r ,
ψ sinh ψ
2 2
1 ∂2
2 − k)ρ +
2 1 ∂ 2 ∂ 2
e.g. s = 2, 3 : ∂k,2 = ρ2 [(ρ ∂R ) + ∂ϕ2 ], ∂k,3 ∂
= ρ1 ( ∂R ρ2 ∂ω22 .
The Laplace–Beltrami kernel is given for the noncompact

y Euclidean spaces
√
and hyperboloids with the Riemannian metric d(x, y) = x g as follows:
k = 0, −1 : ∂k,s
2 1 s
2 (x, y) = δk (x, y),
∂k,s
1 1
d(x,y) dR
∂2
(x, y) = |Ωs−1 | 1 ρs−1 (R) .
k,s
The explicit expressions for the inverse of the Laplacian ∂s2 for Euclidean
positions Rs with the radial function ρ(R) = r involve the Kepler potential
r in R -position:
1 3
⎧
⎪ − |Ω 2 | (x) = s−2
s−1
⎪
⎪
1 1
( rs−2 − 1)
⎪
⎪ ∂
ds q 1 −iqx
⎨
s
|Ωs−1 | |x−y| dr = |Ω | (2π)s q 2 e

s−1
− 2−s
1
,
(
x ,
y ) = ⇒
⎪ − 2 (x) = log r ,
2π 1
s−1
∂s2 1 r
⎪
⎪
⎪
⎪
∂2
⎩ − 4π x) = 1r − 1.
2 ( ∂3
The homogeneous term gives the two-dimensional Laplace kernel in the limit
s → 2. For the hyperboloids Y s one uses as integrals over the radial function
ρ(R) = sinh ψ:
⎧ +
⎪
⎪ s = 3, 4, · · · : − sinhdψ 1
s−1 ψ = s−2
cosh ψ
⎪
⎪ sinhs−2 ψ
,
⎨
|Ωs−1 | +(s − 3) sinhdψ s−3 ψ ,
for ∂ 2 :
⎪ dψ
−1,s
⎪
⎪ s=2: − sinh ψ = log sinh ψ ,
cosh ψ
⎪
⎩ dψ
s=3: − sinh cosh ψ
2 ψ = sinh ψ .
The Laplace kernels for flat spaces with general signature can be taken
from
⎧ n
⎪ Γ( n2 −ν)
⎪
d q
⎪ it π n2 (−q2 −io) n2 −ν e
iqx
= −xΓ(ν)
2 +io ν ,
⎨ (
d4n q
)
SO0 (t, s) ×
R :
n
e.g. ∂ 2 −∂ 2 (x) = (2π)n −q21−io eiqx
1
⎪
⎪ t s
⎪
⎩
t
= in 2 −1)
Γ( n
n .
4π 2 (−x2 +io) 2 −1
For noncompact groups, the distributions with q12 have to be defined by a

complex integration prescription, e.g., by Feynman q 2 ± io.
10.4 Tangent Kernels and Interactions 287
10.4 Tangent Kernels and Interactions

For a manifold with cosets G/K ∼ = M ⊆ Rn , a linear transformation of
n
d x
the bounded functions L∞ (Rn ) ∼= C˙b (Rn ) by the pointwise product with a
−1
Green’s kernel D = κ ,0
M(M) κ0 : L∞ (M) −→ M(M), d −→ κ0 · d = κd ,
gives tangent kernels κ ∈ M(M × M) that act on group functions,
M(M) κd : Lp (M) −→ Lp (M), f −→ κd ∗ f.
For physics, the Radon distributions κ0 · L∞ (M) are called M-interactions.

κ0
Of special interest is the orbit L∞ (G)+ d −→ κd ∈ M(M) of the
cone with the positive-type functions for the cyclic group representations: In
analogy to the transition from a Lie group representation to the associated
Lie algebra representation, a Green’s kernel associates a tangent kernel (an
interaction) to each cyclic group representation. A Green’s kernel κ0 = κ0 · 1
as the image of the unit function d = 1 is associated with the trivial group
representation.
The Fourier transforms κ̃0 ∗ d˜ of tangent kernels κ0 · d are functions of the
translation eigenvalues ((energy-)momenta),
· M(Rn ) L∞ (Rn )
∞
L (R ) d
n
−→ κ = κ · d ∈ M(R )
d 0 n
M(Rn ) − M(Rn )
L∞ (Rn ) M(Rn ) L∞ (Rn )
F$ $F
∗ L∞ (Řn ) M(Řn )
M(Ř ) d˜
n
−→ κ̃ = κ̃ ∗ d˜ ∈ L∞ (Řn )
d 0 ∞
L (Ř ) n
− L∞ (Řn )
M(Řn ) L∞ (Řn ) M(Řn )
10.4.1 Position Interactions

The abelian group D(1) ∼ = R, used for causal time, is isomorphic to its Lie
algebra. The Green’s kernel, inverting the invariant derivative dxd 0 , acts as
identity; the abelian causal group, more generally. the abelian translations
Rn , induces only trivial interactions,
d
dx0 ϑ(x0 ) = δ(x0 ),
ϑ(x0 ) dq
ϑ(x0 )eimx0 −→ ϑ(x0 )eimx0 = 2iπ 1
q−io−m e
iqx0
= ( dxd 0 − im)−1 ,
κ̃0 (q) = 1q , m 1
κ̃ (q) = q−m .
1 1
q ∗ D 1 = log D 1
1 ∗
(∗, q) = ( 2iπ , q − io)
1
1
q ∗ 1
q−m = 1
q−m
Time kernels
Nontrivial interactions are given by tangent kernels of nonabelian (curved)

positions: For odd-dimensional hyperboloids Y 2R−1 with nonabelian degrees
of freedom R ≥ 2, the invariant kernels invert the powers of the Laplacian:
(4π)R Γ(R−N ) 2R−1 Γ(R−N ) −iqx Γ(N − 12 ) 2 2N −1

N = 1, . . . , R − 1 : = d πR q (
q 2 )R−N e = √π ( r ) .
2π(−∂2 )R−N
The action on the positive-type functions x −→ e−|m|r gives Yukawa-like

potentials for N = 1, . . . , R − 1 as inverse Laplacian with invariant m2 :
d2R−1 q Γ(R−N ) N −|m|r
−i (4π)R Γ(R−N )
π R ( 2 2
q +m ) R−N e q
x
= 2 = − d
2
e
|m|
2π(−∂ +m ) d4
2 R−N r
1 2 4(1+|m|r) −|m|r
= |m| , r , r3 ,... e ,
for N = 0, 1, 2, . . . , R − 1.
N = 0 characterizes positive-type functions, i.e., SO0 (1, 2R − 1)-representa-

d2R−1 q
tion coefficients. The kernel for N = 1 with integration ( q | d2(R−1) ω
q 2 )R−1 = d|
is the Kepler potential 21 R−1 ∼ 1r . The Yukawa potentials as position ker-
(−∂ )
nels constitute the position interactions:
, κ̃−m (
2
κ̃0 (
q 2) = 1
q 2 )R−N
( q 2) = 1
q 2 +m2 )R−N
( , N = 1, . . . , R − 1.
2R−1
1
∗ Y 2R−1 = log Y 2R−1
q 2 )R−N
(
2R−1 ∗ 2
∗ = |Ω2R−1 |
, 2R − 1 = 3, 5, . . .
2R−1 |m|
(
1
q 2 )R−N
∗ q 2 +m2 )R
(
= 1
q 2 +m2 )R−N
(
Position kernels (interactions)
10.5 Duality and Normalization

Bi- or sesquilinear forms and dual or scalar products of (generalized) func-
tions, e.g., essentially bounded functions L∞ (Rn ), put in duality with Radon
distributions M(Rn ),
10.5 Duality and Normalization 289
∗
M(Rn ) ∗ L∞ (Rn ) ⊆ L∞ (Rn ), κ, d = κ ∗ d(0) = κ̃− (2π) ˜
n d(0),
are given by the Schur product, i.e., by convolution products at the neutral
element (see Chapter 8). Two elements that are not orthogonal, κ, d = 0,
can be normalized as a dual pair κ, d = 1.
10.5.1 Schur Products of Feynman Measures

Schur products of O(t, s)-scalar Feynman measures, t+s = n, with real invari-
ants m2 ∈ R, e.g., for one factor with a trivial invariant, are (see Chapter 9):
dn q Γ( n Γ( n
2 +ν1 ) 2 +ν2 )
n n +ν n
it π 2 (−q −io+m1 ) 2
2 2 1 (−q −io+m22 ) 2 +ν2
2
1 ζ2
n +ν −1
1
n +ν −1
(1−ζ) 2 2 Γ( n 2 +ν1 +ν2 )
= dζ n
0 [ζm21 +(1−ζ)m22 −io] 2 +ν1 +ν2
m21 dκ2
n n
(κ2 −m22 ) 2 +ν1 −1 (m21 −κ2 ) 2 +ν2 −1 Γ( n2 +ν1 +ν2 )
= m2 m1 −m2
2 2 2 n ,
(m1 −m2 )n+ν1 +ν2 −2 (κ2 −io) 2 +ν1 +ν2
2 2
dn q Γ( n
2 +ν1 ) 1 2 −ν2 )
Γ( n Γ(ν1 +ν2 )
n n +ν
(−q 2 −io)ν2 = Γ( n (m2 −io)ν1 +ν2
.
t
i π 2 (−q −io+m )
2 2 2 1 2)
They can be written with a measure over an invariant line:

1 m21 dκ2
0
dζ = m22 m21 −m22
with κ2 = ζm21 + (1 − ζ)m22 ∈ [m22 , m21 ].
Schur products with orthogonal harmonic O(t, s)-polynomials (q)L (see

Chapter 8) are proportional to the harmonic units (projectors) (1n )L . They
involve characteristic multiplicity factors,
Γ(1+L) Γ( n2)

dn q μ(q 2 )(q)L ⊗ (q)L = δ LL 2L Γ( n dn q μ(q 2 )(q 2 )L (1n )L ,
2 +L)
with the simplest examples for L = 1, 2:

n n
d q μ(q 2 )q a qb = n1 d q μ(q 2 )q 2 δba ,
n ac
d q μ(q 2 )(q a q c − ηn q 2 )(qb qd − ηnbd q 2 )
n δ a δ c +δ a δ c ac
2
= n(n+2) d q μ(q 2 )(q 2 )2 ( b d 2 d b − η nηbd ).
The distributions μ are characterized by invariants of the O(t, s)-representa-

tions acting on (q)L . There remain scalar integrals, for example:
dn q 2 +ν1 ) (−q −io)
Γ( n 2 λ
Γ( n
2 +ν2 )
n n n
it π 2 (−q2 −io+m21 ) 2 +ν1 (−q2 −io+m22 ) 2 +ν2
Γ( n
2 +λ)
1 n +ν −1
2 ζ 1
n
(1−ζ) 2 +ν2 −1 Γ( n
2 +ν1 +ν2 −λ)
= Γ( n dζ n ,
2) 0 [ζm21 +(1−ζ)m22 −io] 2 +ν1 +ν2 −λ
dn q 2 +ν+λ)(−q −io)
Γ( n 2 λ
Γ( n
2 +λ) Γ(ν)
n n =
Γ( n (m2 −io)ν .
it π 2 (−q2 −io+m2 ) 2 +ν+λ 2)
10.5.2 Representation Normalizations for Spheres

and Hyperboloids
The definite orthogonal groups O(0, s) use −q 2 = q 2 for the Schur products:
s q )L ⊗(
ds q Γ( 2 +ν+L) ( q )L

O(s) : s s +ν+L = δ LL Γ(1+L)
2L
Γ(ν)
(m2 −io)ν
(1s )L ,
π2 q 2 −io+m2 ) 2
(
2d2R−1 q q )L ⊗(
q )L

O(2R − 1) : |Ω2R−1 | (
(
q 2 −io+m2 )R+L
= δ LL Γ(1+L) Γ(R)
2L Γ(R+L)
1
1 (12R−1 )L ,
(m2 −io) 2
2d2R q q )L ⊗( L
O(2R) : |Ω2R−1 | (
( q)
q 2 −io+m2 )1+R+L
= δ LL Γ(1+L) Γ(R)
2L
1
Γ(R+L) m2 −io
(12R )L .
ωs−1 ∈ Ωs−1 and the spherical harmonics

The products for the spheres
L
ωs−1 ) are
(
ds q s
Γ(1+L) Γ(1− 2 +N)
s −1 δ (N+L) (m2 − q )L ⊗ (
q 2 )( q)L = δ LL 2L 1− s +N (1s )
L
π2 i(−m2 ) 2
Γ(1+L)
= δ LL 2L π L
1− s +N (1s ) ,
Γ( s −N)(m2 ) 2
2ds q L s
Γ(1+L) Γ( 2 )
2
2 s −1−N
|Ωs−1 |
δ(N+L)
(m −
2
q ) ⊗ (
q )(2 L
q) = δ LL 2L s −N) (m ) 2
Γ( 2
(1s )L ,
dω s−1 s
Γ( 2 )
N +L=0: |Ωs−1 |
ωs−1 )L ⊗ (
( ωs−1 )L = δ LL Γ(1+L)
2L Γ( s +L)
(1s )L .
2
The energy invariants for the bound waves of the nonrelativistic hydrogen
atom are given by momentum multipole singularities 2E = q 2 = −Q2 in the
Fourier-transformed wave functions with the harmonic SO(3)-polynomials
q )L :
(

Y3 ∼
= SO0 (1, 3)/SO(3) x −→ | − Q2L , L}3 (x) ∼ d3 q q )L
(
q 2 +Q2L )2+L
π 2 ( eiqx ,
1
Q2
= (1 + L)2 , L = 2J = 0, 1, . . . .
L
In a noncompact–compact reciprocity, the continuous curvature invariants Q2

of the SO0 (1, 3)-representations for rank-1 position Y 3 ⊃ SO0 (1, 1) are de-
termined by the dimensions (1 + L)2 of SO(4)-representation spaces. As
will be shown in the following, this relation between continuous and in-
teger invariants reflects a representation normalization of the hyperboloid
functions.
The harmonic representation coefficients of odd-dimensional hyperboloids
Y 2R−1 ∼
= SO0 (1, 2R − 1)/SO(2R − 1), R = 1, 2, . . . ,
2d2R−1 q ρ2R−1,L (2q )L i
Y 2R−1 x −→ | − Q2L , L}2R−1 (x ) = |Ω2R−1 | (
q 2 +Q2L )R+L
e qx ,
with SO0 (1, 1)-representation invariants Q2L and normalization factors

ρ2R−1,L , have as dual products with harmonic O(2R − 1)-polynomials (q )L
and harmonic projectors (12R−1 )L :

2d2R−1 q q )L
{L | − Q2L , L}2R−1 = ρ2R−1,L |Ω2R−1 | (
(2
q 2 +Q2L )R+L
⊗ (q )L

= δ LL ρ2R−1,L Γ(1+L)Γ(R)
Γ(R+L)
1 L
|QL | (12R−1 ) .
10.6 Kernel Resolvents and Eigenvalues 291
The normalization conditions for the dual products, starting with the
scalar case | − Q20 , 0}2R−1 (0) = 1, give equations for the invariants:
{L| − Q2L , L}2R−1 = (12R−1 )L ⇒ ρ2R−1,L Γ(1+L)Γ(R) 1

Γ(R+L) |QL | = 1.
The representation normalizations differ from the wave function normal-

izations by the Schur products {−Q2L , L | − Q2L , L}2R−1 (see Chapter 8).
For the abelian case, R = 1, without rotation degrees of freedom, all
invariants are required to coincide:
Y 1 = SO0 (1, 1) ∼
= R : ρ1,L = |QL | = |Q| for all L.
If self-dual residual normalizations, where the powers 2L match the denom-
inator powers ( q 2 + Q2L )L , are used for all hyperboloids, the representation
invariants QL for SO0 (1, 2R − 1) are determined by rotation SO(2R − 1)-
2
invariants via dimensions of symmetric vector space products:

⎧
⎪
⎪ 1, R = 1,
⎨ 1 + L, R = 2,
|Q| Γ(R+L)
|QL | = Γ(1+L)Γ(R) = ⎪ (2+L)(1+L) , R = 3,
⎪
⎩ 2
...,
R+L−1
Γ(R+L)
-L
L = Γ(1+L)Γ(R) = dimK KR ,
L = 0 : |Q0 | = |Q| for all R.
The representation-normalized coefficients are
2R−1 q (2q )L |Q0 | iqx
x −→ | − Q2L , L}2R−1 (x ) = 2d |Ω2R−1 | (
q 2 +Q2L )R+L
e ,
/
q )L |Q0 |
q )L /q=0 = (12R−1 )L , L = 0, 1, 2, . . . .
(2 2R−1
q 2 +Q2 )R+L
(
∗ (
L
The invariants for the nonrelativistic hydrogen atom (Q2L , L) show an

equipartition Q20 = (1 + L)2 Q2L , similar to a flux quantization, of the ba-
sic invariant for the scalar L = 0 case to the dimensions of the symmetric
products of Pauli spinor spaces C2 :
|Q0 |
-
L
R=2: |QL | = 1 + L = dimC C2 .
The basic invariant Q20 is not determined. It can be normalized to 1 or used

as the energy (-momentum) unit, e.g., in the hydrogen atom given by the
Rydberg energy 12 MR c2 ∼ 14 eV with MR = me (αS )2 (see Chapter 4).
10.6 Kernel Resolvents and Eigenvalues

The Fourier transform of the convolutive action of an interaction, given by
a tangent kernel κ from the convolution algebra with the Radon group mea-
sures, on a vector space with group representations coefficients, given by
Lebesgue functions f on the group, parametrized by translations x ∈ Rn ⊇

M∼= G/K,
M(Rn ) κ : Lp (Rn ) −→ Lp (Rn ), f −→ κ ∗ f,
is a pointwise action of an (energy-)momentum function on the corresponding

Fourier-transformed functions, if defined, e.g., between the convolution and
F
pointwise product algebra as Fourier partners L1 (Rn ) f −→ f˜ ∈ C˙0 (Řn ).
With the eigenvalue equations,
κ ∗ f = f, κ̃ · f˜ = f˜,
the invariant eigenvalues, linear or quadratic, are given by the singularities

of the kernel resolvent:

kernel resolvent: q −→ Res κ̃(q) = 1−κ̃(q)
1
,
linear: ⎪
⎪
invariants: spec κ̃ = {I ⎪
⎪κ̃(I) = 1},

kernel resolvent: q 2 −→ Res κ̃(q 2 ) = 1−κ̃(q
1
2) ,
quadratic: 2⎪
⎪
⎪
invariants: spec κ̃ = {I ⎪κ̃(I ) = 1}.
2
The transition from kernel to resolvent, connected with a geometric series,

is parallel to the exponential transition from Lie algebra to group represen-
tation,
⏐l̃ →
− ⏐l
⏐ ⏐

$ $k
1
1−l̃
= l̃ −
k
→ e =
l l
k! ,
k≥0 k≥0
as illustrated for the abelian group D(1),

.
m
⏐ −→ imt = dq m iqt
2iπ⏐q2 e
q
⏐ ⏐ .
with dq 1 m 1
q , 1− m eiqt .
. 2iπ q q
1
1− m −→ eimt = dq 1
2iπ q−m e
iqt
q
A decomposition with respect to translation representations is given by

the principal parts of the resolvent at the invariants with the residues as
normalizations,
ρ(I)
linear: ResI κ̃(q) = q−I with 1
ρ(I) = − ∂κ̃
∂q (I),
ρ(I 2 )
quadratic: ResI 2 κ̃(q 2 ) = q −I 2
2 with 1
ρ(I 2 ) = − ∂q
∂κ̃ 2
2 (I ).
10.6 Kernel Resolvents and Eigenvalues 293
10.6.1 Spacetime Normalization

in the Nambu–Jona-Lasinio Model
The composite massless chiral Goldstone modes in the model of Nambu–Jona-
Lasinio, as described in Chapter 9, arise in a product involving a Feynman
2
propagator of interacting fields dM ,m (x − y) = C|Ψ
Ψ(y)Ψ
Ψ(x)|CFeynman with
a cyclic ground-state vector |C from the degeneracy manifold U(1). It is an
L∞ (R4 )-function of the spacetime translations,
2 d4 q (m2 −M 2 )2
R4 x −→ dM ,m
(x) = ρ(m2 ) πi (2π) γq+m
3 (q 2 +io−M 2 )2 q 2 +io−m2 e
iqx
(dipole regularization),
2 2
and comes as convolution dM ,m = −dM ∗ κm of a dipole with the Green’s
kernel κm of the equation of motion with the basic Dirac field for a free spin- 12
massive particle. The Green’s kernel, i.e., the free-field Feynman propagator,
is a spacetime distribution M(R4 ):
d4 q
(iγ∂ + m)κm (x) = δ(x) ⇒ κm (x) = − γq+m
(2π)4 q2 +io−m2 e
iqx
.
The scalar part of the propagator of the interacting Dirac field is Schur-
normalized as the spacetime representation coefficient by the mass consis-
tency equation for m = 0:
1 2
i
d4 q (m2 −M 2 )2 ρ(m2 )
1= 4m tr dM ,m
(0) = π (2π)3 (q2 +io−M 2 )2 q2 +io−m2
m2 ρ(m2 ) M 2 2
= 8π 2 ( m2 − 1 − log M m2 ).
The lowest-order equations for the translation invariants of the space-

time product representations (bound-states) are given for the functions of
the spacetime translation R4 x −→ Γ(x) = C|Ψ ΨΓΨΨ(x)|B (see Chapter 9):

Γ1 (x) = κΓΓ21 ∗ Γ2 (x),
for Γ1,2 ∈ {14 , γa , γ5 γa , γ5 } :
Γ̃1 (q) = κ̃ΓΓ21 · Γ̃2 (q).
The eigenvalue equations involve a convolution with the spacetime kernel

κ∗
matrix Lp (R4 ) −→ Lp (R4 ) or a multiplication with its energy-momentum
function matrix κ̃:
2
κΓΓ21 (x) = − tr Γ1 κm (x)Γ2 dM ,−m
(x),

ρ(m ) 2
γq−m 4 (m2 −M 2 )2 γq+m
κ̃ΓΓ21 (q) = 16π 2 tr (Γ1 ⊗ Γ2 ) ◦ q −m
2 2 (q −M ) q −m
2 2 2 2 2 .
The translation invariants are given by the singularities of the resolvent,

i.e., the solutions of the characteristic equation of the eigenvalue-dependent
(4 × 4)-matrix κ̃ΓΓ21 :
@ ⎪ A
⎪
1
1−κ̃ ⇒ I 2⎪
⎪ det [δΓΓ12 − κ̃ΓΓ21 (I)] = 0 .
The R4 -function for the composite chiral pseudoscalar field has as equation
of motion in a first-order approximation,

γ5 = κγγ55 ∗ γ5 ,
for C|Ψ
Ψγ5Ψ (x)|π = γ5 (x) :
γ̃5 = κ̃γγ55 · γ̃5 .
It involves the action of the chiral spacetime interaction (kernel) with an

energy-momentum function for the translation invariants:
2
,m κ
m 2 d4 q
dM −→ κγγ55 , with κγγ55 (x) = − 14 tr γ5 κm (x)γ5 dM ,−m
(x) = κ̃γ5 (q 2 )eiqx .
(2π)4 γ5
It can be simplified with the chiral properties,
γ5 (γp + m)γ5 = γp − m, γ5 κm γ5 = κ−m .
The eigenvalue equation [κ̃γγ55 (q 2 ) − 1]γ̃5 (q) = 0 has a solution κ̃γγ55 (0) = 1
for a Poincaré group representation with mass q 2 = 0, if compared with the
consistency equation for the chiral breakdown parameter m = 0,
d4 q
− d4 x κ−m (−x)dM ,m (x) = − κ̃−m (q)d˜M
2 2
,m 1 M 2 ,m
(2π)4 (q) = md (0) = 14 .
This equation is interpretable as the Schur normalization of the regular-

ized Feynman propagator as the spacetime representation coefficient with
the Green’s kernel of the free equation of motion in the form of a dual pair:
∗
κ̃γγ55 (0) = −κ−m ∗ dM (0) = −κ̃−m
2 2
,m
(2π)4 d˜M ,m (0) = 14 .
Chapter 11
Electroweak Spacetime
Unitary relativity in two complex dimensions is parametrized by electroweak

spacetime D(2) ∼ = GL(C2 )/U(2) as the noncompact real four-dimensional
base manifold of a coset bundle U(2)(D(2)) for the global group GL(C2 )
with the typical fiber U(2) = U(12 )◦ SU(2) as local group, i.e., the real four-
dimensional internal (chargelike) hyperisospin operations (see Chapter 6).
In the classical manifold interpretation, electroweak spacetime D(1) × Y 3
is a flat Friedmann universe with a hyperbolically curved position. Its re-
presentations are characterized by two continuous invariants for causal time
D(1) ∼= R+ and hyperbolic position Y 3 ∼ = R3 . The Hilbert representations,
as used for quantum theory, are infinite-dimensional for a nontrivial action of
the external operations D(1) × SO0 (1, 3). They will be related to spacetime
particles and interactions in Chapter 12.
11.1 Operational Spacetime

Metrical tensors contain representation coefficients of the motion (global)
group of a Riemannian manifold. For example, the metrical position coeffi-
cients in Friedmann universes, parametrizing classes of the local group SO(3)
(see Chapters 1 and 2),
dθ2 sin2 θ
x2
g = dτ 2 − R2 (τ )dσk2 , dσk2 = d
2 = dr 2 + r2 dω22 ,
(1+k
x
4 )
2
⎧
2 2
dψ sinh ψ
⎨ SO(4)/SO(3) ∼ = Ω3 , k = 1,
with SO(3) × R /SO(3) ∼
3
= R3 , k = 0,
⎩ ∼
SO0 (1, 3)/SO(3) = Y 3 , k = −1,
are from a real 10-dimensional Hilbert representation ( 12 , 12 ) ∨ ( 12 , 12 ) of the

compact motion group SO(4) for the spherical case Ω3 with an irreducible
one-dimensional contribution (0, 0) and a nine-dimensional one (1, 1) and
from a trivial representation of the Euclidean group SO(3) × R3 for flat
296 Chapter 11 Electroweak Spacetime
position R3 . For the noncompact hyperboloid Y 3 , the metrical coefficients

come from the real 10-dimensional Lorentz group SO0 (1, 3)-representation,
SO(3)
[ 12 | 12 ] ∨ [ 12 | 12 ] = [0|0] ⊕ [1|1] ∼
= 2 × [0] ⊕ [1] ⊕ [2]
SO(2)
∼
= 4 × (0) ⊕ 2 × (±1) ⊕ (±2),
which is not a Hilbert representation.

The irreducible Hilbert representations of the motion group SO(1 + s),
used for spheres Ωs ωs , act on the spherical harmonics (ωs )L with natural
winding number invariants L ∈ N (see Chapter 8), e.g., for Einstein’s static
universe (see Chapter 1), the SO(4)-representations ( L2 , L2 ),
2
R × Ω3 (τ, ω3 ) −→ (1, (ω3 )L ) ∈ C(1+L) ⊂ L2 (Ω3 ).
Positive-type functions for cyclic Hilbert representations of the noncom-

pact positions R3 and Y 3 , isomorphic as manifolds, are given by Fourier-
transformed momentum Dirac distributions (nonrelativistic scattering) and
dipoles (nonrelativistic hydrogen), in general for odd-dimensional maximally
symmetric positions (see Chapter 8):
2d2R−1 q
R2R−1 : x −→ 1
|Ω2R−2 | |P |2R−3
q 2 − P 2 )e−iqx = (−2 dPd2 r2 )R−1 cos P r
δ(
= cos P r, sinP Pr r , . . . ,
2d2R−1 q |Q| −i
Y 2R−1 : x −→ |Ω2R−1 | q 2 +Q2 )R e
(
qx

= e−|Q|r ,
√
R2R−1 ∼
= Y 2R−1 from R2R
x2 + 1

x = cosh ψ
sinh ψ ω2R−2
∈ Y 2R−1 .
The continuous momentum translation invariants P 2 ∈ R+ are for flat

Euclidean positions R2R−1 , and the continuous curvature invariants Q2 ∈ R+
are for hyperbolic positions Y 2R−1 .
In Hilbert representations of spacetime, the positive-type functions for
position have to be considered together with Hilbert representations of time
τ −→ R(τ ), built by the irreducible ones τ −→ eimτ .
Electroweak spacetime is constituted by the classes of the hyperisospin
operations U(2) = U(12 ) ◦ SU(2) in the extended Lorentz group operations
GL(C2 ) = D(12 ) × U(12 ) ◦ SL(C2 ) with the dilation group D(1) = exp R,
the phase group U(1) = exp iR, and the Lorentz (cover) group SL(C2 ):
D(2) ∼
= GL(C2 )/U(2) ∼
= D(1) × Y 3 ,
Y3 ∼
= SL(C2 )/SU(2) ∼
= SO0 (1, 3)/SO(3).
It is the real four-dimensional member D(2) = D4 in two series: the causal

unitary series with the real n2 -dimensional unitary classes of general linear
groups,
D(n) ∼= GL(Cn )/U(n) ∼ = D(1) × SL(Cn )/SU(n),
11.2 Representations of Electroweak Spacetime 297
and the causal orthogonal series with real (1 + s)-dimensional manifolds,

embedding hyperboloids,
D1+s ∼
= D(1) × Y s , Y s ∼
= SO0 (1, s)/SO(s).
One direct factor is the totally ordered abelian dilation group, called causal
group or (eigen)time,
n = 1, s = 0 : D(1) = D1 = D(1) ∼
= eR ∼
= R.
The position factors are globally symmetric spaces (see Chapter 2) of non-
compact type K × K ∗ /K in the case of SL(Cn )/SU(n) and of type BD I
for SO0 (1, s)/SO(s).
The two series meet exactly for (n, s) = (1, 0) and for (n, s) = (2, 3).
Abelian unitary relativity in one dimension, D(1) ∼
= GL(C)/U(1), can be re-
lated to quantum mechanics as a theory of time D(1) realizations in Hilbert
spaces with probability amplitudes. Correspondingly, nonabelian unitary rel-
ativity in two dimensions, D(2) ∼ = GL(C2 )/U(2), will be connected with
quantum field theory as a theory of spacetime realizations.
The Iwasawa factorizations G = K ◦ A ◦ N of the involved special linear
groups into maximal compact group, maximal abelian noncompact group and
a subgroup with a nilpotent Lie algebra, and the minimal parabolic subgroups
G ⊇ (K0 × A) ◦ N are
SL(Cn ) = SU(n) ◦ SO0 (1, 1)n−1 ◦ exp Rn(n−1)

⊇ [SO(2) × SO0 (1, 1)]n−1 ◦ exp Rn(n−1) ,
SO0 (1, s) = SO(s) ◦ SO0 (1, 1) ◦ exp Rs−1
⊇ [SO(s − 1) × SO0 (1, 1)] ◦ exp Rs−1 ,
SL(C ) ∼ SO0 (1, 3) = SO(3) ◦ SO0 (1, 1) ◦ exp R2
2
⊇ [SO(2) × SO0 (1, 1)] ◦ exp R2 .
Therefore, the manifolds D(n) for n ≥ 1 and D1+s for s ≥ 1 have, respec-
tively, real rank n and 2 as the dimensions of a maximal abelian noncompact
subgroup D(1) × SO0 (1, 1)n−1 and D(1) × SO0 (1, 1). The real rank gives
the maximal number of the representation-characterizing invariants from a
continuous spectrum.
11.2 Representations of Electroweak Spacetime

In the following, electroweak spacetime will be considered as a member of the
orthogonal series with the real rank-2 causal spacetimes D1+s ∼ = D(1) × Y s ,
s ≥ 1. A parametrization of D 1+s
is possible by the open future cone of
Minkowski spacetime with Lorentz group action SO0 (1, s) × R1+s :
@ ⎪ A
⎪
⎪
D1+s = x = xx0 = eψ0 cosh ψ
⎪ ω
s−1 ∈ Ω s−1
, ψ 0 , ψ ∈ R
⎪
sinh ψ ωs−1
1+s ⎪

= R1+s
+ = {x ∈ R ⎪
⎪x 2
> 0, x0 > 0}.
The future cone translation parametrization of causal spacetime is a finite-

dimensional and nonunitary representation, e.g., the Weyl representation of
electroweak spacetime,

x1 − ix2
D4 x = xx10 ++ ix x3
= u(
x
) ◦ x0 + r 0
◦ u ( xr ), x0 ± r > 0,
2
x0 − x3 r ψ00+ψ x0 − r
D(1) × SO0 (1, 1) x0 0+ r x0 0− r = e 0 0
ψ0 −ψ , eψ = xx00 −r +r
,
−iϕ
e
Ω2 xr −→ u( xr ) = eiϕcossin2 θ −e cos sin

θ θ
2
2
θ
2
−x1 + ix2
=√ 1 x3 + r
x1 + ix2 x3 + r ∈ SU(2).
2r(x3 +r)
The dilations D(1) ∼ = R+ constitute, in a relativistic framework, the group

for strictly positive “eigentime”:
√
eψ0 = τ = |x| = ϑ(x) x2 ∈ D(1), with ϑ(x) = ϑ(x0 )ϑ(x2 ).
The hyperboloids Y s ∼ = Rs with eψ0 | sinh ψ| = r are the position submanifolds.

In contrast to the D(1) × SO0 (1, 1)-related group parameters (ψ0 , ψ), the
translation parametrization (x0 , r) is appropriate for a transition R4+ ⊂ R4
to tangent space structures with particles.
D1+s is acted on by D(1) × SO0 (1, s). For the future cone, foliated by
position hyperboloids Y s , the action of the causal group D(1) may be called
hyperbolic hopping, from position hyperboloid to position hyperboloid,

x0 ψ0 cosh ψ λ0 x0 λ0 +ψ0 cosh ψ
D(1) : x3 = e sinh ψ −
→ e x3 = e sinh ψ ,
and the action of the dilative Lorentz subgroup SO0 (1, 1) ⊆ SO0 (1, s) hy-
perbolic stretching, inside each position hyperboloid,

x0 ψ0 cosh ψ cosh λ sinh λ x0 ψ0 cosh(λ + ψ)
SO0 (1, 1) : x3 = e sinh ψ −
→ sinh λ cosh λ x3 = e sinh(λ + ψ) .
The rotations SO(s − 1) act on each position hyperboloid.

As classical manifold, D1+s = D(1) × Y s has a Robertson–Walker metric
for a hyperbolic Friedmann universe (see Chapter 1):
g = T 2 (ψ0 )(dψ02 − dy2s ) = dτ 2 − R2 (τ )dy2s .
Spacetime D4 ∼
= R4+ has trivial curvature:

Rdabc ∼
[( TT )2 − TT ]13 0
= − T12 0

[( TT )2 − 1]13

∼
= − R12
R̈R1 3 0
= 0 for T (ψ0 ) = eψ0 , R(τ ) = τ.
0 (1 − Ṙ )13
2
The linear expansion factor (position “radius”), R(τ ) = τ ∈ D(1), cannot

be unitarily analyzed with invariant masses τ −→ eimτ . Also, the metrical
coefficients of the hyperboloid do not belong to Hilbert representations.
In the following, D4 is not considered as a classical manifold, e.g., as a

cosmological model, but as a parametrization of the spacetime operations.
The complex functions of the future cone,
GL(C2 )/U(2) ∼
= R4+ ϑ(x)x −→ ϑ(x)f (x) ∈ C,
are the coefficients of unitary relativity. They arise in all representations of

GL(C2 ) containing trivial hyperisospin U(2)-representations.
The hyperisospin U(2)-induced representations of the extended Lorentz
group GL(C2 ) are subrepresentations of the two-sided regular represen-
tation of GL(C2 ). They act on U(2)-intertwiners, w : GL(C2 ) −→ W
with w(gu−1 ) = d(u)w(g), i.e., on mappings of the causal cone R4+ −→ W
that connect spacetime points, parametrizing unitary classes, with hy-
perisospin orbits in a Hilbert space W with U(2)-representation
d(u) ∈ U(W ).
The action group SL(C2 ) × U(2) with external and internal transfor-
mations as subgroup of the both-sided regular GL(C2 ) × GL(C2 ) action is
realized in the electroweak standard model (see Chapter 6): The representa-
tions are faithful for SL(C2 )×U(2) on the left-handed isodoublet lepton field,
for SO0 (1, 3) × SO(3) on the isospin gauge vector field, for SL(C2 ) × U(1)
on the right-handed isosinglet lepton field, and for SO0 (1, 3) × {1} on the
hypercharge gauge vector field. With the notable exception of the Higgs field,
all isospin SU(2)-representations of the standard model fields are isomorphic
to subrepresentations of their Lorentz group SL(C2 )-representations.
11.2.1 Harmonic Analysis of the Causal Cartan Plane

Causal spacetime D1+s ∼= D(1)× Y s has real rank 2 as dimension of a Cartan
plane D = D(1) × SO0 (1, 1), with causal operations and Lorentz dilations
2
as maximal noncompact abelian group.

The characteristic function of the causal Cartan line D(1) and plane D2
can be unitarily analyzed (Fourier-transformed) with an advanced energy-
momentum measure with a pole at a trivial invariant q0 = 0 and q 2 = 0,
respectively,
dq0 1 iq0 x0
for D(1) and SO0 (1, 1) : ϑ(x0 ) = 2iπ q0 −io e ,
d2 q
for D(1) × SO0 (1, 1) : ϑ(x) = 1
2π 2 −(q−io)2 e
iqx
,
with (q − io)2 = (q0 − io)2 − q32 .
The Fourier transform of an energy-momentum function, holomorphic in

the lower (upper) complex energy q0 plane, i.e., with q0 ∓ io, is val-
ued in the future (past) cone, i.e., supported by causal line and plane
with ϑ(±x0 ).
In a Lorentz group action-compatible translation parametrization, the

harmonic analysis of the Cartan plane uses SO0 (1, 1)-invariants q 2 = m2 as
energy-momentum singularities:
dq0 1
eiq0 x0 = ϑ(x0 )eimx0 ,
D(1) and SO0 (1, 1) : dq0 2iπ q0q−io−m
0 iq0 x0
2iπ (q0 −io)2 −m2 e = ϑ(x0 ) cos mx0 ,
d2 q
D(1) × SO0 (1, 1) : 1
2π 2 −(q−io)2 +m2 e
iqx
= ϑ(x)J0 (|mx|).
For more than one dimension, t + s ≥ 2, the (anti-)Feynman energy-mo-

1
mentum measures with −q2 ±io+m 2 , possible for any SO0 (t, s)-metric with
any real invariant m ∈ R, cannot be combined by causal (advanced and

2
retarded) measures with −(q0 ∓io)12 +q 2 +m2 that require an SO0 (1, s)-metric
and a positive invariant m2 ≥ 0.
The Bessel functions with half-integer index, starting with
∞
$
−imx
(−m2 x2 )k
cos mx = 1 imx
2 (e +e )= dq |q|δ(q − m )e
2 2 iqx
= (2k)! ,
k=0
and an invariant m2 , normalized with |Ω0 | = 2 for the two irreducible D(1)-
representations on a 0-sphere {±im} ∼ = Ω0 , are for odd-dimensional spaces
with (energy-)momenta as eigenvalues, the corresponding functions with in-
teger index for even-dimensional spaces. They start with J0 and an invariant
circle {im cos θ} ∼
= Ω1 , normalized with |Ω1 | = 2π (see Chapter 8). The
integer-index Bessel functions integrate over R-representation coefficients,
π
J0 (mx) = 2π 1 1
−π dθ cos(mx cos θ) = π dψ sin(|mx| cosh ψ)
$∞
2 2
(− m 4x )k
= π1 −m √ dq
m
cos qx = (k!)2 ,
m −q
2 2
k=0
with, for causal Cartan (1, 1)-spacetime, the projections on oscillating repre-
sentation coefficients of the causal operation D(1) (time) and on exponen-
tially decreasing representation coefficients of the Lorentz dilations SO0 (1, 1)
(position):

D(1) : dx3 ϑ(x)J0 (|mx|) = ϑ(x0 ) sinm
mx0
,
D −→
2 e−|mx3 |
SO0 (1, 1) : dx0 ϑ(x)J0 (|mx|) = |m| .
The invariants for the D(1)-oscillations and the SO0 (1, 1)-fall-off must not
to coincide. Two continuous invariants for the rank-2 causal plane are
implemented, in a residual representation, by two poles in the complex
energy-momentum plane for the L2 (D2 )-functions:
d2 q
D(1) × SO0 (1, 1) ϑ(x)x −→ 1
2π 2 [−(q−io)2 +M 2 ][(q−io)2 −m2 ] e
iqx
.
The two poles can be taken as the endpoints of a finite SO0 (1, 1)-invariant
singularity line q 2 = κ2 ∈ [M 2 , m2 ], characteristic for even-dimensional
spaces,
d2 q 1 d2 q
2π 2 [−(q−io)2 +M 2 ][(q−io)2 −m2 ] e = − 0 dζ
1 iqx 1 iqx
2π 2 [−(q−io)2 +ζM 2 +(1−ζ)m2 ]2 e
m2 dκ2 d2 q
= − M 2 m2 −M 2 2π2 [−(q−io) 1
2 +κ2 ]2 e
iqx
m2 dκ2
2 J0 (|κx|)
∂
= ϑ(x) M 2 m2 −M 2 ∂κ
= ϑ(x) J0 (|mx|)−J0 (|Mx|)
m2 −M 2 .
Ahead, the pointwise product of the two energy-momentum poles, i.e., the
convolution product of the two spacetime functions,
d2 q 1
2π 2 [−(q−io)2 +M 2 ][(q−io)2 −m2 ] e
iqx
= −ϑ(x)J0 (|mx|) ∗2 ϑ(x)J0 (|M x|),
will be related to the product structure of the represented group, for eigentime
D(1) and for 1-position SO0 (1, 1) ∼ = Y 1.
In the non-Lorentz-compatible direct product form, the residual represen-
tations of the group for Cartan spacetime give the product of two functions,
for example,
0 dq3
D(1) × SO0 (1, 1) (ϑ(x0 )x0 , x3 ) −→ dq 2iq0
(2iπ)2 [q2 +M 2 ][(q0 −io)2 −m2 ]
3
e−|M x3 |
ei(q0 x0 −q3 x3 ) = ϑ(x0 ) cos mx0 |M| .
11.2.2 Harmonic Analysis of Even-Dimensional

Causal Spacetimes
A causal Cartan plane is a maximal noncompact abelian group in the product
of the causal group (eigentime) and a Lorentz group for nontrivial position
dimension,
D(1) × SO0 (1, 1) ⊆ D(1) × SO0 (1, s), s = 1, 2, . . . .
As familiar from Cartan’s B- and D-series, the orthogonal groups SO0 (1, s)
come in two basically different types, those for odd dimensions and those for
even dimensions.
The lowest-dimensional nonabelian Lorentz groups for odd and even di-
mensions are locally isomorphic, respectively, to the lowest-dimensional non-
trivial special real and special complex groups:
SO0 (1, 2) ∼ SL(R2 ), SO0 (1, 3) ∼ SL(C2 ).
For even spacetime dimensions 1 + s = 2R, the noncompact Cartan sub-

groups SO(2)R−1 × SO0 (1, 1) nontrivially fill the whole diagonal (diagonal-
ized in the complex), whereas for the more complicated odd-dimensional
case 1 + s = 2R − 1 with two different types of Cartan subgroups, either

noncompact SO(2)R−2 × SO0 (1, 1) or compact SO(2)R−1 (tori), there re-
mains one trivial diagonal unit, starting with
⎛ψ ⎞
ψ e 0 0 0
SO0 (1, 3) ⎝ 0 e 0 ⎠
iϕ
SO0 (1, 1) e 0
−ψ ,
0 e
0
−iϕ
0
, 0 e 0
e−ψ
1 0
0 0
eψ 0 0 0 0
SO0 (1, 2) 0 1 0
−ψ
, 0 eiϕ
−iϕ
0.
0 0 e 0 0 e
The Cartan subgroup structure is revelant for the existence and nonexistence
of discrete series representations in the case of SO0 (1, 2R) and SO0 (1, 2R−1),
respectively (see Chapter 8).
The universal cover group of an odd-dimensional group SO0 (t, s), t + s =
2R − 1 = 3, 5, . . . , has one fundamental spinor representation with dimension
2R−1 , e.g., the SU(2)-Pauli spinors for the rotations SO(3), whereas for even-
dimensional groups t + s = 2R = 4, 6, . . . , there are two fundamental spinor
representations with dimension 2R−1 , e.g., the left- and right-handed SL(C2 )-
Weyl spinors for the Lorentz group SO0 (1, 3).
The characteristic future functions in the translation cone parametri-
zations have the harmonic analysis with an advanced energy-momentum mea-
sure (q − io)2 = (q0 − io)2 − q 2:
⎧ x 2R−1
⎪
⎪ ϑ(x) |x||x|
= |Ωd q 1 iq
1 e
iqx
,
⎪
⎪
2R−1 | [−(q−io)2 ]R
⎨ 1 + s = 2R − 1 = 1, 3,
SO0 (1, s) : d2R q . . . 1,
⎪
⎪
1
ϑ(x) x = 1 iqx
⎪ π|Ω2R−1 | [−(q−io)2 ]R −(q−io)2 e ,
2iqR
⎪
⎩
1 + s = 2R = 2, 4, . . . .
The spherical degrees of freedom show up in the order of the pole and in the
normalization of the integration d1+s q by the measures of the n-dimensional
1+n
unit spheres |Ωn | = 2π 2
Γ( 1+n
with the full-dimensional |Ω1+s | = |Ω2R−1 | for
2 )
odd dimensions and the product |Ω1 | |Ω 2 | for even dimensions.

2R−1
The analogue to the one-dimensional dilation-invariant residual energy- or

momentum measure i|Ωdq1 |q , normalized with the circle |Ω1 | = 2π, is, for even
dimensions, the dilation and SO0 (1, 2R − 1)-invariant doubled spherically
2R
d2 q
normalized energy-momentum measure i|Ω1 ||Ω2d q
2R−1 |(q 2 )R , starting with 2iπ 2 q 2
for the Cartan plane. The hyperbolic SO0 (1, 2R − 1)-invariant measures,
not dilation-invariant for m2 = 0, contain the derivative ( ∂q∂ 2 )R−1 of the
Cartan plane measures, “compensating” the corresponding factor (q 2 )R−1 in
the measure d2R q:

Γ(R) ∂ R−1 1
d2R q Γ(R)
: (−q2 +m2 )R = ( ∂q2 ) −q2 +m2 ,
R 2 2 R
π (−q +m )
2ϑ(q)d2R q = d2R−1 y (q 2 )R−1 dq 2 .
The invariant measures of unitary relativity,

2
dn q
D(n) = GL(Cn )/U(n) : ( det q)n , with q = q ∗ ∈ GL(Cn ),
4 4
q for real q ∈ D(1) and the Lorentz invariant ( det q)2 = (q2 )2 for
start with dq d q d q
energy-momenta q ∈ D(2).
The causal spacetime coefficients with one invariant are
d2R−1 q
iq
SO0 (1, 2R − 2) : |Ω 1
2R−1 | [−(q−io)2 +m2 ]R 1 e
iqx
x cos |mx|
= ϑ(x) |x|sin m|x| ,
2R
m 1

SO0 (1, 2R − 1) : π|Ωd 2R−1
q 1
| [−(q−io)2 +m2 ]R −(q−io)2 +m2 e
2iqR
iqx

= ϑ(x) x1 J0 (|mx|).
The translation representation xϑ(x) ∈ D2R has an order-(1 + R) pole with

trivial invariant m2 = 0.
The Fourier transform of the advanced integration is, up to the order
function, the corresponding principal value off-shell integration:
d1+s q Γ(1+N ) 1+s
π [−(q−io)2 +m2 ]1+N e
iqx
= 2ϑ(x0 ) d π q (−qΓ(1+N
2
)
2 1+N e
iqx
,
P +m )
d1+s q Γ(1+N )
π (−q2 +m2 )1+N
e iqx
= i(x0 ) d q (q0 )δ (q − m )e .
1+s (N ) 2 2 iqx
P
Up to the causal order function ϑ(x), the SO0 (1, 2R − 1)-representation

coefficients for noncompact spacetime with hyperbolic position Y 2R−1 coin-
cide with the SO(2R)-coefficients for its compact spherical partner Ω2R−1 .
The coefficients for the spheres are given by the Dirac contribution:
2d2R q
SO(2R) : 1
q 2 −io−m2 )R e
2R−1 | (
i
q
x
= [J0 + iN0 ](|mx|),
iπ|Ω
2R
d q (R−1)
πR δ (q − m )e
2 2 i q
x
= J0 (|mx|).
Even-dimensional spacetime D2R with two continuous invariants (real rank
r = 2) is represented as the Fourier-transformed product of two energy-mo-
mentum distributions, one with a simple pole and the other one with a pole
of order R = 1, 2, . . . , in the Lorentz scalar L2 (D2R )-functions:
2R
D2R ϑ(x)x −→ π|Ωd 2R−1 q 1
| [−(q−io)2 +M 2 ]R [(q−io)2 −m2 ] e
iqx
1 2R
= − 0 dζ ζ R−1 π|Ωd 2R−1 q R
| [−(q−io)2 +ζM 2 +(1−ζ)m2 ]R+1 e
iqx
m2 R 2 d2R q
= − M2 d κ R
π|Ω2R−1 | [−(q−io)2 +κ2 ]R+1 e
iqx
m2 R 2 ∂
= ϑ(x) M 2 d κ ∂κ2 J0 (|κx|).
The spacetime coefficients contain the characteristic rank-dependent integra-
tion with a measure over a line with SO0 (1, 2R − 1)-invariants:
1 m2
dζ ζ R−1 = M 2 dR κ2 with κ2 = ζM 2 + (1 − ζ)m2 ,
0 2 2 R−1
2
m −κ
where dR κ2 = m2dκ−M 2 m2 −M 2 , R = 1, 2, . . . .
Lorentz group SO0 (1, 2R − 1)-properties, and nontrivial properties with

respect to the maximal compact group SO(2R − 1) for R ≥ 2, are obtained
2 ∼ iq and harmonic SO0 (1, 2R − 1)-polynomials
∂ ∂
by derivations ∂x = 2x ∂x
(see Chapter 12).
The simple pole embeds the representation of the abelian time oper-
ations D(1) ∼ = R with rank 1. The representation of hyperbolic position
Y 2R−1 ∼ = R2R−1 with the rank-R Lorentz group SO0 (1, 2R − 1) as motion
group is embedded by the order R pole. With the r = 2 (real rank) contin-
uous invariants and the rc = R − 1 (imaginary rank) discrete invariants of
the orthogonal group SO(2R − 1) with a Cartan torus SO(2)R−1 , the acting
group D(1) × SO0 (1, 2R − 1) has rank r + rc = R + 1.
Obviously, the representations of spacetime D2R differ from those of cor-
responding flat spacetime SO0 (1, 2R − 1) × R2R /SO0 (1, 2R − 1) ∼
= R2R , used
with R = 2 for particles. They are not Feynman propagators. Higher-order
poles, here at q 2 = M 2 for nonflat position with R ≥ 2, cannot be connected
1 2
with particle propagators like q2 +io−m 2; M cannot be used as particle mass,
i.e., as invariant for tangent translations R .
2R
Unitary relativity D4 = D(1) × Y 3 with three space dimensions s = 3

as the minimal nonabelian case has imaginary rank 1 for the SO(3)-rotation
degrees of freedom. The D4 -representation coefficients with two continuous
invariants (real rank 2) involve a position-representing dipole [33], e.g., with
Lorentz group vector properties,
d4 q
GL(C2 )/U(2) ∼
= R4+ ϑ(x)x −→ 2iπ3 [−(q−io)2 +M 2 ]2 [(q−io)2 −m2 ] eiqx
q
∂
m 2
2 2
d4 q 2 iqx
= ∂x M2 d κ 2π 3 [−(q−io)2 +κ2 ]3 e
m 2 2 ∂
2
= −ϑ(x) ∂x ∂
M2 B ∂κ2 J0 (|κx|)
d κ C
= ϑ(x) m2 −M 2 x2 J0 (|M x|) − ∂x2 J0 (|Mx|)−J
x ∂
M −m
2
0 (|mx|)
2 .
∂ 4 ∂ 4
The normalization of harmonic spacetime representations with tensor

properties (q)L will be discussed below and in the next chapter.
11.3 Time and Position Subrepresentations

A group representation represents all subgroups. The projections of the re-
presentations of unitary relativity GL(C2 )/U(2), i.e., of causal spacetime
D4 = D(1) × Y 3 with the action of D(1) × SO0 (1, 3), to those of the factors
causal group and Lorentz group lead, respectively, to representations of free
particles and of interactions: Free particles are related to representations of
the causal group D(1) ∼= R, Lorentz compatibly embedded in representations
of causal spacetime D4 , whereas interactions are related to embedded repre-
sentations of hyperbolic position Y 3 , the isospin SU(2)-classes of the Lorentz
group SL(C2 ).
11.3 Time and Position Subrepresentations 305
11.3.1 Projection to Subgroup Representations

For a closed Lie subgroup H ⊆ G and Haar measures, the (generalized)
G-functions can be projected [36] to (generalized) G/H-functions by integra-
tion over the subgroup (where defined):

μ(g) −→ μ(gH) = H dh μ(gh),
e.g., δk (g) −→ δk (gH) = H dh δ(k −1 gh).
Under certain conditions, related to unimodularity [36], there is the inte-

gral decomposition with respect to the subgroup with suitably normalized
invariant measures:

G
dg μ(g) = G/H dgH μ(gH) = G/H dgH H dh μ(gh).
Examples are representations of translations Rn , which are projected to

representations of subgroups Rn−k by integrations over the subgroup Rk :
dx iqx
R −→ {0} : eiqx −→ 2π e = δ(q) ∼= 1,
dk x ipy+iqx
R −→ R
n n−k
: e ipy+iqx
−→ (2π)k e = [δ(q)]k eipy ∼
= eipy .
Other examples are (semi)direct product groups G = K × H with normal

subgroups H and subgroups K ∼ = G/H: Representations of Euclidean groups
can be projected to lower-dimensional ones, e.g., for three-dimensional posi-
tion translations with the chain of positive-type functions for the subgroups
j0 −→ J0 −→ cos,

SO(3) × R3 −→ SO(2) × R2 : dx3 j0 (P r)
2π 3 d3 q
= dx 2πP δ( q 2 − P 2 )eiqx
d2π2 q
dx22πP δ(q − P cos
= 2
)e qx = J0 (|P
2 i
2P
x|)

,
−→ R : 2π J 0 (|P
x |) = P x1
|P | .
Particle representations of the Poincaré group have nontrivial projections for

time translations and trivial ones for the Euclidean group with momentum
q = 0:

4
d3 x
SO0 (1, 3) ×
R4 −→ R : d q δ(q 2 − m2 )eiqx = cos|m|
mx0
,
4
(2π)3
SO0 (1, 3) ×
R4 −→ SO(3) ×
R3 : dx0 d q δ(q − m )e
2 2 iqx
= 0.
The decomposition with respect to time representation shows the positive-

type functions (spherical Bessel function) of irreducible representations of the
Euclidean group for nontrivial momenta q 2 = q02 − m2 > 0:
SO0 (1, 3) ×
R4 ⊃ [SO(3) × R3 ] × R,
√
4 ∞ sin q02 −m2 r
d q
2π δ(q 2 − m2 )eiqx = m dq0 r cos q0 x0 .
11.3.2 Projection to and Embedding of Particles

and Interactions
The representation coefficients of the acting product group for even-
dimensional causal spacetime D2R , R = 1, 2, . . . ,
⎧
⎨ D(1) × Y 2R−1 = D2R
D(1) × SO0 (1, 2R − 1) for ∼
= R+ × R2R−1 = R2R + ,
⎩
M(R2R
+ ) ∗ L2 (R2R
+ ) = L 2
(R2R
+ ),
are the convolution products of a causally embedded coefficient d2R for the
Lorentz group SO0 (1, 2R − 1) with a bounded function |J0 (r)| ≤ J0 (0) = 1
and a Lorentz compatibly embedded Radon measure κ1 for the causal group
(eigentime) D(1):

d2R (x) d2R q 1
κ1 (x) ∼ π|Ω2R−1 | [−(q−io)2 +m2 ]R

iq eiqx
1

−(q−io)2 +m2

= x2 Γ(R) ∂ 2 R ϑ(x)J0 (|mx|),
∂x
d2R q 4
d2R ∗ κ (x) ∼ π|Ω2R−1 | [−(q−io)2 +M 2 ]iq

1
R [−(q−io)2 +m2 ] e
iqx
.
The Radon distributions for N = 0, 1, . . . have the explicit form
1+N
∂
x2
ϑ(x2 )J0 (|mx|)
∂ 4
$ 0 1+N
(m2 )k (N +k) x2
= k! δ (− 4 ) + ϑ(x2
) ∂
x 2 J0 (|mx|).
∂ 4
k=−N
The factors in the non-Lorentz-compatible direct product representa-

tions of Cartan spacetime D(1) × SO0 (1, 1) (ϑ(x0 )x0 , x3 ) −→ ϑ(x0 )
cos mx0 e−|Mx3 | are obtained from the Lorentz-compatible (generalized)
functions of causal spacetime by projection to (generalized) functions of time
and of hyperbolic position:

D(1) ϑ(x0 )x0 −→ d2R−1 x ϑ(x)μ(x),
D2R ϑ(x)x −→ ϑ(x)μ(x) ⇒
Y 2R−1 x −→ dx0 ϑ(x)μ(x).
The time projections by integration over position, i.e., for trivial momenta
x −i
d2R−1 2π e qx = δ(q ),

π|Ω2R−1 |d2R−1 x 1
R
for D(1) : (2π)2R
x
2 Γ(R)
∂ ϑ(x)J0 (|mx|)
x2
∂ 4 R−1
0 [−(q −io)2 +m2 ]R iq x
1
sin mx0
Γ(R) − ∂m2
1 ∂
= dq 2π
0
iq0 e 0 0
= ϑ(x0 ) m ,
−(q0 −io)2 +m2 cos mx0
contain in the lower component a representation coefficient ϑ cos for time:

dq0
D(1) ϑ(x0 )x0 −→ ϑ(x0 ) cos mx0 = 2iπ q0
(q0 −io)2 −m2 e
iq0 x0
,
R
→ x2 Γ(R) ∂x2 ϑ(x)J0 (|mx|)
∂ 4
d2R q iq iqx
= π|Ω 2R−1 | −(q−io)2 +m2 e .
11.4 Hilbert Spaces for Causal Spacetimes 307
The real invariants q0 = ±m can be used for free particles as translation

invariant q 2 = m2 in a Poincaré group representation. For R ≥ 2, the upper
component displays matrix elements of indefinite unitary faithful translation
representations [9] (half-integer-index spherical Bessel functions [57]):
R−1 R−2
sin mx0 −mx0 cos mx0
− ∂m
∂
2
sin mx0
m = − ∂
∂m 2 2m3
R−1
= x2R−1
0 − ∂t∂2 sin t
t , t = mx0 .
The projections on the position hyperboloid with the SO0 (1, 2R − 1)-
coefficients by integration over time, i.e., for trivial energy dx 0 iq0 x0
2π e =
δ(q0 ),
1
R
for Y 2R−1
: dx0 2 Γ(R) 2
x ∂ ϑ(x)J0 (|mx|)
4 1
∂x

2d2R−1 q (q 2 +m2 )R −iqx
= |Ω2R−1 | i
q e
q 2 +m2

e−|m|r

|m|
= ,
x e−|m|r
R−1 R−1
4m2
Γ(R) r2
|m| − ∂m2
∂
r |m|
give, in the upper component, an exponential potential as a positive-type

function exp ∈ L∞ (Y 2R−1 )+ for hyperbolic position with imaginary invari-
ants |
q | = ±im. The inverse invariant of the SO0 (1, 1)-representations is the
1
characteristic interaction range |m| :
2d2R−1 q |m|
Y 2R−1 x −→ e−|m|r = |Ω2R−1 | (q 2 +m2 )R e
−i
q
x
,
2R
d q 1 iqx
→ ϑ(x)J0 (|mx|) = π|Ω2R−1 | [−(q−io)2 +m2 ]R e .
The lower component involves generalized Yukawa forces (half-integer-index

hyperbolic Macdonald functions [57]), i.e., exponential forces (x)e−|mx|
−|m|r
for Cartan spacetime R = 1 and Yukawa forces proper xr 1+|m|r 2r 2 e for
Minkowski spacetime R = 2:
R−1 −|m|r R−2
1+|m|r −|m|r
2
(r )

x
R−1 − ∂
∂m 2
e
|m|r = 2
(r )

x
R−2 − ∂
∂m 2 2|m|3 r 3 e
R−1 −ρ
= x − ∂ρ∂ 2 e
ρ , ρ = |m|r.
11.4 Hilbert Spaces for Causal Spacetimes

The Hilbert space L2 (D2R ) with square-integrable functions of causal space-
time ϑ(x)x ∈ D2R = D(1) × Y 2R−1 extends the Hilbert space L2 (Y 2R−1 ) for
hyperbolic position, used for position x ∈ R3 ∼ = Y 3 in the case of the bound
states of nonrelativistic hydrogen atom (see Chapter 8).
The Hilbert spaces for the Poincaré group and free particles, built with
the Fock state for the translations, e.g., L2 (Y 3 ) × C1+2J for a massive spin-J
particle with the square-integrable functions of the momentum hyperboloid

(see Chapters 5 and 8), are inappropriate for electroweak spacetime. This is
in analogy to the different Hilbert spaces for free nonrelativistic scattering
L2 (Ω2 )×C2 and the Hilbert space L2 (Y 3 ) for the nonrelativistic bound states
of the periodic table.
11.4.1 Hardy Spaces for the Future and Past

The projection from all time translations R to future and past R± goes with
the transition from Dirac measure to the advanced and retarded measures,
± iπ
1
q∓io = δ(q) ± iπ qP , as seen in the Fourier-transformed step and sign
1 1 1
functions:
⎧
⎪
⎪
F ∓i
ϑ± (t) ↔ q∓io = ϑ̃± (q),
⎨
1±(t)
ϑ(±t) = 2 = ϑ± (t) = ± 2iπ q∓io e ,
dq 1 iqt F
(t) = |t| ↔ − qP ,
t 2i
⎪
⎪
⎩ F
1 ↔ 2πδ(q).
The Lebesgue spaces for the future and past can be obtained from the
Lebesgue spaces for all time translations by projections with ϑ± = ϑ± · ϑ± .
Their Fourier transforms constitute the Hardy spaces H±r
(Ř) for energy func-
tions f˜± . The Hardy functions arise from the Lebesgue functions by a con-
volution with the Fourier-transformed step functions (Radon measures),
1
p + 1
r = 1, 1 ≤ p ≤ 2, ∞ ≥ r ≥ 2,
F
p
L⏐ (R) −→ L⏐ (Ř)
r
dq
⏐ ⏐ f (t) = 2π f˜(q)eiqt= f+ (t) + f− (t),
ϑ± · ϑ̃± ∗, dq ˜
f± (t) = ϑ(±t)f (t) = 2π f± (q ∓ io)eiqt ,
p
L (R± ) −→ H± (Ř)
r
˜ ˜
f± (q ∓ io) = ϑ̃± ∗ f (q ∓ io)
F
= ± 2iπdp 1 ˜
q∓io−p f (p).
The Hardy Hilbert spaces H± 2

(Ř) are Fourier-isomorphic to the square-
integrable functions of the positive and negative lines (future and past),
L2 (R) ⊃ L2 (R± ) ∼ 2
= H± (Ř) ⊂ L2 (Ř),
2 ∼ 2 2 ∼
L (R) = L (R+ ) ⊕ L (R− ) = H+ (Ř) ⊕ H−
2 2
(Ř) ∼
= L2 (Ř),
with the scalar products
dq
f |f ± = dt f± (t)f±

(t) = 2π f˜± (q ∓ io)f˜±
(q ∓ io)

= dt ϑ(±t)f (t)f (t) = ± (2π)2 f (q) i q−q1 ∓io f˜ (q ).
dq dq ˜ 1
11.4.2 Hardy Spaces for Spacetime Cones

For Minkowski translations SO0 (1, s) ×
R1+s , s ≥ 1, the sum of the Hardy
spaces of advanced and retarded energy-momentum functions is a proper
11.4 Hilbert Spaces for Causal Spacetimes 309
subspace of the full Lebesgue space Lp (Ř1+s ). The sum of the characteristic
functions ϑ± for the future and past is the characteristic function ϑc for the
causal bicone (subindex c), equal to 1 for all translations R1+s only for total
order s = 0:
⎪
⎪ 2
1+s ⎪
c = {x ∈ R
s = 1, 2, · · · : R1+s ⎪x > 0} = R1+s
+ ∪ R−
1+s
= R1+s ,
= ϑ(±x0 )ϑ(x2 ), ϑc (x) = ϑ(x2 ), c (x) = (x0 )ϑ(x2 ),
ϑ± (x)
ϑ±= ϑc ±
2 , ϑ±,c · ϑ±,c = ϑ±,c , c · c = ϑc , c · ϑc = c .
c
The causal distributions for even-dimensional spacetimes have the Fourier

transforms, with (q − io)2 = (q0 − io)2 −
q 2,
⎧
⎪
⎪
d2R q 1
eiqx ,
⎨ ϑ± (x) = π|Ω2R−1 | [−(q∓io)2 ]R
2R
SO0 (1, 2R − 1) : ϑ (x) = d q 2
2 )R e
iqx
,
⎪ c
⎪
π|Ω2R−1 | (−qP
⎩ d2R q (R−1) 2 iqx
c (x) = π R i(q0 )δ (q )e ,
with the general relation between advanced (retarded) and principal value
measures:
1+s + ,
Γ(1+N ) Γ(1+N )
2 )1+N ± iπ(q0 )δ
iqx
d q [(q∓io) 2 ]1+N e = d1+s q (q (N )
(−q 2 ) eiqx
P Γ(1+N ) iqx
= 2ϑ(±x0 ) d1+s q (q 2 )1+N e .
P
One obtains the Hardy spaces H± r

(Ř2R ) for energy-momentum functions
of even-dimensional future and past cones:
F f± (x)= ϑ± (x)f (x)

⏐ ) −→ L (Ř
Lp (R 2R r 2R
⏐ ) d2R q
= (2π) ˜
2R f± (q ∓ io)e
iqx
,
⏐ ⏐
ϑ± · ϑ̃± ∗ , ˜ ˜
f± (q ∓ io)= ϑ̃± ∗ f (q ∓ io)
L (R± ) −→ H± (Ř2R )
p 2R r d2R p
F = π|Ω 1 ˜
2R−1 | [−(q∓io−p)2 ]R f (p).
The Hardy Hilbert spaces L2 (D2R ) ∼ ∼ 2 2R ) have the

± ) = H± (Ř
= L2 (R2R
following scalar products:
d2R q ˜
f |f ± =
d2R x f± (x)f± (x) = ˜
(2π)2R f± (q ∓ io)f± (q ∓ io)
2R 2R
= d2R x ϑ± (x)f (x)f (x) = d q d q ˜ 1 ˜
(2π)2R π|Ω2R−1 | f (q) [−(q−q ∓io)2 ]R f (q )
Functions with different Lorentz group SO0 (1, 2R − 1)-properties via har-
monic polynomials (q)L are orthogonal to each other (see Chapter 12),
Γ(1+L) Γ(R)
d2R q μ(q 2 )(q)L ⊗ (q)L = δ LL 2L Γ(R+L) d2R q μ(q 2 )(q 2 )L (12R )L .
The remaining convolutions for scalar Feynman measures at the neutral ele-
ment are as follows (see Chapter 10):
2R ∗ 2
( ∗ , q2 ) = (± , q2 ± io), 2R = 2, 4, . . .
/ i|Ω2R−1 |
(−q2 )λ 2R / Γ(R+λ)Γ(R+ν1 +ν2 −λ)
R+ν1 ∗ 1
R+ν2 / =
(−q2 +m2
1) (−q2 +m22) q=0
Γ(R+ν1 )Γ(R+ν2 )
1 ζ R+ν1 −1 (1−ζ)R+ν2 −1
dζ 2 R+ν1 +ν2 −λ
0 [ζm2
/ 1 +(1−ζ)m2 ]
(−q2 )λ 2R / Γ(R+λ)Γ(R+ν1 +ν2 −λ)
∗ 1
/ = 1
(−q2 +m2 )R+ν1 (−q2 +m2 )R+ν2 q=0 Γ(R+ν1 +ν2 ) (m2 )R+ν1 +ν2 −λ
/
2R / Γ(R+λ)Γ(ν)
(−q ) 2 λ
∗ 2 2 R+ν+λ /
1
= Γ(R+ν+λ) (m2 )ν 1
(−q +m ) q=0
11.5 Spacetime Interactions (Kernels)

Causal spacetime D2R = D(1)×Y 2R−1 has the Lorentz compatibly embedded
kernels of time and hyperbolic position.
The inverse spacetime derivatives involve lightcone-supported Dirac dis-
tributions:
1
d2R q Γ(R−N ) iqx
(4π)R Γ(R−N ) (∂ 2 )R−N = π R [−(q−io)2 ]R−N e

ϑ(x2 ), N = 0,
= ϑ(x0 )2π × (N −1) 2
δ (− x4 ), N = 1, . . . , R − 1,
∂
d2R q Γ(R+1−N ) iq iqx
(4π)R Γ(R+1−N ) (∂ 2 )R+1−N = π R [−(q−io)2 ]R+1−N e

ϑ(x2 ), N = 0,
= ϑ(x0 )πx × 2
δ (N −1) (− x4 ), N = 1, . . . , R.
The time and position projections of the spacetime interactions (kernels)

lead back to the embedded kernels, e.g., to the characteristic causal function
ϑ(x0 ) for time and, for the R = 1 position, to the sign function (x) = xr ,
and, for R ≥ 2, to the Kepler potential 1r :
2R−1 2R−1
|Ω | d x
1
∂ ∼ 1
q = q
q2 : (2π)
dx0
2R
d2R q 2π ϑ(x )

q iqx 0
i|Ω2R−1 | (q−io)2 e = x Γ(2R−1)
2 R−1
,
r 2(r )

|Ω2R−1 | d2R−1 x
1
(∂ 2 )R−1 ∼ 1
(q2 )R−1 : (2π)
dx0
2R
2π
2R
d q 1 iqx
x 0
2R−3
ϑ(x0 ) Γ(2R−2)
|Ω2R−1 | [−(q−io)2 ]R−1 e = 1
.
r
For R ≥ 2, the position projection of the time kernel gives a Kepler-like force

x
r 2R−1 .
The spacetime interactions (kernels) in general are convolution products
of spacetime Green’s kernels with spacetime representation coefficients. They
11.6 Normalization of Electroweak Spacetime 311
are computed by the following convolutions, wherever defined for ν ∈ R (see

Chapter 9):
2R ∗ 2
( ∗ , q2 ) = (± i|Ω2R−1 , q2 ± io), 2R = 2, 4, . . .
L 2R
L L| +L
1 Γ(ν1 ) 2 Γ(R+ν2 ) 1 2
∂
ν1 ∗ ∂q
∂
R+ν2 = Γ(R) ∂q
∂
∂q (−q2 +m21) (−q2 +m2
2)
1 ν
ζ 1 −1 (1−ζ)R+ν2 −1 Γ(ν1 +ν2 )
0
dζ 2 2 2 ν1 +ν2
(−ζ(1−ζ)q +ζm1 +(1−ζ)m2 )
L
Γ(L1 +ν1 )(q)L1 2R 2 Γ(R+ν2 ) L2
∗ ∂q
∂
= Γ(R) ∂
⊗ (q)L1
(−q2 )L1 +ν1 (−q2 +κ2 )R+ν2 ∂q
1 ζ L1 +ν1 −1 (1−ζ)R−ν1 −1 Γ(L1 +ν1 +ν2 )
dζ
0 (−ζq2 +κ2 )L1 +ν1 +ν2
(q)L1 can be any homogeneous energy-momentum polynomial. Nontrivial

Lorentz group properties arise by derivations:
∂
∂q = 2q ∂q∂ 2 , ∂
∂q ⊗ q = 12R + 2q ⊗ q ∂q∂ 2 , . . .
11.6 Normalization of Electroweak Spacetime

The representations of time D(1) and position Y 2R−1 ⊇ SO0 (1, 1), both with
real rank 1, come as Lorentz-compatible products in representations of real
rank-2 spacetime D2R = D(1) × Y 2R−1 . The representation normalization
of spacetime coefficients determines the ratio of the continuous invariants
(masses) for time and position. The spacetime normalization is in analogy to
the mass consistency (“gap”) equation in the chiral model, i.e., to the repre-
sentation normalization of the dipole-regularized propagator of the massive
Dirac field. It extends the position normalization in the nonrelativistic hy-
drogen atom.
11.6.1 Central Correlation of Hypercharge and Isospin

An example for a relation of two discrete invariants, characterizing the rep-
resentations of a rank-2 compact group product, is given by hyperisospin.
The winding numbers of the U(2)-embedded representations of hypercharge
U(1) and isospin SU(2) are centrally correlated (see Chapters 6 and 7). The
nontrivial intersection I(2) ∼
= U(12 ) ∩ SU(2) is the center of SU(2):
eiα0 12 ◦ eiατ ∈ U(2) = U(12 ) ◦ SU(2) ∼

=
U(1)×SU(2)
I(2) .
This can be seen at the two parameter pairs (α0 , α3 ) = (π, 0) and (α0 , α3 ) =
(0, π), which yield the same group element −12 in the two-ality group I(2) ∼ =
{±12 }. Any U(2)-representation, denoted by hypercharge and isospin invari-
ants [y|T ], has to be compatible with this central correlation. A direct product
Cartan subgroup of U(2) is not given by U(12 ) ◦ U(1)3 eiα0 12 +iα3 τ3 , but
12 +τ3 12 −τ3
by a product of projector subgroups U(1)+ × U(1)− eiα+ 2 +iα− 2 .
If the defining representation and its dual e−iα0 12 +iατ ∈ U(2) are denoted
by [± 12 | 12 ], all irreducible U(2)-representations display a correlation of hyper-
charge and isospin invariant as given by
⎪
⎪
⎪ 2T = 0, 1, 2, . . . }.
irrep U(2) = {[y|T ] = [±n + T |T ]⎪n,
The representation-characterizing invariants y for hypercharge and T for
isospin are either both integer or both half-integer; i.e., y + T ∈ Z.
As a consequence in the electroweak standard model, the representations
of the electromagnetic subgroup U(1)+ are characterized by integer winding
(charge) numbers y + T3 = ±n + T + T3 ∈ Z.
All groups U(n) = U(1n ) ◦ SU(n) have a central correlation with a cyclo-
tomic group U(1n )∩SU(n) ∼ = I(n) (n-ality group), with similar consequences
for the U(n)-representations.
11.6.2 The Mass Ratio of Spacetime

The embedded representations of time D(1) come with Lorentz SO0 (1,
2R − 1)-properties, the fundamental one given by a vector:
d2R q
D2R = D(1) × Y 2R−1 ϑ(x)x −→ iπ|Ω 2q
2R−1 | [−(q−io)2 +M 2 ]R [(q−io)2 −m2 ] e
iqx
.
The doubled spherical normalization |Ω1 ||Ω22R−1 | contains the residual Y 2R−1 -
normalization by the corresponding sphere (see Chapter 9).
2
The mass ratio μ2 = M m2 characterizes the representation of even-dimen-
sional causal spacetime D2R with real rank 2. It relates the two invariants
for the Lorentz-compatibly embedded representations of the causal group
(eigentime) D(1) with q02 = m2 (causal dilation-invariant) and of hyperbolic
position Y 2R−1 with q 2 = −M 2 (Lorentz dilation- or curvature-invariant).
The dual product with the vector kernel 1q is given by the Schur product,
i.e., by the convolution at trivial energy-momenta (see Chapter 9) with
Feynman integration,
2R /
/ 2R

2q
(−q +M ) (q −m )
2 2 R 2 2 1
q / with ∗ = i|Ω
2
2R−1 | , q 2
+ io .
q=0
2R 2R
It involves a tensor product = ⊗ ∗ of spaces with a Lorentz group repre-
sentation:
2R 1 2R
2q
(−q2 +μ2 )R (q2 −1)
1q = μ2 dR κ2 (−q2(−qR)
+κ2 )R+1
1q
1 R 2 2R
= − ∂q
∂
μ2
d κ (−q2 +κ1
2 )R q
1
1 1 R−1
∂
= ∂q ⊗ q μ2 dR κ2 0 dξ (1−ξ)
−ξq2 +κ2 .
The representation normalization condition comes with the vector unit ∂

∂q ⊗
q = 12R ∼
= δca ,
2R /
/ 2
12R = (−q2 +μ22q)R (q2 −1) 1q / = − R1 logR M
m2 12R ,
q=0
and the “R-tail” of the logarithm, which integrates the SO0 (1, 2R − 1)-
invariant inverse mass square κ12 over the line [M 2 , m2 ] ∼ [μ2 , 1]:
2d2R q
logR μ2 = − i|Ω2R−1 |(−q2 −io+μ 2 R 2
1 R 2 )1 (qdκ+io−1)
2 2 1 dζ ζ R−1
= − μ2 d κ2κ = − 1−μ 1
2 μ2 κ2 1−μ2
( 1−κ )R−1 = − 0 ζμ 2 +1−ζ
$ (1−μ2 )k−R
R−1 $∞
2 2 k−R
(1−μ )
log μ
= (1−μ 2 )R + k =− k ,
k=1 k=R
log μ2 1
log1 μ2 = 1−μ2 , μ2 = logR μ2 + R1 ,
1−μ2 logR+1
$
R−1
logR 1 = − R
1
, logR μ2 = log μ2 + k + . . . for μ ) 1.
1 2
k=1
0 1 dκ2
For the nonabelian case, the residue μ2 = log μ2 d log κ2 comes with the
κ2
characteristic factor (1−κ2 )R−1 from a Cartan torus SO(2)R−1 ⊂ SO(2R−1).
“Tail” functions for exponents and logarithms are typical for represen-
tations of real rank-1 hyperboloids. For example, the Plancherel measure
dκ Π2R−1 (κ2 ) of the irreducible SO0 (1, 2R − 1)-representations, character-
ized by a continuous positive invariant κ2 , for the harmonic analysis [55] of
functions L2 (Y 2R−1 ) on nonabelian odd-dimensional position hyperboloids
contains the “R-tail” of the hyperbolic function sinh πκ
πκ
(see Chapter 8):
⎧
⎪
⎪ / Γ(R−1+iκ) /2 ;
R−2
⎪
⎪ Π2R−1 2
(κ ) = / / = Γ(R − 1)2 κ2 1 + κ2
,
⎨ Γ(iκ) k 2
Y 2R−1
: k=1
∞
⎪
⎪ ;
⎪
⎪ 1
= 1
= sinh πκ
=
2
1 + κk2 .
⎩ Γ(1−iκ)Γ(1+iκ) κ2 |Γ(iκ)|2 πκ
k=1
Also, the normalization equation in the chiral model of Nambu and

Jona-Lasinio, where the representation of four-dimensional spacetime by the
propagator of the interacting Dirac field is used with the dipole regularization
m2
2 )2 · q 2 −m2 ,
1
(see Chapter 10), contains the Lorentz invariant factors (q2 −M
multiplied by a Fourier-transformed derivative γq + m and its inverse
1 1
γ5 γq−m γ5 = γq+m . Both normalizations involve the 2-tail of the logarithm,
Spacetime representation Kernel Representation normalization
+ ,
d4 q 2
dM
iπ 2m2
1
(q) (γq + m)γ5 γq−m γ5
d4 q 2
dM (q)(γq + m)eiqx 1 d4 q 2
Chiral U(1) iπ 2 m2 γq−m = 14 iπ 2
dM
m2
(q)
ρ(m2 )(m2 −M 2 )2 2
− log M = 1
d48π
2 m2
2
+ 2 m2,
iπ
q
2 dMm2
(q) q ⊗ 1q
d4 q 2
D(1) × Y 3 dM (q) q eiqx 1 14 d4 q 2
iπ 2 m2 q = 4 iπ 2
dM
m2
(q)
M2
− 1
2 log2 m2
=1
2 2 d4 q 2 M2
dM
m2
1
(q)= (q2 −M 2 )2 · q2 −m2 ,
m
iπ 2 m4
dM
m2
(q)=− log2 m2
.
Usually, the invariant mass M 2 in the chiral model is considered to be

2
a cutoff mass M m2 → ∞ for the regularization of the interacting Dirac field.
It is related, by the gap equation, to the translation invariant m2 (particle
mass) and the normalization factor ρ(m2 ) of the basic representation of the
spacetime translations R4 . The mass M 2 in the electroweak spacetime model
is not introduced as an ad hoc regularization mass — it is used as curvature
for the representation of nonflat hyperbolic position, i.e., as invariant for the
Lorentz dilations SO0 (1, 1) ∼
= Y 1 ⊂ Y 3.
In the normalization condition, the logarithm of the mass ratio goes with
$
R−1
the rank, log μ2 ∼ −R − 1
k ,
k=1

log μ2
1−μ2 ⇒ μ2 = 1, R = 1,
−R = logR μ = 2
log μ2 +1−μ2 −3
(1−μ2 )2 ⇒μ ∼e2
, R = 2, . . . .
11.6.3 Internal Multiplicities

Spacetime D4 ∼
=GL(C2 )/U(2) parametrizes hyperisospin classes. A spacetime
coefficient for an induced GL(C2 )-representation can represent nontrivial in-
ternal U(2)-properties; i.e., the matrix elements can come with representa-
tions (matrices) I ∼
= IB
A for spaces with a nontrivial action of the internal
group,
d2R q
D2R ϑ(x)x −→ iπ|Ω 2q⊗I
2R−1 | [−(q−io)2 +μ2 ]R [(q−io)2 −1] e
iqx
.
q ⊗ I is a matrix for an SL(C2 ) × SU(2)-representation. For D4 and a

four-dimensional vector representation [ 12 | 12 ] of the Lorentz group SL(C2 ),
q∼ = q a (σa )ḂA , the hyperisospin representation has to be isomorphic to a sub-
representation; i.e., it has to involve a singlet or triplet SU(2)-representation
[ 12 ] ⊗ [ 12 ] = [0] ⊕ [1].
The corresponding kernel has to come with the dual representation Ǐ, e.g.,
I⊗q = q2 ⊗ Ǐ. A convolution for a dual product with
1 q
1
q⊗I q⊗I
rearranges the energy-momenta of the factors to the sum (center-of-mass)

energy-momenta q1 + q2 by integrating over the relative ones q1 − q2 . Cor-
respondingly, the tensor product of the internal factors (matrices) also has
to be rearranged by Fierz recoupling F into the trace part and the trace-
less part, e.g., for isospin scalar couplings with SU(2)-Pauli matrices τA B ,
A, B = 1, 2,
F
Ǐ ⊗ I = Γj ⊗ Γj , ǏB D D B
A IC = (Γj )A (Γj )C ,
F
e.g., 12 ⊗ 12 = 12 (12 ⊗ 12 + τ ⊗ τ ), δA δC = 12 (δA
B D
δC + τA D τC B ).
D B
In general, with identity 1d ∼ B

= δA , A, B = 1, . . . , d, and normalized traceless
generalized Pauli matrices, there arises the internal multiplicity factor d1 in
the Fierz rearrangement:
⎫
d2 −1
(τ a )a=1 , ⎬
F
a
with tr τ = 0 ⇒ 1d ⊗ 1d = d1 1d ⊗ 1d + 12 τ a ⊗ τ a .
⎭
and tr τ a ◦ τ b = 2δ ab
The internal multiplicity factors contribute to the equations for the mass
ratios in the spacetime representation normalizations. For the Schur product,
only the contribution with the unit matrix in the Fierz recoupling has to be
taken; here d1 1d ⊗ 1d = 1d 1d2 , etc. One obtains with internal identities 1d the
normalization conditions for the vectoral spacetime representations:
/
2R q⊗1 /
(−q2 +μ2 )R (q2 −1) = 12R ⊗ 1d2
2q⊗1d
q2 /
d
q=0
2
⇐⇒ − Rd
1
logR M
m2 = 1.
The arising of multiplicities in normalizations — here the external and

internal dimensions in the product Rd — is familiar, not only from the
Plancherel measure of compact groups (see Chapter 8), but also, e.g., from
the vacuum polarization 0|Ja (x)Jb (y)|0 in perturbative quantum electro-
dynamics (or more general, in gauge theories, abelian and nonabelian) with
the number of the6 charged fields γ −→ e+ e− , μ+ μ− , . . . and the electromag-
d
netic current J = ι=1 Jι . In the latter case, there arise also renormalization
logarithms (see Chapter 5).
Chapter 12
Masses and Coupling

Constants
The nonrelativistic hydrogen atom and the atomic spectrum characterize

cyclic Hilbert representations for the analysis of hyperbolic position Y 3 , i.e., of
the homogeneous space SO0 (1, 3)/SO(3) for rotation relativity. The particle
spectrum is proposed to arise in an analysis of electroweak spacetime D4 as
the homogeneous space for unitary relativity D(1) × Y 3 ∼ = GL(C2 )/U(2),
i.e., for position Y with additional causal (dilation) operations D(1).
3
The particle spectrum (m2 , J, z) ∈ R+ × N2 × Z for flat spacetime is

characterized by the continuous mass as invariant for translations R4 in the
D(1) × U(1)-extended Poincaré group GL(C2 ) × R4 , by (half-)integer spin
or polarization for rotations SU(2) or SO(2) as translation fixgroup in the
Lorentz cover group SL(C2 ), and by an integer charge number for electro-
magnetic windings U(1). The basic interactions are implemented by massless
fields with characteristic coupling constants.
The eigentime D(1) ∼ = R invariants for the representations of electroweak
spacetime D4 ∼ = D(1) × Y 3 ∼= R4+ are proposed to determine the mass of re-
lativistic particles. Since the causal spacetime group GL(C2 ) has real rank 2,
i.e., two characterizing continuous invariants {m2 , M 2 }, the translation in-
variants are related to both the embedded causal group D(1) and Lorentz
group SO0 (1, 3)-representations of 3-position Y 3 .
A classically oriented remark of Born (1962) with respect to a connection
of curvature and particle masses:
The masses of elementary particles, nuclei and electrons repre-
sent enormous concentrations of energy in very small regions of space.
Hence one should presume that they will produce considerable local
curvatures of space and corresponding gravitational fields. Can these
fields explain the cohesive forces, which keep the particles together . . . ?

318 Chapter 12 Masses and Coupling Constants
Invariants for compact groups like electric charges, spin or polarization are
taken from a rational spectrum; they are “quantum numbers” in the original
sense. The particle masses and their ratios seem to come from a continuous
spectrum, not from a rational one, i.e., in a group representation interpre-
tation, as invariants of a noncompact group. If that is true, there has to be
a structure that picks discrete values from a continuum, i.e., a structure for
the “quantization” of continuous invariants.
Such a “discrete picking” arises for products of one basic representation, in
the simplest case for the harmonic oscillator where the energy E1 ∈ R of a 1-
quantum state vector as continuous invariant for the abelian time translations
D(1) ∼= R −→ U(1) comes in integer multiples Ek = kE1 , k ∈ N, for the
-
k
product representations |Ek = |E1 (see Chapter 4).
A nonabelian example is the nonrelativistic hydrogen atom with its
Hilbert representations of hyperbolic position Y 3 ∼= SO0 (1, 3)/SO(3). Here
(Chapters 4 and 8), the bound-states ( L2 , L2 ) (Kepler or harmonic SO(4)-
representations) are the “leading” irreducible representations in the totally
-
L
symmetric products ( 12 , 12 ) of the fundamental SO(4)-representation, the
quartet ( 12 , 12 ) with SO(3)-singlet and -triplet. The continuous negative
energies (or imaginary “momenta” q 2 = −Q2 = 2E < 0) as time trans-
lation invariants are, simultaneously, also invariants for the position group
SO0 (1, 1) ⊂ SO0 (1, 3). They are not equidistanced, but “quantized” in an
equipartition (“flux quantization”) of a basic energy E0 = EL (1 + L)2 to the
dimensions of SO(4)-product representation spaces.
As we will show for electroweak spacetime D4 = D(1) × Y 3 , the ratios of
the particle masses as D(1)-related invariants to one fixed SO0 (1, 1)-invariant
as position curvature are “quantized” in a corresponding noncompact–
compact reciprocity by multiplicities for product representations of external
and internal spin-isospin SU(2) × SU(2)-operations.
For a cyclic representation of a group G, the coset operations G/H, acting
on a cyclic vector with fixgroup H, may be the origin of what, in a more
phenomenological language, is called the degeneracy manifold of a ground-
state with the Nambu–Goldstone degrees of freedom.
Gravity involves the dynamics of dilations, which occur in the Iwasawa
decomposition GL(R4 ) = SO(4) ◦ D(1)4 ◦ exp R6 . In the tetrad, the dilations
come as Lorentz vectors (eaj ) ∈∈GL(R4 )/O(1, 3). For example, the tetrad for
the metric of Reissner spacetime (see Chapters 1 and 3) is built in the form
of an SO0 (1, 1)-dilation by the Newton potential for a mass point and an
electromagnetic contribution for a point charge:
2

2
∂ 2 −
rm + 2rz2 = 4πm δ(x ) + ∂
rz
tt

2z
e2σ3 λ3 = g0 −g0rr , with e2λ3 (r) = 1 − 2
rm + r2

2z
for 1 − 2
m
r + r2 > 0.
12.1 Harmonic Coefficients of Spacetime 319
Masses induce massless interactions: The phenomenon of nontrivial

masses itself is interpretable as a ground-state degeneracy under dilations
D(1) (“breakdown” of dilation symmetry). For the rearrangement of the
spacetime dilation properties, the corresponding dilation degree of freedom
is implemented via multiplets of the external–internal SO0 (1, 3) × SO(3) for
hyperisospin relativity GL(C2 )/U(2) as massless Nambu–Goldstone modes
in, respectively, vector and tensor Lorentz group representations. Similarly
to chiral Nambu–Goldstone modes, which are not chiral gauge fields, dilation
Nambu–Goldstone modes [39] are not dilation gauge fields. A massless gauge
field goes with the “unbroken” symmetries, a massless Nambu–Goldstone
field with the “broken” ones. The dilation rearranging massless scalar fields
are the polarization trivial nonparticle Coulomb and Newton degrees of free-
dom in the electroweak vector and gravitational tensor interactions. They are
supplemented to full-fledged Lorentz group multiplets by gauge and polarized
particle degrees of freedom. The representation normalizations determine the
residues of the massless interactions, i.e., their coupling constants.
12.1 Harmonic Coefficients of Spacetime

The representation matrix elements of electroweak spacetime are convolution
products (group products) of representation coefficients of eigentime and hy-
perbolic position, given by pointwise products in energy-momentum space.
For nontrivial Lorentz group SO0 (1, 2R − 1)-properties, they come with har-
monic energy-momentum polynomials (q)L (see Chapter 8):
D(1) × Y 2R−1 ϑ(x)x −→ |m2L , −ML2 , L}2R (x)

2R
2L−n(L) (q)L
= i|Ω12d q
||Ω2R−1 | [−(q−io)2 +M 2 ]R+n(L) [(q−io)2 −m2 ] e
iqx
,
L L
L = 0, 1, 2, 3, . . . .
The coefficients are from Lorentz compatibly embedded representations of the

causal group D(1) with invariant m2L and of SO0 (1, 2R−1) with invariant ML2
as the position Y 2R−1 -curvature. For the proper Lorentz group SO0 (1, 3), the
harmonic polynomials (q)L are acted on by the (1+L)2 -dimensional represen-
tations [J|J], J = L2 = 0, 12 , 1, . . . . The natural number L is a characterizing
invariant for the maximal compact rotation group SO(2R − 1), relevant for
the nonabelian case R ≥ 2.
The residual normalization of the harmonic representation coefficients
with the powers n(L) has been determined as follows: The additional mul-
1
tipole order (−q2 +M 2 )n(L) for hyperbolic position has to distinguish between
L
L
even and odd powers (q)L , i.e., between integer and half-integer spin J = 2,
L

n(L) = 0, 1, 2, . . . = 2, L − 2n(L) =
0,
L−1
2 ,
1;
2
ML
therefore, with mass ratios μ2L = m2L
,
⎧ L
⎪
⎪ 2 2 (q)L
, L = 0, 2, . . . ,
⎨ (−q2 +μ2L )R+ 2 (q2 −1)
L
2L−n(L) (q)L
=
(−q2 +μ2L )R+n(L) (q2 −1) ⎪
⎪
L+1
(q)L
⎩ 2 2
L−1 , L = 1, 3, . . . .
(−q2 +μ2L )R+ 2 (q2 −1)
Similar to the nonrelativistic hydrogen atom with the normalized coeffi-

cients for the harmonic SO(4)-representations ( L2 , L2 ) (see Chapter 10),
2R−1 q L
Y 2R−1 x −→ | − Q2L , L}2R−1 (x )= 2d|Ω 2R−1 | ( 2
1
2
q +QL ) R 2
2
q +QL
2 (q )L eiqx ,
L= 0, 1, 2, . . . ,
the additional order of the hyperbolic pole and the power of 2 in the nu-
merator coincide, ( −q22+μ2 )n(L) (self-dual residual normalization). However,
L
for relativistic spacetime D(1) × Y 2R−1 , they come with the additional D(1)-
factors ( q21−1 , q22q−1 ) in steps of two: order R-pole for J = 0, 12 , order R+1-pole
for J = 1, 32 , etc.:
L L2
2 2 (q)L
L ∼ 1
(−q2 +μ2L )R
2
−q2 +μ2L
(q)L 1
q2 −1 , L = 0, 2, . . . ,
(−q2 +μ2L )R+ 2 (q2 −1)
L+1 L−1
2 (q)L 2
∼ (q)L−1 q22q−1 , L = 1, 3, . . . .
2 1 2
(−q2 +μ2L )R+
L−1
2 (q2 −1) (−q2 +μ2L )R −q2 +μ2L
It is useful to bring both poles together: The harmonic polynomials (q)L

∂ L
arise as leading terms by harmonic derivations ( ∂q ) ,
(q)L Γ(k) 1 (q)L Γ(k+1)

(−q2 +μ2 )k (q2 −1) =− μ2 dk κ2 (−q2 +κ2 )k+1
1
= − 21L ( ∂q
∂ L
) μ2
dk κ2 (−qΓ(k+1−L)
2 +κ2 )k+1−L + . . . ,
with the line-supported measures for the SO0 (1, 2R − 1)-invariants (see
Chapter 11):
k−1
dκ2 1−κ2
dk κ2 = 1−μ 2 1−μ2 , k = 1, 2, . . .
This is used for the harmonic spacetime representation coefficients:

L L
2 2 (q)L Γ(R− L−2 ) 1 L
R+ L
=− L
2 ∂
∂q μ2
dR+ 2 κ2 1
R− L−2
+ ...,
(−q 2 +μ2
L)
2 (q 2 −1) 22 Γ(R+ L
2
) L (−q 2 +κ2 ) 2
for L = 0, 2, . . . ,
L+1 L
(q)L Γ(R− L−1 ) 1 L−1
=−
2 2
R+ L−1 L−1
2 ∂
∂q μ2
dR+ 2 κ2 1
R− L−1
+. . . ,
(−q 2 +μ2
L)
2 (q 2 −1) 2 2 Γ(R+ L−1
2
) L (−q 2 +κ2 ) 2
for L = 1, 3, . . . ,
12.2 Translation Invariants as Particle Masses 321
explicit for the fundamental vector L = 1 and for L = 0, 2, 3,

1
L=0: 1
(−q2 +μ20 )R (q2 −1)
= −R 2 dR κ2 1
(−q2 +κ2 )R+1 ,
μ10
L=1: 2q
(−q2 +μ21 )R (q2 −1)
= −∂ 2
1
dR κ2 (−q2 +κ 2 )R ,
2
∂q2 μ1
1
L = 2 : (−q2 +μ2(q)
2 )R+1 (q 2 −1) = − 2R
1 ∂
∂q μ22
dR+1 κ2 1
(−q2 +κ2 )R
+ ...,
2
3
3
1
L = 3 : (−q2 +μ4(q) 1
2 )R+1 (q 2 −1) =− 2(R−1)R
∂
∂q μ23 dR+1 κ2 1
(−q2 +κ2 )R−1 + ....
3
12.2 Translation Invariants as Particle Masses

In contrast to the linear energy spacing for time D(1) (harmonic oscillator),
. dq 1 iqt
D(1) ∼
= R t −
⎪→ eimt = 2iπ q−m e −→ (eimt )k ∈ U(1),
⎪
{q = Ek = km⎪ ⎪k = 0, 1, 2, . . . },
there is no such simple regularity for the masses of spacetime particles as

invariants for spacetime D(2). The equidistant energy pole structure of the
product representations for the causal group is a peculiarity of abelian oper-
ations, exemplified by the Fock space for free particles.
For the causal group D(1), the convolution powers of the energy distri-
bution for the defining representation with intrinsic unit m,
˜ = ˜ 1 ˜ 1
d˜k = d
1
˜
D ∗ d ∗EF · · · ∗ dG =
1 1
d(q) q−1 , q−k ,
k times
are acted on by the fundamental kernel:
1
1 k−1 1
κ̃ ∗ D˜1 = log D̃1 : ω k (q) = 1
q ∗ 1
q−1 = 1
q ∗ 1
q−(k−1) = 1
q−(k−1) .
The eigenvalue of a D(1)-product representation d˜k (q) = q−k

1
is given by the
singularity q = k with the eigenvalue equation d̃k (q) = q − k = 0. With
1
tangent kernels and log D̃1 -functions q−(k−1)

1
, the eigenvalue is not given by
the singularity q = k − 1 of ω but by the condition that, there, the resolvent
k
1
1−ω k (q)
is singular (see Chapter 10):
ω k (q) = 1 ⇒ q = k = 1, 2, . . . .
In contrast to the product representations of the group with the invariants

⎪
⎪ 1
{q ⎪
⎪ d˜k (q) = 0}, the normalized products with the kernels, leading to the
⎪
⎪ k
invariant resolvent singularities {q ⎪
⎪ω (q) = 1}, can be generalized to the
nonabelian case.
Analogously to the Schur products for the normalization of the position

representations coefficients with the harmonic rotation group SO(2R − 1)-
polynomials (see Chapter 10),
/
{L| − Q2L , L}2R−1 = (
q )L
2R−1
∗ 2L (
q )L / = (12R−1 )L , L = 0, 1, 2, . . . ,
q 2 +Q2L )R+L
( q=0
the energy-momentum distributions for the spacetime representations with

harmonic Lorentz group SO0 (1, 2R−1)-polynomials (q)L are convoluted with
the corresponding kernels, different for even and odd degree L, to yield the
D(1)-kernels of the spacetime tangent module:
2R
κ̃ ∗ D̃2R ⊆ log D̃2R :
(q)L 2R 2L−n(L) (q)L
L
ω2R (q) = (q2 )L−n(L)
∗ (−q2 +μ 2 )R+n(L) (q 2 −1)
⎧ L
L
⎪
⎨
(q)L 2R
L ∗ 2 2 (q)L
L , L = 0, 2, . . . ,
(q2 ) 2 (−q2 +μ2L )R+ 2 (q2 −1)
=
⎪
L+1
(q)L 2R (q)L
⎩ L+1 ∗ 2 2
L−1 , L = 1, 3, . . . .
(q2 ) 2 (−q2 +μ2L )R+ 2 (q2 −1)
In all cases, the energy-momentum powers in the convolution products of

kernels and representation coefficients are those of the scalar case L = 0:
2R−1 2L (
q )L 2R−1
Y 2R−1 : q )L
( ∗ q 2 +Q2L )R+L
(
∼ 1 ∗ 1
q 2 +Q2L )R
(
,
(q)L 2R 2L−n(L) (q)L 2R
D2R : (q2 )L−n(L)
∗ (−q2 +μ 2 )R+n(L) (q 2 −1) ∼ 1 ∗ 1
(−q2 +μ2L )R (q2 −1)
.
L
The fundamental vector case L = 1 was considered in the last chapter:
2R (q)2 2R 2(q)2
1 ∗ (−q2 +μ21)R (q2 −1) , q2 ∗ (−q2 +μ22 )R+1 (q2 −1) ,
0
L = 0 (scalar), L = 2 (tensor),
q 2R (q)3 2R 4(q)3
q2 ∗ (−q2 +μ21 )R (q2 −1) , (q2 )2 ∗ (−q2 +μ23 )R+1 (q2 −1) ,
2q
L = 1 (vector), L = 3.
The representation normalizations give D(1)-eigenvalue equations. The

D(1)-invariants μ2ι are given by those energy-momenta where the resolvent
is singular,
$ ⎪
⎪ ι
L
ω2R (q) = PιL (q)ω2R
ι
(q 2 ), {μ2ι ⎪
⎪ω2R (μ2ι ) = 1},
ι
with a decomposition on the right-hand side in irreducible representations ι,

which are SO0 (1, 2R−1)-representations (q)2J , J = 0, . . . , L, for the massless
case μ2ι = 0, and embedded SO(2R−1)-representations for nontrivial masses.
12.3 Correlation of Spin and Masses 323
12.3 Correlation of Spin and Masses

The normalization of the harmonic representations of spacetime D2R re-
lates, in a noncompact–compact reciprocity, the characterizing continuous
and discrete invariants of the acting group D(1) × SO0 (1, 2R − 1) to each
other.
The convolution products of the harmonic spacetime representation coef-
ficients with the D(1)-kernels contain as leading terms:
⎧
⎪ (q)L 2R Γ(R− L−2
2 )
⎨ L ∗ 1
Γ(R+ L
∂ L
( ∂q ) L−2 , L = 0, 2, . . . ,
L (q2 ) 2 2) (−q2 +κ2 )R− 2
ω2R (q, κ2 ) =
⎪
⎩ (q)L 2R
∗ 1 ∂
( ∂q ) L Γ(R− L−12 )
, L = 1, 3, . . . ,
L+1
(q2 ) 2 Γ(R+ L−1
2 ) (−q2 +κ2 )R−
L−1
2
with tensor coefficients and scalar integrals:
L
∂ L
( ∂q ) ⊗(q)L L
ω2R (q, κ2 ) = 2 2
Γ(L+1) ω2R (q , κ )
⎧ 1 L−2 L−2
⎪
⎨ Γ(L+1)Γ(R)
dζ ζ 2 (1−ζ)R+ 2
, L = 0, 2, . . . ,
Γ( L L
2 )Γ(R+ 2 )
0 −ζq2 +κ2
L
ω2R (q 2 , κ2 ) =
⎪
⎩ Γ(L+1)Γ(R) 1 ζ
L−1
2 (1−ζ)R+ 2
L−3
Γ( L+1 L−1 0 dζ −ζq +κ

2 2 , L = 1, 3, . . . .
2 )Γ(R+ 2 )
For the Schur products, i.e., at q = 0, the harmonic projectors (units) (12R )L
for the Lorentz group SO0 (1, 2R − 1) survive (see Chapter 8):
⎧
⎨ ) ⊗(q)L
∂ L
( ∂q
= (12R )L ,
L = 0, 1, 2, 3, · · · : Γ(L+1)
⎩ L
ω2R (0, κ2 ) = Γ(L+1)Γ(R) 1
Γ(R+L) κ2 .
In the normalization conditions of the harmonic spacetime representations,

⎧ /
⎪ (q)L 2R
L
2 2 (q)L /
⎪
⎪ ∗ / ,
⎪
⎪
L
(q2 ) 2
L
(−q2 +μ2L )R+ 2 (q2 −1) q=0
⎪
⎪
⎨ L = 0, 2, . . . ,
(12R )L = {L|m2L, −ML2 , L}2R =
⎪
⎪ /
⎪ /
L+1
(q)L 2R (q)L
⎪
⎪ ∗ 2 2
/ ,
⎪
⎪
L+1 L−1
(−q2 +μ2L )R+ 2 (q2 −1) q=0
⎩ (q2 ) 2
L = 1, 3, . . . ,
L
the scalar contribution ω2R (0, κ2 ) with the Lorentz invariant inverse mass
1
square κ2 is integrated over the line, leading to logarithmic “tails” (see
Chapter 11):
1 1 k−1
d k κ2 dκ2 1−κ2
k = 1, 2, · · · : − logk μ2 = μ2 κ2 = 1
1−μ2 μ2 κ2 1−μ2 .
The resulting conditions for the mass ratios μ2L ,

1 L
dR+ 2 κ2
1 = − Γ(L+1)Γ(R)
Γ(R+L)
1
L μ2L κ2 = − Γ(L+1)Γ(R)
Γ(R+L)
1
L logR+ L μ2L ,
22 22 2
L = 0, 2, . . . ,
1 dR+
L−1
κ2
1 = − Γ(L+1)Γ(R)
Γ(R+L) L−1
1
μ2L
2
κ2 = − Γ(L+1)Γ(R)
Γ(R+L)
1
L−1 logR+ L−1 μ2L ,
2 2 2 2 2
L = 1, 3, . . . ,
contain the multiplicity factor with the dimension of the totally symmetric
Lth power of a vector space KR :
L
d2R q μ(q 2 ) (q)(q2⊗(q)
L
)L
= δ LL Γ(L+1)Γ(R) 1
Γ(R+L) 2L (12R )
L
d2R q μ(q 2 ),
-L

Γ(R+L)
dimK KR = R+L−1 L = Γ(L+1)Γ(R) .
The complete normalization conditions have to take into account addi-

tional internal multiplicity factors dint (L) (more ahead):

− logR+n(L) μ2L = 2n(L) R+L−1L dint (L),
with, respectively, n(L) = L L−1

2, 2 ∈ N.
12.4 Massless Interactions

Nontrivial masses rearrange (“break”) the dilation properties (“invariance”)
(x, q) −→ (eψ0 x, e−ψ0 q). The dilation rearrangement (“breakdown”) for space-
time D2R involves both the mass m2 , characterizing the embedded causal
D(1)-representations, and the mass M 2 , characterizing the embedded Lorentz
group SO0 (1, 2R − 1)-representations. The representation coefficients come
with the energy-momentum measure
@ ⎪ A
d2R q ⎪
⎪
(−q2 +M 2 )R (q2 −m2 ) ⎪ m 2
, M 2
> 0 .
On the dilation degeneracy hyperbola Y 1 = SO0 (1, 1) for the two invariant
masses with product e2λ0 , a fixed ratio μ2 determines one point:
2 λ 2
m 0 λ0 e 0 e−2λ = M 2
m2 = μ ,
0 M 2 = e 0 e−λ ⇐⇒ 2λ0 2 2
e =M m .
As a consequence of the degeneracy, one has to expect massless scalar modes

for the dilation degree of freedom, which, for four-dimensional spacetime,
are implemented by nonparticle degrees of freedom (Coulomb and Newton
potential) in the vectorial electroweak and tensorial gravitational long-range
interactions (see Chapter 5).
12.4 Massless Interactions 325
This is in analogy to the chiral phase degeneracy circle Ω1 = SO(2) with

⎪
⎪ 2
radius 2m and scalar propagator { (−q2 +Md2 )2q(q2 −m2 ) ⎪
4
⎪m , M 2 > 0} where
the chiral degree of freedom is realized by the pseudoscalar massless Nambu–
Goldstone field (see Chapter 9).
12.4.1 Massless Vector Modes

The simplest nontrivial equation for D(1)-eigenvalues contains the prod-
2 ∼
2q
uct of the vectorially embedded time representations, the cosines q2 −m
cos mx0 (time projection, see Chapter 11) and, for trivial mass, the kernel
q ∼ 1, with the embedded hyperboloid representation, an exponential poten-
1
−Mr
2 )R ∼ e
1
tial (−q2 +M (position projection),
2R 1 1 (1−ξ)R−1
1
ω2R (q) = 2q
(−q2 +μ21 )R (q2 −1)
1q = ∂
∂q ⊗q μ21 dR κ2 0 dξ −ξq2 +κ2 .
For q 2 = 0, it is decomposable with two projectors for the “spin” SO(2R − 1)-
scalar and -vector, respectively (see Chapter 5),

V = 12R − q⊗q
q2 ,
12R = S + V :
S = q⊗q ,

q 2
∂q ⊗ q = 12R + 2q ⊗ q ∂q2 = V + S 1 + 2q ∂q2 ,

∂ ∂ 2 ∂
for the q 2 -dependent kernels,

⎧ 1 1 R−1
⎨ ω2R
V
(q 2 ) = μ2 dR κ2 0 dξ (1−ξ)
−ξq 2 +κ2 ,
1
ω2R V
(q) = Vω2R S
(q 2 ) + Sω2R (q 2 ) : 1
⎩ ω S (q 2 ) = 1 + 2q 2 ∂ 2 ω V (q 2 ).
2R ∂q 2R
The resolvent with the two related eigenvalue equations is given as follows:

S
S V
ω2R (q 2 ) = 1,
q = 0 : 12R −ω1 (q) = 1−ωS (q2 ) + 1−ωV (q2 ) ,
2 12R
V
2R 2R 2R ω2R (q 2 ) = 1.
The condition for the masslessness of the chiral U(1) Nambu–Goldstone

mode coincides with the consistency condition (“gap equation”) for the Dirac
fermion mass as chiral breakdown parameter, which is a representation nor-
malization for the regularized propagator (see Chapter 10). Similarily, the
1 q=0
normalization condition above ω2R (q) = 12R for the vector spacetime re-
presentation, leading to the determination of the mass ratio μ21 , can be read
as an eigenvalue equation with a solution at q 2 = 0 for a Nambu–Goldstone
dilation mode, i.e., for a Poincaré group representation with mass zero,
S V
q 2 = 0 ⇒ ω2R (0) = ω2R (0) = 1.
The massless solution has the resolvent

12R ρ12R (0)
12R −ω2R
1 (q) = q2 12R + ....
The residue is the inverse of the negative derivative at the singularity. It

can be simplified with the mass ratio condition logR μ21 = −R:
∂ω V 1 d R κ2 1
− ρ1 1(0) = ∂q2R
2 (0) = μ2 (κ2 )2 dξ ξ(1 − ξ)R−1
2R 1 + 0 ,
1
= R(R+1) 1
μ2
+ (R − 1) logR μ21
1
1−R(R−1)μ21
= R(R+1)μ21
.
1
Here, (κ2 )2has been integrated over the invariant line,
⎧ 1
1 d k κ2 1 dζ ζ k−1 ⎨ k, l = 0,
k = 1, 2, · · · : μ2 (κ2 )l = 0 (ζμ2 +1−ζ)l = ⎩
− log k μ 2
, l = 1,
1
μ2 + (k − 1) log k μ2
, l = 2.
For small mass ratios, the residue of the massless D(1)-representation is, up to
M2
a multiplicity factor, the ratio μ21 = m12 of SO0 (1, 2R − 1) to D(1)-invariant:
1
for μ21 ) 1 : − ρ12R (0) ∼ R(R + 1)μ21 .

There is an additional internal multiplicity factor dint (1) (more ahead).
12.4.2 Massless Tensor Modes

Generalizing the structures of the massless vector mode L = 1 in the fore-
going subsection, the normalizations of the massless tensor modes use the
derivatives of the kernels with the leading terms:
⎧
⎪ 1 L R+ L−2
⎪ Γ(L+1)Γ(R)
⎪ Γ( L )Γ(R+ L ) dζ ζ (−ζq
2 (1−ζ) 2
, L = 2, 4, . . . ,
L
⎨ 2 2
0 2 +κ2 )2
∂ω2R 2 2
∂q2 (q , κ ) = ⎪
⎪
⎪ 1 L+1
ζ 2 (1−ζ)R+ 2
L−3
⎩ Γ(L+1)Γ(R)
L+1 L−1 dζ 2 +κ2 )2 , L = 1, 3, . . . .
Γ( 2 )Γ(R+ )
2
0 (−ζq
The scalar L = 0 convolution has no nontrivial q 2 -dependence. For L ≥ 1,

the values at q 2 = 0,
⎧ L
L
⎨ 2, L = 2, 4, . . . ,
∂ω2R Γ(L+1)Γ(R)
(0, κ 2
) = 1
×
⎩ L+1
∂q 2 2
Γ(R+L+1) (κ ) 2
2 , L = 1, 3, . . . ,
have to be integrated over the invariant line,

⎧ 1 dR+ L2 κ2
⎪
⎪
L
, L = 2, 4, . . . ,
L
⎨ L+2
2 2
μ2L (κ2 )2
∂ω2R Γ(L+1)Γ(R)
∂q2 (0) = Γ(R+L+1) × ⎪
⎩ L+1 1 dR+ 2 κ2 ,
⎪ L−1
L+1 μ2L (κ2 )2 L = 1, 3, . . . .

2 2
12.5 Spacetime Masses and Normalizations 327
For small mass ratios μ2L and internal multiplicities dint (L), the residues
ρL
2R (0), i.e., the coupling constants of the massless dilation modes, are
⎧ L+2
⎪
⎨ 2 L μ2L , L = 2, 4, . . . ,
2
Γ(R+L+1)
for μL ) 1 : − ρ2R (0) ∼ dint (L) Γ(L+1)Γ(R) ×
2 L
⎪
⎩ 2 L+12 2
L+1 μL , L = 1, 3, . . . .
12.5 Spacetime Masses and Normalizations

For electroweak spacetime D4 = D(2) ∼= GL(C2 )/U(2) with real dimension 4,
real rank 2, and tangent Minkowski translations R1+3 , the normalizations of
M2
the harmonic representations relate the ratios μ2L = m2L of the continuous
L
invariants for the Cartan plane D(1) × SO0 (1, 1) and the dimensions of the
spaces with the product representations of external–internal spin-isospin.
The product representations of position Y 3 ∼
= SO0 (1, 3)/SO(3) with real
dimension 3, real rank 1, and tangent translations R3 show an equipartition
Q20 = Q2L (1 + L)2 of the energies as SO0 (1, 1) ⊂ SO0 (1, 3)-boost invari-
ants to the dimensions of SO(4)-representation spaces (see Chapter 10). The
noncompact–compact reciprocity of the nonrelativistic hydrogen atom,
SO0 (1, 3)-invariants from Q20 1
= = (1 + L)2 , L = 2J = 0, 1, . . . ,
position Y 3 -normalization: Q2L κ2L
is extended by integration with the measure of the Lorentz group invariants

m2 1
on a line M 2 ∼ μ2 to a “logarithmic” noncompact–compact reciprocity:
1 d k κ2
μ2 κ2 = − logk μ2 ,
ratio of ML2
− log2+n(L) = 2n(L) (1 + L)dint (L),
D(1) × SO0 (1, 3)-invariants m2L
from spacetime L
D(2)-normalization: n(L) = 2, L = 2J = 0, 2, . . . ,
L−1
2 , L = 2J = 1, 3, . . . .
The Lorentz group properties of the spacetime D(2)-representations [J|J]

lead to the external factors 1+2J as dimension of the maximal spin represen-
tation space. The internal multiplicities dint (L) are determined by bi-regular
SL(C2 ) × SU(2)-representations: In the case of the Lorentz vector L = 1 as
the product of two SL(C2 )-Weyl spinors, the internal multiplicity is taken
F
for the product of two SU(2)-isospinors [ 12 ], with Fierz recoupling =:
⎧
⎨ [ 2 ] ⊗ [ 2 ] = [0] ⊕ [1],
⎪ 1 1
L = 2J = 1, F
12 ⊗ 12 = dint1(1) 14 + . . .
[ 12 | 12 ], ⎪
⎩ ⇒ d (1) = 2,
int
and in the case of the Lorentz tensor L = 2 as product of four SL(C2 )-Weyl
spinors for the product of four SU(2)-isospinors:
⎧ 1
⎪ [ 2 ] ⊗ [ 2 ] ⊗ [ 2 ] ⊗ [ 2 ] = ([0] ⊕ [1]) ⊗ ([0] ⊕ [1])
⎪ 1 1 1
⎪
⎪
⎪
⎪ = 2 × [0] ⊕ 3 × [1] ⊕ [2],
L = 2J = 2, ⎪ ⎨
F
[2|2] ∨ [2|2]
1 1 1 1
12 ⊗ 12 ⊗ 12 ⊗ 12 = ( 12 14 + . . . ) ⊗ ( 12 14 + . . . )
∼ ⎪
⎪
= [1|1] ⊕ [0|0], ⎪⎪ F
⎪
⎪ = dint1(2) 116 + . . .
⎪
⎩
⇒ dint (2) = 24 = 16.
In general, the Fierz reordering of the product of 2L units to the unit 1d

leads to the following internal multiplicities:
F
L = 2J = 0, 1, 2, · · · : 12 ⊗ · · · ⊗ 12 = dint1(L) 1d + . . . ,
D EF G
2L times L
d = 2(2 ) = 2, 4, 16, . . . ,
with
dint (L) = 2 2 ·2 = 1, 2, 16, . . .
L L
It is not obvious how to connect the invariants (m2L , ML2 , L) of the re-
presentations of electroweak spacetime D(2) = D(1) × Y 3 , on the one hand,
with, on the other hand, the experimental masses and rotation invariants
(spin, polarization) J = L2 of four-dimensional tangent spacetime R1+3 . In the
following proposal for an orientation, it will be assumed that causal invariants
m2L for embedded D(1) representations with spin L are considered for one
basic hyperbolic mass, i.e., for one fixed position Y 3 -curvature ML2 = M 2 .
This is in contrast to the position Y 3 -representations (Q2L , L) for the Kepler
L
(2
q)
potential in the nonrelativistic hydrogen atom, where multipoles (q 2 +Q 2 )2+L
L
for L-dependent singularities are used.
The mass M 2 in the higher-order poles (m2L , M 2 , L) for spacetime D4
cannot be related to a flat spacetime R4 -particle; it may be connected with
the invariant used as quark mass:
M2 2L−n(L) (q)L ⊗1dint (L)

μ2L = m2L
, with M 2 = Mquark
2
⇒ (−q2 +M 2 )2+n(L) (q2 −m2L )
.
Quarks and gluons are introduced for the parametrization of the strong in-
teractions. With Wigner’s particle definition, confined quarks and gluons are
not particles. The mass of confined quarks and gluons is not a translation
invariant.
The nonparticle degrees of freedom in the multipoles for electroweak
spacetime D(2) represent three-dimensional hyperbolic position Y 3 . If, by
abuse of language, they are also called “quarks,” the quark mass is the posi-
tion curvature. They have no additional internal SU(3)-color symmetry. The
theory could be extended by introducing, ad hoc, additional color degrees of
freedom. Possibly, however, there exists an effective flat spacetime lineariza-
tion of the degrees of freedom in nonparticle higher-order poles by simple
poles, e.g., for dipoles with (qq2 )2 ∼ q13 by the convolution product with three
identical factors,
id4 q 1 iqx
(2π)4 q3 e = ∂13 = ∂1 ∗ ∂1 ∗ ∂1 .
To endow the simple pole parametrization with a property that enforces the
poles into the original tripole product, additional internal (“color”) degrees
d4 q 13 iqx
of freedom can be introduced in the “quark propagator” 1∂3 = i(2π) 4 q e ,
which are confined for particles in the totally antisymmetric tensor product
∗ ∗
with convolution 1∂3 ∧ 1∂3 ∧ 1∂3 = ∂13 . In this case, confined color SU(3) ⊂
U(3) is motivated by the higher-order poles for nonabelian curved three-
dimensional position and stems from the antisymmetric cubic root of U(1).
Color SU(3) is introduced as a unitary continuous generalization of the cubic
roots and the cyclotomic group I(3).
In general: U(s) can be considered to be the sth antisymmetric root of
U(1). U(1) embeds the cyclotomic group I(s) with the sth roots of 1, i.e.,
the U(s)-center,
"
s
⎪
⎪ s
U(s) ∼ ⎪z = 1} ∼
= U(1) ⊃ I(s) = {z ∈ C⎪ = centr U(s).
Position Y s , s = 2R − 1 = 3, 5, . . . , would come with SU(s)-quark multi-

plets — triplets, quintets, septets, etc., with SU(s)-singlets in the totally
"
s
antisymmetric convolution product ∗ 1∂s = ∂1s . The connection between po-
sition dimension s and unitary group U(s) reminds us of the “color” degrees
of freedom of an isotropic s-dimensional harmonic quantum oscillator (see
Chapter 4).
For the vector representation of electroweak spacetime with multiplicity
factor 2 × 2 = 4, the D(1) simple pole singularity at m21 may be related to a
Poincaré group representation for a lepton mass with spin-isospin J = T = 12 ,
i.e., to a lepton spinor-isospinor field l:

2q⊗12 − log2 μ21 = 2dint (1) = 4,
L = 2J = 1, (−q2 +M 2 )2 (q2 −m2 ) :
∼ e−2.5 ∼ 12
M Mquark 1
1 μ1 = m 1
= mlepton .
The corresponding nonparticle Y 3 dipole degrees of freedom may be called

spinor-isospinor quark q.
For the tensor representation, there arises a multiplicity factor 2 × 3 × 24
= 96 for the logarithmic tail, which, via the exponent, gives a huge mass ratio.
Therefore, the D(1) simple pole singularity at m22 may be related to Poincaré
group representations for the Planck mass with spin-isospin J = T = {0, 1}
in a vector-isovector field P (for Planck):

1
2(q⊗q− 44 q2 )⊗(12 ⊗12 ) − log3 μ22 = 6dint (2) = 96,
L = 2J = 2, (−q2 +M 2 )3 (q2 −m2 ) :
∼ e−49 ∼ 1.9×10
M Mquark 1
2 μ2 = m 2
= mPlanck 21 .
The corresponding nonparticle Y 3 tripole degrees of freedom may be called

vector-isovector quark G.
For electroweak spacetime D(2), the residues (normalizations) of the

massless solutions are
⎧ L+2
⎨ 2 2 (1+L)(2+L) dint (L)μ2L , L = 2, 4, . . . ,
L
−ρ4 (0) ∼
L
⎩ 2 L+1
2 (2 + L) dint (L)μ2L , L = 1, 3, . . . .
The massless vector field with four external Lorentz group degrees of
freedom [ 12 | 12 ] (see Chapter 5) and four internal ones (isospin singlet and
triplet),
14 ⊗14
L = 1 : 14 ⊗1 4 −ω (q)
1 = −ρ14 (0) 14q⊗1
2
4
+ ...,
4
−ρ14 (0) ∼ 6dint (1)μ21 = 12μ21 ∼ 1

12 ,
can be related to the electroweak vector-isovector gauge fields A with the

residue −ρ14 (0) as gauge coupling constants. The experimental values in the
electroweak U(2)-standard model are (see Chapter 6)

(g12 , g22 |g 2 , γ 2 ) ∼ 8.4
1 1
, 2.5 | 10.9
1 1
, 1.9 .
The electroweak standard model linearizes the L = 2J = 1 sector of elec-

troweak spacetime with the basic equipment of flat spacetime fields:
Field Symbol SL(C2 ) SU(2) Mass

[L|R] [T ] m2
Spinor lepton l [ 12 |0] [ 12 ] m21 M

m1 ∼ 1
12 ,
Spinor quark q [ 12 |0] [ 12 ] M2 −ρ14 (0) ∼ 1

12 .
Vector gauge A [ 12 | 12 ] [0] ⊕ [1] 0
Electroweak linearization for vector L = 2J = 1
In a sense, the dichotomy of leptons and quarks reflects the embedded

causal group for time and the embedded hyperbolic position. The quark-
parametrized strong interactions describe the transition from a nonparticle
multipole representation of curved position Y 3 to simple pole particle struc-
tures, representing flat position R3 .
Correspondingly, the massless tensor-isotensor field with 10 external de-
grees of freedom for Lorentz group multiplets [ 12 | 12 ] ∨ [ 12 | 12 ] = [0|0] ⊕ [1|1]
and 16 internal ones (two isospin singlets, three triplets, one quintet) may be
related to the gravitational interaction E in the flat spacetime approach (see
Chapter 5):
110 ⊗116
L=2: 110 ⊗116 −ω42 (q)
= −ρ24 (0) 110q⊗1
2
16
+ ...,
−ρ24 (0) ∼ 24dint (2)μ2 = 384μ22 ∼
2 1
1040 .
For the L = 2J = 2 sector of electroweak spacetime, a gravitoweak

linearization can be constructed, in analogy to the L = 2J = 1 sector, with
the basic equipment of flat spacetime fields:
Field Symbol SL(C2 ) SU(2) Mass

[L|R] [T ] m2
Vector particle P [ 12 | 12 ] [0] ⊕ [1] m22 M

m2 ∼ 1
1.9×1021
,
Vector quark G [ 12 | 12 ] [0] ⊕ [1] M2 −ρ24 (0) ∼ 1

1040
.
Tensor gauge E [0|0] ⊕ [1|1] 2 × [0] ⊕ 3 × [1] ⊕ [2] 0
Gravitoweak linearization for tensor L = 2J = 2
Analogously to the embedding of the electromagnetic interactions in an

electroweak isoquartet with three short-range weak interactions, the gravi-
tational interactions are embedded in a gravitoweak iso-16-plet, which, with
Goldstone degeneracy G 3 ∼ = U(2)/U(1)+ of the ground-state, involves 15
short-range interactions in addition to the familiar massless gravity fields.
Bibliography
[1] F. Bacon, Novum Organon (1620), P. Urban and J. Gibson (tr. eds.),
Open Court Publishing Company, Peru, IL, 1994.
[2] V. Bargmann, Irreducible representations of the Lorentz group, Ann.
Math., 48 (1947), 568–640.
[3] C. Becchi, A. Rouet, R. Stora, Ann. Phys. 98 (1976), 287.
[4] H. Behnke, F. Sommer, Theorie der analytischen Funktionen einer kom-
plexen Veränderlichen (1962), Springer-Verlag, Berlin.
[5] G. Birkhoff, J.von Neumann, Ann. Math. 37 (1936), 823 (The Logic of
Quantum Mechanics).
[6] J.D. Bjorken, S. Drell, Relativistische Quantenmechanik (1966), BI
Hochschultaschenbücher, Mannheim.
[7] J.D. Bjorken, S. Drell, Relativistische Quantenfelder (1966), BI
Hochschultaschenbücher, Mannheim.
[8] S. Bochner, Monotone Funktionen, Stieltjessche Integrale und harmoni-
sche Analyse, Math. Ann., 108 (1933), 378–410.
[9] H. Boerner, Darstellungen von Gruppen (1955), Springer-Verlag, Berlin.
[10] N.N. Bogliubov, D.V. Shirkov, Introduction to the Theory of Quantized
Fields (1959), Interscience, New York.
[11] M. Born, Einstein’s Theory of Relativity (1962), Dover Publications, New
York.
[12] N. Bourbaki, Algebra I, Chapters 1–3 (1989), Springer-Verlag, Berlin.
[13] N. Bourbaki, Algebra II, Chapters 4–7 (1990), Springer-Verlag, Berlin.
[14] N. Bourbaki, Algèbre, Chapitre 9 (Formes sesquilineaires et formes
quadratiques) (1959), Hermann, Paris .
[15] N. Bourbaki, Lie Groups and Lie Algebras, Chapters 1–3 (1989),
Springer-Verlag, Berlin.
[16] N. Bourbaki, Groupes et Algèbres de Lie, Chapitres 4–8 (1968–1975),
Hermann, Paris.
[17] N. Bourbaki, Intégration, Chapitres 7–8 (1963), Hermann, Paris.
333
334 Bibliography
[18] N. Bourbaki, Topological Vector Spaces, Chapter 1–5 (1987), Springer-

Verlag, Berlin.
[19] N. Bourbaki, Théories Spectrales (1967), Hermann, Paris.
[20] H. van Dam, M. Veltman, Massive and Massless Yang–Mills and Gravi-
tational Fields, Nuclear Physics B22 (1970), 397–411.
[21] J. Dixmier, C ∗ -Algebras (1969, Engl. Trans. 1977), North-Holland Pub-
lishing Company, Amsterdam.
[22] A. Einstein, Ideas and Opinions (1954), Three River Press, New York
(1995).
[23] V. Fock, Zur Theorie des Wasserstoffatoms, Zeits. Phys. 98 (1935),
145–154.
[24] G.B. Folland, A Course in Abstract Harmonic Analysis (1995), CRC
Press, Boca Raton, FL.
[25] F.G. Frobenius, Gesammelte Abhandlungen (1968), Springer-Verlag,
Berlin.
[26] W. Fulton, J. Harris, Representation Theory (1991), Springer.
[27] I.M. Gelfand and D.A. Raikov, Irreducible unitary representations of
locally bicompact groups, Mat. Sb. 13 (55) (1942), 301–316.
[28] I.M. Gel’fand, G.E. Shilov, Generalized Functions I (Properties and
Operations) (1958, Engl. trans. 1963), Academic Press, New York.
[29] I.M. Gel’fand, G.E. Shilov, Generalized Functions II (Spaces of Funda-
mental and Generalized Functions) (1958, Engl. trans. 1968), Academic
Press, New York.
[30] I.M. Gel’fand, M.I. Graev, N.Ya. Vilenkin, Generalized Functions V (In-
tegral Geometry and Representation Theory) (1962, Engl. trans. 1966),
Academic Press, New York.
[31] I.M. Gelfand, M.A. Neumark, Unitäre Darstellungen der klassischen
Gruppen (1950, German Trans. 1957), Akademie Verlag, Berlin.
[32] R. Gilmore, Lie Groups, Lie Algebras and Some of Their Applications
(1974), John Wiley & Sons, New York.
[33] W. Heisenberg, Einführung in die einheitliche Feldtheorie der Elemen-
tarteilchen (1967), Hirzel, Stuttgart.
[34] W. Heisenberg, Gesammelte Werke Abteilung C: Allgemeinverständliche
Schriften, I–V, (1984–1989), Piper, München.
[35] S. Helgason, Differential Geometry, Lie Groups and Symmetric Spaces
(1978), Academic Press, New York.
[36] S. Helgason, Groups and Geometric Analysis (1984), Academic Press,
New York.
Bibliography 335
[37] N.J. Hicks, Notes on Differential Geometry, (1965) D. van Nostrand

Company, Princeton, NJ.
[38] E. Inönü, E. P. Wigner, On the contraction of groups and their repre-
sentations, Proc. NAS 39 (1952), 510–525.
[39] C.J. Isham, A. Salam, J. Strathdee, Nonlinear realizations of space-time
symmetries. Scalar and tensor gravity, Ann. Phys. 62 (1971), 98–119.
[40] A.A. Kirillov, Elements of the Theory of Representations (1976),
Springer-Verlag, Berlin.
[41] A. Knapp, Representation Theory of Semisimple Groups (1986),
Princeton University Press, Princeton, NJ.
[42] T. Kugo, I. Ojima, Progr. Theor. Phys. 60 (1978), 1869.
[43] G.W. Mackey, On induced representations of groups (1951), Am. J.
Math. 73, 576–592.
[44] A. Messiah, Quantum Mechanics I (1965), North-Holland, Amsterdam.
[45] Y. Nambu, G. Jona-Lasinio, Phys. Rev. 122 (1961), 345–358; 124 (1961),
246–254.
[46] M.A. Neumark, Lineare Darstellungen der Lorentzgruppe (1958, German
Trans. 1963), VEB Deutscher Verlag der Wissenschaften, Berlin.
[47] F. Peter, H. Weyl, Die Vollständigkeit der primitiven Darstellungen einer
geschlossenen kontinuierlichen Gruppe, Math. Ann. 97 (1927), 737–755.
[48] H. Reichenbach, The Philosophy of Space and Time (1927), Dover, New
York.
[49] C.E. Rickart, General Theory of Banach Algebras (1960), D. van
Nostrand, Princeton, NJ.
[50] H. Saller, Operational Quantum Theory I — Nonrelativistic Structures
(2006), Springer, New York.
[51] H. Saller, Operational Quantum Theory II — Relativistic Structures
(2006), Springer, New York.
[52] I. Schur, Neue Begründung der Gruppencharaktere, Sitzungsber. Preuss.
Akad. 1905, 406.
[53] R.U. Sexl, H.K. Urbantke, Gravitation und Kosmologie (1983), 2. Au-
flage, Bibliographisches Institut, Mannheim.
[54] T.O. Sherman, Fourier analysis on the sphere, Trans. AMS 209 (1975),
1–31
[55] R.S. Strichartz, Harmonic analysis on hyperboloids, J. Funct. Anal. 12
(1973), 341–383.
[56] F. Treves, Topological Vector Spaces, Distributions and Kernels (1967),
Academic Press, New York.
336 Bibliography
[57] N.Ja. Vilenkin, A.U. Klimyk, Representations of Lie Groups and Special
Functions (1991), Kluwer Academic Publishers, Dordrecht.
[58] S. Weinberg, A model of leptons, Phys. Rev. Lett. 18 (1967), 507.
[59] H. Weyl, Raum, Zeit, Materie (1923), Wissenschaftliche Buchgesell-
schaft, Darmstadt.
[60] H. Weyl, The Theory of Groups and Quantum Mechanics (1931), Dover,
New York.
[61] H. Weyl, The Classical Groups (1946), Princeton University Press,
Princeton, NJ.
[62] E. P. Wigner, On unitary representations of the inhomogeneous Lorentz
group, Ann. Math. 40 (1939), 149–204.
[63] C. N. Yang, R. Mills, Phys. Rev. 96 (1954), 191.
Index
“Prokrustes transformations”, 7 canonically complexified, 75

Cartan
absolute derivative, 43 plane, 202
adjoint representation, 45 spacetime, 7
angular momentum, 83 torus, 201
annihilation operator, 97, Casimir operator, 279
121 causal
Cartan plane, 299
Becchi–Rouet–Stora
charge, 132 ordering, 120
Becchi–Rouet-Stora orthogonal spacetime, 297
transformations, 130 spacetime, 304
Bessel functions, 230, 246 support, 120
spherical, 227 unitary spacetime, 297
bi-regular, 206 center of mass, 84
group action, 169 central correlation, 160
Bianchi identity, 42 centrifugal potential, 83
Born, M., 3, 317 chargelike, 1, 13, 147, 201
Bose–Fermi twin structure, 132 chiral
Breit–Wigner chirality, 257
functions, 222 degeneracy, 258
states, 222 Goldstone modes, 293
bundle transformations, 257
associated, 148, 149 classical physics, 82
connection, 152 Clifford algebra, 257
coset, 174 color, 97, 328
Lie algebra, 168 compact Lie group, 201
Lie group coset, 167 compact–noncompact
nonrelativistic, 150 duality, 74
principal, 167 reciprocity, 290, 323, 327
section, 149 composite gauge fields, 156
special relativistic, 150 confluent hypergeometric equation,
structural group, 150 253
structural Lie algebra, conformal vector field, 53
153 conjugacy classes, 205
trivialization, 149 connection
vector space, 148 adjoint, 49
337
338 Index
affine, 40 scalar, 54
internal, 156 tensor, 41
Killing, 50 cyclic
Riemannian, 40, 51 Hilbert space, 8
conserved quantities, 55 vector, 210
contraction
Inönü–Wigner, 25 de Sitter universes, 28
convolution, 255 decomposition
abelian groups, 261 Iwasawa, 202
Euclidean spaces, 263
Langlands, 232
Feynman measures, 267
definite metric, 106
Feynman propagators, 272
degeneracy manifold, 210
free particles, 273
degenerate ground-state, 210
hyperboloids, 264
Descartes, R., 3
linear invariants, 262
diagonal group, 44
position representations, 263
differential operators, 279
product, 205, 206
dilation
selfdual invariants, 262
degeneracy, 324
spacetimes, 269
group, 7
spheres, 265
coordinates in gravity, 19
Cartesian, 59, 281 rearrangement, 319
Eddington–Finkelstein, 22 dipole, 109, 251
Eulerian, 58 Dirac
geodesic polar, 58, 281 distribution, 120
Kruskal, 22 equation, 183
stationary, 54 field, 121
weak, 176 measure, 207
Copernicus, N., 5 distinguished frames, 156
coset space, 8 distribution
cosmological constant, 18, 26 advanced, 244
Coulomb causal, 244
degree of freedom, 126 Dirac, 207
interaction, 11 Feynman, 244
potential, 84, 128 hyperbolic, 244
covariant Radon, 207
derivation, 40 residual, 246
derivative, 114, 153 retarded, 244
covariantly constant, 41 spherical, 244
creation operator, 97, 121 distributive basis, 195, 220
current, 156 divergence, 53
electromagnetic, 114 divergencies, 141, 255
curvature, 17, 154 double trace, 204
constant, 57 dual derivation, 283
covariantly constant, 73 dual normalization, 288
Index 339
duality of vector spaces, 209 quantum, 111

dyad, 184 field strengths, 154
Fierz recoupling, 257
eigenvalues, 291 fine structure constant, 6, 84
eigenvector notation, 9 fixgroup, 38
Einstein tensor, 18, 34 flat
Einstein, A., 1, 2, 111
position, 9
electrodynamics, 86, 113
spaces, 198
current, 87
Fock
field strength, 87, 113
space, 106
vector potential, 113
state, 97, 107
electroweak
Fock–Iwanenkov
and strong gauge interactions,
158 coefficients, 52
coupling constants, 163 connection, 115
mass unit, 165 Fock–Mehler transformation, 235
spacetime, 170 Fourier
triangle, 163, 164 components, 217
energy-momentum tensor, 88, 116, inversion, 214
117 transformation, 214, 217
enveloping algebra, 45 free particles, 119
epistemology, 4, 93 Frobenius’ reciprocity, 188, 196
equations of motion, 81 future cone, 297
equivalence principle, 19
Erlangen program, 1 Galilei group, 9
Euclidean space, 58 gauge
external, 1, 13, 147, 201 -fixing constant, 135
external derivative, 32 coupling constants, 164, 330
degree of freedom, 126
Fadeev–Popov
field, 153, 159
fields, 133
ghosts, 130 field normalization, 158
Faraday, M., 4 fixing constant, 108
Feynman interactions, 157
measures, 289 transformation, 43, 153
propagator, 108, 120, 271 vertices, 157
fiber Gell–Mann matrices, 98
bundle, 148 geodesics, 81, 87
bundle base, 148 ghost pair, 109, 129
bundle morphism, 149 global group, 37
bundle trivial, 148 Goldstone manifold, 166, 176
local, 148 gradient, 52
typical, 148 Grassmann algebra, 31
field, 4 gravity
Hilbert metric, 121 Einstein, 17
quantization, 121 of fields, 115
340 Index
Green’s hypergeometric functions, 252

distribution, 259, 283 hyperisospin correlation, 311
kernel, 283
ground state Inönü–Wigner contraction, 25
degenerate, 161 Inönü-Wigner contraction, 253
electroweak, 161 indefinite
manifold, 166 metric, 109
ground-state unitary, 108
degenerate, 256 induced representations, 175, 194
vector, 210 information catalogue, 93
group algebra, 205 inner product
group doubling, 169 invariant, 203
group dual, 205 interactions, 277, 287
internal, 1, 13, 147, 201
Haar measure, 206 multiplicities, 314, 327
Hankel functions, 246 intertwiner, 194
Hardy spaces, 308 invariants, 56
harmonic involution of functions, 208
analysis, 213 isometry group, 37
components, 214 isotropy group, 38
expansion of a field, 121 Iwasawa
polynomials, 95, 226 decomposition, 202, 231
representations, 104 factorization, 297
representations orthogonal
groups, 225 Kant, I., 10
units (projectors), 226 Kepler
harmonic oscillator, 8, 96 dynamics position representa-
Bose, 99 tion, 101
Fermi, 99 dynamics quantum algebraic,
Heisenberg 102
group, 94 potential, 84, 238, 278
Lie algebra, 93 representations, 104
Heisenberg, W., 10, 12 Kepler, J., 5
Helgason, S., 1 kernel, 260, 277
helicity, 9 Green’s, 283
Higgs field, 159, 162, 177 group, 282
homogeneous space, 8 Laplace–Beltrami, 285
horizontal, 148 linear, 284
hydrogen atom, 10, 100, regular, 281
238 regularized, 281
hyperbolic resolvents, 291
hopping, 298 semiregular, 281
stretching, 298 tangent, 287
hyperboloid, 58, 244 Killing form, 204, 279
Y, 7 Klein space, 1
Index 341
Klein, F., 1 Macdonald functions, 230, 246

Knapp, A.W., 2 manifold
affine locally symmetric, 73
Laguerre polynomials, 96 differential, 29
Langlands decomposition, 232 Einstein, 57
Laplace globally symmetric, 76
–Beltrami operator, 53, 279, 285 maximal global symmetry, 57
-Beltrami operator, 115 parallelizable, 31
kernel, 85 Riemannian, 33
Lebesgue Banach spaces, 207 rotation symmetric, 60
left–right tangent structures, 30
multiplication, 169 with dimension 1, 62
left-invariant differential operator, with dimension 2, 62
234 with dimension 3, 64
left-right with dimension 4, 67
translations, 44 mass points, 3
Legendre mass-hyperboloid, 178
functions, 235, 252 massless
polynomials, 228, 252
fields, 126
Lenz-Runge
interactions, 324
classes, 13
tensor modes, 326
vector, 85
vector modes, 325
leptons, 159
Maurer–Cartan formula, 32
Lie algebra
maximally symmetric
compact, 205
position, 24, 86
densities, 156
spacetime, 27
orthogonal symmetric, 73
Lie derivation, 31 Maxwell equations, 113
Lie–Jacobi isomorphisms, 47 Maxwell, J.C., 4, 5
lifetime, 223 measure
light ray bending, 91 Haar, 206
linearized Plancherel, 213
connection coefficients, 118 metric
Einstein equations, 118 Kerr–Newman, 23
gravitative interaction, 119 operational, 55
little group, 112 Reissner, 23
local group, 38, 114 Robertson–Walker, 24, 70
London, F., 13 Schwarzschild, 23
Lorentz covariant derivative, 115 signature, 35
Lorentz force, 87 tangent space, 33, 55
Lorentz group, 6 metric-inducing function, 202, 210
extended, 296 metrical reflection, 34
of a Lie group, 46 metrical structures, 6
representations, 123 motion group, 37
Lorentz invariant measure, 121 multipoles, 238
342 Index
n-bein, 30, 176 perihelion

Nambu invariant, 85
–Goldstone field, 261 shift, 90
–Goldstone modes, 319 periodic table, 13
–Jona Lasinio model, 256, 293 perturbative corrections, 141
Neumann functions, 246 Peter–Weyl decomposition, 188, 216
Newton photons, 11
constant, 6 Plancherel density, 215, 216
potential, 84, 128 Plancherel distribution, 197, 213
Newton, I., 3 Euclidean groups, 228
noncompact Lie group, 202 Harish–Chandra, 236
nondecomposable representation, 108 Poincaré groups, 231
nonparticle degrees of freedom, 328 Planck
nonrelativistic scattering, 9, 227 area, 18
normalization constant, 94
hyperboloids, 290 mass, 6, 84
spacetime, 293 Planck, M., 6
spheres, 290 Plato, 10
Poincaré group, 9, 86
off-shell, 120
of a Lie group, 46
off-shell convolution, 274
representations, 136
on-shell, 120
Wigner classification, 138
operation
pointlike particle, 93, 112
chargelike, 1, 147
pointwise product, 207
external, 1, 147
Poisson–Lie-bracket, 83
internal, 1, 147
position
spacetimelike, 1, 147
orientation manifold, 177 distributions, 277
orthogonal symmetric, 73 flat, 9, 24
hyperbolic, 13, 24, 240, 328
parabolic subgroup, 231 interactions, 287
Parseval formula, 214 kernels, 288
particle, 106 maximally symmetric, 24, 86
definition, 9 measure, 94
masses, 321 paraboloid, 21
spectrum, 317 spherical, 24, 26
particle analysis, 181, 198 positive-type functions, 202, 210
Dirac field, 121 for hyperboloids, 221
scalar field, 121 for spheres, 221
spin-1 field, 124 Gaussian, 224
spin-2 field, 124 principal value, 120
vector field, 134 probability amplitudes, 4
Pauli exclusion principle, 99 product representations, 255
Pauli-Villars regularization, pure gauge, 155
142 pure state, 212
Index 343
quantization principal series, 137

of fields, 121 residual, 291
second, 112 self-dual, 203
quantum algebra spacetime, 112
Bose, 106 spherical, 234
Fermi, 106 square-integrable, 233
quantum gauge fields, 130 supplementary series, 137
quantum theory, 93 translations, 217
quark, 11, 14, 159 residual normalization, 262, 266
masses, 11 resolvents, 291
propagator, 329 Ricci tensor, 18, 42, 54
Radon distribution, 207 Schrödinger

positive-type, 212 equation, 95
relativistic particles, 229 wave function, 93
relativity Schur
electromagnetic, 173, 177, 188 lemma, 203
electroweak, 173 orthogonality, 197, 216
linear representations, 185 product, 209, 216, 289
Lorentz group, 191 self-dual, 203
orthogonal, 198 singularity
particle, 173 hyperboloids, 263
perpendicular, 173, 178, 187 line, 301
rotation, 178, 190 naked, 92
special, 173 spheres, 263
transitions, 180 Sommerfeld, A., 6
unitary, 173, 179, 191 spacetime
renormalizable, 141 (anti-)de Sitter, 28
renormalization-regularization, 141 Cartan, 7, 299
representation Einstein, 26
bi-regular, 206 electroweak, 14, 170, 175, 295
coefficients, 112, 201, 205 Friedmann, 24, 295
cyclic, 210 future cone, 179
cyclic translation, 219 global group, 295
discrete series, 137, 233 harmonic analysis, 301
dual, 204, 212 harmonic coefficients, 319
Euclidean groups, 227 interaction projection, 306
finite-dimensional, 202 interactions, 310
Hilbert, 204 Kerr–Newman, 20
induced, 175 Kruskal, 20
left-right regular, 206 local group, 295
metric, 202 mass ratio, 312
nonamenable, 233 masses, 327
normalization, 211, 290 maximally symmetric, 27
Poincaré groups, 229 motion group, 82
344 Index
normalization, 311, 327 translation

particle projection, 306 group, 7
Reissner, 20 invariants, 293, 321
Schwarzschild, 20, 88 transmutator, 176, 180, 195
spacetime quantization, 3 Higgs, 180, 183, 189
spacetimelike, 1, 13, 147, 201 Pauli, 181
special functions, 8, 244 rectangular, 186
spectral distribution, 215 Weyl, 181, 182
sphere, 58, 244 two-particle
Ω, 7 convolution, 264
spherical functions, 233 threshold factor, 274
Euclidean groups, 236 unimodular group, 206
spherical harmonics, 95, 188, 226
spin, 9 vacuum polarization, 142
spin-mass correlation, 323 vector potential, 87
spinor-isospinor vector-isovector
lepton field, 329 quark field, 329
quark field, 329 field, 329
spontaneous symmetry breakdown, vertical, 148
161 vierbein, 184
standard model, 158 virtual particle, 11
minimal, 159 virtual particles, 120
stationary coordinates, 54 volume element, 33
structural group for frames, 43 wave functions, 93
symmetric space, 8 weak
symmetry rearrangement, 162, 319 boson masses, 165
coordinates, 176
tangent weight diagrams, 103
Euclidean group, 34 Weinberg
groups, 226 angle, 163
kernel, 260 rotation, 163
Poincaré group, 34 Weyl, H., 13
tensor algebra, 31 Wigner
tetrad, 19, 184 axial rotation, 182
as Goldstone field, 127 element, 168
theorem group element, 180
Bochner, 218 rotation, 180
Gel’fand and Raikov, 211 spin rotation, 183
kernel (Schwartz), 281 Wigner, E., 9
Lebesgue, 218 winding number, 201
Weyl, 205 Witt pair, 109, 129
time reflection, 222 Yukawa potential, 120, 278
time translations, 81
torsion tensor, 41 zweibein, 184

Operational Spacetime PDF

Uploaded by

Operational Spacetime PDF

Uploaded by

Operational Spacetime

Fundamental Theories of Physics

An International Book Series on The Fundamental Theories of Physics:

ISBN 978-1-4419-0897-1 e-ISBN 978-1-4419-0898-8

Library of Congress Control Number: 2009939325

c Springer Science+Business Media, LLC 2010

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

0 Introduction and Orientation 1

2.7 Maximally Symmetric Manifolds . . . . . . . . . . . . . . . . 57

5 Quantum Fields of Flat Spacetime 111

5.5 Free Particles for Flat Spacetime . . . . . . . . . . . . . . . . 119

6 External and Internal Operations 147

7 Relativities and Homogeneous Spaces 173

7.2.5 Real Tetrads (Vierbeins) . . . . . . . . . . . . . . . . . 184

8 Representation Coeﬀﬁcients 201

8.10.3 Eigenfunctions of Euclidean Groups . . . . . . . . . . 236

9 Convolutions and Product Representations 255

10 Interactions and Kernels 277

10.6 Kernel Resolvents and Eigenvalues . . . . . . . . . . . . . . . 291

11 Electroweak Spacetime 295

12 Masses and Coupling Constants 317

Introduction and Orientation

Einstein’s theory of general relativity identiﬁes spacetime curvature with

H. Saller, Operational Spacetime: Interactions and Particles, 1

on the coordinates has no separate existence . . . There is no such thing as an

orbits (Copernicus) were improved by Kepler’s ellipses with small eccentric-

cosmological models of general relativity as exempliﬁed by the Friedmann

from the orthochronous, i.e., causality-compatible Lorentz group SO0 (1, 3)

The later-used complex representations for half-integer spin, faithful for

The nontrivial Hilbert representations, i.e., with deﬁnite unitary boosts,

phic abelian groups R3 and R4 .

More complicated are the representations of the semidirect product

orthochronous Lorentz group SO0 (1, 3) and R4 as translations for relativistic

Dynamics Manifold Group Diﬀ. eq.

Flat manifolds come with operation groups, especially translations, whose

considered in the following.

A related quotation from the last public talk of Heisenberg (1975, my

Laguerre polynomials L the hyperbolic operations SO0 (1, 1) ⊂ SO0 (1, 3) in

tized curvature of the position hyperboloid Y . 3

The belief in the universality and completeness of ﬂat Minkowski space-

a consequence of an unbroken SU(3)-gauge interaction is still missing.

they are not unitary Poincaré group representations, no particles according

dilations by the compact U(1)-phase transformations. As of today, all internal

In this chapter, Einstein’s gravity with the identiﬁcation of the curvature of

1.1 Geometrization of Gravity

dimensional internal” curvature of the orange skin described above.

H. Saller, Operational Spacetime: Interactions and Particles, 17

Ř• + Λg = −κT, with Ř• = R• − g2 R•• ,

The spacetime properties, as given in R, determine the matter properties,

The real four-dimensional spacetime manifold M(1,3) is assumed with

2 A real matrix is orthogonally diagonalizable, O ◦ M ◦ O T = diag M , O ∈ SO(n), if

and only if it is symmetric M = M T .

1.2 Schwarzschild–Kruskal Spacetime

It is remarkable that the time-independent nonrelativistic Newton potential

e2λ3 (ρ) = gtt (ρ) = 1 − 2

can be written as a symmetric (6×6)-matrix with 2 −1 = 20 independent

Schwarzschild spacetime embeds as position manifold a rotation paraboloid

Via the constants one transforms to either advanced or retarded gton–

leading to the Kruskal coordinates (u, v, ω

The radial coordinate ρ is a function of u2 − v 2 . The metric involves a hyper-

r0 is a function of λ3 and ξ0 . The Schwarzschild analogue coordinates (λ3 , ξ0 )

Globally SO(3)-invariant Reissner spacetime with a trivial curvature

1.3 Friedmann and de Sitter Universes

g = dt2 − R2 (t)dσk2 , dθ2 sin2 θ

With the introduction of half-integer spin structures after the Stern–

The Lorentz group SO0 (1, 3) is familiar as a special relativistic spacetime

= (iR)3 ⊕ (iR)3 and log SO0 (1, 3) ∼

The Euclidean position arises by “ﬂattening,” i.e., an Inönü-Wigner contrac-

R(t) ∼ R(t0 )e H0 (t−t0 )

t = H0 c ∼ 1.2 × 1026 m, and a characteristic length for position, x =

With the energy-momentum tensor of an ideal cosmological ﬂuid, charac-

the time dependence of the position “radius” t −→ R(t) of a Robertson–

g = dt2 − R2 dσk2 , R2 = const., p = 0 ⇒ k

the ground-state density coincides, up to a factor, with the cosmological con-

3-sphere is isomorphic to the spin group Ω3 ∼

R = 0 ⇐⇒ Ṙ2 = −k = 2 ⇒ R(t) = R0 ± t,

idT(U) = ej ⊗ ěj , ěi , ej = δij .

ea (x) = eaj (x)ej (x), eaj (x) = ěj (x), ea (x),

d : C(M) −→ TT (M), f −→ df with df, v = v(f ), v ∈ T(M),

2děk (ei , ej ) = −ěk ([ei , ej ]) = −ij

g = eT ◦ g ◦ e, e(x) ∈∈GL(n, R)/H.

The torsion equations reﬁne the Maurer-Cartan equations děk = − 12 ij